1 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
5 DF_REF_INSN if DF_REF_INSN_INFO is false.
7 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10 xmm_move instead of sse_move.
11 (sse_store_index): New function.
12 (ix86_register_move_cost): Be more sensible about mismatch stall;
13 model AVX moves correctly; make difference between sse->integer and
15 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
16 moves; make difference between SSE and AVX.
17 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
18 and zmm_move. Increase size of sse load and store tables;
19 add unaligned load and store tables; add ssemmx_to_integer.
20 * x86-tune-costs.h: Update all entries according to real
21 move latencies from Agner Fog's manual and chip documentation.
23 2017-10-23 Jakub Jelinek <jakub@redhat.com>
26 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
27 * config/i386/constraints.md (Wf): New constraint.
28 * config/i386/i386.md (UNSPEC_SBB): New unspec.
29 (cmp<dwi>_doubleword): Removed.
30 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
31 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
32 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
33 expand with cmp<dwi>_doubleword. For LTU and GEU use
34 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
36 * common.opt (gcolumn-info): Enable by default.
37 * doc/invoke.texi (gcolumn-info): Document new default.
39 2017-10-23 Richard Biener <rguenther@suse.de>
41 PR tree-optimization/82672
42 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
43 Fold the stmt if we propagated into it.
45 2017-10-23 Richard Biener <rguenther@suse.de>
47 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
48 (bitmap_remove_expr_from_set): ... this. All callers call this
49 for non-constant values.
50 (bitmap_set_subtract): Rename to...
51 (bitmap_set_subtract_expressions): ... this. Adjust and
53 (bitmap_set_contains_value): Remove superfluous check.
54 (bitmap_set_replace_value): Inline into single caller ...
55 (bitmap_value_replace_in_set): ... here and simplify.
56 (dependent_clean): Merge into ...
57 (clean): ... this using an overload. Adjust.
58 (prune_clobbered_mems): Adjust.
59 (compute_antic_aux): Likewise.
60 (compute_partial_antic_aux): Likewise.
62 2017-10-23 Richard Biener <rguenther@suse.de>
64 PR tree-optimization/82129
66 2017-08-01 Richard Biener <rguenther@suse.de>
68 PR tree-optimization/81181
69 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
70 (compute_antic): ... end of iteration here.
72 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
74 * target.def (starting_frame_offset): New hook.
75 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
76 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
77 * doc/tm.texi.in: Regenerate.
78 * hooks.h (hook_hwi_void_0): Declare.
79 * hooks.c (hook_hwi_void_0): New function.
80 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
81 STARTING_FRAME_OFFSET.
82 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
83 * reload1.c (reload): Likewise.
84 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
85 instead of STARTING_FRAME_OFFSET.
86 * function.c (try_fit_stack_local): Likewise.
87 (assign_stack_local_1): Likewise
88 (instantiate_virtual_regs): Likewise.
89 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
90 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
91 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
92 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
93 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
94 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
95 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
96 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
97 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
98 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
99 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
100 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
101 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
102 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
103 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
104 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
105 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
106 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
107 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
108 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
109 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
110 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
111 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
112 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
113 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
114 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
115 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
116 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
117 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
118 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
119 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
120 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
121 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
122 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
123 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
124 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
125 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
126 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
127 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
128 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
129 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
130 * config/avr/avr.c (avr_starting_frame_offset): Make static and
131 return a HOST_WIDE_INT.
132 (avr_builtin_setjmp_frame_value): Use it instead of
133 STARTING_FRAME_OFFSET.
134 (TARGET_STARTING_FRAME_OFFSET): Redefine.
135 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
136 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
138 (TARGET_STARTING_FRAME_OFFSET): Redefine.
139 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
140 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
141 (TARGET_CONSTANT_ALIGNMENT): Redefine.
142 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
143 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
144 (TARGET_STARTING_FRAME_OFFSET): Redefine.
145 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
146 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
147 (TARGET_STARTING_FRAME_OFFSET): Redefine.
148 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
149 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
151 (TARGET_STARTING_FRAME_OFFSET): Redefine.
152 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
153 * config/mips/mips.c (mips_compute_frame_info): Refer to
154 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
155 (mips_starting_frame_offset): New function.
156 (TARGET_STARTING_FRAME_OFFSET): Redefine.
157 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
158 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
159 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
160 and return a HOST_WIDE_INT.
161 (TARGET_STARTING_FRAME_OFFSET): Redefine.
162 (mmix_initial_elimination_offset): Refer to
163 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
164 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
165 * config/pa/pa.c (pa_starting_frame_offset): New function.
166 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
167 (pa_expand_prologue): Likewise.
168 (TARGET_STARTING_FRAME_OFFSET): Redefine.
169 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
170 !FRAME_GROWS_DOWNWARD handling to...
171 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
172 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
173 !FRAME_GROWS_DOWNWARD handling to...
174 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
175 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
176 !FRAME_GROWS_DOWNWARD handling to...
177 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
178 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
180 (rs6000_starting_frame_offset): New function.
181 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
182 !FRAME_GROWS_DOWNWARD handling to...
183 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
184 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
185 !FRAME_GROWS_DOWNWARD handling to...
186 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
187 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
188 !FRAME_GROWS_DOWNWARD handling to...
189 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
190 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
191 (rs6000_starting_frame_offset): New function.
192 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
193 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
194 * config/vax/vax.c (vax_starting_frame_offset): New function.
195 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
196 (TARGET_STARTING_FRAME_OFFSET): Redefine.
197 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
198 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
199 (TARGET_STARTING_FRAME_OFFSET): Redefine.
200 * system.h (STARTING_FRAME_OFFSET): Poison.
202 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
205 SCALAR_TYPE_MODE instead of TYPE_MODE.
207 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
208 Alan Hayward <alan.hayward@arm.com>
209 David Sherwood <david.sherwood@arm.com>
211 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
213 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
219 2017-10-23 Richard Biener <rguenther@suse.de>
221 PR tree-optimization/82129
222 * tree-ssa-pre.c (bitmap_set_and): Remove.
223 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
224 canonicalizing expressions in the set to those with lowest
225 ID rather than taking that from the first edge.
227 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
229 * combine.c (rtx_equal_for_field_assignment_p): Use
232 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
237 to a SUBREG_PROMOTED_VAR.
239 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
240 Alan Hayward <alan.hayward@arm.com>
241 David Sherwood <david.sherwood@arm.com>
243 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
244 (expand_debug_source_expr): Likewise.
245 * combine.c (combine_simplify_rtx): Likewise.
246 * cse.c (fold_rtx): Likewise.
247 * optabs.c (expand_float): Likewise.
248 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
249 (simplify_binary_operation_1): Likewise.
251 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
252 Alan Hayward <alan.hayward@arm.com>
253 David Sherwood <david.sherwood@arm.com>
255 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
256 (record_promoted_value): Likewise.
257 * expr.c (expand_expr_real_2): Likewise.
258 * ree.c (update_reg_equal_equiv_notes): Likewise.
259 (combine_set_extension): Likewise.
260 * rtlanal.c (low_bitmask_len): Likewise.
261 * simplify-rtx.c (neg_const_int): Likewise.
262 (simplify_binary_operation_1): Likewise.
264 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
265 Alan Hayward <alan.hayward@arm.com>
266 David Sherwood <david.sherwood@arm.com>
268 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
269 * regcprop.c (maybe_mode_change): Likewise.
270 * reload1.c (alter_reg): Likewise.
272 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
274 * inchash.h (inchash::hash::add_wide_int): New function.
275 * lto-streamer-out.c (hash_tree): Use it.
277 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
279 * inchash.h (inchash::hash::add_wide_int): Rename to...
280 (inchash::hash::add_hwi): ...this.
281 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
282 (polymorphic_call_target_hasher::hash): Likewise.
283 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
284 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
285 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
286 * lto-streamer-out.c (hash_tree): Likewise.
287 * optc-save-gen.awk: Likewise.
288 * tree.c (add_expr): Likewise.
290 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
293 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
294 for ordered inequality comparisons even with TARGET_IEEE_FP.
296 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
299 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
300 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
301 Expand with cmp<dwi>_doubleword.
303 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
305 * extend.texi: Add x86 specific to 'nocf_check' attribute.
307 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
308 specific to -fcf-protection option.
310 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
312 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
313 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
314 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
315 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
316 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
317 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
318 (extra_objs): Add cet.o for Linux/x86 targets.
319 (tmake_file): Add i386/t-cet for Linux/x86 targets.
320 * config/i386/cet.c: New file.
321 * config/i386/cetintrin.h: Likewise.
322 * config/i386/t-cet: Likewise.
323 * config/i386/cpuid.h (bit_SHSTK): New.
325 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
326 pass IBT and SHSTK bits.
327 * config/i386/i386-builtin-types.def
328 (VOID_FTYPE_UNSIGNED_PVOID): New.
329 (VOID_FTYPE_UINT64_PVOID): Likewise.
330 * config/i386/i386-builtin.def: Add CET intrinsics.
331 * config/i386/i386-c.c (ix86_target_macros_internal): Add
332 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
333 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
334 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
336 * config/i386/i386.c (rest_of_insert_endbranch): New.
337 (pass_data_insert_endbranch): Likewise.
338 (pass_insert_endbranch): Likewise.
339 (make_pass_insert_endbranch): Likewise.
340 (ix86_notrack_prefixed_insn_p): Likewise.
341 (ix86_target_string): Add -mibt, -mshstk flags.
342 (ix86_option_override_internal): Add flag_cf_protection
344 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
345 (ix86_print_operand): Add 'notrack' prefix output.
346 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
347 (ix86_expand_builtin): Expand CET intrinsics.
348 (x86_output_mi_thunk): Add 'endbranch' instruction.
349 * config/i386/i386.h (TARGET_IBT): New.
350 (TARGET_IBT_P): Likewise.
351 (TARGET_SHSTK): Likewise.
352 (TARGET_SHSTK_P): Likewise.
353 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
354 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
355 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
356 (builtin_setjmp_setup): New pattern.
357 (builtin_longjmp): Likewise.
358 (rdssp<mode>): Likewise.
359 (incssp<mode>): Likewise.
360 (saveprevssp): Likewise.
361 (rstorssp): Likewise.
362 (wrss<mode>): Likewise.
363 (wruss<mode>): Likewise.
364 (setssbsy): Likewise.
365 (clrssbsy): Likewise.
366 (nop_endbr): Likewise.
367 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
369 * config/i386/immintrin.h: Include <cetintrin.h>.
370 * config/i386/linux-common.h
371 (file_end_indicate_exec_stack_and_cet): New prototype.
372 (TARGET_ASM_FILE_END): New.
374 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
376 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
377 latencies instead of having separate table; make difference between
378 integer and float costs.
379 * i386.h (processor_costs): Remove scalar_stmt_cost,
380 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
381 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
383 * x86-tune-costs.h: Remove entries which has been removed in
384 procesor_costs from all tables; make cond_taken_branch_cost
385 and cond_not_taken_branch_cost COST_N_INSNS based.
387 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
389 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
391 2017-10-20 Jakub Jelinek <jakub@redhat.com>
393 * config/i386/i386.md (isa): Remove fma_avx512f.
394 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
395 <avx512>_fmadd_<mode>_mask3<round_name>,
396 <avx512>_fmsub_<mode>_mask<round_name>,
397 <avx512>_fmsub_<mode>_mask3<round_name>,
398 <avx512>_fnmadd_<mode>_mask<round_name>,
399 <avx512>_fnmadd_<mode>_mask3<round_name>,
400 <avx512>_fnmsub_<mode>_mask<round_name>,
401 <avx512>_fnmsub_<mode>_mask3<round_name>,
402 <avx512>_fmaddsub_<mode>_mask<round_name>,
403 <avx512>_fmaddsub_<mode>_mask3<round_name>,
404 <avx512>_fmsubadd_<mode>_mask<round_name>,
405 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
406 (*vec_widen_umult_even_v16si<mask_name>,
407 *vec_widen_smult_even_v16si<mask_name>): Likewise.
408 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
410 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
412 * extend.texi: Add 'nocf_check' documentation.
413 * gimple.texi: Add second parameter to
414 gimple_build_call_from_tree.
415 * invoke.texi: Add -fcf-protection documentation.
416 * rtl.texi: Add REG_CALL_NOTRACK documenation.
418 2017-10-20 Richard Biener <rguenther@suse.de>
420 PR tree-optimization/82473
421 * tree-vect-loop.c (vectorizable_reduction): Properly get at
422 the largest input type.
424 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
426 * c-attribs.c (handle_nocf_check_attribute): New function.
427 (c_common_attribute_table): Add 'nocf_check' handling.
428 * gimple-parser.c: Add second argument NULL to
429 gimple_build_call_from_tree.
430 * attrib.c (comp_type_attributes): Check nocf_check attribute.
431 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
433 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
434 * common.opt: Add fcf-protection flag.
435 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
436 * flag-types.h: Add enum cf_protection_level.
437 * gimple.c (gimple_build_call_from_tree): Add second parameter.
438 Add 'nocf_check' attribute propagation to gimple call.
439 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
440 (gimple_build_call_from_tree): Update prototype.
441 (gimple_call_nocf_check_p): New function.
442 (gimple_call_set_nocf_check): Likewise.
443 * gimplify.c: Add second argument to gimple_build_call_from_tree.
444 * ipa-icf.c: Add nocf_check attribute in statement hash.
445 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
446 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
447 * toplev.c (process_options): Add flag_cf_protection handling.
449 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
451 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
453 2017-10-20 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/82603
456 * tree-if-conv.c (predicate_mem_writes): Make sure to only
457 remove false predicated stores.
459 2017-10-20 Richard Biener <rguenther@suse.de>
461 * graphite-isl-ast-to-gimple.c
462 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
463 Remove return value and simplify, dump copied stmt after lhs
465 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
466 Reduce dump verbosity.
467 (gsi_insert_earliest): Likewise.
468 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
469 * graphite.c (print_global_statistics): Adjust dumping.
470 (print_graphite_scop_statistics): Likewise.
471 (print_graphite_statistics): Do not dump loops here.
472 (graphite_transform_loops): But here.
474 2017-10-20 Nicolas Roche <roche@adacore.com>
476 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
477 * configure: Regenerate.
479 2017-10-20 Jakub Jelinek <jakub@redhat.com>
482 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
483 functions when optimizing replace GIMPLE_RETURN stmts with
484 calls to __builtin_unreachable ().
487 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
488 for -fsanitize=thread link of executables.
489 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
493 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
495 (<shift_insn><mode>3<mask_name>): Change the last of the 3
496 define_insns for logical vector shifts to use VI248_AVX512BW
497 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
498 condition, useless isa and prefix attributes. Change the first
499 2 of these define_insns to ...
500 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
501 define_insn for avx512vl.
502 (<shift_insn><mode>3): ... and this, new define_insn without
503 masking for non-avx512vl.
506 * config/i386/sse.md (*andnot<mode>3,
507 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
508 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
509 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
510 not applied use empty suffix even for TARGET_AVX512VL.
511 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
512 is applied, supply evex,evex or evex,evex,evex instead of just
515 2017-10-20 Julia Koval <julia.koval@intel.com>
517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
518 (OPTION_MASK_ISA_GFNI_UNSET): New.
519 (ix86_handle_option): Handle OPT_mgfni.
520 * config/i386/cpuid.h (bit_GFNI): New.
521 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
522 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
523 * config/i386/i386.c (ix86_target_string): Add -mgfni.
524 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
525 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
526 * config/i386/i386.opt: Add mgfni.
528 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
530 * config/msp430/msp430.c (msp430_option_override): Disable
531 -fdelete-null-pointer-checks.
532 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
534 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
536 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
537 of x87 and SSE instructions.
539 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
541 * asan.c (create_cond_insert_point): Do not update edge count.
542 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
543 (afdo_propagate_circuit): Likewise.
544 (afdo_calculate_branch_prob): Likewise.
545 (afdo_annotate_cfg): Likewise.
546 * basic-block.h (struct edge_def): Remove count.
547 (edge_def::count): New accessor.
548 * bb-reorder.c (rotate_loop): Update.
549 (find_traces_1_round): Update.
550 (connect_traces): Update.
551 (sanitize_hot_paths): Update.
552 * cfg.c (unchecked_make_edge): Update.
553 (make_single_succ_edge): Update.
554 (check_bb_profile): Update.
555 (dump_edge_info): Update.
556 (update_bb_profile_for_threading): Update.
557 (scale_bbs_frequencies_int): Update.
558 (scale_bbs_frequencies_gcov_type): Update.
559 (scale_bbs_frequencies_profile_count): Update.
560 (scale_bbs_frequencies): Update.
561 * cfganal.c (connect_infinite_loops_to_exit): Update.
562 * cfgbuild.c (compute_outgoing_frequencies): Update.
563 (find_many_sub_basic_blocks): Update.
564 * cfgcleanup.c (try_forward_edges): Update.
565 (try_crossjump_to_edge): Update
566 * cfgexpand.c (expand_gimple_cond): Update
567 (expand_gimple_tailcall): Update
568 (construct_exit_block): Update
569 * cfghooks.c (verify_flow_info): Update
570 (redirect_edge_succ_nodup): Update
572 (make_forwarder_block): Update
573 (duplicate_block): Update
574 (account_profile_record): Update
575 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
576 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
577 * cfgloopmanip.c (scale_loop_profile): Update.
579 (lv_adjust_loop_entry_edge): Update.
580 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
581 (force_nonfallthru_and_redirect): Update.
582 (purge_dead_edges): Update.
583 (rtl_flow_call_edges_add): Update.
584 * cgraphunit.c (init_lowered_empty_function): Update.
585 (cgraph_node::expand_thunk): Update.
586 * gimple-pretty-print.c (dump_probability): Update.
587 (dump_edge_probability): Update.
588 * gimple-ssa-isolate-paths.c (isolate_path): Update.
589 * haifa-sched.c (sched_create_recovery_edges): Update.
590 * hsa-gen.c (convert_switch_statements): Update.
591 * ifcvt.c (dead_or_predicable): Update.
592 * ipa-inline-transform.c (inline_transform): Update.
593 * ipa-split.c (split_function): Update.
594 * ipa-utils.c (ipa_merge_profiles): Update.
595 * loop-doloop.c (add_test): Update.
596 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
597 * lto-streamer-in.c (input_cfg): Update.
598 (input_function): Update.
599 * lto-streamer-out.c (output_cfg): Update.
600 * modulo-sched.c (sms_schedule): Update.
601 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
602 * predict.c (maybe_hot_edge_p): Update.
603 (unlikely_executed_edge_p): Update.
604 (probably_never_executed_edge_p): Update.
605 (dump_prediction): Update.
606 (drop_profile): Update.
607 (propagate_unlikely_bbs_forward): Update.
608 (determine_unlikely_bbs): Update.
609 (force_edge_cold): Update.
610 * profile.c (compute_branch_probabilities): Update.
611 * reg-stack.c (better_edge): Update.
612 * shrink-wrap.c (handle_simple_exit): Update.
613 * tracer.c (better_p): Update.
614 * trans-mem.c (expand_transaction): Update.
615 (split_bb_make_tm_edge): Update.
616 * tree-call-cdce.c: Update.
617 * tree-cfg.c (gimple_find_sub_bbs): Update.
618 (gimple_split_edge): Update.
619 (gimple_duplicate_sese_region): Update.
620 (gimple_duplicate_sese_tail): Update.
621 (gimple_flow_call_edges_add): Update.
622 (insert_cond_bb): Update.
623 (execute_fixup_cfg): Update.
624 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
625 * tree-complex.c (expand_complex_div_wide): Update.
626 * tree-eh.c (lower_resx): Update.
627 (unsplit_eh): Update.
628 (cleanup_empty_eh_move_lp): Update.
629 * tree-inline.c (copy_edges_for_bb): Update.
630 (freqs_to_counts): Update.
631 (copy_cfg_body): Update.
632 * tree-ssa-dce.c (remove_dead_stmt): Update.
633 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
634 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
635 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
636 (unloop_loops): Update.
637 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
638 * tree-ssa-loop-split.c (connect_loops): Update.
639 (split_loop): Update.
640 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
641 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
642 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
643 * tree-ssa-reassoc.c (branch_fixup): Update.
644 * tree-ssa-tail-merge.c (replace_block_by): Update.
645 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
646 (compute_path_counts): Update.
647 (update_profile): Update.
648 (recompute_probabilities): Update.
649 (update_joiner_offpath_counts): Update.
650 (estimated_freqs_path): Update.
651 (freqs_to_counts_path): Update.
652 (clear_counts_path): Update.
653 (ssa_fix_duplicate_block_edges): Update.
654 (duplicate_thread_path): Update.
655 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
656 (case_bit_test_cmp): Update.
657 (collect_switch_conv_info): Update.
658 (gen_inbound_check): Update.
659 (do_jump_if_equal): Update.
660 (emit_cmp_and_jump_insns): Update.
661 * tree-tailcall.c (decrease_profile): Update.
662 (eliminate_tail_call): Update.
663 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
664 (vect_do_peeling): Update.
665 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
666 * ubsan.c (ubsan_expand_null_ifn): Update.
667 (ubsan_expand_ptr_ifn): Update.
668 * value-prof.c (gimple_divmod_fixed_value): Update.
669 (gimple_mod_pow2): Update.
670 (gimple_mod_subtract): Update.
672 (gimple_stringop_fixed_value): Update.
674 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
677 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
679 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
681 PR rtl-optimization/82395
682 * ira-color.c (allocno_priority_compare_func): Fix comparison step
683 based on non_spilled_static_chain_regno_p.
685 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
687 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
688 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
689 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
691 2017-10-19 Martin Sebor <msebor@redhat.com>
693 PR tree-optimization/82596
694 * tree.c (array_at_struct_end_p): Handle STRING_CST.
696 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
698 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
699 (get_mem_refs_of_builtin_call): Likewise.
700 * builtins.c (expand_builtin_apply): Adjust call to
701 allocate_dynamic_stack_space.
702 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
703 the third argument to allocate_dynamic_stack_space, otherwise -1.
704 (expand_builtin): Deal with all alloca variants.
705 (is_inexpensive_builtin): Likewise.
706 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
707 * calls.c (special_function_p): Deal with all alloca variants.
708 (initialize_argument_information): Adjust call to
709 allocate_dynamic_stack_space.
710 (expand_call): Likewise.
711 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
712 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
713 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
714 use it for the stack usage computation.
715 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
716 * function.c (gimplify_parameters): Call build_alloca_call_expr.
717 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
718 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
719 (in_loop_p): Remove first argument and useless check.
720 (pass_walloca::execute): Remove useless test and adjust call to above.
721 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
722 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
723 (gimplify_call_expr): Deal with all alloca variants.
724 * hsa-gen.c (gen_hsa_alloca): Likewise.
725 (gen_hsa_insns_for_call): Likewise.
726 * ipa-pure-const.c (special_builtin_state): Likewise.
727 * tree-chkp.c (chkp_build_returned_bound): Likewise.
728 * tree-object-size.c (alloc_object_size): Likewise.
729 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
730 (call_may_clobber_ref_p_1): Likewise.
731 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
732 (ccp_fold_stmt): Likewise.
733 (optimize_stack_restore): Likewise.
734 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
735 (mark_all_reaching_defs_necessary_1): Likewise.
736 (propagate_necessity): Likewise.
737 (eliminate_unnecessary_stmts): Likewise.
738 * tree.c (build_common_builtin_nodes): Build
739 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
740 (build_alloca_call_expr): New function.
741 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
742 (CASE_BUILT_IN_ALLOCA): Likewise.
743 (build_alloca_call_expr): Declare.
744 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
746 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
749 * dwarf2out.c (new_die_raw): New static inline function.
750 (new_die): Use it to create the DIE.
751 (add_AT_external_die_ref): Likewise.
752 (clone_die): Likewise.
753 (clone_as_declaration): Likewise.
754 (dwarf2out_vms_debug_main_pointer): Likewise.
755 (base_type_die): Likewise. Remove early return for corner cases.
756 Do not call add_pubtype on the DIE here.
757 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
758 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
759 typedefs for base types with DW_AT_endianity. Make sure a DIE with
760 native order exists for base types, attach the DIE manually and call
761 add_pubtype on it. Do not equate a reverse order DIE to the type.
763 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
765 * config/arm/arm.c (align_ok_ldrd_strd): New function.
766 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
768 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
770 2017-10-19 Jakub Jelinek <jakub@redhat.com>
772 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
773 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
774 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
775 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
776 * opts.c (sanitizer_opts): Add builtin.
777 * ubsan.c (instrument_builtin): New function.
778 (pass_ubsan::execute): Call it.
779 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
780 * doc/invoke.texi: Document -fsanitize=builtin.
782 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
783 builtins, store max (log2 (align), 0) into uchar field instead of
784 align into uptr field.
785 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
786 store uchar 0 field instead of uptr 0 field.
787 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
788 instead of passing one address of struct with 2 locations pass
789 two addresses of structs with 1 location each.
790 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
791 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
792 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
793 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
794 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
795 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
796 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
797 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
799 2017-10-19 Martin Liska <mliska@suse.cz>
802 * file-find.c (remove_prefix): Remove.
803 * file-find.h (remove_prefix): Likewise.
804 * gcc-ar.c: Remove smartness of lookup.
806 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
808 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
809 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
810 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
812 2017-10-19 Jakub Jelinek <jakub@redhat.com>
815 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
816 (setcc + and to xor + setcc): New peephole2.
818 2017-10-19 Tom de Vries <tom@codesourcery.com>
820 * doc/sourcebuild.texi (Test Directives, Variants of
821 dg-require-support): Add dg-require-stack-size.
823 2017-10-19 Martin Liska <mliska@suse.cz>
826 * gimplify.c (gimplify_decl_expr): Do not instrument variables
827 that have a large alignment.
828 (gimplify_target_expr): Likewise.
830 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
832 PR rtl-optimization/82602
833 * ira.c (rtx_moveable_p): Return false for volatile asm.
835 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
838 * config/i386/i386-modes.def (CCGZ): New CC mode.
839 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
840 * config/i386/predicates.md (ix86_comparison_operator):
842 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
843 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
844 with double-word subtraction.
845 (put_condition_code): Handle CCGZmode.
847 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
849 * wide-int.cc (debug (const wide_int &)): New.
850 (debug (const wide_int *)): New.
851 (debug (const widest_int &)): New.
852 (debug (const widest_int *)): New.
854 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
857 * lra-constraints.c (curr_insn_transform): Use non-input operand
858 instead of output one for matched reload.
860 2017-10-18 Bin Cheng <bin.cheng@arm.com>
862 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
863 (tree-ssa-loop-ivopts.h): New header file.
864 (struct builtin_info): New fields.
865 (classify_builtin_1): Compute and record base and offset parts for
866 memset builtin partition by calling strip_offset.
867 (offset_cmp, fuse_memset_builtins): New functions.
868 (finalize_partitions): Fuse adjacent memset partitions by calling
870 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
871 Expose the interface.
872 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
874 2017-10-18 Bin Cheng <bin.cheng@arm.com>
876 PR tree-optimization/82574
877 * tree-loop-distribution.c (find_single_drs): New parameter. Check
878 that data reference must be executed exactly once per iteration
879 against the outermost loop in nest.
880 (classify_partition): Update call to above function.
882 2017-10-18 Richard Biener <rguenther@suse.de>
884 PR tree-optimization/82591
885 * graphite.c (graphite_transform_loops): Move code gen message
887 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
888 Here. Handle scop_to_isl_ast failing.
889 (scop_to_isl_ast): Limit the number of ISL operations.
891 2017-10-18 Richard Biener <rguenther@suse.de>
893 * graphite-isl-ast-to-gimple.c
894 (translate_isl_ast_to_gimple::set_rename): Simplify.
895 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
896 (graphite_copy_stmts_from_block): ... here.
897 (copy_bb_and_scalar_dependences): Simplify.
898 (add_parameters_to_ivs_params): Canonicalize.
899 (generate_entry_out_of_ssa_copies): Simplify.
900 * graphite-sese-to-poly.c (extract_affine_name): Simplify
901 by passing in ISL dimension.
902 (parameter_index_in_region_1): Rename to ...
903 (parameter_index_in_region): ... this.
904 (extract_affine): Adjust assert, pass down parameter index.
905 (add_param_constraints): Use range-info when available.
906 (build_scop_context): Adjust.
907 * sese.c (new_sese_info): Adjust.
908 (free_sese_info): Likewise.
909 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
910 Remove unused typedefs.
911 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
913 2017-10-18 Martin Liska <mliska@suse.cz>
915 * combine.c (simplify_compare_const): Add gcc_fallthrough.
917 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
919 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
920 (s390_sched_init): Do not reset s390_sched_state if we entered the
921 current basic block via a fallthru edge and all others are unlikely.
923 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
925 * config/s390/s390.c (NUM_SIDES): New variable.
926 (LONGRUNNING_THRESHOLD): New variable.
927 (LATENCY_FACTOR): New variable.
928 (s390_sched_score): Decrease score for long-running instructions on
930 (s390_sched_variable_issue): Perform bookkeeping for long-running
933 2017-10-18 Richard Biener <rguenther@suse.de>
935 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
936 Simplify with removal of the parameter rename map.
937 (set_rename): Likewise.
938 (should_copy_to_new_region): Likewise.
939 (graphite_copy_stmts_from_block): Likewise.
940 (copy_bb_and_scalar_dependences): Remove initialization of
941 unused copied_bb_map.
943 (copy_internal_parameters): Likewise.
944 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
945 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
947 (parameter_index_in_region_1): Rename to ...
948 (assign_parameter_index_in_region): ... this. Assert we have
949 a parameter we handle.
950 (scan_tree_for_params): Adjust.
951 * sese.h (parameter_rename_map_t): Remove.
952 (struct sese_info_t): Remove unused parameter_rename_map and
953 copied_bb_map members.
954 * sese.c (new_sese_info): Adjust.
955 (free_sese_info): Likewise.
957 2017-10-18 Martin Liska <mliska@suse.cz>
960 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
963 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
965 * doc/invoke.texi (ffunction-sections and fdata-sections):
968 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
970 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
971 the use statement can throw internally.
973 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
975 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
976 any RTX present on the RHS of a SET.
977 * compare-elim.c (try_eliminate_compare): Restore comment.
979 2017-10-17 Jakub Jelinek <jakub@redhat.com>
981 * langhooks.h (struct lang_hooks): Document that tree_size langhook
982 may be also called on tcc_type nodes.
983 * langhooks.c (lhd_tree_size): Likewise.
985 2017-10-17 David Malcolm <dmalcolm@redhat.com>
987 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
988 format_warning_at_substring.
989 (maybe_warn): Convert source_range * param to a location_t. Pass
990 UNKNOWN_LOCATION rather than NULL to fmtwarn.
991 (format_directive): Remove code to extract source_ranges and
992 source_range * in favor of just a location_t.
993 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
995 * substring-locations.c (format_warning_va): Convert
996 source_range * param to a location_t.
997 (format_warning_at_substring): Likewise.
998 * substring-locations.h (format_warning_va): Likewise.
999 (format_warning_at_substring): Likewise.
1001 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1003 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1005 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1006 and scatter/gather ops.
1008 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1009 vec_gather_load and vec_scatter_store.
1010 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1011 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1012 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1013 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1014 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1015 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1017 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1019 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1020 Detect FTST instruction and handle its register pops. Only pop
1021 second operand if can_pop_second_op is true.
1022 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1023 set can_pop_second_op to false in the compare_for_stack_reg call.
1025 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1026 output_fp_compare for stack register operands.
1027 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1028 instructions here. Do not emit stack register pops here. Assert
1029 that FCOMPP pops next to top stack register. Rewrite function.
1031 2017-10-17 Nathan Sidwell <nathan@acm.org>
1034 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1035 use DECL_ASSEMBLER_NAME_RAW.
1038 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1041 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1042 Wilco Dijkstra <wilco.dijkstra@arm.com>
1044 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1045 converstion to Pmode from the buf_addr.
1047 2017-10-17 Richard Biener <rguenther@suse.de>
1049 * graphite-dependences.c (scop_get_reads_and_writes): Change
1050 output parameters to references.
1052 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1054 PR 71026/tree-optimization
1055 * fold-const.c (distribute_real_division): Removed.
1056 (fold_binary_loc): Remove calls to distribute_real_divison.
1058 2017-10-17 Richard Biener <rguenther@suse.de>
1060 * graphite-scop-detection.c
1061 (scop_detection::stmt_has_simple_data_refs_p): Always use
1062 the full nest as region.
1063 (try_generate_gimple_bb): Likewise.
1064 * sese.c (scalar_evolution_in_region): Simplify now that
1065 SCEV can handle instantiation in regions.
1066 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1067 in the non-loop part of a function if requested.
1069 2017-10-17 Richard Biener <rguenther@suse.de>
1071 PR tree-optimization/82563
1072 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1074 (graphite_regenerate_ast_isl): Call it.
1075 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1077 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1079 PR tree-optimization/82549
1080 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1081 Formatting fixes. Instead of calling make_bit_field_ref with negative
1084 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1086 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1087 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1088 _mm_maskz_reduce_ss): New.
1089 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1090 __builtin_ia32_reducess_mask): Ditto..
1091 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1092 * config/i386/sse.md (reduces<mode>): Renamed to ...
1093 (reduces<mode><mask_scalar_name>): ... this.
1094 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1096 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1097 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1099 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1101 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1102 * selftest-run-tests.c (selftest::run_tests): Call
1103 selftest::unique_ptr_tests_cc_tests.
1104 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1105 * unique-ptr-tests.cc: New file.
1107 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1110 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1112 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1113 (make_hard_regno_born, make_hard_regno_dead): Update
1114 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1116 2017-10-16 Jeff Law <law@redhat.com>
1118 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1120 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1122 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1124 2017-10-16 Olivier Hainque <hainque@adacore.com>
1126 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1127 with_cpu if we were configured for an e500v2 target cpu name.
1129 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1131 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1132 * doc/invoke.texi: Document +nodsp as a valid extension for
1135 2017-10-16 Martin Liska <mliska@suse.cz>
1137 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1138 (test_set_range): Likewise.
1139 (test_range_functions): Rename to ...
1140 (test_bit_in_range): ... this.
1141 (sbitmap_c_tests): Add new test.
1143 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1145 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1146 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1147 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1149 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1151 * config/aarch64/aarch64-builtins.c
1152 (aarch64_types_quadopu_lane_qualifiers): New.
1153 (TYPES_QUADOPU_LANE): New.
1154 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1155 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1156 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1157 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1158 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1159 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1160 (Vdottype, DOTPROD): New.
1161 (sur): Add SDOT and UDOT.
1163 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1165 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1166 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1167 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1168 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1169 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1170 (cortex-a75.cortex-a55): Likewise.
1171 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1173 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1175 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1176 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1177 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1178 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1179 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1180 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1181 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1182 * config/arm/types.md (neon_dot, neon_dot_q): New.
1183 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1185 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1187 * config/arm/arm.h (TARGET_DOTPROD): New.
1188 * config/arm/arm.c (arm_arch_dotprod): New.
1189 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1190 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1191 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1192 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1193 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1194 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1195 * doc/invoke.texi (armv8.2-a): Document dotprod
1197 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1199 * i386.c (ix86_vec_cost): New function.
1200 (ix86_rtx_costs): Handle vector operations better.
1201 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1202 * x86-tune-costs.h: Add new costs to all tables.
1204 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1206 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1208 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1209 divsd, sqrtss and sqrtsd
1210 * x86-tune-costs.h: Add new entries to all costs.
1211 (znver1_cost): Fix to match real instruction latencies.
1213 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1214 Michael Collison <michael.collison@arm.com>
1216 * compare-elim.c: Include emit-rtl.h.
1217 (can_merge_compare_into_arith): New function.
1218 (try_validate_parallel): Likewise.
1219 (try_merge_compare): Likewise.
1220 (try_eliminate_compare): Call the above when no previous clobber
1222 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1225 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1229 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1230 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1234 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1235 to be any operand_equal_p operands. For & (B - 1) require
1236 B to be power of 2. Recognize
1237 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1239 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1242 * optabs.c (expand_memory_blockage): Fix call of
1243 targetm.have_memory_blockage.
1245 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1248 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1249 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1250 objects to extra_objs instead of overwriting it.
1252 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1254 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1255 Use any_fp_register_operand as operand[3] predicate. Simplify
1256 equality test for operands[2] and operands[4] memory location.
1257 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1258 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1259 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1260 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1261 any_fp_register_operand as operand[1] predicate. Simplify
1262 equality test for operands[0] and operands[3] memory location.
1263 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1264 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1265 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1267 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1269 * target-insns.def: Add memory_blockage.
1270 * optabs.c (expand_memory_blockage): New function.
1271 (expand_asm_memory_barrier): Rename ...
1272 (expand_asm_memory_blockage): ... to this.
1273 (expand_mem_thread_fence): Call expand_memory_blockage
1274 instead of expand_asm_memory_barrier.
1275 (expand_mem_singnal_fence): Ditto.
1276 (expand_atomic_load): Ditto.
1277 (expand_atomic_store): Ditto.
1278 * doc/md.texi (Standard Pattern Names For Generation):
1279 Document memory_blockage instruction pattern.
1281 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1283 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1284 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1285 * config/rl78/rl78.md: New define_expand "adddi3".
1287 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1289 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1292 2017-10-13 Jeff Law <law@redhat.com>
1294 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1296 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1299 * internal-fn.c (expand_mul_overflow): If both operands have
1300 the same highpart of -1 or 0 and the topmost bit of lowpart
1301 is different, overflow is if res <= 0 rather than res < 0.
1303 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1305 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1306 TARGET_P9_VECTOR code for unaligned_load case.
1308 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1310 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1312 2017-10-13 Nathan Sidwell <nathan@acm.org>
1314 * tree-core.h (tree_contains_struct): Make bool.
1315 * tree.c (tree_contains_struct): Likewise.
1316 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1317 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1318 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1319 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1320 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1322 2017-10-13 Richard Biener <rguenther@suse.de>
1324 * graphite-isl-ast-to-gimple.c
1325 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1326 parameters and dominance check.
1327 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1328 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1329 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1330 Do not update SSA form here or do intermediate IL verification.
1331 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1332 (graphite_initialize): Remove check on the number of loops in
1333 the function and inline into graphite_transform_loops.
1334 (graphite_finalize): Inline into graphite_transform_loops.
1335 (graphite_transform_loops): Perform SSA update and IL verification
1337 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1339 2017-10-13 Richard Biener <rguenther@suse.de>
1341 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1342 graphite_expression_type_precision): Avoid global constructor
1344 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1345 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1346 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1347 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1348 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1349 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1351 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1354 * config/i386/i386.h (ix86_red_zone_size): New.
1355 * config/i386/i386.md (push peephole2s): Replace
1356 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1358 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * combine.c (can_change_dest_mode): Reject changes in
1363 REGMODE_NATURAL_SIZE.
1365 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1366 Alan Hayward <alan.hayward@arm.com>
1367 David Sherwood <david.sherwood@arm.com>
1369 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1370 (expand_debug_source_expr): Likewise.
1371 * combine.c (combine_simplify_rtx): Likewise.
1372 * cse.c (fold_rtx): Likewise.
1373 * fwprop.c (canonicalize_address): Likewise.
1374 * targhooks.c (default_shift_truncation_mask): Likewise.
1376 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1377 Alan Hayward <alan.hayward@arm.com>
1378 David Sherwood <david.sherwood@arm.com>
1380 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1381 (widened_mode): Likewise.
1382 (expand_unop): Likewise.
1383 * ree.c (transform_ifelse): Likewise.
1384 (merge_def_and_ext): Likewise.
1385 (combine_reaching_defs): Likewise.
1386 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1388 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1389 Alan Hayward <alan.hayward@arm.com>
1390 David Sherwood <david.sherwood@arm.com>
1392 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1393 * combine.c (gen_lowpart_for_combine): Likewise.
1394 * dwarf2out.c (rtl_for_decl_location): Likewise.
1395 * final.c (alter_subreg): Likewise.
1396 * rtlhooks.c (gen_lowpart_general): Likewise.
1397 (gen_lowpart_if_possible): Likewise.
1399 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1400 Alan Hayward <alan.hayward@arm.com>
1401 David Sherwood <david.sherwood@arm.com>
1403 * calls.c (expand_call): Use subreg_lowpart_offset.
1404 * cse.c (cse_insn): Likewise.
1405 * regcprop.c (copy_value): Likewise.
1406 (copyprop_hardreg_forward_1): Likewise.
1408 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1411 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1412 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1413 =Q constraints to +Q and into insn condition add check
1414 that operands[0] and operands[1] are equal.
1415 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1416 =Q constraints to +Q and into insn condition add check
1417 that operands[0] is equal to either operands[1] or operands[2].
1420 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1421 instead of handling MINUS_EXPR twice (once for each argument),
1422 canonicalize operand order and handle just once, use rtype where
1423 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1426 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1427 any values of __C while still being pattern recognizable as a simple
1430 2017-10-13 Richard Biener <rguenther@suse.de>
1432 PR tree-optimization/82451
1434 2017-10-02 Richard Biener <rguenther@suse.de>
1436 PR tree-optimization/82355
1437 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1438 a mapping for the enclosing loop but avoid generating one for
1440 (copy_bb_and_scalar_dependences): Remove premature codegen
1441 error on PHIs in blocks duplicated into multiple places.
1442 * graphite-scop-detection.c
1443 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1444 in the region use it as loop and nest to analyze the DR in.
1445 (try_generate_gimple_bb): Likewise.
1446 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1447 (add_loop_constraints): For blocks in a loop not in the region
1448 create a dimension with a single iteration.
1449 * sese.h (gbb_loop_at_index): Remove assert.
1451 * cfgloop.c (loop_preheader_edge): For the loop tree root
1452 return the single successor of the entry block.
1453 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1454 Reset the SCEV hashtable and niters.
1455 * graphite-scop-detection.c
1456 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1457 assert that we only have POLYNOMIAL_CHREC that vary in loops
1458 contained in the region.
1459 (scop_detection::graphite_can_represent_expr): Adjust.
1460 (scop_detection::stmt_has_simple_data_refs_p): For loops
1461 not in the region set loop to NULL. The nest is now the
1462 entry edge to the region.
1463 (try_generate_gimple_bb): Likewise.
1464 * sese.c (scalar_evolution_in_region): Adjust for
1465 instantiate_scev change.
1466 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1467 Make nest parameter the edge into the region.
1468 (create_data_ref): Likewise.
1469 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1470 entry edge into a region and adjust instantiate_scev calls.
1471 (create_data_ref): Likewise.
1472 (graphite_find_data_references_in_stmt): Likewise.
1473 (find_data_references_in_stmt): Pass the loop preheader edge
1474 from the nest argument.
1475 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1476 parameter the edge into the region.
1477 (instantiate_parameters): Use the loop preheader edge as entry.
1478 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1480 (get_instantiated_value_entry): Make instantiate_below parameter
1481 the edge into the region.
1482 (instantiate_scev_name): Likewise. Adjust dominance checks,
1483 when we cannot use loop-based instantiation instantiate by
1484 walking use-def chains.
1485 (instantiate_scev_poly): Adjust.
1486 (instantiate_scev_binary): Likewise.
1487 (instantiate_scev_convert): Likewise.
1488 (instantiate_scev_not): Likewise.
1489 (instantiate_array_ref): Remove.
1490 (instantiate_scev_3): Likewise.
1491 (instantiate_scev_2): Likewise.
1492 (instantiate_scev_1): Likewise.
1493 (instantiate_scev_r): Do not blindly handle N-operand trees.
1494 Do not instantiate array-refs. Handle all constants and invariants.
1495 (instantiate_scev): Make instantiate_below parameter
1496 the edge into the region.
1497 (resolve_mixers): Use the loop preheader edge for the region
1498 parameter to instantiate_scev_r.
1499 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1501 2017-10-13 Richard Biener <rguenther@suse.de>
1503 PR tree-optimization/82525
1504 * graphite-isl-ast-to-gimple.c
1505 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1507 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1508 Fail code generation when we cannot represent the isl integer.
1509 (binary_op_to_tree): Elide modulo operations that are no-ops
1510 in the type we code generate. Remove now superfluous code
1513 2017-10-13 Richard Biener <rguenther@suse.de>
1515 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1516 (scop_detection::harmful_loop_in_region): Remove premature
1517 IV type restriction.
1518 (scop_detection::graphite_can_represent_scev): We can handle
1519 pointer IVs just fine.
1521 2017-10-13 Alan Modra <amodra@gmail.com>
1523 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1524 "Clobbers and Scratch Registers". Add paragraph on
1525 alternative to clobbers for scratch registers and OpenBLAS
1528 2017-10-13 Alan Modra <amodra@gmail.com>
1530 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1531 example of a memory input for a string of known length. Move
1532 commentary out of table. Add a number of new examples
1533 covering array memory inputs.
1535 2017-10-12 Martin Liska <mliska@suse.cz>
1537 PR tree-optimization/82493
1538 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1539 (test_range_functions): New function.
1540 (sbitmap_c_tests): Likewise.
1541 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1542 * selftest.h (sbitmap_c_tests): New function.
1544 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1546 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1548 * config/rs6000/amo.h: Fix spacing issue.
1550 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1553 * config/i386/i386.md (*ashl<mode>3_mask_1,
1554 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1555 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1558 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1560 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1561 (profile_probability): Set max_probability
1562 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1564 * profile-count.c (profile_probability::differs_from_p): Do not
1565 rely on max_probaiblity == 10000
1567 2017-10-12 Jeff Law <law@redhat.com>
1569 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1572 2017-10-12 Martin Sebor <msebor@redhat.com>
1576 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1577 (handle_alias_pairs): Call it.
1578 * common.opt (-Wattribute-alias): New option.
1579 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1580 * doc/invoke.texi (-Wattribute-alias): Document.
1582 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1585 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1587 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1589 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1590 (make_hard_regno_born, make_hard_regno_dead): Update
1591 bb_killed_pseudos and bb_gen_pseudos.
1593 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1595 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1597 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1599 * config/alpha/alpha.c (alpha_split_conditional_move):
1600 Use std::swap instead of manually swapping.
1601 (alpha_stdarg_optimize_hook): Ditto.
1602 (alpha_canonicalize_comparison): Ditto.
1604 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1606 * tree-loop-distribution.c (struct builtin_info): New struct.
1607 (struct partition): Refactor fields into struct builtin_info.
1608 (partition_free): Free struct builtin_info.
1609 (build_size_arg_loc, build_addr_arg_loc): Delete.
1610 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1611 information from struct builtin_info.
1612 (find_single_drs): New function refactored from classify_partition.
1613 Also moved builtin validity checks to this function.
1614 (compute_access_range, alloc_builtin): New functions.
1615 (classify_builtin_st, classify_builtin_ldst): New functions.
1616 (classify_partition): Refactor code into functions find_single_drs,
1617 classify_builtin_st and classify_builtin_ldst.
1618 (distribute_loop): Don't do runtime alias check when distributing
1620 (find_seed_stmts_for_distribution): New function.
1621 (pass_loop_distribution::execute): Refactor code finding seed
1622 stmts into above function. Support distribution for the innermost
1623 two-level loop nest. Adjust dump information.
1625 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1627 * tree-loop-distribution.c: Adjust the general comment.
1628 (NUM_PARTITION_THRESHOLD): New macro.
1629 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1630 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1631 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1632 in call to build_partition_graph.
1633 (finalize_partitions): New parameter. Make loop distribution more
1634 conservative by fusing more partitions.
1635 (distribute_loop): Don't do runtime alias check in case of loop nest
1637 (find_seed_stmts_for_distribution): New function.
1638 (prepare_perfect_loop_nest): New function.
1639 (pass_loop_distribution::execute): Refactor code finding seed stmts
1640 and loop nest into above functions. Support loop nest distribution.
1641 Adjust dump information accordingly.
1643 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1645 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1646 and set PTYPE_SEQUENTIAL for merged partition.
1648 2017-10-12 Richard Biener <rguenther@suse.de>
1650 PR tree-optimization/69728
1652 2017-09-19 Richard Biener <rguenther@suse.de>
1654 PR tree-optimization/69728
1655 * graphite-sese-to-poly.c (schedule_error): New global.
1656 (add_loop_schedule): Handle empty domain by failing the
1658 (build_original_schedule): Handle schedule_error.
1660 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1661 domain by returning an unchanged schedule.
1663 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1665 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1666 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1668 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1670 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1673 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1676 * expr.c (store_field): Don't optimize away bitsize == 0 store
1677 from CALL_EXPR with addressable return type.
1679 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1681 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1682 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1683 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1684 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1685 TARGET_ISEL instead of TARGET_ISEL<sel>.
1687 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1689 * config/rs6000/rs6000.c
1690 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1692 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1694 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1696 (reg_or_cint_operand): Fix comment.
1697 (reg_or_zero_operand): New predicate.
1698 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1699 * config/rs6000/rs6000.c (output_isel): Delete.
1700 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1701 instead of reg_or_cint_operand. Output instruction directly (not via
1703 (isel_unsigned_<mode>): Ditto.
1704 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1705 gpc_reg_operand. Add an instruction alternative for this. Output
1706 instruction directly.
1707 (*isel_reversed_unsigned_<mode>): Ditto.
1709 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1711 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1712 (TARGET_CANONICALIZE_COMPARISON): Define.
1714 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1717 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1718 Check whether the dest is REG before adding REG_EQUIV note.
1720 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1723 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1725 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1726 (make_hard_regno_born, make_hard_regno_dead): Update
1727 bb_killed_pseudos and bb_gen_pseudos.
1729 2017-10-11 Nathan Sidwell <nathan@acm.org>
1731 * incpath.h (enum incpath_kind): Name enum, prefix values.
1732 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1733 * incpath.c (heads, tails): Use INC_MAX.
1734 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1735 (merge_include_chains, split_quote_chain,
1736 register_include_chains): Update incpath_kind names.
1737 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1738 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1740 (add_framework_path, darwin_register_objc_includes): Likewise.
1741 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1743 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1745 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1746 Do not use float_operator operator predicate.
1747 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1748 * config/i386/predicates.md (float_operator): Remove predicate.
1750 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1752 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1753 (*jccxf_i387): Ditto.
1754 (*jcc<mode>_i387): Ditto.
1755 (*jccu<mode>_i387): Ditto.
1756 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1757 (*jcc_*_i387 splitters): Remove.
1758 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1759 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1760 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1763 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1765 * profile-count.h (slow_safe_scale_64bit): New function.
1766 (safe_scale_64bit): New inline.
1767 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1768 * profile-count.c: Include wide-int.h
1769 (slow_safe_scale_64bit): New.
1771 2017-10-11 Nathan Sidwell <nathan@acm.org>
1773 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1774 HAS_DECL_ASSEMBLER_NAME_P.
1775 * gimple-expr.c (gimple_decl_printable_name: Check
1776 HAS_DECL_ASSEMBLER_NAME_P too.
1777 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1778 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1779 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1780 * passes.c (rest_of_decl_compilation): Check
1781 HAS_DECL_ASSEMBLER_NAME_P too.
1782 * recog.c (verify_changes): Likewise.
1783 * tree-pretty-print.c (dump_decl_name): Likewise.
1784 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1786 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1787 (SET_DECL_ASSEMBLER_NAME): Use it.
1788 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1789 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1790 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1792 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1794 * config.gcc (i386, x86_64): Add extra objects.
1795 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1796 (ix86_min_insn_size): Declare.
1797 (ix86_issue_rate): Declare.
1798 (ix86_adjust_cost): Declare.
1799 (ia32_multipass_dfa_lookahead): Declare.
1800 (ix86_macro_fusion_p): Declare.
1801 (ix86_macro_fusion_pair_p): Declare.
1802 (ix86_bd_has_dispatch): Declare.
1803 (ix86_bd_do_dispatch): Declare.
1804 (ix86_core2i7_init_hooks): Declare.
1805 (ix86_atom_sched_reorder): Declare.
1806 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1807 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1808 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1809 (rip_relative_addr_p): Rename to ...
1810 (ix86_rip_relative_addr_p): ... this one; export.
1811 (memory_address_length): Update.
1812 (ix86_issue_rate): Move to x86-tune-sched.c.
1813 (ix86_flags_dependent): Move to x86-tune-sched.c.
1814 (ix86_agi_dependent): Move to x86-tune-sched.c.
1815 (exact_dependency_1): Move to x86-tune-sched.c.
1816 (exact_store_load_dependency): Move to x86-tune-sched.c.
1817 (ix86_adjust_cost): Move to x86-tune-sched.c.
1818 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1819 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1820 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1821 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1822 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1823 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1824 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1825 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1826 (min_insn_size): Rename to ...
1827 (ix86_min_insn_size): ... this one; export.
1828 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1829 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1830 (core2i7_first_cycle_multipass_backtrack): Move to
1831 x86-tune-sched-core.c.
1832 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1833 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1834 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1835 (ix86_avoid_jump_mispredicts): Update.
1836 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1837 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1838 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1839 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1840 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1841 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1842 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1843 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1844 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1845 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1846 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1847 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1848 (BIG): Move to ix86-tune-sched-bd.c.
1849 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1850 (enum insn_path): Move to ix86-tune-sched-bd.c.
1851 (get_mem_group): Move to ix86-tune-sched-bd.c.
1852 (is_cmp): Move to ix86-tune-sched-bd.c.
1853 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1854 (is_branch): Move to ix86-tune-sched-bd.c.
1855 (is_prefetch): Move to ix86-tune-sched-bd.c.
1856 (init_window): Move to ix86-tune-sched-bd.c.
1857 (allocate_window): Move to ix86-tune-sched-bd.c.
1858 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1859 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1860 (process_end_window): Move to ix86-tune-sched-bd.c.
1861 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1862 (find_constant): Move to ix86-tune-sched-bd.c.
1863 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1864 (has_immediate): Move to ix86-tune-sched-bd.c.
1865 (get_insn_path): Move to ix86-tune-sched-bd.c.
1866 (get_insn_group): Move to ix86-tune-sched-bd.c.
1867 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1868 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1869 (add_insn_window): Move to ix86-tune-sched-bd.c.
1870 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1871 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1872 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1873 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1874 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1875 (do_dispatch): Move to ix86-tune-sched-bd.c.
1876 (has_dispatch): Move to ix86-tune-sched-bd.c.
1877 * config/i386/t-i386: Add new object files.
1878 * config/i386/x86-tune-costs.h: New file.
1879 * config/i386/x86-tune-sched-atom.c: New file.
1880 * config/i386/x86-tune-sched-bd.c: New file.
1881 * config/i386/x86-tune-sched-core.c: New file.
1882 * config/i386/x86-tune-sched.c: New file.
1884 2017-10-11 Liu Hao <lh_mouse@126.com>
1886 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1887 the generic version below instead.
1888 (should_colorize): Recognize Windows consoles as terminals
1890 * pretty-print.c [__MINGW32__] (write_all): New function.
1891 [__MINGW32__] (find_esc_head): Likewise.
1892 [__MINGW32__] (find_esc_terminator): Likewise.
1893 [__MINGW32__] (eat_esc_sequence): Likewise.
1894 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1896 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1899 2017-10-11 Richard Biener <rguenther@suse.de>
1901 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1902 Properly call analyze_scalar_evolution with the loop of the stmt.
1904 2017-10-11 Richard Biener <rguenther@suse.de>
1906 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1907 * tree-core.h (tree_base): Add chrec_var union member.
1908 * tree.h (CHREC_VAR): Remove.
1909 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1910 * tree-chrec.h (build_polynomial_chrec): Adjust.
1911 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1912 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1914 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1916 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1917 * match.pd: ... here.
1918 ((T) X == (T) Y): Relax condition.
1920 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1922 PR tree-optimization/82472
1923 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1925 (break_alias_scc_partitions): Update postorder number.
1927 2017-10-11 Martin Liska <mliska@suse.cz>
1930 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1932 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1934 2017-10-11 Martin Liska <mliska@suse.cz>
1936 * print-rtl.c (print_insn): Move declaration of idbuf
1937 to same scope as name.
1939 2017-10-11 Martin Liska <mliska@suse.cz>
1944 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1947 2017-10-11 Martin Liska <mliska@suse.cz>
1950 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1953 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1955 * config.gcc (default_gnu_indirect_function): Default to yes for
1956 arm*-*-linux* with glibc.
1958 2017-10-11 Richard Biener <rguenther@suse.de>
1960 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1961 default-defs and types we do not want to analyze.
1962 (interpret_loop_phi): Replace unreachable code with an assert.
1963 (compute_scalar_evolution_in_loop): Remove and inline ...
1964 (analyze_scalar_evolution_1): ... here, replacing condition with
1965 what makes the intent clearer. Remove handling of cases
1966 get_scalar_evolution now handles.
1968 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1970 PR rtl-optimization/81434
1971 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1972 comment for main loop. In sched_group_found if, also add checks for
1973 pass and min_cost_group.
1975 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1977 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1978 (rs6000_insn_cost): New function.
1979 * config/rs6000/rs6000.md (cost): New attribute.
1981 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1982 H.J. Lu <hongjiu.lu@intel.com>
1986 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1987 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1988 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1989 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1990 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1991 OPTION_MASK_ISA_MMX and some other bit set require both
1992 mmx and the ISAs without the mmx bit.
1993 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1994 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1995 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1996 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1997 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1998 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1999 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2000 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2001 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2002 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2003 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2004 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2005 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2006 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2007 Add OPTION_MASK_ISA_MMX.
2009 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2011 * config.gcc (armv7*-*-freebsd*): New target.
2012 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2014 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2016 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2017 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2020 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2022 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2023 a template parameter.
2024 (WIDE_INT_REF_FOR): Update accordingly.
2025 * tree.h (wi::int_traits <const_tree>): Delete.
2026 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2027 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2028 (wi::tree_to_wide_ref): New typedef.
2029 (wi::to_wide): New function.
2030 * calls.c (get_size_range): Use wi::to_wide when operating on
2032 * cgraph.c (cgraph_node::create_thunk): Likewise.
2033 * config/i386/i386.c (ix86_data_alignment): Likewise.
2034 (ix86_local_alignment): Likewise.
2035 * dbxout.c (stabstr_O): Likewise.
2036 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2037 * expr.c (const_vector_from_tree): Likewise.
2038 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2039 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2040 (fold_negate_expr_1, int_const_binop_1, const_binop)
2041 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2042 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2043 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2044 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2045 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2046 (fold_not_const, round_up_loc): Likewise.
2047 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2048 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2049 (alloca_call_type): Likewise.
2050 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2051 * godump.c (go_output_typedef): Likewise.
2052 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2053 * internal-fn.c (get_min_precision): Likewise.
2054 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2055 * ipa-polymorphic-call.c
2056 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2057 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2058 (ipa_modify_call_arguments): Likewise.
2059 * match.pd: Likewise.
2060 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2061 * print-tree.c (print_node_brief, print_node): Likewise.
2062 * stmt.c (expand_case): Likewise.
2063 * stor-layout.c (layout_type): Likewise.
2064 * tree-affine.c (tree_to_aff_combination): Likewise.
2065 * tree-cfg.c (group_case_labels_stmt): Likewise.
2066 * tree-data-ref.c (dr_analyze_indices): Likewise.
2067 (prune_runtime_alias_test_list): Likewise.
2068 * tree-dump.c (dequeue_and_dump): Likewise.
2069 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2070 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2071 * tree-pretty-print.c (dump_generic_node): Likewise.
2072 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2073 (simple_iv_with_niters): Likewise.
2074 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2075 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2076 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2077 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2078 (refine_value_range_using_guard, number_of_iterations_ne_max)
2079 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2080 (get_cst_init_from_scev, record_nonwrapping_iv)
2081 (scev_var_range_cant_overflow): Likewise.
2082 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2083 * tree-ssa-pre.c (compute_avail): Likewise.
2084 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2085 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2086 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2087 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2088 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2089 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2090 (dump_case_nodes, try_switch_expansion): Likewise.
2091 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2092 (vect_do_peeling): Likewise.
2093 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2094 * tree-vect-stmts.c (vectorizable_load): Likewise.
2095 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2096 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2097 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2098 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2099 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2100 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2101 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2102 (evrp_dom_walker::before_dom_children): Likewise.
2103 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2104 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2105 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2106 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2107 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2108 (get_range_pos_neg): Likewise.
2109 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2110 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2111 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2112 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2113 * config/avr/avr.c (avr_fold_builtin): Likewise.
2114 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2115 * config/msp430/msp430.c (msp430_attr): Likewise.
2116 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2117 * config/powerpcspe/powerpcspe-c.c
2118 (altivec_resolve_overloaded_builtin): Likewise.
2119 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2120 (rs6000_expand_ternop_builtin): Likewise.
2121 * config/rs6000/rs6000-c.c
2122 (altivec_resolve_overloaded_builtin): Likewise.
2123 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2124 (rs6000_expand_ternop_builtin): Likewise.
2125 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2127 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2129 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2130 when copying loop nest with only one inner loop.
2132 2017-10-10 Richard Biener <rguenther@suse.de>
2134 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2135 blocks if SCEV is active.
2136 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2138 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2139 (scev_initialize): Assert we are not yet initialized.
2141 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2143 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2144 inner loop's exit stmt by making it always exit the loop, otherwise
2145 we would generate an infinite empty loop.
2147 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2149 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2150 renaming variables in new preheader if it's deleted.
2152 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2154 * tree-loop-distribution.c (struct partition): Remove unused field
2155 loops of the structure.
2156 (partition_alloc, partition_free): Ditto.
2157 (build_rdg_partition_for_vertex): Ditto.
2159 2017-10-09 Jeff Law <law@redhat.com>
2161 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2162 return type to match prototype and documentation.
2164 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2166 * config/rs6000/rs6000.c (processor_costs): Move to ...
2167 * config/rs6000/rs6000.h: ... here.
2168 (rs6000_cost): Declare.
2170 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2172 * except.c (setjmp_fn): New global variable.
2173 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2174 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2175 if DONT_USE_BUILTIN_SETJMP is defined.
2177 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2179 * target.def (insn_cost): New hook.
2180 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2181 * doc/tm.texi: Regenerate.
2182 * rtlanal.c (insn_cost): Use the new hook.
2184 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2186 * combine.c (combine_validate_cost): Compute the new insn_cost,
2187 not just pattern_cost.
2188 (try_combine): Adjust comment.
2190 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2192 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2194 * combine.c (uid_insn_cost): Adjust comment.
2195 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2197 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2198 * dse.c (find_shift_sequence): Ditto.
2199 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2200 (bb_valid_for_noce_process_p): Use pattern_cost.
2201 * rtl.h (insn_rtx_cost): Delete.
2202 (pattern_cost): New prototype.
2203 (insn_cost): New prototype.
2204 * rtlanal.c (insn_rtx_cost): Rename to...
2205 (pattern_cost): ... this.
2208 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2210 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2211 (*jcc<mode>_0_r_i387): Ditto.
2212 (*jccxf_r_i387): Ditto.
2213 (*jcc<mode>_r_i387): Ditto.
2214 (*jccu<mode>_r_i387): Ditto.
2215 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2216 (*jcc): Rename from *jcc_1.
2218 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2220 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2221 deferred rescans after the lvx/stvx recombination pre-pass.
2223 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2225 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2226 memory operation instruction support.
2227 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2228 (rs6000-ibm-aix[789]*): Likewise.
2229 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2230 Document new functions.
2232 2017-10-09 Richard Biener <rguenther@suse.de>
2234 PR tree-optimization/82397
2235 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2236 equality only for semantically equal trees.
2238 2017-10-09 Richard Biener <rguenther@suse.de>
2240 PR tree-optimization/82449
2241 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2242 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2243 allow constant addresses.
2244 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2247 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2249 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2252 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2255 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2258 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2261 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2263 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2266 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2267 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2269 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2271 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2272 (WI_BINARY_PREDICATE_RESULT): Likewise.
2273 (wi::binary_traits::operator_result): New type.
2274 (wi::binary_traits::predicate_result): Likewise.
2275 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2276 (generic_wide_int::operator==, generic_wide_int::operator!=)
2277 (generic_wide_int::operator&, generic_wide_int::and_not)
2278 (generic_wide_int::operator|, generic_wide_int::or_not)
2279 (generic_wide_int::operator^, generic_wide_int::operator+
2280 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2282 (operator~, unary operator-, operator==, operator!=, operator&)
2283 (operator|, operator^, operator+, binary operator-, operator*): New
2285 * expr.c (get_inner_reference): Use wi::bit_and_not.
2286 * fold-const.c (fold_binary_loc): Likewise.
2287 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2288 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2289 (bit_value_binop): Likewise.
2290 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2291 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2292 (extract_range_from_binary_expr_1): Likewise.
2293 (masked_increment): Likewise.
2294 (simplify_bit_ops_using_ranges): Likewise.
2296 2017-10-09 Martin Jambor <mjambor@suse.cz>
2299 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2300 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2301 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2303 (hsa_fixup_mov_insn_type): New function.
2304 (hsa_op_with_type::get_in_type): Use it.
2305 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2306 immediates in an assert.
2307 (hsa_op_with_type::extend_int_to_32bit): New method.
2308 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2309 types. Convert to dest type if necessary.
2310 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2311 (reg_for_gimple_ssa): Pass false as min32int to
2312 hsa_type_for_scalar_tree_type.
2313 (gen_hsa_addr): Fixup type when creating addresable temporary.
2314 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2315 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2316 necessary. Call hsa_fixup_mov_insn_type.
2317 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2318 extend operands and convert to dest type if necessary.
2319 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2320 to dest type if necessary.
2321 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2322 if conversion nt necessary and size matches.
2323 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2324 to dest type if necessary.
2325 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2326 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2328 (gen_hsa_clrsb): Likewise.
2329 (gen_hsa_ffs): Likewise.
2330 (gen_hsa_divmod): Extend operands and convert to dest type if
2332 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2334 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2336 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2337 Remove empty default arguments. Use a brace block as output
2339 (conditional return): Ditto.
2341 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2342 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2344 (group_ending_nop): Ditto.
2345 (doloop_end): Ditto.
2346 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2347 (splitters for those): Ditto.
2349 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2351 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2352 a conditional jump (and the compare for it) so that pc_rtx is the
2354 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2355 for the deleted and renamed ctr<mode>_internal[234] patterns.
2356 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2357 Delete second conditional return pattern.
2358 (ctr<mode>_internal2): Delete this second bdnz pattern.
2359 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2360 (ctr<mode>_internal4): Delete this second bdz pattern.
2362 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2364 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2365 (always_initialized_rtx_for_ssa_name_p): New predicate.
2366 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2367 (finish_out_of_ssa): Free new field of SA.
2368 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2369 * tree-ssa-coalesce.c: Include tree-ssa.h.
2370 (get_parm_default_def_partitions): Remove extern keyword.
2371 (get_undefined_value_partitions): New function.
2372 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2373 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2375 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2377 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2379 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2381 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2383 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2384 for targets that preffer 128bit.
2386 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2388 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2390 2017-10-08 Olivier Hainque <hainque@adacore.com>
2392 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2393 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2395 (thumb_set_return_address): Likewise.
2397 2017-10-08 Olivier Hainque <hainque@adacore.com>
2399 * common/config/arm/arm-common.c (arm_except_unwind_info):
2400 Handle DWARF2_UNWIND_INFO.
2402 2017-10-07 Michael Collison <michael.collison@arm.com>
2404 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2407 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2409 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2410 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2411 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2412 defined, force the creation of a new block for a dispatch label.
2414 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2416 * invoke.texi (Wsuggest-attribute=cold): Document.
2417 * common.opt (Wsuggest-attribute=cold): New
2418 * ipa-pure-const.c (warn_function_cold): New function.
2419 * predict.c (compute_function_frequency): Use it.
2420 * predict.h (warn_function_cold): Declare.
2422 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2424 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2427 2017-10-06 Martin Liska <mliska@suse.cz>
2429 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2430 keyword for member functions.
2431 (struct sanopt_tree_couple): New struct.
2432 (struct sanopt_tree_couple_hash): New function.
2433 (struct sanopt_ctx): Add new hash_map.
2434 (has_dominating_ubsan_ptr_check): New function.
2435 (record_ubsan_ptr_check_stmt): Likewise.
2436 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2437 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2438 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2440 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2443 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2444 aarch64_simd_valid_immediate on CONST_VECTORs.
2445 (aarch64_reg_or_bic_imm): Likewise.
2447 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2449 PR rtl-optimization/82396
2450 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2452 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2454 * graphite-dependences.c (scop_get_reads): Move code to...
2455 (scop_get_must_writes): Move code to...
2456 (scop_get_may_writes): Move code to...
2457 (scop_get_reads_and_writes): ... here.
2458 (scop_get_dependences): Call scop_get_reads_and_writes.
2460 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2462 PR tree-optimization/82434
2463 * fold-const.h (can_native_encode_type_p,
2464 can_native_encode_string_p): Remove.
2465 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2466 don't encode anything, just return what would be otherwise returned.
2467 (native_encode_fixed, native_encode_complex, native_encode_vector):
2469 (native_encode_string): Likewise. Inline by hand
2470 can_native_encode_string_p.
2471 (can_native_encode_type_p): Remove.
2472 (can_native_encode_string_p): Remove.
2473 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2474 STRING_CSTs using can_native_encode_string_p, test all
2475 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2476 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2477 argument from native_encode_expr.
2478 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2479 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2482 2017-10-06 Richard Biener <rguenther@suse.de>
2484 PR tree-optimization/82397
2485 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2486 operand_equal_p but rely on data_ref_compare_tree for detecting
2488 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2489 to match up with dr_group_sort_cmp.
2491 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2494 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2496 * config/s390/s390-builtin-types.def: Regenerate.
2498 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2501 * config/s390/s390-builtin-types.def: Regenerate.
2502 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2503 Change flag from B_VXE to B_VX.
2504 (s390_vec_min_dbl): Remove B_VXE flag.
2506 2017-10-06 Richard Biener <rguenther@suse.de>
2508 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2509 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2510 translate_isl_ast_to_gimple::is_valid_rename,
2511 translate_isl_ast_to_gimple::get_rename,
2512 translate_isl_ast_to_gimple::get_def_bb_for_const,
2513 translate_isl_ast_to_gimple::get_new_name,
2514 translate_isl_ast_to_gimple::collect_all_ssa_names,
2515 translate_isl_ast_to_gimple::copy_loop_phi_args,
2516 translate_isl_ast_to_gimple::collect_all_ssa_names,
2517 translate_isl_ast_to_gimple::copy_loop_phi_args,
2518 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2519 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2520 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2521 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2522 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2523 translate_isl_ast_to_gimple::copy_cond_phi_args,
2524 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2525 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2526 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2527 translate_isl_ast_to_gimple::rename_uses,
2528 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2529 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2530 (set_rename_for_each_def): Likewise.
2531 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2532 here. Handle rewriting SCEV analyzable uses here.
2533 (copy_bb_and_scalar_dependences): Generate code for PHI
2535 (graphite_regenerate_ast_isl): Adjust.
2536 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2537 (add_write, add_read): New functions.
2538 (build_cross_bb_scalars_def): Use it and simplify.
2539 (build_cross_bb_scalars_use): Likewise.
2540 (graphite_find_cross_bb_scalar_vars): Inline into...
2541 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2542 simulating out-of-SSA. Compute liveout and add dependencies.
2543 (build_scops): Force an empty entry block.
2544 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2546 (sese_build_liveouts): Declare.
2547 (sese_trivially_empty_bb_p): Likewise.
2548 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2549 compute liveout and debug_liveout.
2550 (sese_bad_liveouts_use): Remove.
2551 (sese_reset_debug_liveouts_bb): Likewise.
2552 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2553 (sese_build_liveouts): Build liveout and debug_liveout and store
2555 (new_sese_info): Adjust.
2556 (free_sese_info): Likewise.
2557 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2558 do not build liveout here.
2559 (move_sese_in_condition): Adjust region entry.
2560 (scev_analyzable_p): Match up with chrec_apply requirements.
2561 (sese_trivially_empty_bb_p): New.
2562 * tree-into-ssa.c (get_reaching_def): Properly support generating
2563 default-defs for incremental rewrite of anonymous names.
2565 2017-10-06 Richard Biener <rguenther@suse.de>
2567 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2568 precision do not perform modulo reduction.
2570 2017-10-06 Richard Biener <rguenther@suse.de>
2572 PR tree-optimization/82436
2573 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2574 conservatively choose the vectorization factor when checking
2575 whether we can perform the required load permutation.
2576 (vect_transform_slp_perm_load): Assert when we may not fail.
2578 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2580 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2581 message for incompatible -msdata=* and -mcall-* options.
2583 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2585 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2586 rate for post-reload scheduling.
2588 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2590 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2592 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2594 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2595 to improve monte carlo in scimark.
2597 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2599 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2600 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2601 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2602 pentium4_cost, nocona_cost): Set reassociation width to 1.
2603 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2604 width to 2 for fp operations and 1 otherwise.
2605 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2607 (atom_cost): Set reassociation width to 2.
2608 (slm_cost, generic_cost): Set fp reassociation width
2609 to 2 and 1 otherwise.
2610 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2611 (core_cost): Set fp reassociation width to 4 and vector to 2.
2612 (ix86_reassociation_width): Rewrite using cost table; special case
2613 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2614 and TARGET_AVX128_OPTIMAL.
2615 * config/i386/i386.h (processor_costs): Add
2616 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2617 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2618 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2619 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2620 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2621 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2623 2017-10-05 Nathan Sidwell <nathan@acm.org>
2625 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2627 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2629 * config/arm/arm.c (arm_test_fpu_data): New.
2630 (arm_run_selftests): Call arm_test_fpu_data.
2632 2017-10-04 Nathan Sidwell <nathan@acm.org>
2634 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2636 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2638 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2640 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2641 check type for aarch64_simd_valid_immediate.
2642 (aarch64_output_simd_mov_immediate): Update prototype.
2643 (aarch64_simd_valid_immediate): Update prototype.
2644 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2645 support for ORR-immediate.
2646 (and<mode>3): modified pattern to add support for BIC-immediate.
2647 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2648 now checks for valid immediate for BIC and ORR based on new enum
2650 (aarch64_output_simd_mov_immediate): Function now used to output
2651 BIC/ORR imm as well based on new enum argument.
2652 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2654 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2655 (aarch64_reg_or_bic_imm): Likewise.
2657 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2659 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2660 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2662 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2666 PR rtl-optimization/82396
2667 * haifa-sched.c (autopref_multipass_init): Simplify
2669 (autopref_rank_data): Simplify sort order.
2670 * sched-int.h (autopref_multipass_data_): Remove
2671 multi_mem_insn_p, min_offset and max_offset.
2673 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2675 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2677 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2679 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2682 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2684 * doc/sourcebuild.texi: Document vect_long_mult.
2686 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2688 PR tree-optimization/82413
2689 * fold-const.c (build_range_check): Use widest_int when comparing
2690 the maximum ETYPE value with HIGH.
2692 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2694 PR rtl-optimization/82396
2695 * haifa-sched.c (autopref_multipass_init): Simplify
2697 (autopref_rank_data): Simplify sort order.
2698 * sched-int.h (autopref_multipass_data_): Remove
2699 multi_mem_insn_p, min_offset and max_offset.
2701 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2703 PR tree-optimization/82381
2704 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2705 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2708 PR tree-optimization/82374
2709 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2710 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2711 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2712 current_function_decl to the new decl.
2714 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2716 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2717 helper macro for IEEE float128 hardware built-in functions.
2718 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2720 (TRUNCF128_ODD): Likewise.
2721 (ADDF128_ODD): Likewise.
2722 (SUBF128_ODD): Likewise.
2723 (MULF128_ODD): Likewise.
2724 (DIVF128_ODD): Likewise.
2725 (FMAF128_ODD): Likewise.
2726 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2727 UNSPEC_TRUNC_ROUND_TO_ODD.
2728 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2729 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2730 floating point round to odd instructions.
2731 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2732 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2733 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2734 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2735 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2736 (trunc<mode>sf2_hw): Change the truncate with round to odd
2737 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2738 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2739 to odd hardware instructions.
2740 (sub<mode>3_odd): Likewise.
2741 (mul<mode>3_odd): Likewise.
2742 (div<mode>3_odd): Likewise.
2743 (sqrt<mode>2_odd): Likewise.
2744 (fma<mode>4_odd): Likewise.
2745 (fms<mode>4_odd): Likewise.
2746 (nfma<mode>4_odd): Likewise.
2747 (nfms<mode>4_odd): Likewise.
2748 (trunc<mode>df2_odd): Change the truncate with round to odd
2749 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2751 * doc/extend.texi (PowerPC built-in functions): Update documentation
2752 for existing IEEE float128-bit built-in functions. Add built-in
2753 functions that generate the IEEE 128-bit floating point round to
2756 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2758 PR rtl-optimization/77729
2759 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2760 to (X&(C1&~C2))|C2 transformations.
2762 2017-10-03 Martin Jambor <mjambor@suse.cz>
2764 PR tree-optimization/82363
2765 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2766 mismatch, mark lacc written regardless of racc.
2768 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2770 PR tree-optimization/82381
2771 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2772 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2773 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2774 fallthrough into reassoc_stmt_dominates_stmt_p.
2777 * combine.c (combine_instructions): Don't combine in unreachable
2780 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2783 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2784 function to not use the have_cpu variable. Do not set cpu_index,
2785 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2787 (rs6000_valid_attribute_p): Remove duplicate initializations of
2788 old_optimize and func_optimize.
2789 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2790 (rs6000_activate_target_options): Make global.
2791 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2794 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2796 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2797 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2798 Set *poffset to 0 and *psize and *pmax_size to -1 if
2799 *poffset + *psize overflows in HOST_WIDE_INT.
2801 PR tree-optimization/82387
2802 PR tree-optimization/82388
2803 PR tree-optimization/82389
2804 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2805 instead of live_bytes non-NULL.
2807 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2810 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2813 2017-10-02 Richard Biener <rguenther@suse.de>
2815 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2816 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2817 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2819 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2821 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2822 requested precision matches the type's.
2823 * calls.c (alloc_max_size): Calculate the new candidate size as
2824 a widest_int and use wi::to_widest when comparing it with the
2825 current candidate size.
2826 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2827 zero rather than integer_zero_node.
2828 * match.pd: Check for a no-op conversion before using wi::add
2829 rather than after. Use tree_to_uhwi when summing small shift
2830 counts into an unsigned int.
2832 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2833 Alan Hayward <alan.hayward@arm.com>
2834 David Sherwood <david.sherwood@arm.com>
2837 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2838 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2839 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2840 POINTER_AND_FP_REGS.
2842 2017-10-02 Richard Biener <rguenther@suse.de>
2844 PR tree-optimization/82355
2845 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2846 a mapping for the enclosing loop but avoid generating one for
2848 (copy_bb_and_scalar_dependences): Remove premature codegen
2849 error on PHIs in blocks duplicated into multiple places.
2850 * graphite-scop-detection.c
2851 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2852 in the region use it as loop and nest to analyze the DR in.
2853 (try_generate_gimple_bb): Likewise.
2854 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2855 (add_loop_constraints): For blocks in a loop not in the region
2856 create a dimension with a single iteration.
2857 * sese.h (gbb_loop_at_index): Remove assert.
2859 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2861 * omp-expand.c (adjust_context_scope): New function.
2862 (expand_parallel_call): Call adjust_context_scope.
2864 2017-10-01 Jeff Law <law@redhat.com>
2866 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2867 dom_opt_dom_walker class with direct access to private members.
2868 Add comments. Call test_for_singularity.
2869 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2870 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2871 m_dummy_cond anymore.
2872 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2874 (pass_dominator:execute): Build the dummy_cond here and pass it
2875 to the dom_opt_dom_walker ctor.
2876 (test_for_singularity): New function.
2878 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2879 Maya Rashish <coypu@sdf.org>
2881 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2882 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2883 (alpha*-*-netbsd*) Use nbsd_tm_file.
2884 (arm*-*-netbsdelf*) Likewise.
2885 (i[34567]86-*-netbsdelf*) Likewise.
2886 (x86_64-*-netbsd*) Likewise.
2887 (mips*-*-netbsd*) Likewise.
2888 (powerpc-*-netbsd*) Likewise.
2889 (sh*-*-netbsd*) Likewise.
2890 (sparc-*-netbsdelf*) Likewise.
2891 (sparc64-*-netbsd*) Likewise.
2892 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2894 (vax-*-netbsdelf*) Likewise.
2895 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2896 (UINT_FAST8_TYPE) Likewise.
2897 (INT_FAST16_TYPE) Check CHAR_FAST16.
2898 (UINT_FAST16_TYPE) Likewise.
2900 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2903 * config/i386/i386.md
2904 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2905 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2906 *divmodsi4_zext_2): New define_insn_and_split.
2907 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2908 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2909 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2910 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2911 New define_insn_and_split.
2912 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2913 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2914 operands[1] having DImode when mode is SImode.
2916 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2917 always SImode for DIV and MOD in REG_EQUAL notes.
2919 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2922 * match.pd: Fix handling of NaNs in pattern.
2924 2017-09-29 Jeff Law <law@redhat.com>
2926 * sbitmap.c (bitmap_bit_in_range_p): New function.
2927 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2928 * tree-ssa-dse.c (live_bytes_read): New function.
2929 (dse_classify_store): Ignore reads of dead bytes.
2931 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2932 typos and whitespace errors.
2933 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2934 * config/s390/s390.c (s390_emit_prologue): Likewise.
2936 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2939 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2940 with a symbol for LRA.
2942 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2944 PR rtl-optimization/82338
2945 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2947 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2949 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2950 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2951 (qsort_chk): Declare.
2952 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2953 (qsort_chk): New function.
2955 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2957 PR tree-optimization/82337
2958 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2959 phi definition if the PHI result appears in an abnormal PHI.
2960 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2961 basis appears in an abnormal PHI.
2963 2017-09-29 Richard Biener <rguenther@suse.de>
2965 * graphite-isl-ast-to-gimple.c
2966 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2967 (binary_op_to_tree): Use it.
2968 (get_rename_from_scev): Likewise.
2969 (copy_loop_phi_nodes): Likewise.
2970 (copy_bb_and_scalar_dependences): Likewise.
2971 (translate_pending_phi_nodes): Likewise.
2973 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2976 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2977 for movabsq $(i32 << shift), r64.
2979 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2981 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2982 index when encoding %esp as %rsp to avoid 0x67 prefix.
2984 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2986 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2987 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2989 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2991 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2992 Extensions with more than 16 double VFP registers.
2993 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2994 to_clear_mask and all code related to it. Replace the remaining
2995 entry by a sbitmap and adapt code accordingly.
2997 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2999 * brig-builtins.def: Change pure attributes to const.
3001 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3003 * config.gcc (default_gnu_indirect_function): Default to yes for
3004 sparc*-*-linux* with glibc.
3006 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3008 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3009 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3010 when creating .init_array and .fini_array sections with priority
3013 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3016 * config/aarch64/aarch64.c
3017 (aarch64_builtin_support_vector_misalignment): Always return false
3018 when misalignment is unknown.
3020 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3022 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3023 this function to return false if the definition used by the swap
3024 instruction is artificial, or if the memory address from which the
3025 constant value is loaded is not represented by a base address held
3026 in a register or if the base address register is a frame or stack
3027 pointer. Additionally, return false if the base address of the
3028 loaded constant is a SYMBOL_REF but is not considered to be a
3030 (replace_swapped_load_constant): New function.
3031 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3032 loaded constant vector with a load of a swapped constant vector.
3034 2017-09-27 Carl Love <cel@us.ibm.com>
3036 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3037 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3038 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3041 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3043 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3044 first, always call autopref_rank_data otherwise.
3046 2017-09-27 Richard Biener <rguenther@suse.de>
3048 * graphite-scop-detection.c (find_scop_parameters): Move
3049 loop bound handling ...
3050 (gather_bbs::before_dom_children): ... here, avoiding the need
3051 to build scop_info->loop_nest.
3052 (record_loop_in_sese): Remove.
3053 * sese.h (sese_info_t::loop_nest): Remove.
3054 * sese.c (new_sese_info): Do not allocate loop_nest.
3055 (free_sese_info): Do not free loop_nest.
3057 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3060 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3061 lhs from calls if the lhs has addressable type.
3063 2017-09-27 Richard Biener <rguenther@suse.de>
3065 * graphite.h (scop::max_alias_set): New member.
3066 * graphite-scop-detection.c: Remove references to non-existing
3067 --param in comments.
3068 (build_alias_sets): Record the maximum alias set used for drs.
3069 (build_scops): Support zero as unlimited for
3070 --param graphite-max-arrays-per-scop.
3071 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3073 (build_poly_sr_1): ... here. Compute alias set based on the
3074 maximum alias set used for drs rather than
3075 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3077 2017-09-27 Richard Biener <rguenther@suse.de>
3079 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3080 --param loop-block-tile-size=0 to disable tiling.
3082 2017-09-27 Richard Biener <rguenther@suse.de>
3084 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3085 (graphite-max-nb-scop-params): Document special value zero.
3086 * domwalk.h (dom_walker::STOP): New symbolical constant.
3087 (dom_walker::dom_walker): Add optional parameter for bb to
3089 (dom_walker::~dom_walker): Declare.
3090 (dom_walker::before_dom_children): Document STOP return value.
3091 (dom_walker::m_user_bb_to_rpo): New member.
3092 (dom_walker::m_bb_to_rpo): Likewise.
3093 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3094 mapping here if not provided by the user.
3095 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3096 provided by the user.
3097 (dom_walker::STOP): Define.
3098 (dom_walker::walk): Do not compute bb to RPO mapping here.
3099 Support STOP return value from before_dom_children to stop
3101 * graphite-optimize-isl.c (optimize_isl): If the schedule
3102 is the same still generate code if -fgraphite-identity
3103 or -floop-parallelize-all are given.
3104 * graphite-scop-detection.c: Include cfganal.h.
3105 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3107 (gather_bbs::before_dom_children): Return STOP for BBs
3109 (build_scops): Compute bb to RPO mapping and pass it to
3110 the domwalk. Treat --param graphite-max-nb-scop-params=0
3111 as not limiting the number of params.
3112 * graphite.c (graphite_initialize): Remove limit on the
3113 number of basic-blocks in a function.
3114 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3115 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3116 default value of 10.
3118 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3120 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3121 Adjust code to eliminate needing to do the shift right 32-bits
3122 operation after XSCVDPSPN.
3124 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3126 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3127 ((X / Y) != 0 -> X >= Y): Likewise.
3129 2017-09-26 Carl Love <cel@us.ibm.com>
3131 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3132 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3133 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3134 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3135 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3136 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3137 definitions and overloading.
3138 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3139 statement for P9V_BUILTIN_XST_LEN_R.
3140 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3141 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3142 define_expand and define_insn for the instructions and builtins.
3143 * doc/extend.texi: Update the built-in documentation file for the new
3145 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3146 define_insn for the instructions
3148 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3151 * gcc/config/netbsd-protos.h: New file.
3152 * gcc/config/netbsd.c: New file.
3153 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3154 * gcc/config/t-netbsd: New file.
3155 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3156 (tmake_file) Add t-netbsd.
3157 (extra_objs) Add netbsd.o.
3159 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3163 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3165 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3167 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3168 sign extension from a vector register to a GPR by doing a 32-bit
3169 direct move and then an EXTSW.
3170 (extendsi<mode>2 splitter): Likewise.
3171 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3172 right or vector extract after doing XSCVDPSPN. Use
3173 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3175 (movdi_from_sf_zero_ext): Likewise.
3176 (reload_gpr_from_vsxsf): Likewise.
3177 (p8_mfvsrd_4_disf): Delete, no longer used.
3178 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3179 then needing to move the SFmode to a GPR to use the XSCVDPSP
3180 instruction instead of FRSP and XSCVDPSPN.
3181 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3182 it is adjacent to the other XSCVSPDP insns.
3183 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3184 SFmode to be in traditional Altivec registers.
3185 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3186 (vsx_xscvspdpn): Likewise.
3187 (vsx_xscvspdpn_scalar): Likewise.
3189 2017-09-26 Martin Jambor <mjambor@suse.cz>
3191 * tree-sra.c (compare_access_positions): Put integral types first,
3192 stabilize sorting of integral types, remove conditions putting
3193 non-full-precision integers last.
3194 (sort_and_splice_var_accesses): Disable scalarization if a
3195 non-integert would be represented by a non-full-precision integer.
3197 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3199 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3200 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3201 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3202 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3203 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3204 conditionals inside the function instead of around it. Call
3205 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3206 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3208 2017-09-26 Richard Biener <rguenther@suse.de>
3210 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3212 (scop_detection::build_scop_breadth): ... this. Removed.
3213 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3214 (scop_detection::harmful_stmt_in_bb): Likewise.
3215 (scop_detection::graphite_can_represent_stmt): Likewise.
3216 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3217 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3218 (scop_detection::can_represent_loop_1): ... this. Removed.
3219 (scop_detection::harmful_loop_in_region): Simplify after inlining
3220 the above and remove more quadraticness.
3221 (build_scops): Adjust.
3222 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3225 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3228 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3229 REGNO (base) == SP_REG if base is a REG.
3232 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3233 if it is different SSA_NAME.
3234 (optimize_range_tests_cmp_bitwise): New function.
3235 (optimize_range_tests): Call it.
3237 2017-09-26 Richard Biener <rguenther@suse.de>
3239 PR tree-optimization/82321
3240 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3241 for the def being inside the loop.
3243 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3245 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3247 * config/s390/s390-builtins.def: Fix constraint on op4.
3249 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3251 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3252 independent expanders.
3253 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3254 ("vec_ordered", "vec_unordered"): New expanders.
3256 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3258 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3261 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3263 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3264 vec_unpacks_lo_v16qi.
3265 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3267 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3269 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3270 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3271 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3273 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3275 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3277 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3279 ("*vec_slb<mode>"): New insn pattern.
3280 ("vec_shr_<mode>"): New expander.
3281 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3282 and force the shift count operand to V16QImode.
3283 ("vec_srb<mode>"): Set shift count mode to V16QI.
3285 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3287 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3288 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3289 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3291 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3294 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3296 2017-09-26 Richard Biener <rguenther@suse.de>
3298 PR tree-optimization/82320
3299 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3302 2017-09-25 Jeff Law <law@redhat.com>
3304 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3305 prototype for new argument.
3306 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3307 mostly extracted from rs6000_emit_allocate_stack.
3308 (rs6000_emit_probe_stack_range_stack_clash): New function.
3309 (rs6000_emit_allocate_stack): Call
3310 rs6000_emit_probe_stack_range_stack_clash as needed.
3311 (rs6000_emit_probe_stack_range): Add additional argument
3312 to call to gen_probe_stack_range{si,di}.
3313 (output_probe_stack_range): New.
3314 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3315 (output_probe_stack_range_stack_clash): New.
3316 (rs6000_emit_prologue): Emit notes into dump file as requested.
3317 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3318 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3319 Add additional operand and pass it to output_probe_stack_range.
3321 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3323 PR tree-optimization/82163
3324 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3325 (checking_verify_loop_closed_ssa): New parameter.
3326 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3327 (check_loop_closed_ssa_stmt): Delete.
3328 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3329 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3330 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3333 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3335 * brig-builtins.def: Treat HSAIL barrier builtins as
3336 setjmp/longjump style functions.
3338 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3340 * target.def (constant_alignment): New hook.
3341 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3342 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3343 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3344 * doc/tm.texi: Regenerate.
3345 * targhooks.h (default_constant_alignment): Declare.
3346 (constant_alignment_word_strings): Likewise.
3347 * targhooks.c (default_constant_alignment): New function.
3348 (constant_alignment_word_strings): Likewise.
3349 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3350 instead of CONSTANT_ALIGNMENT.
3351 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3352 (force_const_mem): Likewise.
3353 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3354 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3355 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3356 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3357 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3359 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3360 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3361 constant_alignment_word_strings.
3362 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3363 (CONSTANT_ALIGNMENT): Likewise.
3364 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3365 (arm_constant_alignment): New function.
3366 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3367 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3368 constant_alignment_word_strings.
3369 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3370 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3371 constant_alignment_word_strings.
3372 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3373 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3374 (cris_constant_alignment): New function.
3375 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3376 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3377 (epiphany_constant_alignment): New function.
3378 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3379 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3380 constant_alignment_word_strings.
3381 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3382 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3383 constant_alignment_word_strings.
3384 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3385 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3386 constant_alignment_word_strings.
3387 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3388 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3389 * config/i386/i386.c (ix86_constant_alignment): Make static.
3390 Use the same interface as the target hook.
3391 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3392 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3393 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3394 constant_alignment_word_strings.
3395 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3396 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3397 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3398 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3399 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3400 constant_alignment_word_strings.
3401 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3402 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3403 constant_alignment_word_strings.
3404 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3405 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3406 constant_alignment_word_strings.
3407 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3408 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3410 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3411 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3412 * config/mips/mips.c (mips_constant_alignment): New function.
3413 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3414 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3415 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3416 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3417 (mmix_constant_alignment): Make static. Use the same interface
3419 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3420 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3421 constant_alignment_word_strings.
3422 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3423 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3424 constant_alignment_word_strings.
3425 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3426 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3427 constant_alignment_word_strings.
3428 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3429 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3430 (rs6000_constant_alignment): New function.
3431 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3432 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3433 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3434 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3435 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3436 (rs6000_constant_alignment): New function.
3437 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3438 * config/s390/s390.c (s390_constant_alignment): New function.
3439 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3440 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3441 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3442 constant_alignment_word_strings.
3443 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3444 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3445 (sparc_constant_alignment): New function.
3446 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3447 * config/spu/spu.c (spu_constant_alignment): New function.
3448 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3449 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3450 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3451 constant_alignment_word_strings.
3452 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3453 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3454 constant_alignment_word_strings.
3455 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3456 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3457 constant_alignment_word_strings.
3458 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3459 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3460 (visium_constant_alignment): New function.
3461 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3462 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3463 (xtensa_constant_alignment): New function.
3464 * system.h (CONSTANT_ALIGNMENT): Poison.
3466 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3469 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3470 (rs6000_builtin_valid_without_lhs): New helper function.
3471 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3472 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3474 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3476 * target.h (vec_perm_indices): Use unsigned short rather than
3478 (auto_vec_perm_indices): Likewise.
3479 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3480 Use unsigned int rather than unsigned char.
3481 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3483 2017-09-25 Richard Biener <rguenther@suse.de>
3485 * cfgloop.h (sort_sibling_loops): Declare.
3486 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3487 (sort_sibling_loops): New function sorting the sibling loop list
3489 * graphite.c (graphite_transform_loops): Sort sibling loops.
3491 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3493 * target.def (vec_perm_const_ok): Change sel parameter to
3495 * optabs-query.c (can_vec_perm_p): Update accordingly.
3496 * doc/tm.texi: Regenerate.
3497 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3498 auto_vec_perm_indices and remove separate nelt field.
3499 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3500 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3501 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3502 (aarch64_expand_vec_perm_const): Update accordingly.
3503 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3504 to vec_perm_indices.
3505 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3506 auto_vec_perm_indices and remove separate nelt field.
3507 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3508 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3509 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3511 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3512 to vec_perm_indices.
3513 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3514 sel to vec_perm_indices.
3515 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3516 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3517 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3519 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3522 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3525 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3526 on the FUNCTION_DECL function context if it has a DIE that is a
3529 2017-09-25 Richard Biener <rguenther@suse.de>
3531 PR tree-optimization/82285
3532 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3535 2017-09-25 Tom de Vries <tom@codesourcery.com>
3539 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3542 2017-09-25 Richard Biener <rguenther@suse.de>
3544 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3545 ISL errors other than isl_error_quota happen. Dump if the
3546 schedule is the same.
3547 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3548 errors instead of aborting inside ISL.
3550 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3553 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3554 of libgcc_eh for m64.
3555 * config/i386/darwin64.h: Likewise.
3557 2017-09-25 Richard Biener <rguenther@suse.de>
3560 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3561 attribute for incomplete types nor twice for complete ones.
3563 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3566 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3567 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3570 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3573 * config/i386/i386.opt (mprefer-avx256): Use
3574 ix86_target_flags variable.
3575 * config/i386/i386.c (ix86_target_string): Move
3576 -mprefer-avx256 to flag2_opts.
3578 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3581 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3582 and x != -1 | y != -1 into (x & y) != -1.
3584 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3586 * config.gcc: Add new case statement to set
3587 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3588 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3589 s390x-*-linux* case statements. Added aarch64 to the list of
3590 supported architectures.
3592 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3594 PR tree-optimization/82289
3595 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3596 STMT_VINFO_RELEVANT_P.
3598 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3599 Alan Hayward <alan.hayward@arm.com>
3600 David Sherwood <david.sherwood@arm.com>
3602 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3603 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3604 (extract_range_from_binary_expr_1): Don't call
3605 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3607 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3608 Alan Hayward <alan.hayward@arm.com>
3609 David Sherwood <david.sherwood@arm.com>
3611 * target.def (preferred_vector_alignment): New hook.
3612 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3614 * doc/tm.texi: Regenerate.
3615 * targhooks.h (default_preferred_vector_alignment): Declare.
3616 * targhooks.c (default_preferred_vector_alignment): New function.
3617 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3619 (DR_TARGET_ALIGNMENT): New macro.
3620 (aligned_access_p): Update commentary.
3621 (vect_known_alignment_in_bytes): New function.
3622 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3624 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3625 Calculate the misalignment based on the target alignment rather than
3627 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3628 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3629 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3630 the target alignment, rather than masking the element misalignment
3631 with the number of elements in a vector. Also use the target
3632 alignment when calculating the maximum number of peels.
3633 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3634 instead of TYPE_ALIGN_UNIT.
3635 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3636 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3637 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3638 (vect_create_data_ref_ptr): Likewise.
3639 (vect_setup_realignment): Realign by ANDing with
3640 -DR_TARGET_MISALIGNMENT.
3641 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3642 the number of peels based on DR_TARGET_ALIGNMENT.
3643 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3644 with the guaranteed alignment boundary when deciding whether
3646 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3647 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3648 (ensure_base_align): Remove stmt_info parameter. Get the
3649 target base alignment from DR_TARGET_ALIGNMENT.
3650 (vectorizable_store): Update call accordingly. Interpret
3651 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3653 (vectorizable_load): Likewise.
3655 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3656 Alan Hayward <alan.hayward@arm.com>
3657 David Sherwood <david.sherwood@arm.com>
3659 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3660 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3661 (vect_enhance_data_refs_alignment): Likewise.
3663 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3665 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3666 error. Only quit immediately if parsing is complete.
3667 (BEGIN): Initialize fatal_err and parse_done.
3668 (begin fpu, end fpu): Check number of arguments.
3669 (begin arch, end arch): Likewise.
3670 (begin cpu, end cpu): Likewise.
3671 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3672 (optalias): Likewise.
3674 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3676 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3677 * config/arm/arm-isa.h: Delete. Move definitions to ...
3678 * arm-cpus.in: ... here. Use new feature and fgroup values.
3679 * config/arm/arm.c (arm_option_override): Use lower case for feature
3681 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3682 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3683 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3685 (print_isa_bits_for): New function.
3686 (gen_isa): New function.
3687 (gen_comm_data): Use print_isa_bits_for.
3688 (define feature): New keyword.
3689 (define fgroup): New keyword.
3690 * config/arm/t-arm (TM_H): Remove.
3691 (GTM_H): Add arm-isa.h.
3692 (arm-isa.h): Add rule to generate file.
3693 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3694 case for feature bit names.
3696 2017-09-22 Richard Biener <rguenther@suse.de>
3698 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3700 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3701 print of no dependency loops ...
3702 * graphite.c (graphite_transform_loops): ... here.
3703 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3705 (same_close_phi_node, remove_duplicate_close_phi,
3706 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3707 (canonicalize_loop_closed_ssa): ... here and simplify.
3708 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3709 (optimize_isl): Use dump_printf_loc to tell when we stopped
3710 optimizing because of an ISL timeout.
3712 2017-09-22 Richard Biener <rguenther@suse.de>
3714 PR tree-optimization/82291
3715 * tree-if-conv.c (predicate_mem_writes): Make sure to
3716 remove writes in blocks predicated with false.
3718 2017-09-22 Richard Biener <rguenther@suse.de>
3720 * sese.c: Include cfganal.h.
3721 (if_region_set_false_region): Remove.
3722 (create_if_region_on_edge): Likewise.
3723 (move_sese_in_condition): Re-implement without destroying
3726 2017-09-22 Richard Biener <rguenther@suse.de>
3728 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3729 Verify both BBs contain loop PHI nodes before dispatching to
3731 (graphite_regenerate_ast_isl): Do not recompute dominators,
3732 do not verify three times. Restructure for clarity.
3733 * graphite-scop-detection.c (same_close_phi_node,
3734 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3735 defined_in_loop_p, canonicalize_loop_closed_ssa,
3736 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3737 checking and SSA rewrite, move to ...
3738 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3739 (graphite_initialize): Do not pass in ctx, do not reset the
3740 SCEV cache, compute only dominators.
3741 (graphite_transform_loops): Allocate ISL ctx after
3742 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3743 Maintain post-dominators only around build_scops.
3744 * sese.c (if_region_set_false_region): Make static. Free
3745 and recompute dominators.
3746 (move_sese_in_condition): Assert we don't get called with
3747 post-dominators computed.
3748 * sese.h (if_region_set_false_region): Remove.
3750 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3752 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3753 mode attribute for TARGET_AVX512VL.
3755 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3757 * config/i386/i386.opt (mprefer-avx256): New option.
3758 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3760 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3761 for TARGET_PREFER_AVX256.
3762 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3764 2017-09-21 Jeff Law <law@redhat.com>
3766 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3767 Fix dump output if the only stack space is for pushed registers.
3769 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3771 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3774 2017-09-21 Martin Sebor <msebor@redhat.com>
3777 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3778 code (in C++) or code that triggers warnings.
3780 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3782 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3784 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3786 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3787 * sched-rgn.c: Ditto.
3788 * sel-sched-ir.c: Ditto.
3790 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3792 * toplev.h (set_random_seed): Adjust return type.
3793 * toplev.c (init_local_tick): Move eager initialization of random_seed
3794 to get_random_seed. Adjust comment.
3795 (init_random_seed): Inline to get_random_seed, delete.
3796 (get_random_seed): Initialize random_seed lazily.
3797 (set_random_seed): Do not return previous value.
3798 (print_switch_value): Do not call get_random_seed.
3800 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3802 * cgraph.c (delete_function_version): New, broken out from...
3803 (cgraph_node::delete_function_version): ...here. Rename to
3804 cgraph_node::delete_function_version_by_decl. Update all uses.
3805 (cgraph_node::remove): Call delete_function_version.
3807 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3810 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3811 VAR_DECLs to which addressable non-volatile parameters are mapped
3812 and for id->retvar after the return value assignment. Clear
3813 id->retval and id->retbnd after inlining.
3815 2017-09-21 Richard Biener <rguenther@suse.de>
3817 PR tree-optimization/82276
3818 PR tree-optimization/82244
3819 * tree-vrp.c (build_assert_expr_for): Set
3820 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3822 (remove_range_assertions): Revert earlier change.
3824 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3827 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3829 2017-09-21 Richard Biener <rguenther@suse.de>
3831 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3832 Restore valid IL after code generation errors.
3833 * graphite.c (graphite_transform_loops): Diagnose code
3834 generation issues as MSG_MISSED_OPTIMIZATION and continue
3835 with processing SCOPs.
3837 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3838 Alan Hayward <alan.hayward@arm.com>
3839 David Sherwood <david.sherwood@arm.com>
3841 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3842 rather than choosing between plus_constant and gen_rtx_<CODE>.
3843 * expr.c (emit_push_insn): Likewise.
3844 (expand_expr_real_2): Likewise.
3846 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3847 Alan Hayward <alan.hayward@arm.com>
3848 David Sherwood <david.sherwood@arm.com>
3850 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3852 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3853 Alan Hayward <alan.hayward@arm.com>
3854 David Sherwood <david.sherwood@arm.com>
3856 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3857 calling tree_to_uhwi.
3859 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3860 Alan Hayward <alan.hayward@arm.com>
3861 David Sherwood <david.sherwood@arm.com>
3863 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3864 INTEGER_CST rather than a negative test for ADDR_EXPR.
3866 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3867 Alan Hayward <alan.hayward@arm.com>
3868 David Sherwood <david.sherwood@arm.com>
3870 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3871 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3873 2017-09-21 Richard Biener <rguenther@suse.de>
3875 PR tree-optimization/71351
3876 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3877 graphite_create_new_loop_guard): Remove, fold remaining parts
3879 (translate_isl_ast_node_for): ... here and simplify.
3881 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3884 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3885 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3886 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3887 alternative always use QI mode, for -Os imov (=R,R) alternative
3888 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3891 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3892 Jeff Law <law@redhat.com>
3894 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3895 (allocate_stack_space): New function, partially extracted from
3897 (s390_emit_prologue): Track offset to most recent stack probe.
3898 Code to allocate space moved into allocate_stack_space.
3899 Dump actions when no stack is allocated.
3900 (s390_prologue_plus_offset): New function.
3901 (s390_emit_stack_probe): Likewise.
3903 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3905 * common.opt (Wa, Wl, Wp, g, gz=): Add
3907 (gno-column-info): Remove.
3908 (gcolumn-info): Drop RejectNegative.
3910 (gno-record-gcc-switches): Remove.
3911 (grecord-gcc-switches): Drop RejectNegative.
3912 (gno-split-dwarf): Remove.
3913 (gsplit-dwarf): Drop RejectNegative.
3914 (gno-strict-dwarf): Remove.
3915 (gstrict-dwarf): Drop RejectNegative.
3916 * config/darwin.opt (gfull, gused): Add RejectNegative.
3917 * dwarf2out.c (gen_producer_string): Drop
3918 gno-record-gcc-switches handler.
3919 * optc-gen.awk: Add g to prefixes with negative forms.
3920 * opts-common.c (remapping_prefix_p): New.
3921 (find_opt): Check it.
3922 (generate_canonical_option): Test g prefix.
3923 (option_map): Add -gno- mapping.
3924 (add_misspelling_candidates): Check remapping_prefix_p.
3926 2017-09-20 Jeff Law <law@redhat.com>
3928 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3929 thinko in stack clash protection support.
3931 * explow.c (compute_stack_clash_protection_loop_data): Use
3932 CONST_INT_P instead of explicit test. Verify object is a
3933 CONST_INT_P before looking at INTVAL.
3934 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3935 instead of explicit test.
3937 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3940 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3941 address instead of to r1 and r11.
3943 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3945 * config.gcc: Support "knm".
3946 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3947 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3949 * config/i386/i386.c (m_KNM): Define.
3950 (processor_target_table): Add "knm".
3952 (ix86_option_override_internal): Add "knm".
3953 (ix86_issue_rate): Add PROCESSOR_KNM.
3954 (ix86_adjust_cost): Ditto.
3955 (ia32_multipass_dfa_lookahead): Ditto.
3956 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3957 (fold_builtin_cpu): Add M_INTEL_KNM.
3958 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3959 (processor_type): Add PROCESSOR_KNM.
3960 * config/i386/x86-tune.def: Add m_KNM.
3961 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3963 2017-09-20 Richard Biener <rguenther@suse.de>
3965 PR tree-optimization/80213
3966 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3967 are allowed in empty BBs as well.
3968 (canonicalize_loop_closed_ssa): Also look for other complex
3970 (scop_detection::get_sese): Include the loop-closed PHI block
3972 (scop_detection::merge_sese): Remove code adding extra blocks.
3973 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3974 (build_scops): Assert the final returned scop is invalid.
3976 2017-09-20 Richard Biener <rguenther@suse.de>
3978 PR tree-optimization/82264
3979 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3981 (vn_phi_lookup): Likewise.
3982 (vn_phi_insert): Likewise.
3984 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3986 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3987 that fits into uhwi or shwi, add DW_AT_const_value regardless
3988 of early_dwarf without going through RTL, using add_AT_unsigned
3991 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3992 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3993 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3994 (DEBUG_MACRO_SECTION): Likewise.
3995 (DEBUG_LTO_MACRO_SECTION): Likewise.
3996 (DEBUG_STR_DWO_SECTION): Likewise.
3997 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3998 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3999 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4000 (DEBUG_LTO_LINE_STR_SECTION): Define.
4001 (init_sections_and_labels): Initialize debug_line_str_section
4002 variable. Initialize debug_loc_section for -gdwarf-5 to
4003 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4005 2017-09-20 Richard Biener <rguenther@suse.de>
4007 * graphite-sese-to-poly.c (extract_affine): Properly handle
4008 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4010 2017-09-20 Richard Biener <rguenther@suse.de>
4012 PR tree-optimization/81373
4013 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4014 Force SESE live-out defs to be handled even if they are
4017 2017-09-19 Jeff Law <law@redhat.com>
4019 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4020 nothing for stack adjustments with REG_STACK_CHECK.
4021 * sched-deps.c (parse_add_or_inc): Reject insns with
4022 REG_STACK_CHECK from dependency breaking.
4023 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4024 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4025 * reg-notes.def (STACK_CHECK): New note.
4027 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4028 (ix86_expand_prologue): Dump stack clash info as needed.
4029 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4031 * function.c (dump_stack_clash_frame_info): New function.
4032 * function.h (dump_stack_clash_frame_info): Prototype.
4033 (enum stack_clash_probes): New enum.
4035 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4036 flag_stack_clash_protection.
4037 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4038 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4039 (arm_frame_pointer_required): Likewise.
4040 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4041 (ia64_expand_prologue): Likewise.
4042 * config/mips/mips.c (mips_expand_prologue): Likewise.
4043 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4044 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4045 (sparc_flat_expand_prologue): Likewise.
4046 * config/spu/spu.c (spu_expand_prologue): Likewise.
4048 * explow.c: Include "params.h".
4049 (anti_adjust_stack_and_probe_stack_clash): New function.
4050 (get_stack_check_protect): Likewise.
4051 (compute_stack_clash_protection_loop_data): Likewise.
4052 (emit_stack_clash_protection_loop_start): Likewise.
4053 (emit_stack_clash_protection_loop_end): Likewise.
4054 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4055 Use anti_adjust_stack_and_probe_stack_clash.
4056 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4057 (emit_stack_clash_protection_loop_start): Likewise.
4058 (emit_stack_clash_protection_loop_end): Likewise.
4059 * rtl.h (get_stack_check_protect): Prototype.
4060 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4061 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4062 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4064 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4066 * doc/tm.texi: Rebuilt.
4067 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4068 get_stack_check_protect.
4069 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4070 * config/arm/arm.c (arm_expand_prologue): Likewise.
4071 (arm_frame_pointer_required): Likewise.
4072 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4073 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4074 * config/mips/mips.c (mips_expand_prologue): Likewise.
4075 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4076 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4077 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4078 (sparc_flat_expand_prologue): Likewise.
4080 * common.opt (-fstack-clash-protection): New option.
4081 * flag-types.h (enum stack_check_type): Note difference between
4082 -fstack-check= and -fstack-clash-protection.
4083 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4084 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4085 * toplev.c (process_options): Issue warnings/errors for cases
4086 not handled with -fstack-clash-protection.
4087 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4088 (-fstack-check): Note additional problem with -fstack-check=generic.
4089 Note that -fstack-check is primarily for Ada and refer users
4090 to -fstack-clash-protection for stack-clash-protection.
4091 Document new params for stack clash protection.
4093 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4095 * config/i386/i386.c (ix86_split_long_move): Do not handle
4096 address used for LEA in a special way.
4098 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4100 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4102 2017-09-19 Martin Sebor <msebor@redhat.com>
4105 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4106 of incompatible types.
4108 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4110 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4111 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4113 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4115 2017-09-19 Richard Biener <rguenther@suse.de>
4117 PR tree-optimization/82244
4118 * tree-vrp.c (remove_range_assertions): Do not propagate
4119 a constant to abnormals but replace the assert with a copy.
4121 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4123 PR rtl-optimization/57878
4124 PR rtl-optimization/68988
4125 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4126 avoidance test involving non_reload_pseudos. Move frequency test
4127 below the general fragmentation avoidance test.
4129 2017-09-19 Richard Biener <rguenther@suse.de>
4131 PR tree-optimization/69728
4132 * graphite-sese-to-poly.c (schedule_error): New global.
4133 (add_loop_schedule): Handle empty domain by failing the
4135 (build_original_schedule): Handle schedule_error.
4137 2017-09-19 Richard Biener <rguenther@suse.de>
4139 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4140 Do not iterate to sibling loops but only to siblings of inner
4143 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4146 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4147 that the registers are different.
4149 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4151 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4152 to processor_model and "amdfam17h" to arch_names_table.
4153 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4155 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4158 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4160 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4161 Alan Hayward <alan.hayward@arm.com>
4162 David Sherwood <david.sherwood@arm.com>
4164 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4166 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4168 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4169 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4170 vect_schedule_slp_instance.
4171 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4172 Update call to vect_slp_analyze_node_operations. Simplify return
4174 (vect_slp_analyze_bb_1): Update call accordingly.
4175 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4176 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4177 (vect_schedule_slp): Update call accordingly.
4179 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4180 Alan Hayward <alan.hayward@arm.com>
4181 David Sherwood <david.sherwood@arm.com>
4183 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4184 with types that aren't in fact scalar.
4186 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4188 * tree-vect-slp.c (vect_record_max_nunits): New function,
4190 (vect_build_slp_tree_1): ...here.
4191 (vect_build_slp_tree_2): Call it for phis too.
4193 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4195 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4196 to vect_get_vec_def_for_operand when getting the mask operand.
4198 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4199 Alan Hayward <alan.hayward@arm.com>
4200 David Sherwood <david.sherwood@arm.com>
4202 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4205 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4206 Alan Hayward <alan.hayward@arm.com>
4207 David Sherwood <david.sherwood@arm.com>
4209 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4210 calculation for vector booleans.
4212 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4213 Alan Hayward <alan.hayward@arm.com>
4214 David Sherwood <david.sherwood@arm.com>
4216 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4218 (vect_transform_stmt): ...here.
4219 (vect_analyze_stmt): Use it instead of calling
4220 vectorizable_live_operation directly.
4222 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4224 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4225 non-SIMT targets in acc vector loops.
4227 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4229 * configure.ac: Add arc and check if assembler supports gdwarf2.
4230 * configure: Regenerate.
4232 2017-09-18 Richard Biener <rguenther@suse.de>
4234 PR tree-optimization/82220
4235 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4236 epilogue niters from the min_profitable_iters compute.
4238 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4241 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4242 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4243 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4245 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4248 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4249 switching to a new text section.
4251 2017-09-18 Richard Biener <rguenther@suse.de>
4253 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4255 (build_alias_set): Reject aliases with no access function.
4257 2017-09-18 Richard Biener <rguenther@suse.de>
4259 PR tree-optimization/79622
4260 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4262 (build_cross_bb_scalars_use): Likewise.
4264 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4266 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4268 2017-09-18 Alan Modra <amodra@gmail.com>
4271 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4272 stack_pointer_rtx for count 0. Update comments. Break up
4273 large rtl expression.
4275 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4277 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4278 Increase to 20 bytes.
4279 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4280 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4281 or avx version of the stub.
4283 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4286 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4287 compute the minimum stack alignment. Also update preferred stack
4288 boundary for leaf functions.
4290 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4292 PR tree-optimization/82228
4293 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4296 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4298 * common/config/nds32/nds32-common.c
4299 (nds32_option_optimization_table): Refine formatting.
4300 (nds32_option_optimization_table): Use -fsched-pressure and
4301 -fomit-frame-pointer for specific optimization level.
4303 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4305 * config/nds32/nds32.c: Refine formatting and comments.
4306 * config/nds32/nds32.h: Likewise.
4307 * config/nds32/nds32.md: Likewise.
4308 * config/nds32/nds32-cost.c: Likewise.
4309 * config/nds32/nds32-isr.c: Likewise.
4310 * config/nds32/nds32-md-auxiliary.c: Likewise.
4311 * config/nds32/nds32-multiple.md: Likewise.
4312 * config/nds32/nds32-predicates.c: Likewise.
4314 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4315 Jakub Jelinek <jakub@redhat.com>
4317 Add support for -std=c++2a.
4318 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4320 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4322 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4325 * doc/extend.texi (Common Function Attributes): Add
4326 references to ARM, AArch64, and S/390 specific attributes.
4327 (Function Specific Option Pragmas): Add AArch64 and S/390
4328 to list of back ends that support the target pragma.
4330 2017-09-15 Nathan Sidwell <nathan@acm.org>
4332 * doc/standards.texi: Fix C++17 description. Update URLs for
4335 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4337 * common.opt (Wcast-align=strict): New warning option.
4338 * doc/invoke.texi: Document -Wcast-align=strict.
4340 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4342 * cgraph.h (cgraph_thunk_info): Add comments.
4343 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4344 assert for VIRTUAL_* arguments stricter.
4346 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4348 PR tree-optimization/71026
4349 * match.pd: Move RDIV patterns from fold-const.c
4350 * fold-const.c (distribute_real_division): Removed.
4351 (fold_binary_loc): Remove calls to distribute_real_divison.
4353 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4355 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4356 c++1z and gnu++1z as deprecated. Change other references to
4357 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4358 Change -Wc++1z-compat to -Wc++17-compat.
4359 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4360 * dwarf2out.c (highest_c_language): Handle C++17.
4361 (gen_compile_unit_die): Likewise.
4363 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4365 PR rtl-optimization/82192
4366 * combine.c (make_extraction): Don't look through non-paradoxical
4367 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4370 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4371 Alan Hayward <alan.hayward@arm.com>
4372 David Sherwood <david.sherwood@arm.com>
4374 * target.def (function_arg_offset): New hook.
4375 * targhooks.h (default_function_arg_offset): Declare.
4376 * targhooks.c (default_function_arg_offset): New function.
4377 * function.c (locate_and_pad_parm): Use
4378 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4379 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4380 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4381 * doc/tm.texi: Regenerate.
4382 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4383 * config/spu/spu.c (spu_function_arg_offset): New function.
4384 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4385 * system.h (FUNCTION_ARG_OFFSET): Poison.
4387 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4388 Alan Hayard <alan.hayward@arm.com>
4389 David Sherwood <david.sherwood@arm.com>
4391 * target.def (truly_noop_truncation): New hook.
4392 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4393 than TRULY_NOOP_TRUNCATION.
4394 * hooks.h (hook_bool_uint_uint_true): Declare.
4395 * hooks.c (hook_bool_uint_uint_true): New function.
4396 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4397 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4398 * doc/tm.texi: Regenerate.
4399 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4400 rather than TRULY_NOOP_TRUNCATION in comments.
4401 (simplify_comparison): Likewise.
4402 (record_truncated_value): Likewise.
4403 * expmed.c (extract_bit_field_1): Likewise.
4404 (extract_split_bit_field): Likewise.
4405 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4406 instead of TRULY_NOOP_TRUNCATION.
4407 * function.c (assign_parm_setup_block): Likewise.
4408 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4409 * rtlhooks.c: Include target.h.
4410 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4411 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4412 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4413 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4414 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4415 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4416 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4417 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4418 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4419 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4420 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4421 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4422 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4423 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4424 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4425 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4426 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4427 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4428 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4429 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4430 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4431 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4432 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4433 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4434 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4435 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4436 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4437 rather than TRULY_NOOP_TRUNCATION in comments.
4438 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4439 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4440 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4441 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4442 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4443 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4444 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4445 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4446 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4447 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4448 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4449 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4450 rather than TRULY_NOOP_TRUNCATION in comments.
4451 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4452 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4453 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4454 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4455 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4456 TRULY_NOOP_TRUNCATION condition.
4457 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4458 (TRULY_NOOP_TRUNCATION): Delete.
4459 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4460 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4461 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4462 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4463 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4464 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4465 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4466 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4467 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4468 rather than TRULY_NOOP_TRUNCATION in comments.
4469 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4470 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4471 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4472 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4473 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4474 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4476 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4479 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4480 (*cmp_ior): Likewise.
4481 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4482 (*ior_scc_scc_cmp): Likewise.
4483 (*and_scc_scc): Likewise.
4484 (*and_scc_scc_cmp): Likewise.
4486 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4487 Alan Hayard <alan.hayward@arm.com>
4488 David Sherwood <david.sherwood@arm.com>
4490 * target.def (can_change_mode_class): New hook.
4491 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4492 (hard_regno_nregs): Likewise.
4493 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4494 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4495 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4496 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4497 (LOAD_EXTEND_OP): Update accordingly.
4498 * doc/tm.texi: Regenerate.
4499 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4500 CANNOT_CHANGE_MODE_CLASS.
4501 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4502 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4503 * combine.c (simplify_set): Update accordingly.
4504 * emit-rtl.c (validate_subreg): Likewise.
4505 * recog.c (general_operand): Likewise.
4506 * regcprop.c (mode_change_ok): Likewise.
4507 * reload1.c (choose_reload_regs): Likewise.
4508 (inherit_piecemeal_p): Likewise.
4509 * rtlanal.c (simplify_subreg_regno): Likewise.
4510 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4511 instead of CANNOT_CHANGE_MODE_CLASS.
4512 (reload_cse_simplify_operands): Likewise.
4513 * reload.c (push_reload): Use targetm.can_change_mode_class
4514 instead of CANNOT_CHANGE_MODE_CLASS.
4515 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4516 REG_CANNOT_CHANGE_MODE_P.
4517 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4518 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4519 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4520 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4521 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4522 (arm_can_change_mode_class): New function.
4523 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4524 than CANNOT_CHANGE_MODE_CLASS in comments.
4525 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4526 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4527 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4528 (ix86_can_change_mode_class): ...this new function, inverting the
4529 sense of the return value.
4530 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4531 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4532 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4533 (ia64_can_change_mode_class): New function.
4534 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4535 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4536 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4537 (m32c_can_change_mode_class): ...this new function, inverting the
4538 sense of the return value.
4539 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4540 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4541 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4542 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4543 (mips_can_change_mode_class): ...this new function, inverting the
4544 sense of the return value.
4545 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4546 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4547 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4548 (msp430_can_change_mode_class): New function.
4549 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4550 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4552 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4553 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4554 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4555 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4556 (pa_cannot_change_mode_class): Replace with...
4557 (pa_can_change_mode_class): ...this new function, inverting the
4558 sense of the return value.
4559 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4560 than CANNOT_CHANGE_MODE_CLASS in comments.
4561 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4562 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4563 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4564 (pdp11_cannot_change_mode_class): Replace with...
4565 (pdp11_can_change_mode_class): ...this new function, inverting the
4566 sense of the return value.
4567 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4568 * config/powerpcspe/powerpcspe-protos.h
4569 (rs6000_cannot_change_mode_class_ptr): Delete.
4570 * config/powerpcspe/powerpcspe.c
4571 (rs6000_cannot_change_mode_class_ptr): Delete.
4572 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4573 (rs6000_option_override_internal): Assign to
4574 targetm.can_change_mode_class instead of
4575 rs6000_cannot_change_mode_class_ptr.
4576 (rs6000_cannot_change_mode_class): Replace with...
4577 (rs6000_can_change_mode_class): ...this new function, inverting the
4578 sense of the return value.
4579 (rs6000_debug_cannot_change_mode_class): Replace with...
4580 (rs6000_debug_can_change_mode_class): ...this new function.
4581 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4582 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4584 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4585 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4587 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4588 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4589 (rs6000_option_override_internal): Assign to
4590 targetm.can_change_mode_class instead of
4591 rs6000_cannot_change_mode_class_ptr.
4592 (rs6000_cannot_change_mode_class): Replace with...
4593 (rs6000_can_change_mode_class): ...this new function, inverting the
4594 sense of the return value.
4595 (rs6000_debug_cannot_change_mode_class): Replace with...
4596 (rs6000_debug_can_change_mode_class): ...this new function.
4597 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4598 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4599 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4600 (s390_can_change_mode_class): ...this new function, inverting the
4601 sense of the return value.
4602 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4603 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4604 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4605 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4606 (sh_cannot_change_mode_class): Replace with...
4607 (sh_can_change_mode_class): ...this new function, inverting the
4608 sense of the return value.
4609 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4610 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4611 (sparc_can_change_mode_class): New function.
4612 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4613 * config/spu/spu.c (spu_can_change_mode_class): New function.
4614 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4615 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4616 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4617 (visium_can_change_mode_class): New function.
4618 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4620 2017-09-15 Richard Biener <rguenther@suse.de>
4622 PR tree-optimization/82217
4623 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4624 but not undefined case.
4626 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4629 * postreload.c (reload_cse_simplify_operands): Skip
4630 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4632 2017-09-15 Richard Biener <rguenther@suse.de>
4634 PR tree-optimization/68823
4635 * graphite-scop-detection.c (build_alias_set): If we have a
4636 possible dependence check whether we can handle them by just
4637 looking at the DRs DR_ACCESS_FNs.
4638 (build_scops): If build_alias_set fails, fail the SCOP.
4640 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4642 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4643 to support float128 built-in functions that require the ISA 3.0
4645 (BU_FLOAT128_3_HW): Likewise.
4646 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4648 (FMAF128): Likewise.
4650 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4651 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4652 floating point instructions.
4653 (rs6000_invalid_builtin): Likewise.
4654 (rs6000_builtin_mask_names): Likewise.
4655 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4656 (RS6000_BTM_FLOAT128_HW): Likewise.
4657 (RS6000_BTM_COMMON): Likewise.
4658 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4660 * doc/extend.texi (RS/6000 built-in functions): Document the
4661 IEEE 128-bit floating point square root and fused multiply-add
4664 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4666 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4667 reg (r2) isn't in the set of registers defined in the prologue.
4669 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4670 Alan Hayward <alan.hayward@arm.com>
4671 David Sherwood <david.sherwood@arm.com>
4673 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4674 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4675 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4676 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4677 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4679 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4680 max_vectorization_factor.
4681 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4683 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4684 Alan Hayward <alan.hayward@arm.com>
4685 David Sherwood <david.sherwood@arm.com>
4687 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4688 (vect_worthwhile_without_simd_p): Declare.
4689 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4690 (vectorizable_reduction): Use it.
4691 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4692 (vectorizable_operation): Likewise.
4694 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4695 Alan Hayward <alan.hayward@arm.com>
4696 David Sherwood <david.sherwood@arm.com>
4698 * tree-vectorizer.h (vect_get_num_copies): New function.
4699 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4700 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4701 (vectorizable_induction): Likewise.
4702 (vectorizable_live_operation): Likewise.
4703 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4704 (vectorizable_bswap): Likewise.
4705 (vectorizable_call): Likewise.
4706 (vectorizable_conversion): Likewise.
4707 (vectorizable_assignment): Likewise.
4708 (vectorizable_shift): Likewise.
4709 (vectorizable_operation): Likewise.
4710 (vectorizable_store): Likewise.
4711 (vectorizable_load): Likewise.
4712 (vectorizable_condition): Likewise.
4713 (vectorizable_comparison): Likewise.
4714 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4716 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4717 Alan Hayward <alan.hayward@arm.com>
4718 David Sherwood <david.sherwood@arm.com>
4720 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4721 of vect_init_vector.
4723 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4724 Alan Hayward <alan.hayward@arm.com>
4725 David Sherwood <david.sherwood@arm.com>
4727 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4728 an inline wrapper that provides a location.
4729 (gimple_build_vector): Likewise.
4730 * gimple-fold.c (gimple_build_vector_from_val): New function.
4731 (gimple_build_vector): Likewise.
4732 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4733 functions to build the initial value. Always return a gimple value.
4734 (get_initial_defs_for_reduction): Likewise. Only compute
4736 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4737 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4738 (vectorizable_induction): Use gimple_build_vector rather than
4741 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4742 Alan Hayward <alan.hayward@arm.com>
4743 David Sherwood <david.sherwood@arm.com>
4745 * target.h (vec_perm_indices): New typedef.
4746 (auto_vec_perm_indices): Likewise.
4747 * optabs-query.h: Include target.h
4748 (can_vec_perm_p): Take a vec_perm_indices *.
4749 * optabs-query.c (can_vec_perm_p): Likewise.
4750 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4751 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4752 * tree-vect-generic.c (lower_vec_perm): Likewise.
4753 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4754 (vect_grouped_load_supported): Likewise.
4755 (vect_shift_permute_load_chain): Likewise.
4756 (vect_permute_store_chain): Use auto_vec_perm_indices.
4757 (vect_permute_load_chain): Likewise.
4758 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4759 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4760 Update uses of can_vec_perm_p.
4761 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4762 mode with a number of elements. Take a vec_perm_indices *.
4763 (vect_create_epilog_for_reduction): Update accordingly.
4764 Use auto_vec_perm_indices.
4765 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4766 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4767 (vect_transform_slp_perm_load): Likewise.
4768 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4769 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4770 (vect_gen_perm_mask_checked): Likewise.
4771 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4772 (vect_gen_perm_mask_checked): Likewise.
4773 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4774 (vectorizable_store): Likewise.
4775 (vectorizable_load): Likewise.
4776 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4777 (vectorizable_bswap): Likewise.
4779 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4780 Alan Hayward <alan.hayward@arm.com>
4781 David Sherwood <david.sherwood@arm.com>
4783 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4784 * tree.c (build_vector): Likewise.
4785 (build_vector_from_ctor): Update accordingly.
4786 (build_vector_from_val): Likewise.
4787 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4788 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4789 * tree-vect-generic.c (add_rshift): Likewise.
4790 (expand_vector_divmod): Likewise.
4791 (optimize_vector_constructor): Likewise.
4792 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4793 (vect_transform_slp_perm_load): Likewise.
4794 (vect_schedule_slp_instance): Likewise.
4795 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4796 (vectorizable_call): Likewise.
4797 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4798 * expmed.c (make_tree): Likewise.
4799 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4800 a vector passed to build_vector.
4801 (fold_convert_const): Likewise.
4802 (exact_inverse): Likewise.
4803 (fold_ternary_loc): Likewise.
4804 (fold_relational_const): Likewise.
4805 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4806 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4807 (const_unop): Likewise. Store the reduction accumulator in a
4808 variable rather than an array.
4809 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4810 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4811 the new vector, rather than constructing it after the input arrays.
4812 (native_interpret_vector): Use auto_vec<tree> when building
4813 a vector passed to build_vector. Add elements in order.
4814 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4815 auto_vec<tree> when building a vector passed to build_vector.
4816 (vect_create_epilog_for_reduction): Likewise.
4817 (vectorizable_induction): Likewise.
4818 (get_initial_def_for_reduction): Likewise. Fix indentation of
4820 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4822 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4823 passed to build_vector.
4825 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4826 Alan Hayward <alan.hayward@arm.com>
4827 David Sherwood <david.sherwood@arm.com>
4829 * tree-core.h (tree_base::u): Add an "nelts" field.
4830 (tree_vector): Use VECTOR_CST_NELTS as the length.
4831 * tree.c (tree_size): Likewise.
4832 (make_vector): Initialize VECTOR_CST_NELTS.
4833 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4834 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4835 TYPE_VECTOR_SUBPARTS.
4836 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4837 as the number of units, setting it from VECTOR_CST_NELTS.
4838 (const_vector_from_tree): Likewise.
4839 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4840 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4841 (fold_negate_expr_1): Likewise.
4842 (fold_convert_const): Likewise.
4843 (const_binop): Likewise. Differentiate the number of output and
4845 (const_unop): Likewise.
4846 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4847 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4848 in cases that did the opposite.
4850 2017-09-14 Richard Biener <rguenther@suse.de>
4852 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4855 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4857 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4859 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4862 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4863 if and where to split a bb, except for splitting before debug insn
4864 sequences followed by non-label real insn. Delete debug insns
4865 in between basic blocks.
4867 * combine.c (make_compound_operation_int): Formatting fixes.
4869 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4870 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4871 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4872 * config/sol2.h (LINK_EH_SPEC): Likewise.
4873 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4874 * config/s390/linux.h (LINK_SPEC): Likewise.
4875 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4876 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4877 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4878 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4879 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4880 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4881 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4882 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4884 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4886 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4888 (ENDFILE_LINUX_SPEC): Likewise.
4889 (LINK_EH_SPEC): Likewise.
4890 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4891 (LINK_OS_LINUX_SPEC32): Likewise.
4892 (LINK_OS_LINUX_SPEC64): Likewise.
4893 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4894 (LINK_OS_LINUX_SPEC): Likewise.
4896 2017-09-13 Martin Liska <mliska@suse.cz>
4899 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4900 CASE_HIGH is NULL_TREE.
4902 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4903 Alan Hayward <alan.hayward@arm.com>
4904 David Sherwood <david.sherwood@arm.com>
4906 * target.def (secondary_memory_needed): New hook.
4907 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4908 instead of SECONDARY_MEMORY_NEEDED.
4909 (secondary_memory_needed_mode): Likewise.
4910 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4911 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4912 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4913 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4914 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4915 * doc/tm.texi: Regenerate.
4916 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4917 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4918 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4919 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4920 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4921 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4922 mode argument first and change the reg_class arguments to reg_class_t.
4923 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4924 Make static. Update the call to inline_secondary_memory_needed.
4925 (ix86_register_move_cost): Update the call to
4926 inline_secondary_memory_needed.
4927 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4928 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4930 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4931 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4933 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4934 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4935 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4936 and match hook interface. Add comment from mips.h.
4937 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4938 * config/mmix/mmix.md (truncdfsf2): Refer to
4939 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4941 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4942 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4943 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4944 (pa_secondary_memory_needed): New function.
4945 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4946 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4947 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4948 (pdp11_secondary_memory_needed): Make static and match hook interface.
4949 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4950 * config/powerpcspe/powerpcspe-protos.h
4951 (rs6000_secondary_memory_needed_ptr): Delete.
4952 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4954 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4955 (rs6000_option_override_internal): Assign to
4956 targetm.secondary_memory_needed rather than
4957 rs6000_secondary_memory_needed_ptr.
4958 (rs6000_secondary_memory_needed): Match hook interface.
4959 (rs6000_debug_secondary_memory_needed): Likewise.
4960 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4961 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4962 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4963 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4964 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4965 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4967 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4968 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4969 (rs6000_option_override_internal): Assign to
4970 targetm.secondary_memory_needed rather than
4971 rs6000_secondary_memory_needed_ptr.
4972 (rs6000_secondary_memory_needed): Match hook interface.
4973 (rs6000_debug_secondary_memory_needed): Likewise.
4974 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4975 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4976 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4977 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4978 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4979 (sparc_secondary_memory_needed): New function.
4980 * lra-constraints.c (check_and_process_move): Refer to
4981 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4983 (curr_insn_transform): Likewise.
4984 (process_alt_operands): Use targetm.secondary_memory_needed
4985 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4986 (check_secondary_memory_needed_p): Likewise.
4987 (choose_split_class): Likewise.
4988 * reload.c: Unconditionally include code that was previously
4989 conditional on SECONDARY_MEMORY_NEEDED.
4990 (push_secondary_reload): Use targetm.secondary_memory_needed
4991 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4992 (push_reload): Likewise.
4993 * reload1.c: Unconditionally include code that was previously
4994 conditional on SECONDARY_MEMORY_NEEDED.
4995 (choose_reload_regs): Use targetm.secondary_memory_needed
4996 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4997 (gen_reload): Likewise.
4998 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5000 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5001 Alan Hayward <alan.hayward@arm.com>
5002 David Sherwood <david.sherwood@arm.com>
5004 * target.def (secondary_memory_needed_mode): New hook:
5005 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5006 * targhooks.h (default_secondary_memory_needed_mode): New function.
5007 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5008 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5009 * doc/tm.texi: Regenerate.
5010 * lra-constraints.c (check_and_process_move): Use
5011 targetm.secondary_memory_needed_mode instead of
5012 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5013 (curr_insn_transform): Likewise.
5014 * reload.c (get_secondary_mem): Likewise.
5015 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5016 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5018 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5019 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5020 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5021 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5022 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5024 * config/powerpcspe/powerpcspe-protos.h
5025 (rs6000_secondary_memory_needed_mode): Delete.
5026 * config/powerpcspe/powerpcspe.c
5027 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5028 (rs6000_secondary_memory_needed_mode): Make static.
5029 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5030 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5032 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5034 (rs6000_secondary_memory_needed_mode): Make static.
5035 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5036 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5037 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5038 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5039 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5041 (sparc_secondary_memory_needed_mode): New function.
5042 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5044 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5046 * config/aarch64/constraints.md (Umq): New constraint.
5047 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5049 (mov<mode>): Update condition.
5051 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5053 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5054 when bitposition is the same.
5056 2017-09-13 Richard Biener <rguenther@suse.de>
5058 * dwarf2out.c (output_die_symbol): Remove.
5059 (output_die): Do not output a DIEs symbol.
5061 2017-09-13 Richard Biener <rguenther@suse.de>
5064 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5065 default-def to avoid breaking iterator update with the weird
5066 interaction with cgraph_update_edges_for_call_stmt_node.
5068 2017-09-13 Richard Biener <rguenther@suse.de>
5070 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5071 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5072 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5073 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5075 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5077 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5078 Disable pc relative literal load irrespective of
5079 TARGET_FIX_ERR_A53_84341 for default.
5081 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5083 * config/sparc/sparc.c (output_return): Output the source location of
5084 the insn in the delay slot, if any.
5085 (output_sibcall): Likewise.
5087 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5090 * common.opt (-static-pie): New alias.
5091 (shared): Negate static-pie.
5092 (-no-pie): Update help text.
5094 (static-pie): New option.
5095 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5096 -static-pie support.
5097 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5098 (LINK_EH_SPEC): Likewise.
5099 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5100 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5101 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5102 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5103 (init_gcc_specs): Likewise.
5104 (init_spec): Likewise.
5105 (display_help): Update help message for -pie.
5106 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5109 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5111 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5112 (movdi_aarch64): Likewise.
5113 (movti_aarch64): Likewise.
5115 2017-09-12 Simon Wright <simon@pushface.org>
5118 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5119 calculation of the minor version, always output as 0.
5121 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5124 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5125 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5126 on it early, rather than manual conversion late. For
5127 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5128 instead of performing manual conversion.
5130 2017-09-12 Carl Love <cel@us.ibm.com>
5132 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5133 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5135 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5136 VMULOSW): Add definitions.
5137 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5138 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5139 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5141 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5143 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5145 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5147 (movti_aarch64): Likewise.
5148 (movdf_aarch64): Likewise.
5149 (movtf_aarch64): Likewise.
5150 (load_pairdi): Likewise.
5151 (store_pairdi): Likewise.
5152 (load_pairdf): Likewise.
5153 (store_pairdf): Likewise.
5154 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5155 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5156 (ldr_got_small_<mode>): Likewise.
5157 (ldr_got_small_28k_<mode>): Likewise.
5158 (ldr_got_tiny): Likewise.
5159 * config/aarch64/iterators.md (ldst_sz): New.
5160 (ldpstp_sz): Likewise.
5161 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5163 (thunderx_load): Split load_8 to load_16.
5164 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5166 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5167 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5168 (xgene1_store_pair): Split store_8 to store_16.
5169 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5170 (falkor_st_0_st_sd): Split store_8 to store_16.
5172 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5174 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5175 and store1/2/3/4 to store_4/8/12/16.
5176 * config/aarch64/aarch64.md: Update for rename.
5177 * config/arm/arm.md: Likewise.: Likewise.
5178 * config/arm/arm.c: Likewise.
5179 * config/arm/thumb1.md: Likewise.
5180 * config/arm/thumb2.md: Likewise.
5181 * config/arm/vfp.md: Likewise.
5182 * config/arm/arm-generic.md: Likewise.
5183 * config/arm/arm1020e.md: Likewise.
5184 * config/arm/arm1026ejs.md: Likewise.
5185 * config/arm/arm1136jfs.md: Likewise.
5186 * config/arm/arm926ejs.md: Likewise.
5187 * config/arm/cortex-a15.md: Likewise.
5188 * config/arm/cortex-a17.md: Likewise.
5189 * config/arm/cortex-a5.md: Likewise.
5190 * config/arm/cortex-a53.md: Likewise.
5191 * config/arm/cortex-a57.md: Likewise.
5192 * config/arm/cortex-a7.md: Likewise.
5193 * config/arm/cortex-a8.md: Likewise.
5194 * config/arm/cortex-a9.md: Likewise.
5195 * config/arm/cortex-m4.md: Likewise.
5196 * config/arm/cortex-m7.md: Likewise.
5197 * config/arm/cortex-r4.md: Likewise.
5198 * config/arm/exynos-m1.md: Likewise.
5199 * config/arm/fa526.md: Likewise.
5200 * config/arm/fa606te.md: Likewise.
5201 * config/arm/fa626te.md: Likewise.
5202 * config/arm/fa726te.md: Likewise.
5203 * config/arm/fmp626.md: Likewise.
5204 * config/arm/iwmmxt.md: Likewise.
5205 * config/arm/ldmstm.md: Likewise.
5206 * config/arm/marvell-pj4.md: Likewise.
5207 * config/arm/xgene1.md: Likewise.
5208 * config/aarch64/thunderx.md: Likewise.
5209 * config/aarch64/thunderx2t99.md: Likewise.
5210 * config/aarch64/falkor.md: Likewise.
5212 2017-09-12 Martin Liska <mliska@suse.cz>
5214 * attribs.c (private_lookup_attribute): New function.
5215 * attribs.h (private_lookup_attribute): Declared here.
5216 (lookup_attribute): Called from this place.
5218 2017-09-12 Richard Biener <rguenther@suse.de>
5220 PR tree-optimization/82157
5221 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5222 stmts with side-effects.
5224 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5225 Alan Hayward <alan.hayward@arm.com>
5226 David Sherwood <david.sherwood@arm.com>
5228 * target.def (hard_regno_nregs): New hook.
5229 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5230 * targhooks.h (default_hard_regno_nregs): Declare.
5231 * targhooks.c (default_hard_regno_nregs): New function.
5232 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5233 (TARGET_HARD_REGNO_NREGS): ...this hook.
5234 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5235 (CLASS_MAX_NREGS): Likewise.
5236 * doc/tm.texi: Regenerate.
5237 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5238 instead of HARD_REGNO_NREGS.
5239 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5240 HARD_REGNO_NREGS in the comment.
5241 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5242 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5243 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5244 Return an unsigned int.
5245 (TARGET_HARD_REGNO_NREGS): Redefine.
5246 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5247 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5248 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5249 (arc_hard_regno_nregs): New function.
5250 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5251 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5252 (arm_hard_regno_nregs): New function.
5253 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5254 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5255 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5256 (TARGET_HARD_REGNO_NREGS): Redefine.
5257 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5258 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5259 (HARD_REGNO_NREGS): Delete.
5260 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5261 (cr16_hard_regno_nregs): New function.
5262 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5263 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5264 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5265 (cris_hard_regno_nregs): New function.
5266 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5267 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5268 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5269 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5270 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5271 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5272 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5273 (frv_hard_regno_nregs): Make static. Take and return an
5275 (frv_class_max_nregs): Remove outdated copy of documentation.
5276 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5277 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5278 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5279 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5280 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5281 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5282 (TARGET_HARD_REGNO_NREGS): Redefine.
5283 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5284 (CLASS_MAX_NREGS): Update comment.
5285 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5286 (ia64_hard_regno_nregs): New function.
5287 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5288 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5289 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5290 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5291 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5293 (m32c_hard_regno_nregs): Likewise. Make static.
5294 (TARGET_HARD_REGNO_NREGS): Redefine.
5295 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5296 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5297 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5298 (m68k_hard_regno_nregs): New function.
5299 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5300 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5301 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5302 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5303 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5304 Take and return an unsigned int.
5305 (TARGET_HARD_REGNO_NREGS): Redefine.
5306 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5307 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5308 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5309 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5310 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5311 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5312 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5313 (msp430_hard_regno_nregs): Make static. Take and return an
5315 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5316 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5317 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5318 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5319 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5320 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5321 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5322 (TARGET_HARD_REGNO_NREGS): Redefine.
5323 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5324 (PA_HARD_REGNO_NREGS): ...this.
5325 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5326 (PA_HARD_REGNO_NREGS): ...this.
5327 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5328 (pa_hard_regno_nregs): New function.
5329 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5330 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5331 (pdp11_hard_regno_nregs): New function.
5332 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5333 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5334 (rs6000_hard_regno_nregs_hook): New function.
5335 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5336 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5337 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5338 Take and return an unsigned int. Move earlier in file.
5339 (TARGET_HARD_REGNO_NREGS): Redefine.
5340 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5341 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5342 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5343 (rl78_hard_regno_nregs): Make static. Take and return an
5345 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5346 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5347 (rs6000_hard_regno_nregs_hook): New function.
5348 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5349 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5350 (TARGET_HARD_REGNO_NREGS): Redefine.
5351 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5352 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5353 instead of HARD_REGNO_NREGS.
5354 (s390_hard_regno_nregs): New function.
5355 (s390_hard_regno_mode_ok): Add comment from s390.h.
5356 (TARGET_HARD_REGNO_NREGS): Redefine.
5357 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5358 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5359 (sh_hard_regno_nregs): New function.
5360 (sh_pass_in_reg_p): Use it.
5361 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5362 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5363 (sparc_hard_regno_nregs): New function.
5364 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5365 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5366 (spu_function_arg_advance): Use it, supplying a valid register number.
5367 (TARGET_HARD_REGNO_NREGS): Redefine.
5368 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5369 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5370 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5371 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5372 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5373 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5374 (CLASS_MAX_NREGS): Remove copy of old documentation.
5375 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5376 (visium_hard_regno_nregs): New function.
5377 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5378 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5379 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5380 xtensa_hard_regno_nregs): New function.
5381 * system.h (HARD_REGNO_NREGS): Poison.
5383 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5385 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5386 hard_regno_nregs instead of HARD_REGNO_NREGS.
5387 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5388 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5389 (c6x_expand_epilogue): Likewise.
5390 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5391 (frv_read_iacc_argument): Likewise.
5392 * config/sh/sh.c: Include regs.h.
5393 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5394 (regs_used): Likewise.
5395 (output_stack_adjust): Likewise.
5396 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5397 * expmed.c: Include regs.h.
5398 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5399 * ree.c: Include regs.h.
5400 (combine_reaching_defs): Use hard_regno_nregs instead of
5402 (add_removable_extension): Likewise.
5404 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5406 * regs.h (hard_regno_nregs): Turn into a function.
5407 (end_hard_regno): Update accordingly.
5408 * caller-save.c (setup_save_areas): Likewise.
5409 (save_call_clobbered_regs): Likewise.
5410 (replace_reg_with_saved_mem): Likewise.
5411 (insert_restore): Likewise.
5412 (insert_save): Likewise.
5413 * combine.c (can_change_dest_mode): Likewise.
5414 (move_deaths): Likewise.
5415 (distribute_notes): Likewise.
5416 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5417 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5418 (rs6000_split_multireg_move): Likewise.
5419 (rs6000_register_move_cost): Likewise.
5420 (rs6000_memory_move_cost): Likewise.
5421 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5422 (rs6000_split_multireg_move): Likewise.
5423 (rs6000_register_move_cost): Likewise.
5424 (rs6000_memory_move_cost): Likewise.
5425 * cselib.c (cselib_reset_table): Likewise.
5426 (cselib_lookup_1): Likewise.
5427 * emit-rtl.c (set_mode_and_regno): Likewise.
5428 * function.c (aggregate_value_p): Likewise.
5429 * ira-color.c (setup_profitable_hard_regs): Likewise.
5430 (check_hard_reg_p): Likewise.
5431 (calculate_saved_nregs): Likewise.
5432 (assign_hard_reg): Likewise.
5433 (improve_allocation): Likewise.
5434 (calculate_spill_cost): Likewise.
5435 * ira-emit.c (modify_move_list): Likewise.
5436 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5437 (ira_hard_reg_in_set_p): Likewise.
5438 * ira.c (setup_reg_mode_hard_regset): Likewise.
5439 (clarify_prohibited_class_mode_regs): Likewise.
5440 (check_allocation): Likewise.
5441 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5442 (lra_setup_reg_renumber): Likewise.
5443 (setup_try_hard_regno_pseudos): Likewise.
5444 (spill_for): Likewise.
5445 (assign_hard_regno): Likewise.
5446 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5447 * lra-constraints.c (in_class_p): Likewise.
5448 (lra_constraint_offset): Likewise.
5449 (simplify_operand_subreg): Likewise.
5450 (lra_constraints): Likewise.
5451 (split_reg): Likewise.
5452 (split_if_necessary): Likewise.
5453 (invariant_p): Likewise.
5454 (inherit_in_ebb): Likewise.
5455 * lra-lives.c (process_bb_lives): Likewise.
5456 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5457 (get_hard_regs): Likewise.
5458 (do_remat): Likewise.
5459 * lra-spills.c (assign_spill_hard_regs): Likewise.
5460 * mode-switching.c (create_pre_exit): Likewise.
5461 * postreload.c (reload_combine_recognize_pattern): Likewise.
5462 * recog.c (peep2_find_free_register): Likewise.
5463 * regcprop.c (kill_value_regno): Likewise.
5464 (set_value_regno): Likewise.
5465 (copy_value): Likewise.
5466 (maybe_mode_change): Likewise.
5467 (find_oldest_value_reg): Likewise.
5468 (copyprop_hardreg_forward_1): Likewise.
5469 * regrename.c (check_new_reg_p): Likewise.
5470 (regrename_do_replace): Likewise.
5471 * reload.c (push_reload): Likewise.
5472 (combine_reloads): Likewise.
5473 (find_dummy_reload): Likewise.
5474 (operands_match_p): Likewise.
5475 (find_reloads): Likewise.
5476 (find_equiv_reg): Likewise.
5477 (reload_adjust_reg_for_mode): Likewise.
5478 * reload1.c (count_pseudo): Likewise.
5479 (count_spilled_pseudo): Likewise.
5480 (find_reg): Likewise.
5481 (clear_reload_reg_in_use): Likewise.
5482 (free_for_value_p): Likewise.
5483 (allocate_reload_reg): Likewise.
5484 (choose_reload_regs): Likewise.
5485 (reload_adjust_reg_for_temp): Likewise.
5486 (emit_reload_insns): Likewise.
5487 (delete_output_reload): Likewise.
5488 * rtlanal.c (subreg_get_info): Likewise.
5489 * sched-deps.c (sched_analyze_reg): Likewise.
5490 * sel-sched.c (init_regs_for_mode): Likewise.
5491 (mark_unavailable_hard_regs): Likewise.
5492 (choose_best_reg_1): Likewise.
5493 (verify_target_availability): Likewise.
5494 * valtrack.c (dead_debug_insert_temp): Likewise.
5495 * var-tracking.c (track_loc_p): Likewise.
5496 (emit_note_insn_var_location): Likewise.
5497 * varasm.c (make_decl_rtl): Likewise.
5498 * reginfo.c (choose_hard_reg_mode): Likewise.
5499 (init_reg_modes_target): Refer directly to
5500 this_target_regs->x_hard_regno_nregs.
5502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5504 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5505 instead of hard_regno_nregs.
5507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5509 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5510 end_hard_regno instead of hard_regno_nregs.
5511 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5512 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5513 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5514 * ira-color.c (improve_allocation): Likewise.
5515 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5516 * lra-lives.c (mark_regno_live): Likewise.
5517 (mark_regno_dead): Likewise.
5518 * lra-remat.c (operand_to_remat): Likewise.
5519 * lra.c (collect_non_operand_hard_regs): Likewise.
5520 * postreload.c (reload_combine_note_store): Likewise.
5521 (move2add_valid_value_p): Likewise.
5522 * reload.c (regno_clobbered_p): Likewise.
5524 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5526 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5528 * config/v850/v850.c (v850_reorg): Likewise.
5529 * reload.c (refers_to_regno_for_reload_p): Likewise.
5530 (find_equiv_reg): Likewise.
5531 * reload1.c (reload_reg_reaches_end_p): Likewise.
5533 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5535 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5537 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5538 * config/arm/arm.c (output_move_neon): Likewise.
5539 (arm_attr_length_move_neon): Likewise.
5540 (neon_split_vcombine): Likewise.
5541 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5542 (c6x_mark_reg_written): Likewise.
5543 (c6x_dwarf_register_span): Likewise.
5544 * config/i386/i386.c (ix86_save_reg): Likewise.
5545 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5546 (rws_access_reg): Likewise.
5547 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5548 * mode-switching.c (create_pre_exit): Likewise.
5549 * ree.c (combine_reaching_defs): Likewise.
5550 (add_removable_extension): Likewise.
5551 * regcprop.c (find_oldest_value_reg): Likewise.
5552 (copyprop_hardreg_forward_1): Likewise.
5553 * reload.c (reload_inner_reg_of_subreg): Likewise.
5554 (push_reload): Likewise.
5555 (combine_reloads): Likewise.
5556 (find_dummy_reload): Likewise.
5557 (reload_adjust_reg_for_mode): Likewise.
5558 * reload1.c (find_reload_regs): Likewise.
5559 (forget_old_reloads_1): Likewise.
5560 (reload_reg_free_for_value_p): Likewise.
5561 (reload_adjust_reg_for_temp): Likewise.
5562 (emit_reload_insns): Likewise.
5563 (delete_output_reload): Likewise.
5564 * sel-sched.c (choose_best_reg_1): Likewise.
5565 (choose_best_pseudo_reg): Likewise.
5567 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5568 Alan Hayward <alan.hayward@arm.com>
5569 David Sherwood <david.sherwood@arm.com>
5571 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5572 * target.def (slow_unaligned_access): New hook.
5573 * targhooks.h (default_slow_unaligned_access): Declare.
5574 * targhooks.c (default_slow_unaligned_access): New function.
5575 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5576 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5577 * doc/tm.texi: Regenerate.
5578 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5579 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5580 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5582 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5583 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5585 (rs6000_slow_unaligned_access): New function.
5586 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5587 (expand_block_compare): Likewise.
5588 (expand_strn_compare): Likewise.
5589 (rs6000_rtx_costs): Likewise.
5590 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5591 (riscv_slow_unaligned_access): Likewise.
5592 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5593 (riscv_slow_unaligned_access_p): ...this and make static.
5594 (riscv_option_override): Update accordingly.
5595 (riscv_slow_unaligned_access): New function.
5596 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5597 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5598 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5599 (rs6000_slow_unaligned_access): New function.
5600 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5601 (rs6000_rtx_costs): Likewise.
5602 * config/rs6000/rs6000-string.c (expand_block_compare)
5603 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5604 of SLOW_UNALIGNED_ACCESS.
5605 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5606 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5607 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5608 of SLOW_UNALIGNED_ACCESS.
5609 * expmed.c (simple_mem_bitfield_p): Likewise.
5610 * expr.c (alignment_for_piecewise_move): Likewise.
5611 (emit_group_load_1): Likewise.
5612 (emit_group_store): Likewise.
5613 (copy_blkmode_from_reg): Likewise.
5614 (emit_push_insn): Likewise.
5615 (expand_assignment): Likewise.
5616 (store_field): Likewise.
5617 (expand_expr_real_1): Likewise.
5618 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5619 * lra-constraints.c (simplify_operand_subreg): Likewise.
5620 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5621 * gimple-ssa-store-merging.c: Likewise in block comment at start
5623 * tree-ssa-strlen.c: Include target.h.
5624 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5625 of SLOW_UNALIGNED_ACCESS.
5626 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5628 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5630 PR rtl-optimization/82185
5631 * expmed.c (emit_store_flag_int): Only test tem if it has been
5634 2017-09-12 Richard Biener <rguenther@suse.de>
5637 * match.pd ((FTYPE) N CMP CST): Fix typo.
5639 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5641 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5643 (mips_near_type_p): Add 'short_call' attribute as a synonym
5645 * doc/extend.texi (short_call): Document new function attribute.
5647 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5650 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5651 assertion check that in the condition.
5652 (get_atomic_generic_size): Likewise. Before testing if parameter
5653 has pointer type, if it has array type, call for C++
5654 default_conversion to perform array-to-pointer conversion.
5656 2017-09-12 Richard Biener <rguenther@suse.de>
5658 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5659 for operations we cannot scalarize.
5661 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5663 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5664 vectors heap vectors. Clean up comments.
5665 Make visited_bbs a reference.
5666 (profitable_jump_thread_path): Make GC
5667 vectors heap vectors. Clean up comments.
5669 (convert_and_register_jump_thread_path): Make GC vectors heap
5671 (check_subpath_and_update_thread_path): Same. Clean up comments.
5672 Make visited_bbs a reference.
5673 (handle_phi): Abstract common code to to
5674 register_jump_thread_path_if_profitable.
5675 Rename VAR_BB to DEF_BB.
5677 Make GC vectors heap vectors.
5678 Make visited_bbs a reference.
5679 (handle_assignment): Same.
5680 (register_jump_thread_path_if_profitable): New.
5681 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5683 Make GC vectors heap vectors. Clean up comments.
5684 Make visited_bbs a reference.
5685 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5686 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5689 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5692 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5693 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5695 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5697 Revert r251800 and r251799.
5699 2017-09-11 Martin Jambor <mjambor@suse.cz>
5702 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5703 arguments in advance.
5704 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5705 use it to find predecessor edges.
5706 (naive_outof_ssa): Collect vector of predecessors.
5708 2017-09-08 Jason Merrill <jason@redhat.com>
5710 PR c++/70029 - ICE with ref-qualifier and -flto
5711 * langhooks.h (struct lang_hooks_for_types): Add
5712 copy_lang_qualifiers.
5713 * attribs.c (build_type_attribute_qual_variant): Use it.
5714 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5716 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5717 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5719 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5722 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5723 (*mulsi3_sp64): New instruction.
5724 (mulsi3): New expander.
5726 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5728 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5730 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5732 * sancov.c: Include memmodel.h.
5734 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5737 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5740 2017-09-07 Carl Love <cel@us.ibm.com>
5742 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5743 the sldi instruction.
5745 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5747 * sancov.c: Include tm_p.h.
5749 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5752 * output.h (switch_to_other_text_partition): New declaration.
5753 * varasm.c (switch_to_other_text_partition): New function.
5754 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5755 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5756 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5757 to the other text partition before emitting LCL label and switch back
5758 after emitting the word after it.
5760 2017-09-07 Richard Biener <rguenther@suse.de>
5762 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5763 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5764 critical edges here, after loop init.
5765 (pass_data_pre): Remove PROP_no_crit_edges flags.
5766 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5767 for valueization of call args to avoid leaking VN_TOP.
5768 (visit_use): Assert we do not visit default defs.
5769 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5770 Use error_mark_node to more easily detect leaking VN_TOP.
5771 All default-defs are varying, not VN_TOP. Mark them visited.
5772 (run_scc_vn): Make code match comment.
5774 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5776 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5777 OPTION_MASK_FLOAT128_KEYWORD.
5778 (POWERPC_MASKS): Likewise.
5779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5780 support for the -mfloat128-type option, and make -mfloat128
5781 default on PowerPC Linux systems. Define or undefine
5782 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5783 Define __float128 to be __ieee128 if IEEE 128-bit support is
5784 enabled, or undefine it.
5785 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5786 Delete defining __FLOAT128_TYPE__.
5787 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5788 -mfloat128-type option and make -mfloat128 default on PowerPC
5790 (TARGET_FLOAT128_TYPE): Likewise.
5791 (-mfloat128-type): Likewise.
5792 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5793 Delete the -mfloat128-type option and make -mfloat128 default on
5794 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5795 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5796 128-bit floating point is enabled. Change tests from using
5797 -mfloat128-type to -mfloat128.
5798 (rs6000_mangle_type): Use the correct mangling for the __float128
5799 type even if normal long double is restricted to 64-bits.
5800 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5802 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5803 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5804 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5806 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5807 documentation for -mfloat128.
5809 2017-09-06 Olivier Hainque <hainque@adacore.com>
5811 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5813 2017-09-06 Wish Wu <wishwu007@gmail.com>
5814 Jakub Jelinek <jakub@redhat.com>
5816 * asan.c (initialize_sanitizer_builtins): Add
5817 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5818 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5819 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5820 BT_FN_VOID_UINT64_PTR variables.
5821 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5822 (BT_FN_VOID_UINT16_UINT16): Likewise.
5823 (BT_FN_VOID_UINT32_UINT32): Likewise.
5824 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5825 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5826 (BT_FN_VOID_UINT64_PTR): Likewise.
5827 * common.opt (flag_sanitize_coverage): New variable.
5828 (fsanitize-coverage=trace-pc): Remove.
5829 (fsanitize-coverage=): Add.
5830 * flag-types.h (enum sanitize_coverage_code): New enum.
5831 * fold-const.c (fold_range_test): Disable non-short-circuit
5832 optimization if flag_sanitize_coverage.
5833 (fold_truth_andor): Likewise.
5834 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5835 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5836 (coverage_sanitizer_opts): New array.
5837 (get_closest_sanitizer_option): Add OPTS argument, handle also
5838 OPT_fsanitize_coverage_.
5839 (parse_sanitizer_options): Adjusted to also handle
5840 OPT_fsanitize_coverage_.
5841 (common_handle_option): Add OPT_fsanitize_coverage_.
5842 * sancov.c (instrument_comparison, instrument_switch): New function.
5843 (sancov_pass): Add trace-cmp support.
5844 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5845 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5846 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5847 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5848 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5849 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5850 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5851 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5852 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5853 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5855 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5857 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5858 error. Only quit immediately if parsing is complete.
5859 (BEGIN): Initialize fatal_err and parse_done.
5860 (begin fpu, end fpu): Check number of arguments.
5861 (begin arch, end arch): Likewise.
5862 (begin cpu, end cpu): Likewise.
5863 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5864 (optalias): Likewise.
5866 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5868 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5869 * config/arm/arm-isa.h: Delete. Move definitions to ...
5870 * arm-cpus.in: ... here. Use new feature and fgroup values.
5871 * config/arm/arm.c (arm_option_override): Use lower case for feature
5873 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5874 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5875 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5877 (print_isa_bits_for): New function.
5878 (gen_isa): New function.
5879 (gen_comm_data): Use print_isa_bits_for.
5880 (define feature): New keyword.
5881 (define fgroup): New keyword.
5882 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5883 (arm-isa.h): Add rule to generate file.
5884 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5885 case for feature bit names.
5887 2017-09-06 Richard Biener <rguenther@suse.de>
5889 * tree-ssa-pre.c (NECESSARY): Remove.
5890 (create_expression_by_pieces): Do not touch pass-local flags.
5891 (insert_into_preds_of_block): Likewise.
5892 (do_pre_regular_insertion): Likewise.
5893 (eliminate_insert): Likewise.
5894 (eliminate_dom_walker::before_dom_children): Likewise.
5895 (fini_eliminate): Do not look at inserted_exprs.
5896 (mark_operand_necessary): Remove.
5897 (remove_dead_inserted_code): Replace with simple work-list
5898 algorithm based on inserted_exprs and SSA uses.
5899 (pass_pre::execute): Re-order fini_eliminate and
5900 remove_dead_inserted_code.
5902 2017-09-06 Olivier Hainque <hainque@adacore.com>
5904 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5905 for VxWorks 7. Adjust surrounding comments.
5907 2017-09-06 Richard Biener <rguenther@suse.de>
5909 * gimple-ssa-strength-reduction.c
5910 (find_candidates_dom_walker::before_dom_children): Also allow
5913 2017-09-06 Richard Biener <rguenther@suse.de>
5915 PR tree-optimization/82108
5916 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5917 for gap in the non-permutation SLP case.
5919 2017-09-06 Martin Jambor <mjambor@suse.cz>
5921 PR tree-optimization/82078
5922 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5923 add_access_to_work_queue...
5924 (build_accesses_from_assign): ...here.
5925 (propagate_all_subaccesses): Make sure racc is the group
5926 representative, if there is one.
5928 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5931 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5934 2017-09-06 Richard Biener <rguenther@suse.de>
5936 * gimple-ssa-strength-reduction.c
5937 (find_candidates_dom_walker::before_doom_children): Use a
5938 type and not a mode check.
5940 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5943 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5945 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5947 2017-09-05 Jeff Law <law@redhat.com>
5949 PR tree-optimization/64910
5950 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5951 cases where we have 3 or more operands.
5953 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5956 * omp-low.c (lower_omp_for): Recompute tree invariant if
5957 gimple_omp_for_initial/final is ADDR_EXPR.
5960 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5961 into gimple val before gimplification fo the COND_EXPR.
5963 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5965 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5966 REGION_COPY argument.
5967 (thread_through_all_blocks): Remove unused argument to
5968 duplicate_thread_path.
5970 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5971 Alan Hayward <alan.hayward@arm.com>
5972 David Sherwood <david.sherwood@arm.com>
5974 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5975 Take a scalar_mode rather than a machine_mode.
5976 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5977 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5978 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5979 (aarch64_gen_adjusted_ldpstp): Likewise.
5980 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5983 Alan Hayward <alan.hayward@arm.com>
5984 David Sherwood <david.sherwood@arm.com>
5986 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5987 Take a scalar_int_mode instead of a machine_mode.
5988 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5989 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5990 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5991 (aarch64_simd_attr_length_rglist): Delete.
5992 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5993 a scalar_int_mode instead of a machine_mode.
5994 (aarch64_add_offset): Likewise.
5995 (aarch64_internal_mov_immediate): Likewise
5996 (aarch64_add_constant_internal): Likewise.
5997 (aarch64_add_constant): Likewise.
5998 (aarch64_movw_imm): Likewise.
5999 (aarch64_rtx_arith_op_extract_p): Likewise.
6000 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6001 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6002 Remove assert that the mode isn't a vector.
6003 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6004 (aarch64_expand_mov_immediate): Update calls after above changes.
6005 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6006 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6007 (aarch64_move_imm): Likewise.
6008 (aarch64_can_const_movi_rtx_p): Likewise.
6009 (aarch64_strip_extend): Likewise.
6010 (aarch64_extr_rtx_p): Likewise.
6011 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6012 a CONST_INT when the mode parameter is VOIDmode.
6013 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6015 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6017 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6018 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6019 (bitwise_type_for_mode): Update accordingly.
6021 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6023 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6024 * stor-layout.c (mode_for_size_tree): Likewise.
6025 (mode_for_array): Update accordingly.
6026 (layout_decl): Likewise.
6027 (compute_record_mode): Likewise. Only set the mode once.
6029 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6031 * target.def (get_mask_mode): Change return type to opt_mode.
6033 * doc/tm.texi: Regenerate.
6034 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6035 * targhooks.c (default_get_mask_mode): Likewise.
6036 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6037 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6038 targetm.get_mask_mode.
6039 * tree.c (build_truth_vector_type): Likewise.
6041 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6043 * machmode.h (mode_for_vector): Return an opt_mode.
6044 * stor-layout.c (mode_for_vector): Likewise.
6045 (mode_for_int_vector): Update accordingly.
6046 (layout_type): Likewise.
6047 * config/i386/i386.c (emit_memmov): Likewise.
6048 (ix86_expand_set_or_movmem): Likewise.
6049 (ix86_expand_vector_init): Likewise.
6050 (ix86_get_mask_mode): Likewise.
6051 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6053 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6054 * expmed.c (extract_bit_field_1): Likewise.
6055 * expr.c (expand_expr_real_2): Likewise.
6056 * optabs-query.c (can_vec_perm_p): Likewise.
6057 (can_vec_mask_load_store_p): Likewise.
6058 * optabs.c (expand_vec_perm): Likewise.
6059 * targhooks.c (default_get_mask_mode): Likewise.
6060 * tree-vect-stmts.c (vectorizable_store): Likewise.
6061 (vectorizable_load): Likewise.
6062 (get_vectype_for_scalar_type_and_size): Likewise.
6064 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6066 * machmode.h (mode_for_int_vector): New function.
6067 * stor-layout.c (mode_for_int_vector): Likewise.
6068 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6069 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6070 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6071 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6072 (s390_expand_vcond): Likewise.
6074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6076 * machmode.h (opt_machine_mode): New type.
6077 (opt_mode<T>): Allow construction from anything that can be
6079 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6080 (mode_for_size): Return an opt_machine_mode.
6081 * stor-layout.c (mode_for_size): Likewise.
6082 (mode_for_size_tree): Update call accordingly.
6083 (bitwise_mode_for_mode): Likewise.
6084 (make_fract_type): Likewise.
6085 (make_accum_type): Likewise.
6086 * caller-save.c (replace_reg_with_saved_mem): Update call
6088 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6089 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6090 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6091 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6092 * expmed.c (extract_bit_field_1): Likewise.
6093 * reload.c (get_secondary_mem): Likewise.
6094 * varasm.c (assemble_integer): Likewise.
6095 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6098 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6100 * machmode.h (decimal_float_mode_for_size): New function.
6101 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6102 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6103 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6104 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6105 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6106 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6108 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6110 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6111 (get_builtin_sync_mode): Likewise.
6112 (expand_ifn_atomic_compare_exchange): Likewise.
6113 (expand_builtin_atomic_clear): Likewise.
6114 (expand_builtin_atomic_test_and_set): Likewise.
6115 (fold_builtin_atomic_always_lock_free): Likewise.
6116 * calls.c (compute_argument_addresses): Likewise.
6117 (emit_library_call_value_1): Likewise.
6118 (store_one_arg): Likewise.
6119 * combine.c (combine_instructions): Likewise.
6120 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6121 * config/arm/arm.c (arm_function_value): Likewise.
6122 (aapcs_allocate_return_reg): Likewise.
6123 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6124 * config/i386/i386.c (construct_container): Likewise.
6125 (ix86_gimplify_va_arg): Likewise.
6126 (ix86_expand_sse_cmp): Likewise.
6127 (emit_memmov): Likewise.
6128 (emit_memset): Likewise.
6129 (expand_small_movmem_or_setmem): Likewise.
6130 (ix86_expand_pextr): Likewise.
6131 (ix86_expand_pinsr): Likewise.
6132 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6133 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6135 * config/mips/mips.c (mips_function_value_1) Likewise.
6136 (mips_block_move_straight): Likewise.
6137 (mips_expand_ins_as_unaligned_store): Likewise.
6138 * config/powerpcspe/powerpcspe.c
6139 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6140 (rs6000_darwin64_record_arg_flush): Likewise.
6141 * config/rs6000/rs6000.c
6142 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6143 (rs6000_darwin64_record_arg_flush): Likewise.
6144 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6145 (sparc_function_value_1): Likewise.
6146 * config/spu/spu.c (adjust_operand): Likewise.
6147 (spu_emit_branch_or_set): Likewise.
6148 (arith_immediate_p): Likewise.
6149 * emit-rtl.c (gen_lowpart_common): Likewise.
6150 * expr.c (expand_expr_real_1): Likewise.
6151 * function.c (assign_parm_setup_block): Likewise.
6152 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6153 * reload1.c (alter_reg): Likewise.
6154 * stor-layout.c (mode_for_vector): Likewise.
6155 (layout_type): Likewise.
6157 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6159 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6160 (spu_convert_move): Likewise.
6161 * lower-subreg.c (resolve_simple_move): Likewise.
6163 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6166 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6167 define_insn to a define_expand.
6168 (altivec_vsum2sws_direct): New define_insn.
6169 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6171 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6173 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6176 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6181 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6182 realign stack if stack alignment needed is less than incoming
6185 2017-09-05 Marek Polacek <polacek@redhat.com>
6188 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6191 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6193 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6195 2017-09-05 Richard Biener <rguenther@suse.de>
6197 PR tree-optimization/82084
6198 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6200 2017-09-05 Richard Biener <rguenther@suse.de>
6202 PR tree-optimization/82102
6203 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6205 2017-09-05 Martin Liska <mliska@suse.cz>
6207 PR tree-optimization/82032
6208 * tree-cfg.c (generate_range_test): New function.
6209 * tree-cfg.h (generate_range_test): Declared here.
6210 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6211 (cleanup_control_expr_graph): Use it.
6212 * tree-switch-conversion.c (try_switch_expansion): Remove
6214 (emit_case_nodes): Use generate_range_test.
6216 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6219 * config/i386/i386.md (*<btsc><mode>_mask): Add
6220 TARGET_USE_BT to insn constraint.
6221 (*btr<mode>_mask): Ditto.
6223 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6225 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6226 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6228 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6231 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6232 TARGET_NEON and TARGET_IWMMXT.
6233 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6234 TARGET_NEON and TARGET_IWMMXT.
6235 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6237 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6239 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6240 (ix86_rewrite_tls_address): Ditto.
6241 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6242 (ix86_rewrite_tls_address_1): Ditto.
6243 (ix86_rewrite_tls_address): Ditto.
6244 * config/i386/predicates.md (tls_address_pattern): New predicate.
6245 * config/i386/i386.md (TLS address splitter): New splitter.
6247 2017-09-04 Richard Biener <rguenther@suse.de>
6249 PR tree-optimization/82084
6250 * fold-const.h (can_native_encode_string_p): Declare.
6251 * fold-const.c (can_native_encode_string_p): Factor out from ...
6252 (native_encode_string): ... here.
6253 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6254 vectorizing stores from constants we later cannot handle.
6256 2017-09-04 Marek Polacek <polacek@redhat.com>
6259 * doc/invoke.texi: Update -Wtautological-compare documentation.
6261 2017-09-04 Jeff Law <law@redhat.com>
6263 PR tree-optimization/64910
6264 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6265 swap the first and last operand if the last is a constant.
6267 2017-09-04 Marek Polacek <polacek@redhat.com>
6270 * convert.c (do_narrow): When sanitizing signed integer overflows,
6271 bail out for signed types.
6272 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6274 2017-09-04 Richard Biener <rguenther@suse.de>
6276 PR tree-optimization/82060
6277 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6278 Move devirtualization after stmt folding and before EH/AB/noreturn
6279 cleanup to get the stmt refs canonicalized. Use a bool instead
6280 of gimple_modified_p since that doesn't work for NOPs. Schedule
6281 NOPs generated by folding for removal.
6283 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6284 Alan Hayward <alan.hayward@arm.com>
6285 David Sherwood <david.sherwood@arm.com>
6287 * coretypes.h (pad_direction): New enum.
6288 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6289 (FUNCTION_ARG_PADDING): Likewise.
6290 * target.def (function_arg_padding): New hook.
6291 * targhooks.h (default_function_arg_padding): Declare.
6292 * targhooks.c (default_function_arg_padding): New function.
6293 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6294 (TARGET_FUNCTION_ARG_PADDING): ...this.
6295 * doc/tm.texi: Regenerate.
6296 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6297 instead of direction.
6298 (compute_argument_addresses): Likewise.
6299 (load_register_parameters): Likewise.
6300 (emit_library_call_value_1): Likewise.
6301 (store_one_arg): Use targetm.calls.function_arg_padding instead
6302 of FUNCTION_ARG_PADDING.
6303 (must_pass_in_stack_var_size_or_pad): Likewise.
6304 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6305 (emit_group_store): Likewise.
6306 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6307 instead of FUNCTION_ARG_PADDING.
6308 (emit_push_insn): Likewise, and propagate enum change throughout
6310 * function.h (direction): Delete.
6311 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6313 * function.c (assign_parm_find_stack_rtl): Likewise.
6314 (assign_parm_setup_block_p): Likewise.
6315 (assign_parm_setup_block): Likewise.
6316 (gimplify_parameters): Likewise.
6317 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6318 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6320 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6321 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6322 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6323 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6324 (aarch64_function_arg_padding): ...this new function.
6325 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6326 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6327 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6328 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6329 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6330 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6331 (arm_pad_arg_upward): Replace with...
6332 (arm_function_arg_padding): ...this new function.
6333 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6335 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6336 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6337 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6338 (ia64_hpux_function_arg_padding): Replace with...
6339 (ia64_function_arg_padding): ...this new function. Use pad_direction
6340 instead of direction. Check for TARGET_HPUX.
6341 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6342 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6343 (iq2000_function_arg_padding): New function.
6344 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6345 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6346 (mips_function_arg_padding): ...this new function.
6347 (mips_pad_reg_upward): Update accordingly.
6348 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6349 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6350 targetm.calls.function_arg_padding.
6351 (FUNCTION_ARG_PADDING): Delete.
6352 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6353 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6354 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6355 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6356 (nios2_block_reg_padding): Return pad_direction instead of direction.
6357 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6358 instead of direction.
6359 (nios2_function_arg_padding): Likewise. Make static.
6360 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6361 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6362 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6363 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6364 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6365 (pa_function_arg_padding): Make static. Return pad_direction instead
6367 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6368 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6369 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6370 instead of direction. Use targetm.calls.function_arg_padding.
6371 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6372 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6373 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6374 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6375 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6377 (function_arg_padding): Rename to...
6378 (rs6000_function_arg_padding): ...this. Make static. Return
6379 pad_direction instead of direction.
6380 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6381 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6382 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6383 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6384 instead of direction. Use targetm.calls.function_arg_padding.
6385 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6386 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6387 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6388 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6389 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6390 (function_arg_padding): Rename to...
6391 (rs6000_function_arg_padding): ...this. Make static. Return
6392 pad_direction instead of direction.
6393 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6394 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6395 * config/s390/s390.c (s390_function_arg_padding): New function.
6396 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6397 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6398 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6399 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6400 (function_arg_padding): Rename to...
6401 (sparc_function_arg_padding): ...this. Make static. Return
6402 pad_direction instead of direction.
6403 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6404 * config/spu/spu.c (spu_function_arg_padding): New function.
6405 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6406 * system.h (FUNCTION_ARG_PADDING): Poison.
6408 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6409 Alan Hayward <alan.hayward@arm.com>
6410 David Sherwood <david.sherwood@arm.com>
6412 * target.def (modes_tieable_p): New hook.
6413 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6414 (TARGET_MODES_TIEABLE_P): ...this.
6415 * doc/tm.texi.in: Regenerate.
6416 * hooks.h (hook_bool_mode_mode_true): Declare.
6417 * hooks.c (hook_bool_mode_mode_true): New function.
6418 * combine.c (subst): Use targetm.modes_tieable_p instead of
6420 * dse.c (find_shift_sequence): Likewise.
6421 * expmed.c (extract_low_bits): Likewise.
6422 * lower-subreg.c: Include target.h.
6423 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6425 * rtlanal.c (rtx_cost): Likewise.
6426 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6427 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6428 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6429 (TARGET_MODES_TIEABLE_P): Redefine.
6430 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6431 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6432 (TARGET_MODES_TIEABLE_P): Redefine.
6433 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6434 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6435 (arc_modes_tieable_p): New function.
6436 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6437 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6438 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6439 (arm_modes_tieable_p): Make static.
6440 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6441 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6442 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6443 (TARGET_MODES_TIEABLE_P): Redefine.
6444 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6445 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6446 (TARGET_MODES_TIEABLE_P): Redefine.
6447 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6448 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6449 (cr16_modes_tieable_p): New function.
6450 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6451 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6452 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6453 (TRULY_NOOP_TRUNCATION): Update comment.
6454 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6455 (TRULY_NOOP_TRUNCATION): Update comment.
6456 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6457 (frv_modes_tieable_p): New function.
6458 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6459 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6460 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6461 (TARGET_MODES_TIEABLE_P): Redefine.
6462 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6463 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6464 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6465 (TARGET_MODES_TIEABLE_P): Redefine.
6466 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6467 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6468 (ia64_modes_tieable_p): New function.
6469 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6470 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6471 (iq2000_modes_tieable_p): New function.
6472 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6473 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6474 (lm32_modes_tieable_p): New function.
6475 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6476 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6477 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6478 (TARGET_MODES_TIEABLE_P): Redefine.
6479 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6480 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6481 (m32r_modes_tieable_p): New function.
6482 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6483 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6484 (m68k_modes_tieable_p): New function.
6485 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6486 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6487 (mcore_modes_tieable_p): New function.
6488 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6489 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6491 (TARGET_MODES_TIEABLE_P): Redefine.
6492 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6493 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6494 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6495 (TARGET_MODES_TIEABLE_P): Redefine.
6496 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6497 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6498 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6499 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6500 (mn10300_modes_tieable_p): ...this and make static.
6501 (TARGET_MODES_TIEABLE_P): Redefine.
6502 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6503 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6504 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6505 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6506 (msp430_modes_tieable_p): Make static.
6507 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6508 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6509 (TARGET_MODES_TIEABLE_P): Redefine.
6510 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6511 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6512 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6513 (TARGET_MODES_TIEABLE_P): Redefine.
6514 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6515 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6516 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6517 (TARGET_MODES_TIEABLE_P): Redefine.
6518 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6519 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6520 (pdp11_modes_tieable_p): New function.
6521 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6522 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6523 (rs6000_modes_tieable_p): New function.
6524 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6525 * config/powerpcspe/powerpcspe.md: Update comment.
6526 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6527 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6528 (TARGET_MODES_TIEABLE_P): Redefine.
6529 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6530 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6531 (rl78_modes_tieable_p): New function.
6532 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6533 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6534 (rs6000_modes_tieable_p): New function.
6535 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6536 * config/rs6000/rs6000.md: Update comment.
6537 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6538 * config/rx/rx.c (rx_modes_tieable_p): New function.
6539 (TARGET_MODES_TIEABLE_P): Redefine.
6540 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6541 * config/s390/s390.c (s390_modes_tieable_p): New function.
6542 (TARGET_MODES_TIEABLE_P): Redefine.
6543 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6544 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6545 (sh_modes_tieable_p): New function.
6546 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6547 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6548 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6549 (sparc_modes_tieable_p): Make static.
6550 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6551 * config/spu/spu.c (spu_modes_tieable_p): New function.
6552 (TARGET_MODES_TIEABLE_P): Redefine.
6553 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6554 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6555 (TARGET_MODES_TIEABLE_P): Redefine.
6556 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6557 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6558 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6559 * config/v850/v850.c (v850_modes_tieable_p): New function.
6560 (TARGET_MODES_TIEABLE_P): Redefine.
6561 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6562 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6563 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6564 (visium_modes_tieable_p): New function.
6565 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6566 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6567 (xtensa_modes_tieable_p): New function.
6568 * system.h (MODES_TIEABLE_P): Poison.
6570 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6571 Alan Hayward <alan.hayward@arm.com>
6572 David Sherwood <david.sherwood@arm.com>
6574 * target.def (hard_regno_mode_ok): New hook.
6575 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6576 (TARGET_HARD_REGNO_MODE_OK): ...this.
6577 * doc/tm.texi.in: Regenerate.
6578 * hooks.h (hook_bool_uint_mode_true): Declare.
6579 * hooks.c (hook_bool_uint_mode_true): New function.
6580 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6582 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6583 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6584 instead of HARD_REGNO_MODE_OK.
6585 * caller-save.c: Include target.h.
6586 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6588 * combine.c (can_combine_p): Likewise.
6589 (combinable_i3pat): Likewise.
6590 (can_change_dest_mode): Likewise.
6591 * expr.c (init_expr_target): Likewise.
6592 (convert_move): Likewise.
6593 (convert_modes): Likewise.
6594 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6595 (setup_prohibited_mode_move_regs): Likewise.
6596 * ira.h (target_ira): Likewise.
6597 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6598 * lra-constraints.c (process_alt_operands): Likewise.
6599 (split_reg): Likewise.
6600 * recog.c (peep2_find_free_register): Likewise.
6601 * ree.c (combine_reaching_defs): Likewise.
6602 * regcprop.c (maybe_mode_change): Likewise.
6603 * reginfo.c (init_reg_sets_1): Likewise.
6604 (choose_hard_reg_mode): Likewise.
6605 (simplifiable_subregs): Likewise.
6606 * regrename.c (check_new_reg_p): Likewise.
6607 * reload.c (find_valid_class): Likewise.
6608 (find_valid_class_1): Likewise.
6609 (reload_inner_reg_of_subreg): Likewise.
6610 (push_reload): Likewise.
6611 (combine_reloads): Likewise.
6612 (find_dummy_reload): Likewise.
6613 (find_reloads): Likewise.
6614 * reload1.c (find_reg): Likewise.
6615 (set_reload_reg): Likewise.
6616 (allocate_reload_reg): Likewise.
6617 (choose_reload_regs): Likewise.
6618 (reload_adjust_reg_for_temp): Likewise.
6619 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6620 (simplify_subreg_regno): Likewise.
6621 * sel-sched.c (init_regs_for_mode): Likewise.
6622 * varasm.c (make_decl_rtl): Likewise.
6623 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6624 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6626 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6627 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6628 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6629 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6630 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6632 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6633 (arc_mode_class): Delete.
6634 (HARD_REGNO_MODE_OK): Delete.
6635 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6636 (arc_hard_regno_mode_ok): Rename old array to...
6637 (arc_hard_regno_mode_ok_modes): ...this.
6638 (arc_conditional_register_usage): Update accordingly.
6639 (arc_mode_class): Make static.
6640 (arc_hard_regno_mode_ok): New function.
6641 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6642 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6643 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6644 (arm_hard_regno_mode_ok): Make static.
6645 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6647 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6648 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6649 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6652 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6653 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6654 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6655 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6657 * config/bfin/predicates.md (valid_reg_operand): Use
6658 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6659 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6660 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6662 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6663 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6664 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6665 (cr16_hard_regno_mode_ok): Make static and return a bool.
6666 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6667 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6668 (cris_hard_regno_mode_ok): New function.
6669 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6670 (epiphany_mode_class): Delete.
6671 (HARD_REGNO_MODE_OK): Delete.
6672 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6673 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6674 (hard_regno_mode_ok): Rename to...
6675 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6676 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6677 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6679 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6680 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6681 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6682 (frv_hard_regno_mode_ok): Make static and return a bool.
6683 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6685 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6686 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6687 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6688 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6691 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6692 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6693 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6696 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6697 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6698 (ia64_hard_regno_mode_ok): New function.
6699 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6700 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6701 (iq2000_hard_regno_mode_ok): New function.
6702 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6703 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6704 (lm32_hard_regno_mode_ok): New function.
6705 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6706 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6707 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6708 instead of HARD_REGNO_MODE_OK.
6709 (m32c_hard_regno_ok): Rename to...
6710 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6711 (m32c_cannot_change_mode_class): Update accordingly.
6712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6713 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6714 (m32r_mode_class): Delete.
6715 (HARD_REGNO_MODE_OK): Delete.
6716 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6717 (m32r_hard_regno_mode_ok): Rename to...
6718 (m32r_hard_regno_modes): ...this.
6719 (m32r_mode_class): Make static.
6720 (m32r_hard_regno_mode_ok): New function.
6721 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6722 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6723 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6724 (m68k_hard_regno_mode_ok): Make static.
6725 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6726 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6727 (mcore_hard_regno_mode_ok): New function.
6728 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6729 (HARD_REGNO_MODE_OK): Delete.
6730 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6732 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6733 (microblaze_hard_regno_mode_ok): New function.
6734 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6735 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6736 (mips_hard_regno_mode_ok): Delete.
6737 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6738 (mips_hard_regno_mode_ok_p): ...this and make static.
6739 (mips_hard_regno_mode_ok_p): Rename to...
6740 (mips_hard_regno_mode_ok_uncached): ...this.
6741 (mips_hard_regno_mode_ok): New function.
6742 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6743 of HARD_REGNO_MODE_OK.
6744 (mips_option_override): Update after above name changes.
6745 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6746 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6747 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6748 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6749 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6750 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6751 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6752 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6753 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6754 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6755 (msp430_hard_regno_mode_ok): Make static and return a bool.
6756 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6757 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6758 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6761 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6762 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6763 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6764 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6765 (PA_HARD_REGNO_MODE_OK): ...this
6766 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6767 (PA_HARD_REGNO_MODE_OK): ...this.
6768 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6769 (pa_hard_regno_mode_ok): New function.
6770 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6771 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6772 (pdp11_hard_regno_mode_ok): New function.
6773 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6774 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6776 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6779 (rs6000_hard_regno_mode_ok): Rename to...
6780 (rs6000_hard_regno_mode_ok_uncached): ...this.
6781 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6782 (rs6000_hard_regno_mode_ok): New function.
6783 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6784 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6785 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6786 (riscv_hard_regno_mode_ok): ...this and make static.
6787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6788 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6789 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6790 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6791 (rl78_hard_regno_mode_ok): Make static and return bool.
6792 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6793 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6795 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6797 (rs6000_hard_regno_mode_ok): Rename to...
6798 (rs6000_hard_regno_mode_ok_uncached): ...this.
6799 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6800 (rs6000_hard_regno_mode_ok): New function.
6801 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6802 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6803 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6804 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6805 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6806 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6807 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6808 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6809 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6810 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6811 (sh_hard_regno_mode_ok): Make static.
6812 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6813 instead of HARD_REGNO_MODE_OK.
6814 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6815 (sparc_mode_class): Delete.
6816 (HARD_REGNO_MODE_OK): Delete.
6817 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6818 (hard_regno_mode_classes): Make static.
6819 (sparc_mode_class): Likewise.
6820 (sparc_hard_regno_mode_ok): New function.
6821 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6822 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6823 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6826 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6827 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6828 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6829 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6831 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6832 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6833 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6834 (visium_hard_regno_mode_ok): New function.
6835 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6836 instead of HARD_REGNO_MODE_OK.
6837 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6838 (HARD_REGNO_MODE_OK): Delete.
6839 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6840 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6841 (xtensa_option_override): Update accordingly.
6842 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6843 (xtensa_hard_regno_mode_ok): New function.
6844 * system.h (HARD_REGNO_MODE_OK): Poison.
6846 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6847 Alan Hayward <alan.hayward@arm.com>
6848 David Sherwood <david.sherwood@arm.com>
6850 * target.def (hard_regno_call_part_clobbered): New hook.
6851 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6852 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6853 * doc/tm.texi: Regenerate.
6854 * hooks.h (hook_bool_uint_mode_false): Declare.
6855 * hooks.c (hook_bool_uint_mode_false): New function.
6856 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6857 * cselib.c (cselib_process_insn): Use
6858 targetm.hard_regno_call_part_clobbered instead of
6859 HARD_REGNO_CALL_PART_CLOBBERED.
6860 * ira-conflicts.c (ira_build_conflicts): Likewise.
6861 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6862 * lra-constraints.c (need_for_call_save_p): Likewise.
6863 * lra-lives.c: Include target.h.
6864 (check_pseudos_live_through_calls): Use
6865 targetm.hard_regno_call_part_clobbered instead of
6866 HARD_REGNO_CALL_PART_CLOBBERED.
6867 * regcprop.c: Include target.h.
6868 (copyprop_hardreg_forward_1): Use
6869 targetm.hard_regno_call_part_clobbered instead of
6870 HARD_REGNO_CALL_PART_CLOBBERED.
6871 * reginfo.c (choose_hard_reg_mode): Likewise.
6872 * regrename.c (check_new_reg_p): Likewise.
6873 * reload.c (find_equiv_reg): Likewise.
6874 * reload1.c (emit_reload_insns): Likewise.
6875 * sched-deps.c (deps_analyze_insn): Likewise.
6876 * sel-sched.c (init_regs_for_mode): Likewise.
6877 (mark_unavailable_hard_regs): Likewise.
6878 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6879 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6880 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6882 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6883 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6884 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6886 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6888 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6889 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6890 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6892 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6893 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6894 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6896 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6897 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6899 * config/powerpcspe/powerpcspe.c
6900 (rs6000_hard_regno_call_part_clobbered): New function.
6901 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6902 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6903 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6905 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6906 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6907 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6909 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6910 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6911 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6913 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6914 Alan Hayward <alan.hayward@arm.com>
6915 David Sherwood <david.sherwood@arm.com>
6917 * rtl.h (subreg_memory_offset): Declare.
6918 * emit-rtl.c (subreg_memory_offset): New function.
6919 * expmed.c (store_bit_field_1): Use it.
6920 * expr.c (undefined_operand_subword_p): Likewise.
6921 * simplify-rtx.c (simplify_subreg): Likewise.
6923 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6925 PR rtl-optimization/57448
6928 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6929 using atomic_load pattern.
6930 (expand_atomic_store): Likewise.
6932 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6935 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6936 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6937 handling. Use replace_call_with_value with NULL instead of
6938 gsi_replace, unlink_stmt_vdef and release_defs.
6940 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6943 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6945 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6948 * rtl.h (emit_library_call_value_1): Declare.
6949 (emit_library_call): Replace declaration with a series of overloads.
6950 Remove the parameter count argument.
6951 (emit_library_call_value): Likewise.
6952 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6953 with an "rtx_mode_t *".
6954 (emit_library_call_value): Delete.
6955 (emit_library_call): Likewise.
6956 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6957 (asan_emit_allocas_unpoison): Likewise.
6958 * builtins.c (expand_builtin_powi): Likewise.
6959 (expand_asan_emit_allocas_unpoison): Likewise.
6960 * cfgexpand.c (expand_main_function): Likewise.
6961 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6962 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6963 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6964 * config/arm/arm.c (arm_trampoline_init): Likewise.
6965 (arm_call_tls_get_addr): Likewise.
6966 (arm_expand_divmod_libfunc): Likewise.
6967 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6968 (smulsi3_highpart): Likewise.
6969 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6970 (c6x_expand_compare): Likewise.
6971 (c6x_expand_movmem): Likewise.
6972 * config/frv/frv.c (frv_trampoline_init): Likewise.
6973 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6974 (ix86_expand_divmod_libfunc): Likewise.
6975 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6976 (ia64_expand_compare): Likewise.
6977 (ia64_profile_hook): Likewise.
6978 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6979 (nonlocal_goto): Likewise.
6980 (restore_stack_nonlocal): Likewise.
6981 * config/m32r/m32r.c (block_move_call): Likewise.
6982 (m32r_trampoline_init): Likewise.
6983 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6984 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6985 (m68k_call_m68k_read_tp): Likewise.
6986 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6987 (microblaze_expand_divide): Likewise.
6988 * config/mips/mips.h (mips_args): Likewise.
6989 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6990 (MIPS_ICACHE_SYNC): Likewise.
6991 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6992 (nios2_trampoline_init): Likewise.
6993 * config/pa/pa.c (hppa_tls_call): Likewise.
6994 (pa_trampoline_init): Likewise.
6995 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6996 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6997 (expand_strn_compare): Likewise.
6998 (rs6000_generate_compare): Likewise.
6999 (rs6000_expand_float128_convert): Likewise.
7000 (output_profile_hook): Likewise.
7001 (rs6000_trampoline_init): Likewise.
7002 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7003 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7004 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7005 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7006 (rs6000_generate_compare): Likewise.
7007 (rs6000_expand_float128_convert): Likewise.
7008 (output_profile_hook): Likewise.
7009 (rs6000_trampoline_init): Likewise.
7010 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7011 * config/sh/sh.c (sh_trampoline_init): Likewise.
7012 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7013 (sparc_emit_float_lib_cmp): Likewise.
7014 (sparc32_initialize_trampoline): Likewise.
7015 (sparc64_initialize_trampoline): Likewise.
7016 (sparc_profile_hook): Likewise.
7017 * config/spu/spu.c (ea_load_store): Likewise.
7018 * config/spu/spu.md (floatunssidf2): Likewise.
7019 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7020 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7021 * config/visium/visium.c (expand_block_move_4): Likewise.
7022 (expand_block_move_2): Likewise.
7023 (expand_block_move_1): Likewise.
7024 (expand_block_set_4): Likewise.
7025 (expand_block_set_2): Likewise.
7026 (expand_block_set_1): Likewise.
7027 (visium_trampoline_init): Likewise.
7028 (visium_profile_hook): Likewise.
7029 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7030 (xtensa_setup_frame_addresses): Likewise.
7031 (xtensa_trampoline_init): Likewise.
7032 * except.c (sjlj_emit_function_enter): Likewise.
7033 (sjlj_emit_function_exit): Likewise.
7034 * explow.c (allocate_dynamic_stack_space): Likewise.
7035 (probe_stack_range): Likewise.
7036 * expr.c (convert_mode_scalar): Likewise.
7037 * optabs.c (expand_binop): Likewise.
7038 (expand_twoval_binop_libfunc): Likewise.
7039 (expand_unop): Likewise.
7040 (prepare_cmp_insn): Likewise.
7041 (prepare_float_lib_cmp): Likewise.
7042 (expand_float): Likewise.
7043 (expand_fix): Likewise.
7044 (expand_fixed_convert): Likewise.
7045 (maybe_emit_sync_lock_test_and_set): Likewise.
7046 (expand_atomic_compare_and_swap): Likewise.
7047 (expand_mem_thread_fence): Likewise.
7048 (expand_atomic_fetch_op): Likewise.
7050 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7052 * doc/generic.texi (OpenACC): Adjust URL.
7053 * doc/invoke.texi (C Dialect Options): Ditto.
7055 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7057 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7058 predicate for operand 1. Add (m,<S>) constraint.
7059 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7060 Prevent memory operand 1 with register operand 2.
7062 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7064 PR rtl-optimization/82024
7065 * combine.c (try_combine): If the combination result is a PARALLEL,
7066 and we only need to retain the SET in there that would be placed
7067 at I2, check that we can place that at I3 instead, before doing so.
7069 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7072 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7074 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7075 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7078 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7079 Jeff Law <law@redhat.com>
7081 * varasm.c (bss_initializer_p): Do not put constants into .bss
7082 (categorize_decl_for_section): Handle bss_initializer_p returning
7083 false when DECL_INITIAL is NULL.
7085 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7088 * config/s390/s390.c (s390_can_inline_p): New function.
7090 2017-09-01 Jeff Law <law@redhat.com>
7092 PR tree-optimization/82052
7093 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7094 Always initialize the returned slot after a hash table miss
7095 when INSERT is true.
7097 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7099 * config/s390/s390.md (mem_signal_fence): Remove.
7100 * doc/md.texi (mem_signal_fence): Remove.
7101 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7103 * target-insns.def (mem_signal_fence): Remove.
7105 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7108 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7111 * asan.c (create_odr_indicator): Strip name encoding from assembler
7112 name before appending it after __odr_asan_.
7114 2017-09-01 Martin Liska <mliska@suse.cz>
7116 PR tree-optimization/82059
7117 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7118 frequency only when an edge is redirected.
7120 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7122 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7123 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7124 (arc_conditional_register_usage): Remove ARC600 lp_count
7126 (arc_file_start): Emit Tag_ARC_CPU_variation.
7127 (arc_can_use_doloop_p): New conditions to use ZOLs.
7128 (hwloop_fail): New function.
7129 (hwloop_optimize): Likewise.
7130 (hwloop_pattern_reg): Likewise.
7131 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7132 (arc_reorg_loops): New function, calls reorg_loops.
7133 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7134 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7136 (arc_loop_hazard): Remove function, functionality moved into
7138 (arc_hazard): Remove arc_loop_hazard call.
7139 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7140 into hwloop_optimize.
7141 (arc_label_align): Remove ZOL handling.
7142 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7143 * config/arc/arc.md (doloop_begin): Remove pattern.
7144 (doloop_begin_i): Likewise.
7145 (doloop_end_i): Likewise.
7146 (doloop_fallback): Likewise.
7147 (doloop_fallback_m): Likewise.
7148 (doloop_end): Reimplement expand.
7149 (arc_lp): New pattern for LP instruction.
7150 (loop_end): New pattern.
7151 (loop_fail): Likewise.
7152 (decrement_and_branch_until_zero): Likewise.
7153 * config/arc/arc.opt (mlpc-width): New option.
7154 * doc/invoke.texi (mlpc-width): Document option.
7156 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7158 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7159 (arc_ccfsm_advance): Fix checking for delay slots.
7160 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7162 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7164 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7166 (movhi_insn): Update store instruction constraint which are saving
7167 6-bit short immediates.
7168 (movsi_insn): Consider also short scaled load operations.
7169 (zero_extendhisi2_i): Use Usd constraint instead of T.
7170 (extendhisi2_i): Add q constraint.
7171 (arc_clzsi2): Add type and length attributes.
7172 (arc_ctzsi2): Likewise.
7173 * config/arc/constraints.md (Usc): Update constraint, the
7174 assembler can parse two relocations for a single instruction.
7176 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7178 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7179 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7181 2017-08-31 Olivier Hainque <hainque@adacore.com>
7183 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7184 match as powerpc-wrs-vxworks*.
7186 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7188 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7189 register constraint for by-element operand.
7190 (aarch64_mls_elt_merge<mode>): Likewise.
7192 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7194 * config/arc/arc.c (arc_can_follow_jump): Check for short
7197 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7199 * config.gcc: Use g.opt for arc.
7200 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7201 functionality moved to ...
7202 (legitimate_scaled_address_p): New function, ...here.
7203 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7204 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7205 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7207 (arc_override_options): Handle G option.
7208 (arc_output_pic_addr_const): Correct function definition.
7209 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7210 (arc_decl_anon_ns_mem_p): Delete.
7211 (arc_in_small_data_p): Overhaul this function to take into
7212 consideration the value given via G option.
7213 (arc_rewrite_small_data_1): Renamed and corrected old
7214 arc_rewrite_small_data function.
7215 (arc_rewrite_small_data): New function.
7216 (small_data_pattern): Don't use pic_offset_table_rtx.
7217 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7218 * config/arc/simdext.md (movmisalignv2hi): Use
7219 prepare_move_operands function.
7221 (movmisalign*): Likewise.
7222 * doc/invoke.texi (ARC options): Document -G option.
7224 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7226 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7228 * config/arc/arc.c (arc_print_operand): Output scalled address for
7229 sdata whenever is possible.
7230 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7231 load/stores are available.
7232 (compact_sda_memory_operand): Check for the alignment required by
7233 code density instructions.
7234 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7236 * config/arc/constraints.md (Usd): Update constraint.
7237 (Us0): New constraint.
7238 (Usc): Update constraint.
7240 2017-08-31 Richard Biener <rguenther@suse.de>
7243 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7246 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7248 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7249 Resize type_signature.
7251 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7252 Alan Hayward <alan.hayward@arm.com>
7253 David Sherwood <david.sherwood@arm.com>
7255 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7256 subregs whose inner modes can be stored in GPRs.
7257 (aarch64_classify_index): Likewise.
7259 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7260 Alan Hayward <alan.hayward@arm.com>
7261 David Sherwood <david.sherwood@arm.com>
7263 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7264 (V_INT_EQUIV): ...this.
7265 (v_cmp_result): Rename to...
7266 (v_int_equiv): ...this.
7267 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7268 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7269 (copysign<mode>3): Likewise.
7270 (aarch64_simd_bsl<mode>_internal): Likewise.
7271 (aarch64_simd_bsl<mode>): Likewise.
7272 (vec_cmp<mode><mode>): Likewise.
7273 (vcond<mode><mode>): Likewise.
7274 (vcond<v_cmp_mixed><mode>): Likewise.
7275 (vcondu<mode><v_cmp_mixed>): Likewise.
7276 (aarch64_cm<optab><mode>): Likewise.
7277 (aarch64_cmtst<mode>): Likewise.
7278 (aarch64_fac<optab><mode>): Likewise.
7279 (vec_perm_const<mode>): Likewise.
7280 (vcond_mask_<mode><v_cmp_result>): Rename to...
7281 (vcond_mask_<mode><v_int_equiv>): ...this.
7282 (vec_cmp<mode><v_cmp_result>): Rename to...
7283 (vec_cmp<mode><v_int_equiv>): ...this.
7285 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7286 Alan Hayward <alan.hayward@arm.com>
7287 David Sherwood <david.sherwood@arm.com>
7289 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7291 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7292 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7293 (UNSPEC_LD4_DREG): New unspecs.
7294 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7295 (aarch64_ld2<mode>_dreg_be): Replace with...
7296 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7298 (aarch64_ld3<mode>_dreg_le)
7299 (aarch64_ld3<mode>_dreg_be): Replace with...
7300 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7302 (aarch64_ld4<mode>_dreg_le)
7303 (aarch64_ld4<mode>_dreg_be): Replace with...
7304 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7307 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7309 PR tree-optimization/81987
7310 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7311 insert an initializer in a location not dominated by the stride
7314 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7316 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7317 on the entire header of the finally block in the fallthru case.
7319 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7321 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7323 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7325 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7326 rs6000_emit_move_from_cr and call renamed function.
7327 (rs6000_emit_prologue): Call renamed functions.
7328 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7329 movesi_from_cr, remove volatile CRs.
7331 2017-08-30 Jon Beniston <jon@beniston.com>
7332 Richard Biener <rguenther@suse.de>
7334 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7335 of VECTOR_MODE_P check.
7336 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7337 element vector types.
7339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7341 * df.h (df_read_modify_subreg_p): Remove in favor of...
7342 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7343 const_rtx instead of an rtx.
7344 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7345 * df-problems.c (df_word_lr_mark_ref): Likewise.
7346 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7347 (mark_pseudo_reg_dead): Likewise.
7348 (mark_ref_dead): Likewise.
7349 * reginfo.c (init_subregs_of_mode): Likewise.
7350 * sched-deps.c (sched_analyze_1): Likewise.
7351 * df-scan.c (df_def_record_1): Likewise.
7352 (df_uses_record): Likewise.
7353 (df_read_modify_subreg_p): Remove in favor of...
7354 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7355 const_rtx instead of an rtx.
7357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7358 Alan Hayward <alan.hayward@arm.com>
7359 David Sherwood <david.sherwood@arm.com>
7361 * rtl.h (partial_subreg_p): New function.
7362 * caller-save.c (save_call_clobbered_regs): Use it.
7363 * calls.c (expand_call): Likewise.
7364 * combine.c (combinable_i3pat): Likewise.
7365 (simplify_set): Likewise.
7366 (make_extraction): Likewise.
7367 (make_compound_operation_int): Likewise.
7368 (gen_lowpart_or_truncate): Likewise.
7369 (force_to_mode): Likewise.
7370 (make_field_assignment): Likewise.
7371 (reg_truncated_to_mode): Likewise.
7372 (record_truncated_value): Likewise.
7373 (move_deaths): Likewise.
7374 * cse.c (record_jump_cond): Likewise.
7375 (cse_insn): Likewise.
7376 * cselib.c (cselib_lookup_1): Likewise.
7377 * expmed.c (extract_bit_field_using_extv): Likewise.
7378 * function.c (assign_parm_setup_reg): Likewise.
7379 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7380 * ira-build.c (create_insn_allocnos): Likewise.
7381 * lra-coalesce.c (merge_pseudos): Likewise.
7382 * lra-constraints.c (match_reload): Likewise.
7383 (simplify_operand_subreg): Likewise.
7384 (curr_insn_transform): Likewise.
7385 * lra-lives.c (process_bb_lives): Likewise.
7386 * lra.c (new_insn_reg): Likewise.
7387 (lra_substitute_pseudo): Likewise.
7388 * regcprop.c (mode_change_ok): Likewise.
7389 (maybe_mode_change): Likewise.
7390 (copyprop_hardreg_forward_1): Likewise.
7391 * reload.c (push_reload): Likewise.
7392 (find_reloads): Likewise.
7393 (find_reloads_subreg_address): Likewise.
7394 * reload1.c (alter_reg): Likewise.
7395 (eliminate_regs_1): Likewise.
7396 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7398 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7400 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7401 back to if statements, including unpack.
7403 2017-08-30 Martin Liska <mliska@suse.cz>
7406 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7408 (func_checker::compare_asm_inputs_outputs): ... this function.
7409 (func_checker::compare_gimple_asm): Use the function to compare
7410 also ASM constrains.
7411 * ipa-icf-gimple.h: Rename the function.
7413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7414 Alan Hayward <alan.hayward@arm.com>
7415 David Sherwood <david.sherwood@arm.com>
7417 * coretypes.h (complex_mode): New type.
7418 * gdbhooks.py (build_pretty_printer): Handle it.
7419 * machmode.h (complex_mode): New class.
7420 (complex_mode::includes_p): New function.
7421 (is_complex_int_mode): Likewise.
7422 (is_complex_float_mode): Likewise.
7423 * genmodes.c (get_mode_class): Handle complex mode classes.
7424 * function.c (expand_function_end): Use is_complex_int_mode.
7426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7427 Alan Hayward <alan.hayward@arm.com>
7428 David Sherwood <david.sherwood@arm.com>
7430 * coretypes.h (scalar_mode_pod): New typedef.
7431 * gdbhooks.py (build_pretty_printer): Handle it.
7432 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7433 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7434 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7435 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7439 Alan Hayward <alan.hayward@arm.com>
7440 David Sherwood <david.sherwood@arm.com>
7442 * machmode.h (mode_for_vector): Take a scalar_mode instead
7444 * stor-layout.c (mode_for_vector): Likewise.
7445 * explow.c (promote_mode): Use as_a <scalar_mode>.
7446 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7449 Alan Hayward <alan.hayward@arm.com>
7450 David Sherwood <david.sherwood@arm.com>
7452 * target.def (preferred_simd_mode): Take a scalar_mode
7453 instead of a machine_mode.
7454 * targhooks.h (default_preferred_simd_mode): Likewise.
7455 * targhooks.c (default_preferred_simd_mode): Likewise.
7456 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7457 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7458 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7459 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7460 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7461 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7462 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7463 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7465 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7466 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7467 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7468 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7469 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7470 * doc/tm.texi: Regenerate.
7471 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7475 Alan Hayward <alan.hayward@arm.com>
7476 David Sherwood <david.sherwood@arm.com>
7478 * target.def (scalar_mode_supported_p): Take a scalar_mode
7479 instead of a machine_mode.
7480 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7481 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7482 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7483 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7484 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7485 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7486 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7487 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7488 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7489 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7490 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7491 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7492 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7493 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7495 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7496 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7497 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7498 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7499 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7501 * doc/tm.texi: Regenerate.
7503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7504 Alan Hayward <alan.hayward@arm.com>
7505 David Sherwood <david.sherwood@arm.com>
7507 * coretypes.h (opt_scalar_mode): New typedef.
7508 * gdbhooks.py (build_pretty_printers): Handle it.
7509 * machmode.h (mode_iterator::get_2xwider): Add overload for
7511 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7513 * expr.c (convert_mode_scalar): Likewise.
7514 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7515 * optabs.c (expand_float): Likewise.
7516 (expand_fix): Likewise.
7517 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7520 Alan Hayward <alan.hayward@arm.com>
7521 David Sherwood <david.sherwood@arm.com>
7523 * optabs.c (expand_float): Explicitly check for scalars before
7524 using a branching expansion.
7525 (expand_fix): Likewise.
7527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7528 Alan Hayward <alan.hayward@arm.com>
7529 David Sherwood <david.sherwood@arm.com>
7531 * expr.c (convert_mode): Split scalar handling out into...
7532 (convert_mode_scalar): ...this new function. Treat the modes
7535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7536 Alan Hayward <alan.hayward@arm.com>
7537 David Sherwood <david.sherwood@arm.com>
7539 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7541 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7548 rather than a machine_mode.
7549 (fixed_from_string): Likewise.
7550 (fixed_convert): Likewise.
7551 (fixed_convert_from_int): Likewise.
7552 (fixed_convert_from_real): Likewise.
7553 (real_convert_from_fixed): Likewise.
7554 * fixed-value.c (fixed_from_double_int): Likewise.
7555 (fixed_from_string): Likewise.
7556 (fixed_convert): Likewise.
7557 (fixed_convert_from_int): Likewise.
7558 (fixed_convert_from_real): Likewise.
7559 (real_convert_from_fixed): Likewise.
7560 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7563 Alan Hayward <alan.hayward@arm.com>
7564 David Sherwood <david.sherwood@arm.com>
7566 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7567 of separate mode class checks. Do not allow vector modes here.
7568 (immed_wide_int_const): Use as_a <scalar_mode>.
7569 * explow.c (trunc_int_for_mode): Likewise.
7570 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7571 (wi::shwi): Likewise.
7572 (wi::min_value): Likewise.
7573 (wi::max_value): Likewise.
7574 * dwarf2out.c (loc_descriptor): Likewise.
7575 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7579 Alan Hayward <alan.hayward@arm.com>
7580 David Sherwood <david.sherwood@arm.com>
7582 * tree.h (SCALAR_TYPE_MODE): New macro.
7583 * expr.c (expand_expr_addr_expr_1): Use it.
7584 (expand_expr_real_2): Likewise.
7585 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7586 (fold_convert_const_fixed_from_int): Likewise.
7587 (fold_convert_const_fixed_from_real): Likewise.
7588 (native_encode_fixed): Likewise
7589 (native_encode_complex): Likewise
7590 (native_encode_vector): Likewise.
7591 (native_interpret_fixed): Likewise.
7592 (native_interpret_real): Likewise.
7593 (native_interpret_complex): Likewise.
7594 (native_interpret_vector): Likewise.
7595 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7596 (simd_clone_adjust_argument_types): Likewise.
7597 (simd_clone_init_simd_arrays): Likewise.
7598 (simd_clone_adjust): Likewise.
7599 * stor-layout.c (layout_type): Likewise.
7600 * tree.c (build_minus_one_cst): Likewise.
7601 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7602 * tree-inline.c (estimate_move_cost): Likewise.
7603 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7604 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7605 (vectorizable_reduction): Likewise.
7606 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7607 (vect_recog_mixed_size_cond_pattern): Likewise.
7608 (check_bool_pattern): Likewise.
7609 (adjust_bool_pattern): Likewise.
7610 (search_type_for_mask_1): Likewise.
7611 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7612 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7613 (vectorizable_load): Likewise.
7614 (vectorizable_store): Likewise.
7615 * ubsan.c (ubsan_encode_value): Likewise.
7616 * varasm.c (output_constant): Likewise.
7618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7619 Alan Hayward <alan.hayward@arm.com>
7620 David Sherwood <david.sherwood@arm.com>
7622 * coretypes.h (scalar_mode): New class.
7623 * machmode.h (scalar_mode): Likewise.
7624 (scalar_mode::includes_p): New function.
7625 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7626 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7627 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7628 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7629 * expmed.c (store_bit_field_1): Likewise.
7630 (extract_bit_field_1): Likewise.
7631 * expr.c (write_complex_part): Likewise.
7632 (read_complex_part): Likewise.
7633 (emit_move_complex_push): Likewise.
7634 (expand_expr_real_2): Likewise.
7635 * function.c (assign_parm_setup_reg): Likewise.
7636 (assign_parms_unsplit_complex): Likewise.
7637 * optabs.c (expand_binop): Likewise.
7638 * rtlanal.c (subreg_get_info): Likewise.
7639 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7640 * varasm.c (output_constant_pool_2): Likewise.
7642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7643 Alan Hayward <alan.hayward@arm.com>
7644 David Sherwood <david.sherwood@arm.com>
7646 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7648 (expmed_mult_highpart_optab): Likewise.
7649 (expmed_mult_highpart): Likewise.
7651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7652 Alan Hayward <alan.hayward@arm.com>
7653 David Sherwood <david.sherwood@arm.com>
7655 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7656 instead of a machine_mode.
7657 (builtin_memset_read_str): Likewise.
7658 * builtins.c (c_readstr): Likewise.
7659 (builtin_memcpy_read_str): Likewise.
7660 (builtin_strncpy_read_str): Likewise.
7661 (builtin_memset_read_str): Likewise.
7662 (builtin_memset_gen_str): Likewise.
7663 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7664 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7665 instead of a machine_mode.
7666 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7668 (make_extraction): Likewise.
7669 (try_widen_shift_mode): Take and return scalar_int_modes instead
7671 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7672 a scalar_int_mode instead of a machine_mode.
7673 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7674 (avr_addr_space_pointer_mode): Likewise.
7675 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7676 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7677 (msp430_unwind_word_mode): Likewise.
7678 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7679 (spu_addr_space_pointer_mode): Likewise.
7680 (spu_addr_space_address_mode): Likewise.
7681 (spu_libgcc_cmp_return_mode): Likewise.
7682 (spu_libgcc_shift_count_mode): Likewise.
7683 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7684 (rl78_addr_space_pointer_mode): Likewise.
7685 (fl78_unwind_word_mode): Likewise.
7686 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7688 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7689 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7690 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7691 (mips_valid_pointer_mode): Likewise.
7692 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7693 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7694 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7696 (ft32_addr_space_address_mode): Likewise.
7697 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7698 scalar_int_mode instead of a machine_mode.
7699 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7701 (m32c_addr_space_address_mode): Likewise.
7702 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7703 (rs6000_eh_return_filter_mode): Likewise.
7704 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7705 (rs6000_eh_return_filter_mode): Likewise.
7706 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7707 (s390_libgcc_shift_count_mode): Likewise.
7708 (s390_unwind_word_mode): Likewise.
7709 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7711 * target.def (mode_rep_extended): Likewise.
7712 (valid_pointer_mode): Likewise.
7713 (addr_space.valid_pointer_mode): Likewise.
7714 (eh_return_filter_mode): Return a scalar_int_mode rather than
7716 (libgcc_cmp_return_mode): Likewise.
7717 (libgcc_shift_count_mode): Likewise.
7718 (unwind_word_mode): Likewise.
7719 (addr_space.pointer_mode): Likewise.
7720 (addr_space.address_mode): Likewise.
7721 * doc/tm.texi: Regenerate.
7722 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7724 (do_jump): Use scalar_int_mode for local variables.
7725 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7726 rather than a machine_mode.
7727 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7728 (scompare_loc_descriptor_wide): Likewise.
7729 (scompare_loc_descriptor_narrow): Likewise.
7730 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7732 * except.c (sjlj_emit_dispatch_table): Likewise.
7733 (expand_builtin_eh_copy_values): Likewise.
7734 * explow.c (convert_memory_address_addr_space_1): Likewise.
7735 Take a scalar_int_mode rather than a machine_mode.
7736 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7737 than a machine_mode.
7738 (memory_address_addr_space): Use scalar_int_mode for local variables.
7739 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7740 rather than a machine_mode.
7741 * expmed.c (mask_rtx): Likewise.
7742 (init_expmed_one_conv): Likewise.
7743 (expand_mult_highpart_adjust): Likewise.
7744 (extract_high_half): Likewise.
7745 (expmed_mult_highpart_optab): Likewise.
7746 (expmed_mult_highpart): Likewise.
7747 (expand_smod_pow2): Likewise.
7748 (expand_sdiv_pow2): Likewise.
7749 (emit_store_flag_int): Likewise.
7750 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7752 (extract_low_bits): Likewise.
7753 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7755 * expr.c (pieces_addr::adjust): Likewise.
7756 (can_store_by_pieces): Likewise.
7757 (store_by_pieces): Likewise.
7758 (clear_by_pieces_1): Likewise.
7759 (expand_expr_addr_expr_1): Likewise.
7760 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7761 (expand_expr_real_1): Likewise.
7762 (try_casesi): Likewise.
7763 * final.c (shorten_branches): Likewise.
7764 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7765 type of "mode" to machine_mode.
7766 * internal-fn.c (expand_arith_overflow_result_store): Take a
7767 scalar_int_mode rather than a machine_mode.
7768 (expand_mul_overflow): Use scalar_int_mode for local variables.
7769 * loop-doloop.c (doloop_modify): Likewise.
7770 (doloop_optimize): Likewise.
7771 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7772 than a machine_mode.
7773 (expand_doubleword_shift_condmove): Likewise.
7774 (expand_doubleword_shift): Likewise.
7775 (expand_doubleword_clz): Likewise.
7776 (expand_doubleword_popcount): Likewise.
7777 (expand_doubleword_parity): Likewise.
7778 (expand_absneg_bit): Use scalar_int_mode for local variables.
7779 (prepare_float_lib_cmp): Likewise.
7780 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7781 rather than a machine_mode.
7782 (convert_memory_address_addr_space): Likewise.
7783 (get_mode_bounds): Likewise.
7784 (get_address_mode): Return a scalar_int_mode rather than a
7786 * rtlanal.c (get_address_mode): Likewise.
7787 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7788 than a machine_mode.
7789 * targhooks.c (default_mode_rep_extended): Likewise.
7790 (default_valid_pointer_mode): Likewise.
7791 (default_addr_space_valid_pointer_mode): Likewise.
7792 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7793 than a machine_mode.
7794 (default_libgcc_cmp_return_mode): Likewise.
7795 (default_libgcc_shift_count_mode): Likewise.
7796 (default_unwind_word_mode): Likewise.
7797 (default_addr_space_pointer_mode): Likewise.
7798 (default_addr_space_address_mode): Likewise.
7799 * targhooks.h (default_eh_return_filter_mode): Likewise.
7800 (default_libgcc_cmp_return_mode): Likewise.
7801 (default_libgcc_shift_count_mode): Likewise.
7802 (default_unwind_word_mode): Likewise.
7803 (default_addr_space_pointer_mode): Likewise.
7804 (default_addr_space_address_mode): Likewise.
7805 (default_mode_rep_extended): Take a scalar_int_mode rather than
7807 (default_valid_pointer_mode): Likewise.
7808 (default_addr_space_valid_pointer_mode): Likewise.
7809 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7811 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7812 rather than a machine_mode.
7813 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7814 for local variables.
7815 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7816 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7817 than a machine_mode.
7819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7820 Alan Hayward <alan.hayward@arm.com>
7821 David Sherwood <david.sherwood@arm.com>
7823 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7824 the mode argument to scalar_int_mode.
7825 (do_jump_by_parts_zero_rtx): Likewise.
7826 (do_jump_by_parts_equality_rtx): Likewise.
7827 (do_jump_by_parts_greater): Take a mode argument.
7828 (do_jump_by_parts_equality): Likewise.
7829 (do_jump_1): Update calls accordingly.
7831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7832 Alan Hayward <alan.hayward@arm.com>
7833 David Sherwood <david.sherwood@arm.com>
7835 * is-a.h (safe_dyn_cast): New function.
7836 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7837 (jump_table_for_label): Likewise.
7838 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7839 instead of an rtx_insn *.
7840 (shorten_branches): Use dyn_cast instead of LABEL_P and
7841 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7842 rtx_jump_table_data::get_data_mode.
7843 (final_scan_insn): Likewise.
7845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7846 Alan Hayward <alan.hayward@arm.com>
7847 David Sherwood <david.sherwood@arm.com>
7849 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7850 trying to combine a full-register integer set with a subreg
7853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7854 Alan Hayward <alan.hayward@arm.com>
7855 David Sherwood <david.sherwood@arm.com>
7857 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7858 that is always either address_mode or pointer_mode.
7860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7861 Alan Hayward <alan.hayward@arm.com>
7862 David Sherwood <david.sherwood@arm.com>
7864 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7865 when the two are known to be equal.
7867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7868 Alan Hayward <alan.hayward@arm.com>
7869 David Sherwood <david.sherwood@arm.com>
7871 * simplify-rtx.c (simplify_const_unary_operation): Use
7872 is_a <scalar_int_mode> instead of checking for a nonzero
7873 precision. Forcibly convert op_mode to a scalar_int_mode
7874 in that case. More clearly differentiate the operand and
7875 result modes and use the former when deciding what the value
7876 of a count-bits operation should be. Use is_int_mode instead
7877 of checking for a MODE_INT. Remove redundant check for whether
7878 this mode has a zero precision.
7880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7881 Alan Hayward <alan.hayward@arm.com>
7882 David Sherwood <david.sherwood@arm.com>
7884 * optabs.c (widen_leading): Change the type of the mode argument
7885 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7886 (widen_bswap): Likewise.
7887 (expand_parity): Likewise.
7888 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7889 (expand_ffs): Likewise.
7890 (epand_unop): Check for scalar integer modes before calling the
7893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7894 Alan Hayward <alan.hayward@arm.com>
7895 David Sherwood <david.sherwood@arm.com>
7897 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7899 (expand_expr_real_1): Update call accordingly.
7901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7902 Alan Hayward <alan.hayward@arm.com>
7903 David Sherwood <david.sherwood@arm.com>
7905 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7906 value_mode arguments. Use scalar_int_mode internally.
7907 (store_bit_field_1): Rename the new integer mode from imode
7908 to op0_mode and use it instead of GET_MODE (op0). Update calls
7909 to store_split_bit_field, store_bit_field_using_insv and
7910 store_fixed_bit_field.
7911 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7912 Use scalar_int_mode internally. Use a bit count rather than a mode
7913 when calculating the largest bit size for get_best_mode.
7914 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7915 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7916 Remove assertion that OP0 has a scalar integer mode.
7917 (store_split_bit_field): Add op0_mode and value_mode arguments.
7918 Update calls to extract_fixed_bit_field.
7919 (extract_bit_field_using_extv): Add an op0_mode argument.
7920 Use scalar_int_mode internally.
7921 (extract_bit_field_1): Rename the new integer mode from imode to
7922 op0_mode and use it instead of GET_MODE (op0). Update calls to
7923 extract_split_bit_field, extract_bit_field_using_extv and
7924 extract_fixed_bit_field.
7925 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7926 to extract_split_bit_field and extract_fixed_bit_field_1.
7927 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7928 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7930 (extract_split_bit_field): Add an op0_mode argument. Update call
7931 to extract_fixed_bit_field.
7933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7934 Alan Hayward <alan.hayward@arm.com>
7935 David Sherwood <david.sherwood@arm.com>
7937 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7938 * explow.c (hard_function_value): Likewise.
7939 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7940 convert_to_mode call outside the loop.
7941 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7942 for the mode iterator. Require the mode specified by max_pieces
7944 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7946 (copy_blkmode_to_reg): Likewise.
7947 (set_storage_via_setmem): Likewise.
7948 * optabs.c (prepare_cmp_insn): Likewise.
7949 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7950 * stor-layout.c (finish_bitfield_representative): Likewise.
7952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7953 Alan Hayward <alan.hayward@arm.com>
7954 David Sherwood <david.sherwood@arm.com>
7956 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7957 * expr.c (convert_move): Use them.
7958 (convert_modes): Likewise.
7959 (store_expr_with_bounds): Likewise.
7961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7962 Alan Hayward <alan.hayward@arm.com>
7963 David Sherwood <david.sherwood@arm.com>
7965 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7966 parameter for the mode of "x". Remove the "known_x", "known_mode"
7967 and "known_ret" arguments. Change the type of the mode argument
7969 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7970 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7971 (reg_num_sign_bit_copies_for_combine): Likewise.
7972 * rtlanal.c (nonzero_bits1): Likewise.
7973 (num_sign_bit_copies1): Likewise.
7974 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7975 (reg_num_sign_bit_copies_general): Likewise.
7976 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7977 (reg_nonzero_bits_general): Likewise.
7979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7980 Alan Hayward <alan.hayward@arm.com>
7981 David Sherwood <david.sherwood@arm.com>
7983 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7984 than in subroutines. Return 1 for non-integer modes.
7985 (cached_num_sign_bit_copies): Change the type of the mode parameter
7987 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7988 classes. Handle CONST_INT_P first and then check whether X also
7989 has a scalar integer mode. Check the same thing for inner registers
7990 of a SUBREG and for values that are being extended or truncated.
7992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7993 Alan Hayward <alan.hayward@arm.com>
7994 David Sherwood <david.sherwood@arm.com>
7996 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7997 in subroutines. Return the mode mask for non-integer modes.
7998 (cached_nonzero_bits): Change the type of the mode parameter
8000 (nonzero_bits1): Likewise. Remove early exit for other mode
8001 classes. Handle CONST_INT_P first and then check whether X
8002 also has a scalar integer mode.
8004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8005 Alan Hayward <alan.hayward@arm.com>
8006 David Sherwood <david.sherwood@arm.com>
8008 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8009 Return a scalar_int_mode and assert that the size is greater than
8011 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8012 (op_by_pieces_d::op_by_pieces_d): Likewise.
8013 (op_by_pieces_d::run): Likewise.
8014 (can_store_by_pieces): Likewise.
8016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8017 Alan Hayward <alan.hayward@arm.com>
8018 David Sherwood <david.sherwood@arm.com>
8020 * combine.c (extract_left_shift): Add a mode argument and update
8022 (make_compound_operation_int): Change the type of the mode parameter
8023 to scalar_int_mode and update the call to extract_left_shift.
8025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8026 Alan Hayward <alan.hayward@arm.com>
8027 David Sherwood <david.sherwood@arm.com>
8029 * combine.c (simplify_and_const_int): Change the type of the mode
8030 parameter to scalar_int_mode.
8031 (simplify_and_const_int_1): Likewise. Update recursive call.
8033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8034 Alan Hayward <alan.hayward@arm.com>
8035 David Sherwood <david.sherwood@arm.com>
8037 * combine.c (simplify_compare_const): Check that the mode is a
8038 scalar_int_mode (rather than VOIDmode) before testing its
8040 (simplify_comparison): Move COMPARISON_P handling out of the
8041 loop and restrict the latter part of the loop to scalar_int_modes.
8042 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8043 and when considering SUBREG_REGs. Use is_int_mode instead of
8044 checking GET_MODE_CLASS against MODE_INT.
8046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8047 Alan Hayward <alan.hayward@arm.com>
8048 David Sherwood <david.sherwood@arm.com>
8050 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8051 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8052 shift_unit_mode and for modes involved in scalar shifts.
8054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8055 Alan Hayward <alan.hayward@arm.com>
8056 David Sherwood <david.sherwood@arm.com>
8058 * combine.c (force_int_to_mode): New function, split out from...
8059 (force_to_mode): ...here. Keep xmode up-to-date and use it
8060 instead of GET_MODE (x).
8062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8063 Alan Hayward <alan.hayward@arm.com>
8064 David Sherwood <david.sherwood@arm.com>
8066 * optabs-query.h (extraction_insn::struct_mode): Change type to
8067 opt_scalar_int_mode and update comment.
8068 (extraction_insn::field_mode): Change type to scalar_int_mode.
8069 (extraction_insn::pos_mode): Likewise.
8070 * combine.c (make_extraction): Update accordingly.
8071 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8072 (get_optab_extraction_insn): Likewise.
8073 * recog.c (simplify_while_replacing): Likewise.
8074 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8075 parameter to opt_scalar_int_mode.
8077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8078 Alan Hayward <alan.hayward@arm.com>
8079 David Sherwood <david.sherwood@arm.com>
8081 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8082 to a scalar_int_mode instead of a machine_mode.
8083 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8084 (get_best_mode): Return a boolean and use a pointer argument to store
8085 the selected mode. Replace the limit mode parameter with a bit limit.
8086 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8087 for the values returned by bit_field_mode_iterator::next_mode.
8088 (store_bit_field): Update call to get_best_mode.
8089 (store_fixed_bit_field): Likewise.
8090 (extract_fixed_bit_field): Likewise.
8091 * expr.c (optimize_bitfield_assignment_op): Likewise.
8092 * fold-const.c (optimize_bit_field_compare): Likewise.
8093 (fold_truth_andor_1): Likewise.
8094 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8095 Update for new type of m_mode.
8096 (get_best_mode): As above.
8098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8103 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8104 (store_bit_field): Check is_a <scalar_int_mode> before calling
8105 strict_volatile_bitfield_p.
8106 (extract_bit_field): Likewise.
8108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8109 Alan Hayward <alan.hayward@arm.com>
8110 David Sherwood <david.sherwood@arm.com>
8112 * target.def (cstore_mode): Return a scalar_int_mode.
8113 * doc/tm.texi: Regenerate.
8114 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8115 * targhooks.h (default_cstore_mode): Likewise.
8116 * targhooks.c (default_cstore_mode): Likewise, using a forced
8118 * expmed.c (emit_cstore): Expect the target of the cstore to be
8121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8122 Alan Hayward <alan.hayward@arm.com>
8123 David Sherwood <david.sherwood@arm.com>
8125 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8127 (niter_desc): Likewise mode.
8128 (iv_analyze): Add a mode parameter.
8130 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8131 and change its type to scalar_int_mode.
8132 * loop-iv.c: Update commentary at head of file.
8133 (iv_constant): Pass the mode paraeter before the rtx it describes
8134 and change its type to scalar_int_mode. Remove VOIDmode handling.
8135 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8136 (iv_extend): Likewise.
8137 (shorten_into_mode): Likewise.
8138 (iv_add): Use scalar_int_mode.
8139 (iv_mult): Likewise.
8140 (iv_shift): Likewise.
8141 (canonicalize_iv_subregs): Likewise.
8142 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8143 it describes and change its mode to scalar_int_mode. Also change
8144 the type of the returned inner_mode to scalar_int_mode.
8145 (get_biv_step): Likewise, turning outer_mode from a pointer
8146 into a direct parameter. Update call to get_biv_step_1.
8147 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8148 iv_constant and get_biv_step.
8149 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8150 and change its type to scalar_int_mode. Don't initialise iv->mode
8151 to VOIDmode and remove later checks for its still being VOIDmode.
8152 Update calls to iv_analyze_op and iv_analyze_expr. Check
8153 is_a <scalar_int_mode> when changing the mode under consideration.
8154 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8155 Update call to iv_analyze_expr.
8156 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8157 inner register is not also a scalar_int_mode. Update call to
8159 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8160 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8161 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8162 separate mode class checks. Update calls to iv_analyze. Remove
8163 fix-up of VOIDmodes after iv_analyze_biv.
8164 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8165 don't have a scalar_int_mode. Update call to biv_p.
8167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8168 Alan Hayward <alan.hayward@arm.com>
8169 David Sherwood <david.sherwood@arm.com>
8171 * cfgexpand.c (convert_debug_memory_address): Use
8172 as_a <scalar_int_mode>.
8173 * combine.c (expand_compound_operation): Likewise.
8174 (make_extraction): Likewise.
8175 (change_zero_ext): Likewise.
8176 (simplify_comparison): Likewise.
8177 * cse.c (cse_insn): Likewise.
8178 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8179 (mem_loc_descriptor): Likewise.
8180 (loc_descriptor): Likewise.
8181 * expmed.c (init_expmed_one_mode): Likewise.
8182 (synth_mult): Likewise.
8183 (emit_store_flag_1): Likewise.
8184 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8185 of a comparison with size.
8186 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8187 (reduce_to_bit_field_precision): Likewise.
8188 * function.c (expand_function_end): Likewise.
8189 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8190 * loop-doloop.c (doloop_modify): Likewise.
8191 * optabs.c (expand_binop): Likewise.
8192 (expand_unop): Likewise.
8193 (expand_copysign_absneg): Likewise.
8194 (prepare_cmp_insn): Likewise.
8195 (maybe_legitimize_operand): Likewise.
8196 * recog.c (const_scalar_int_operand): Likewise.
8197 * rtlanal.c (get_address_mode): Likewise.
8198 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8199 (simplify_cond_clz_ctz): Likewise.
8200 * tree-nested.c (get_nl_goto_field): Likewise.
8201 * tree.c (build_vector_type_for_mode): Likewise.
8202 * var-tracking.c (use_narrower_mode): Likewise.
8204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8205 Alan Hayward <alan.hayward@arm.com>
8206 David Sherwood <david.sherwood@arm.com>
8208 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8209 * builtins.c (expand_builtin_signbit): Use it.
8210 * cfgexpand.c (expand_debug_expr): Likewise.
8211 * dojump.c (do_jump): Likewise.
8212 (do_compare_and_jump): Likewise.
8213 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8214 * expmed.c (make_tree): Likewise.
8215 * expr.c (expand_expr_real_2): Likewise.
8216 (expand_expr_real_1): Likewise.
8217 (try_casesi): Likewise.
8218 * fold-const-call.c (fold_const_call_ss): Likewise.
8219 * fold-const.c (unextend): Likewise.
8220 (extract_muldiv_1): Likewise.
8221 (fold_single_bit_test): Likewise.
8222 (native_encode_int): Likewise.
8223 (native_encode_string): Likewise.
8224 (native_interpret_int): Likewise.
8225 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8226 * internal-fn.c (expand_addsub_overflow): Likewise.
8227 (expand_neg_overflow): Likewise.
8228 (expand_mul_overflow): Likewise.
8229 (expand_arith_overflow): Likewise.
8230 * match.pd: Likewise.
8231 * stor-layout.c (layout_type): Likewise.
8232 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8233 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8234 * tree-ssanames.c (get_range_info): Likewise.
8235 * tree-switch-conversion.c (array_value_type) Likewise.
8236 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8237 (vect_recog_divmod_pattern): Likewise.
8238 (vect_recog_mixed_size_cond_pattern): Likewise.
8239 * tree-vrp.c (extract_range_basic): Likewise.
8240 (simplify_float_conversion_using_ranges): Likewise.
8241 * tree.c (int_fits_type_p): Likewise.
8242 * ubsan.c (instrument_bool_enum_load): Likewise.
8243 * varasm.c (mergeable_string_section): Likewise.
8244 (narrowing_initializer_constant_valid_p): Likewise.
8245 (output_constant): Likewise.
8247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8248 Alan Hayward <alan.hayward@arm.com>
8249 David Sherwood <david.sherwood@arm.com>
8251 * machmode.h (NARROWEST_INT_MODE): New macro.
8252 * expr.c (alignment_for_piecewise_move): Use it instead of
8253 GET_CLASS_NARROWEST_MODE (MODE_INT).
8254 (push_block): Likewise.
8255 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8257 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * postreload.c (move2add_valid_value_p): Change the type of the
8264 mode parameter to scalar_int_mode.
8265 (move2add_use_add2_insn): Add a mode parameter and use it instead
8267 (move2add_use_add3_insn): Likewise.
8268 (reload_cse_move2add): Update accordingly.
8270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8271 Alan Hayward <alan.hayward@arm.com>
8272 David Sherwood <david.sherwood@arm.com>
8274 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8276 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8277 * optabs.c (expand_unop): Likewise.
8279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8280 Alan Hayward <alan.hayward@arm.com>
8281 David Sherwood <david.sherwood@arm.com>
8283 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8284 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8285 (popcount_loc_descriptor): Likewise.
8286 (bswap_loc_descriptor): Likewise.
8287 (rotate_loc_descriptor): Likewise.
8288 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8289 calling the functions above.
8291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8292 Alan Hayward <alan.hayward@arm.com>
8293 David Sherwood <david.sherwood@arm.com>
8295 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8297 (try_combine): Likewise.
8298 (simplify_if_then_else): Likewise.
8299 * cse.c (cse_insn): Likewise.
8300 * dwarf2out.c (mem_loc_descriptor): Likewise.
8301 * emit-rtl.c (gen_lowpart_common): Likewise.
8302 * simplify-rtx.c (simplify_truncation): Likewise.
8303 (simplify_binary_operation_1): Likewise.
8304 (simplify_const_relational_operation): Likewise.
8305 (simplify_ternary_operation): Likewise.
8306 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8309 Alan Hayward <alan.hayward@arm.com>
8310 David Sherwood <david.sherwood@arm.com>
8312 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8313 * reload.c (push_reload): Likewise.
8314 (find_reloads): Likewise.
8316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8317 Alan Hayward <alan.hayward@arm.com>
8318 David Sherwood <david.sherwood@arm.com>
8320 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8321 (make_compound_operation_int): Likewise.
8322 (change_zero_ext): Likewise.
8323 * expr.c (convert_move): Likewise.
8324 (convert_modes): Likewise.
8325 * fwprop.c (forward_propagate_subreg): Likewise.
8326 * loop-iv.c (get_biv_step_1): Likewise.
8327 * optabs.c (widen_operand): Likewise.
8328 * postreload.c (move2add_valid_value_p): Likewise.
8329 * recog.c (simplify_while_replacing): Likewise.
8330 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8331 (simplify_binary_operation_1): Likewise. Remove redundant
8332 mode equality check.
8334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * combine.c (combine_simplify_rtx): Add checks for
8339 is_a <scalar_int_mode>.
8340 (simplify_if_then_else): Likewise.
8341 (make_field_assignment): Likewise.
8342 (simplify_comparison): Likewise.
8343 * ifcvt.c (noce_try_bitop): Likewise.
8344 * loop-invariant.c (canonicalize_address_mult): Likewise.
8345 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8348 Alan Hayward <alan.hayward@arm.com>
8349 David Sherwood <david.sherwood@arm.com>
8351 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8352 is_a <scalar_int_mode> instead of != BLKmode.
8354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8355 Alan Hayward <alan.hayward@arm.com>
8356 David Sherwood <david.sherwood@arm.com>
8358 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8359 instead of != VOIDmode.
8360 * combine.c (if_then_else_cond): Likewise.
8361 (change_zero_ext): Likewise.
8362 * dwarf2out.c (mem_loc_descriptor): Likewise.
8363 (loc_descriptor): Likewise.
8364 * rtlanal.c (canonicalize_condition): Likewise.
8365 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8368 Alan Hayward <alan.hayward@arm.com>
8369 David Sherwood <david.sherwood@arm.com>
8371 * simplify-rtx.c (simplify_binary_operation_1): Use
8372 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8375 Alan Hayward <alan.hayward@arm.com>
8376 David Sherwood <david.sherwood@arm.com>
8378 * wide-int.h (int_traits<unsigned char>) New class.
8379 (int_traits<unsigned short>) Likewise.
8380 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8381 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8383 * combine.c (set_nonzero_bits_and_sign_copies): Use
8384 is_a <scalar_int_mode>.
8385 (find_split_point): Likewise.
8386 (combine_simplify_rtx): Likewise.
8387 (simplify_logical): Likewise.
8388 (expand_compound_operation): Likewise.
8389 (expand_field_assignment): Likewise.
8390 (make_compound_operation): Likewise.
8391 (extended_count): Likewise.
8392 (change_zero_ext): Likewise.
8393 (simplify_comparison): Likewise.
8394 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8395 (ucompare_loc_descriptor): Likewise.
8396 (minmax_loc_descriptor): Likewise.
8397 (mem_loc_descriptor): Likewise.
8398 (loc_descriptor): Likewise.
8399 * expmed.c (init_expmed_one_mode): Likewise.
8400 * lra-constraints.c (lra_constraint_offset): Likewise.
8401 * optabs.c (prepare_libcall_arg): Likewise.
8402 * postreload.c (move2add_note_store): Likewise.
8403 * reload.c (operands_match_p): Likewise.
8404 * rtl.h (load_extend_op): Likewise.
8405 * rtlhooks.c (gen_lowpart_general): Likewise.
8406 * simplify-rtx.c (simplify_truncation): Likewise.
8407 (simplify_unary_operation_1): Likewise.
8408 (simplify_binary_operation_1): Likewise.
8409 (simplify_const_binary_operation): Likewise.
8410 (simplify_const_relational_operation): Likewise.
8411 (simplify_subreg): Likewise.
8412 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8413 * var-tracking.c (adjust_mems): Likewise.
8414 (prepare_call_arguments): Likewise.
8416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8417 Alan Hayward <alan.hayward@arm.com>
8418 David Sherwood <david.sherwood@arm.com>
8420 * machmode.h (is_int_mode): New fuction.
8421 * combine.c (find_split_point): Use it.
8422 (combine_simplify_rtx): Likewise.
8423 (simplify_if_then_else): Likewise.
8424 (simplify_set): Likewise.
8425 (simplify_shift_const_1): Likewise.
8426 (simplify_comparison): Likewise.
8427 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8428 * cse.c (notreg_cost): Likewise.
8429 (cse_insn): Likewise.
8430 * cselib.c (cselib_lookup_1): Likewise.
8431 * dojump.c (do_jump_1): Likewise.
8432 (do_compare_rtx_and_jump): Likewise.
8433 * dse.c (get_call_args): Likewise.
8434 * dwarf2out.c (rtl_for_decl_init): Likewise.
8435 (native_encode_initializer): Likewise.
8436 * expmed.c (emit_store_flag_1): Likewise.
8437 (emit_store_flag): Likewise.
8438 * expr.c (convert_modes): Likewise.
8439 (store_field): Likewise.
8440 (expand_expr_real_1): Likewise.
8441 * fold-const.c (fold_read_from_constant_string): Likewise.
8442 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8443 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8444 * optabs.c (expand_binop): Likewise.
8445 (expand_unop): Likewise.
8446 (expand_abs_nojump): Likewise.
8447 (expand_one_cmpl_abs_nojump): Likewise.
8448 * simplify-rtx.c (mode_signbit_p): Likewise.
8449 (val_signbit_p): Likewise.
8450 (val_signbit_known_set_p): Likewise.
8451 (val_signbit_known_clear_p): Likewise.
8452 (simplify_relational_operation_1): Likewise.
8453 * tree.c (vector_type_mode): Likewise.
8455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8456 Alan Hayward <alan.hayward@arm.com>
8457 David Sherwood <david.sherwood@arm.com>
8459 * machmode.h (smallest_mode_for_size): Fix formatting.
8460 (smallest_int_mode_for_size): New function.
8461 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8462 instead of smallest_mode_for_size.
8463 * combine.c (make_extraction): Likewise.
8464 * config/arc/arc.c (arc_expand_movmem): Likewise.
8465 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8466 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8467 * config/s390/s390.c (s390_expand_insv): Likewise.
8468 * config/sparc/sparc.c (assign_int_registers): Likewise.
8469 * config/spu/spu.c (spu_function_value): Likewise.
8470 (spu_function_arg): Likewise.
8471 * coverage.c (get_gcov_type): Likewise.
8472 (get_gcov_unsigned_t): Likewise.
8473 * dse.c (find_shift_sequence): Likewise.
8474 * expmed.c (store_bit_field_1): Likewise.
8475 * expr.c (convert_move): Likewise.
8476 (store_field): Likewise.
8477 * internal-fn.c (expand_arith_overflow): Likewise.
8478 * optabs-query.c (get_best_extraction_insn): Likewise.
8479 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8480 * stor-layout.c (layout_type): Likewise.
8481 (initialize_sizetypes): Likewise.
8482 * targhooks.c (default_get_mask_mode): Likewise.
8483 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8486 Alan Hayward <alan.hayward@arm.com>
8487 David Sherwood <david.sherwood@arm.com>
8489 * machmode.h (opt_mode::else_blk): New function.
8490 (int_mode_for_mode): Declare.
8491 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8492 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8494 * cfgexpand.c (expand_debug_expr): Likewise.
8495 * combine.c (gen_lowpart_or_truncate): Likewise.
8496 (gen_lowpart_for_combine): Likewise.
8497 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8498 * config/avr/avr.c (avr_to_int_mode): Likewise.
8499 (avr_out_plus_1): Likewise.
8500 (avr_out_plus): Likewise.
8501 (avr_out_round): Likewise.
8502 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8503 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8504 (s390_expand_vcond): Likewise.
8505 * config/spu/spu.c (spu_split_immediate): Likewise.
8506 (spu_expand_mov): Likewise.
8507 * dse.c (get_stored_val): Likewise.
8508 * expmed.c (store_bit_field_1): Likewise.
8509 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8511 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8512 (extract_low_bits): Likewise.
8513 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8514 handling rather than repeating the check.
8515 (emit_group_store): Likewise.
8516 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8517 * optabs.c (expand_absneg_bit): Likewise.
8518 (expand_copysign_absneg): Likewise.
8519 (expand_copysign_bit): Likewise.
8520 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8521 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8522 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8523 * var-tracking.c (prepare_call_arguments): Likewise.
8524 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8525 int_mode_for_mode instead of mode_for_size.
8526 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8529 Alan Hayward <alan.hayward@arm.com>
8530 David Sherwood <david.sherwood@arm.com>
8532 * machmode.h (int_mode_for_size): New function.
8533 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8534 instead of mode_for_size.
8535 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8537 * combine.c (expand_field_assignment): Use int_mode_for_size
8538 instead of mode_for_size.
8539 (make_extraction): Likewise.
8540 (simplify_shift_const_1): Likewise.
8541 (simplify_comparison): Likewise.
8542 * dojump.c (do_jump): Likewise.
8543 * dwarf2out.c (mem_loc_descriptor): Likewise.
8544 * emit-rtl.c (init_derived_machine_modes): Likewise.
8545 * expmed.c (flip_storage_order): Likewise.
8546 (convert_extracted_bit_field): Likewise.
8547 * expr.c (copy_blkmode_from_reg): Likewise.
8548 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8549 * internal-fn.c (expand_mul_overflow): Likewise.
8550 * lower-subreg.c (simple_move): Likewise.
8551 * optabs-libfuncs.c (init_optabs): Likewise.
8552 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8553 * tree.c (vector_type_mode): Likewise.
8554 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8555 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8556 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8557 * tree-vect-stmts.c (vectorizable_load): Likewise.
8558 (vectorizable_store): Likewise.
8560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8561 Alan Hayward <alan.hayward@arm.com>
8562 David Sherwood <david.sherwood@arm.com>
8564 * coretypes.h (pod_mode): New type.
8565 (scalar_int_mode_pod): New typedef.
8566 * machmode.h (pod_mode): New class.
8567 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8568 * genmodes.c (emit_mode_int_n): Update accordingly.
8569 * lower-subreg.h (target_lower_subreg): Change type to
8570 scalar_int_mode_pod.
8571 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8572 scalar_int_mode_pod.
8574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8575 Alan Hayward <alan.hayward@arm.com>
8576 David Sherwood <david.sherwood@arm.com>
8578 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8579 machine_mode to scalar_int_mode.
8580 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8581 (rs6000_option_override_internal): Remove cast to int.
8582 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8583 machine_mode to scalar_int_mode.
8584 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8585 (rs6000_option_override_internal): Remove cast to int.
8586 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8587 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8589 * config/s390/s390.c (s390_expand_builtin): Likewise.
8590 * coretypes.h (scalar_int_mode): New type.
8591 (opt_scalar_int_mode): New typedef.
8592 * machmode.h (scalar_int_mode): New class.
8593 (scalar_int_mode::includes_p): New function.
8594 (byte_mode): Change type to scalar_int_mode.
8595 (word_mode): Likewise.
8596 (ptr_mode): Likewise.
8597 * emit-rtl.c (byte_mode): Likewise.
8598 (word_mode): Likewise.
8599 (ptr_mode): Likewise.
8600 (init_derived_machine_modes): Update accordingly.
8601 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8602 and MODE_PARTIAL_INT.
8603 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8604 opt_scalar_int_mode.
8606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8607 Alan Hayward <alan.hayward@arm.com>
8608 David Sherwood <david.sherwood@arm.com>
8610 * target.def (libgcc_floating_mode_supported_p): Take a
8612 * doc/tm.texi: Regenerate.
8613 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8615 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8616 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8620 Alan Hayward <alan.hayward@arm.com>
8621 David Sherwood <david.sherwood@arm.com>
8623 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8624 * doc/tm.texi: Regenerate.
8625 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8626 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8627 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8628 * targhooks.h (default_floatn_mode): Likewise.
8629 * targhooks.c (default_floatn_mode): Likewise.
8630 * tree.c (build_common_tree_nodes): Update accordingly.
8632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8633 Alan Hayward <alan.hayward@arm.com>
8634 David Sherwood <david.sherwood@arm.com>
8636 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8637 (mode_iterator::iterate_p): Likewise.
8638 (mode_iterator::get_wider): Likewise.
8639 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8642 Alan Hayward <alan.hayward@arm.com>
8643 David Sherwood <david.sherwood@arm.com>
8645 * coretypes.h (opt_scalar_float_mode): New typedef.
8646 * machmode.h (float_mode_for_size): New function.
8647 * emit-rtl.c (double_mode): Delete.
8648 (init_emit_once): Use float_mode_for_size.
8649 * stor-layout.c (layout_type): Likewise.
8650 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8653 Alan Hayward <alan.hayward@arm.com>
8654 David Sherwood <david.sherwood@arm.com>
8656 * output.h (assemble_real): Take a scalar_float_mode.
8657 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8658 * config/arm/arm.md (consttable_4): Likewise.
8659 (consttable_8): Likewise.
8660 (consttable_16): Likewise.
8661 * config/mips/mips.md (consttable_float): Likewise.
8662 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8663 * varasm.c (assemble_real): Take a scalar_float_mode.
8664 (output_constant_pool_2): Update accordingly.
8665 (output_constant): Likewise.
8667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8668 Alan Hayward <alan.hayward@arm.com>
8669 David Sherwood <david.sherwood@arm.com>
8671 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8672 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8673 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8674 (native_encode_real): Likewise.
8675 (native_interpret_real): Likewise.
8676 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8677 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8680 Alan Hayward <alan.hayward@arm.com>
8681 David Sherwood <david.sherwood@arm.com>
8683 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8684 <scalar_float_mode>. Simplify.
8685 (gen_extend_conv_libfunc): Likewise.
8687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8688 Alan Hayward <alan.hayward@arm.com>
8689 David Sherwood <david.sherwood@arm.com>
8691 * coretypes.h (scalar_float_mode): New type.
8692 * machmode.h (mode_traits::from_int): Use machine_mode if
8693 USE_ENUM_MODES is defined.
8694 (is_a): New function.
8696 (dyn_cast): Likewise.
8697 (scalar_float_mode): New class.
8698 (scalar_float_mode::includes_p): New function.
8699 (is_float_mode): Likewise.
8700 * gdbhooks.py (MachineModePrinter): New class.
8701 (build_pretty_printer): Use it for scalar_float_mode.
8702 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8703 (format_helper::format_helper): Turn into a template.
8704 * genmodes.c (get_mode_class): New function.
8705 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8706 or machine_mode if none.
8707 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8708 as_a <scalar_float_mode>.
8709 * dwarf2out.c (mem_loc_descriptor): Likewise.
8710 (insert_float): Likewise.
8711 (add_const_value_attribute): Likewise.
8712 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8713 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8714 (expand_unop): Update accordingly.
8715 (expand_abs_nojump): Likewise.
8716 (expand_copysign_absneg): Take a scalar_float_mode.
8717 (expand_copysign_bit): Likewise.
8718 (expand_copysign): Update accordingly.
8720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8721 Alan Hayward <alan.hayward@arm.com>
8722 David Sherwood <david.sherwood@arm.com>
8724 * coretypes.h (opt_mode): New class.
8725 * machmode.h (opt_mode): Likewise.
8726 (opt_mode::else_void): New function.
8727 (opt_mode::require): Likewise.
8728 (opt_mode::exists): Likewise.
8729 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8730 (GET_MODE_2XWIDER_MODE): Likewise.
8731 (mode_iterator::get_wider): Update accordingly.
8732 (mode_iterator::get_2xwider): Likewise.
8733 (mode_iterator::get_known_wider): Likewise, turning into a template.
8734 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8735 forcing a wider mode to exist.
8736 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8737 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8738 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8739 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8740 * lower-subreg.c (init_lower_subreg): Likewise.
8741 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8742 on the final iteration.
8743 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8744 a wider mode exists before asking for a move pattern.
8745 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8746 forcing a wider mode to exist.
8747 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8748 returning false if no such mode exists.
8749 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8750 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8751 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8752 Avoid checking for a MODE_INT if we already know the mode is not a
8754 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8755 forcing a wider mode to exist.
8756 (expmed_mult_highpart_optab): Likewise.
8757 (expmed_mult_highpart): Likewise.
8758 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8760 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8761 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8762 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8763 * internal-fn.c (expand_mul_overflow): Update use of
8764 GET_MODE_2XWIDER_MODE.
8765 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8766 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8767 GET_MODE_WIDER_MODE.
8768 (convert_plusminus_to_widen): Likewise.
8769 * tree-switch-conversion.c (array_value_type): Likewise.
8770 * var-tracking.c (emit_note_insn_var_location): Likewise.
8771 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8772 Return false inside rather than outside the loop if no wider mode
8774 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8775 and GET_MODE_2XWIDER_MODE
8776 (can_compare_p): Use else_void.
8777 * gdbhooks.py (OptMachineModePrinter): New class.
8778 (build_pretty_printer): Use it for opt_mode.
8780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8781 Alan Hayward <alan.hayward@arm.com>
8782 David Sherwood <david.sherwood@arm.com>
8784 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8785 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8788 Alan Hayward <alan.hayward@arm.com>
8789 David Sherwood <david.sherwood@arm.com>
8791 * machmode.h (mode_traits): New structure.
8792 (get_narrowest_mode): New function.
8793 (mode_iterator::start): Likewise.
8794 (mode_iterator::iterate_p): Likewise.
8795 (mode_iterator::get_wider): Likewise.
8796 (mode_iterator::get_known_wider): Likewise.
8797 (mode_iterator::get_2xwider): Likewise.
8798 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8799 (FOR_EACH_MODE): Likewise.
8800 (FOR_EACH_MODE_FROM): Likewise.
8801 (FOR_EACH_MODE_UNTIL): Likewise.
8802 (FOR_EACH_WIDER_MODE): Likewise.
8803 (FOR_EACH_2XWIDER_MODE): Likewise.
8804 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8805 * combine.c (simplify_comparison): Likewise
8806 * config/i386/i386.c (type_natural_mode): Likewise.
8807 * cse.c (cse_insn): Likewise.
8808 * dse.c (find_shift_sequence): Likewise.
8809 * emit-rtl.c (init_derived_machine_modes): Likewise.
8810 (init_emit_once): Likewise.
8811 * explow.c (hard_function_value): Likewise.
8812 * expmed.c (extract_fixed_bit_field_1): Likewise.
8813 (extract_bit_field_1): Likewise.
8814 (expand_divmod): Likewise.
8815 (emit_store_flag_1): Likewise.
8816 * expr.c (init_expr_target): Likewise.
8817 (convert_move): Likewise.
8818 (alignment_for_piecewise_move): Likewise.
8819 (widest_int_mode_for_size): Likewise.
8820 (emit_block_move_via_movmem): Likewise.
8821 (copy_blkmode_to_reg): Likewise.
8822 (set_storage_via_setmem): Likewise.
8823 (compress_float_constant): Likewise.
8824 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8825 * optabs-query.c (get_best_extraction_insn): Likewise.
8826 * optabs.c (expand_binop): Likewise.
8827 (expand_twoval_unop): Likewise.
8828 (expand_twoval_binop): Likewise.
8829 (widen_leading): Likewise.
8830 (widen_bswap): Likewise.
8831 (expand_parity): Likewise.
8832 (expand_unop): Likewise.
8833 (prepare_cmp_insn): Likewise.
8834 (prepare_float_lib_cmp): Likewise.
8835 (expand_float): Likewise.
8836 (expand_fix): Likewise.
8837 (expand_sfix_optab): Likewise.
8838 * postreload.c (move2add_use_add2_insn): Likewise.
8839 * reg-stack.c (reg_to_stack): Likewise.
8840 * reginfo.c (choose_hard_reg_mode): Likewise.
8841 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8842 * stor-layout.c (mode_for_size): Likewise.
8843 (smallest_mode_for_size): Likewise.
8844 (mode_for_vector): Likewise.
8845 (finish_bitfield_representative): Likewise.
8846 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8847 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8848 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8849 * var-tracking.c (prepare_call_arguments): Likewise.
8851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8852 Alan Hayward <alan.hayward@arm.com>
8853 David Sherwood <david.sherwood@arm.com>
8855 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8856 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8857 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8858 * machmode.h (mode_size): Move earlier in file.
8859 (mode_precision): Likewise.
8860 (mode_inner): Likewise.
8861 (mode_nunits): Likewise.
8862 (mode_unit_size): Likewise.
8863 (unit_unit_precision): Likewise.
8864 (mode_wider): Likewise.
8865 (mode_2xwider): Likewise.
8866 (machine_mode): New class.
8867 (mode_to_bytes): New function.
8868 (mode_to_bits): Likewise.
8869 (mode_to_precision): Likewise.
8870 (mode_to_inner): Likewise.
8871 (mode_to_unit_size): Likewise.
8872 (mode_to_unit_precision): Likewise.
8873 (mode_to_nunits): Likewise.
8874 (GET_MODE_SIZE): Use mode_to_bytes.
8875 (GET_MODE_BITSIZE): Use mode_to_bits.
8876 (GET_MODE_PRECISION): Use mode_to_precision.
8877 (GET_MODE_INNER): Use mode_to_inner.
8878 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8879 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8880 (GET_MODE_NUNITS): Use mode_to_nunits.
8881 * system.h (ALWAYS_INLINE): New macro.
8882 * config/powerpcspe/powerpcspe-c.c
8883 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8884 int for arg1_mode and arg2_mode.
8886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8887 Alan Hayward <alan.hayward@arm.com>
8888 David Sherwood <david.sherwood@arm.com>
8890 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8891 Prefix mode names with E_ in case statements.
8892 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8893 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8894 (aarch64_split_simd_move): Likewise.
8895 (aarch64_gen_storewb_pair): Likewise.
8896 (aarch64_gen_loadwb_pair): Likewise.
8897 (aarch64_gen_store_pair): Likewise.
8898 (aarch64_gen_load_pair): Likewise.
8899 (aarch64_get_condition_code_1): Likewise.
8900 (aarch64_constant_pool_reload_icode): Likewise.
8901 (get_rsqrte_type): Likewise.
8902 (get_rsqrts_type): Likewise.
8903 (get_recpe_type): Likewise.
8904 (get_recps_type): Likewise.
8905 (aarch64_gimplify_va_arg_expr): Likewise.
8906 (aarch64_simd_container_mode): Likewise.
8907 (aarch64_emit_load_exclusive): Likewise.
8908 (aarch64_emit_store_exclusive): Likewise.
8909 (aarch64_expand_compare_and_swap): Likewise.
8910 (aarch64_gen_atomic_cas): Likewise.
8911 (aarch64_emit_bic): Likewise.
8912 (aarch64_emit_atomic_swap): Likewise.
8913 (aarch64_emit_atomic_load_op): Likewise.
8914 (aarch64_evpc_trn): Likewise.
8915 (aarch64_evpc_uzp): Likewise.
8916 (aarch64_evpc_zip): Likewise.
8917 (aarch64_evpc_ext): Likewise.
8918 (aarch64_evpc_rev): Likewise.
8919 (aarch64_evpc_dup): Likewise.
8920 (aarch64_gen_ccmp_first): Likewise.
8921 (aarch64_gen_ccmp_next): Likewise.
8922 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8923 (alpha_emit_xfloating_libcall): Likewise.
8924 (emit_insxl): Likewise.
8925 (alpha_arg_type): Likewise.
8926 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8927 (arc_preferred_simd_mode): Likewise.
8928 (arc_secondary_reload): Likewise.
8929 (get_arc_condition_code): Likewise.
8930 (arc_print_operand): Likewise.
8931 (arc_legitimate_constant_p): Likewise.
8932 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8933 * config/arc/arc.md (casesi_load): Likewise.
8934 (casesi_compact_jump): Likewise.
8935 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8936 (cc_use_register): Likewise.
8937 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8938 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8939 (arm_init_iwmmxt_builtins): Likewise.
8940 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8941 (neon_expand_vector_init): Likewise.
8942 (arm_attr_length_move_neon): Likewise.
8943 (maybe_get_arm_condition_code): Likewise.
8944 (arm_emit_vector_const): Likewise.
8945 (arm_preferred_simd_mode): Likewise.
8946 (arm_output_iwmmxt_tinsr): Likewise.
8947 (thumb1_output_casesi): Likewise.
8948 (thumb2_output_casesi): Likewise.
8949 (arm_emit_load_exclusive): Likewise.
8950 (arm_emit_store_exclusive): Likewise.
8951 (arm_expand_compare_and_swap): Likewise.
8952 (arm_evpc_neon_vuzp): Likewise.
8953 (arm_evpc_neon_vzip): Likewise.
8954 (arm_evpc_neon_vrev): Likewise.
8955 (arm_evpc_neon_vtrn): Likewise.
8956 (arm_evpc_neon_vext): Likewise.
8957 (arm_validize_comparison): Likewise.
8958 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8959 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8960 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8961 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8962 (c6x_preferred_simd_mode): Likewise.
8963 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8964 (epiphany_rtx_costs): Likewise.
8965 * config/epiphany/predicates.md (proper_comparison_operator):
8967 * config/frv/frv.c (condexec_memory_operand): Likewise.
8968 (frv_emit_move): Likewise.
8969 (output_move_single): Likewise.
8970 (output_condmove_single): Likewise.
8971 (frv_hard_regno_mode_ok): Likewise.
8972 (frv_matching_accg_mode): Likewise.
8973 * config/h8300/h8300.c (split_adds_subs): Likewise.
8974 (h8300_rtx_costs): Likewise.
8975 (h8300_print_operand): Likewise.
8976 (compute_mov_length): Likewise.
8977 (output_logical_op): Likewise.
8978 (compute_logical_op_length): Likewise.
8979 (compute_logical_op_cc): Likewise.
8980 (h8300_shift_needs_scratch_p): Likewise.
8981 (output_a_shift): Likewise.
8982 (compute_a_shift_length): Likewise.
8983 (compute_a_shift_cc): Likewise.
8984 (expand_a_rotate): Likewise.
8985 (output_a_rotate): Likewise.
8986 * config/i386/i386.c (classify_argument): Likewise.
8987 (function_arg_advance_32): Likewise.
8988 (function_arg_32): Likewise.
8989 (function_arg_64): Likewise.
8990 (function_value_64): Likewise.
8991 (ix86_gimplify_va_arg): Likewise.
8992 (ix86_legitimate_constant_p): Likewise.
8993 (put_condition_code): Likewise.
8994 (split_double_mode): Likewise.
8995 (ix86_avx256_split_vector_move_misalign): Likewise.
8996 (ix86_expand_vector_logical_operator): Likewise.
8997 (ix86_split_idivmod): Likewise.
8998 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8999 (ix86_build_const_vector): Likewise.
9000 (ix86_build_signbit_mask): Likewise.
9001 (ix86_match_ccmode): Likewise.
9002 (ix86_cc_modes_compatible): Likewise.
9003 (ix86_expand_branch): Likewise.
9004 (ix86_expand_sse_cmp): Likewise.
9005 (ix86_expand_sse_movcc): Likewise.
9006 (ix86_expand_int_sse_cmp): Likewise.
9007 (ix86_expand_vec_perm_vpermi2): Likewise.
9008 (ix86_expand_vec_perm): Likewise.
9009 (ix86_expand_sse_unpack): Likewise.
9010 (ix86_expand_int_addcc): Likewise.
9011 (ix86_split_to_parts): Likewise.
9012 (ix86_vectorize_builtin_gather): Likewise.
9013 (ix86_vectorize_builtin_scatter): Likewise.
9014 (avx_vpermilp_parallel): Likewise.
9015 (inline_memory_move_cost): Likewise.
9016 (ix86_tieable_integer_mode_p): Likewise.
9017 (x86_maybe_negate_const_int): Likewise.
9018 (ix86_expand_vector_init_duplicate): Likewise.
9019 (ix86_expand_vector_init_one_nonzero): Likewise.
9020 (ix86_expand_vector_init_one_var): Likewise.
9021 (ix86_expand_vector_init_concat): Likewise.
9022 (ix86_expand_vector_init_interleave): Likewise.
9023 (ix86_expand_vector_init_general): Likewise.
9024 (ix86_expand_vector_set): Likewise.
9025 (ix86_expand_vector_extract): Likewise.
9026 (emit_reduc_half): Likewise.
9027 (ix86_emit_i387_round): Likewise.
9028 (ix86_mangle_type): Likewise.
9029 (ix86_expand_round_sse4): Likewise.
9030 (expand_vec_perm_blend): Likewise.
9031 (canonicalize_vector_int_perm): Likewise.
9032 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9033 (expand_vec_perm_1): Likewise.
9034 (expand_vec_perm_interleave3): Likewise.
9035 (expand_vec_perm_even_odd_pack): Likewise.
9036 (expand_vec_perm_even_odd_1): Likewise.
9037 (expand_vec_perm_broadcast_1): Likewise.
9038 (ix86_vectorize_vec_perm_const_ok): Likewise.
9039 (ix86_expand_vecop_qihi): Likewise.
9040 (ix86_expand_mul_widen_hilo): Likewise.
9041 (ix86_expand_sse2_abs): Likewise.
9042 (ix86_expand_pextr): Likewise.
9043 (ix86_expand_pinsr): Likewise.
9044 (ix86_preferred_simd_mode): Likewise.
9045 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9046 * config/i386/sse.md (*andnot<mode>3): Likewise.
9047 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9048 (*<code><mode>3): Likewise.
9049 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9050 (ia64_expand_atomic_op): Likewise.
9051 (ia64_arg_type): Likewise.
9052 (ia64_mode_to_int): Likewise.
9053 (ia64_scalar_mode_supported_p): Likewise.
9054 (ia64_vector_mode_supported_p): Likewise.
9055 (expand_vec_perm_broadcast): Likewise.
9056 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9057 (iq2000_function_arg_advance): Likewise.
9058 (iq2000_function_arg): Likewise.
9059 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9060 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9061 (m68k_libcall_value): Likewise.
9062 (m68k_function_value): Likewise.
9063 (sched_attr_op_type): Likewise.
9064 * config/mcore/mcore.c (mcore_output_move): Likewise.
9065 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9067 (microblaze_function_arg): Likewise.
9068 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9069 (mips_print_operand): Likewise.
9070 (mips_mode_ok_for_mov_fmt_p): Likewise.
9071 (mips_vector_mode_supported_p): Likewise.
9072 (mips_preferred_simd_mode): Likewise.
9073 (mips_expand_vpc_loongson_even_odd): Likewise.
9074 (mips_expand_vec_unpack): Likewise.
9075 (mips_expand_vi_broadcast): Likewise.
9076 (mips_expand_vector_init): Likewise.
9077 (mips_expand_vec_reduc): Likewise.
9078 (mips_expand_msa_cmp): Likewise.
9079 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9080 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9081 (cc_flags_for_mode): Likewise.
9082 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9083 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9084 (nds32_output_casesi_pc_relative): Likewise.
9085 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9086 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9087 (nvptx_gen_unpack): Likewise.
9088 (nvptx_gen_pack): Likewise.
9089 (nvptx_gen_shuffle): Likewise.
9090 (nvptx_gen_wcast): Likewise.
9091 (nvptx_preferred_simd_mode): Likewise.
9092 * config/pa/pa.c (pa_secondary_reload): Likewise.
9093 * config/pa/predicates.md (base14_operand): Likewise.
9094 * config/powerpcspe/powerpcspe-c.c
9095 (altivec_resolve_overloaded_builtin): Likewise.
9096 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9098 (rs6000_preferred_simd_mode): Likewise.
9099 (output_vec_const_move): Likewise.
9100 (rs6000_expand_vector_extract): Likewise.
9101 (rs6000_split_vec_extract_var): Likewise.
9102 (reg_offset_addressing_ok_p): Likewise.
9103 (rs6000_legitimate_offset_address_p): Likewise.
9104 (rs6000_legitimize_address): Likewise.
9105 (rs6000_emit_set_const): Likewise.
9106 (rs6000_const_vec): Likewise.
9107 (rs6000_emit_move): Likewise.
9108 (spe_build_register_parallel): Likewise.
9109 (rs6000_darwin64_record_arg_recurse): Likewise.
9110 (swap_selector_for_mode): Likewise.
9111 (spe_init_builtins): Likewise.
9112 (paired_init_builtins): Likewise.
9113 (altivec_init_builtins): Likewise.
9114 (do_load_for_compare): Likewise.
9115 (rs6000_generate_compare): Likewise.
9116 (rs6000_expand_float128_convert): Likewise.
9117 (emit_load_locked): Likewise.
9118 (emit_store_conditional): Likewise.
9119 (rs6000_output_function_epilogue): Likewise.
9120 (rs6000_handle_altivec_attribute): Likewise.
9121 (rs6000_function_value): Likewise.
9122 (emit_fusion_gpr_load): Likewise.
9123 (emit_fusion_p9_load): Likewise.
9124 (emit_fusion_p9_store): Likewise.
9125 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9126 (fusion_gpr_mem_load): Likewise.
9127 (fusion_addis_mem_combo_load): Likewise.
9128 (fusion_addis_mem_combo_store): Likewise.
9129 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9130 (fusion_gpr_mem_load): Likewise.
9131 (fusion_addis_mem_combo_load): Likewise.
9132 (fusion_addis_mem_combo_store): Likewise.
9133 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9135 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9136 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9137 (rs6000_preferred_simd_mode): Likewise.
9138 (output_vec_const_move): Likewise.
9139 (rs6000_expand_vector_extract): Likewise.
9140 (rs6000_split_vec_extract_var): Likewise.
9141 (reg_offset_addressing_ok_p): Likewise.
9142 (rs6000_legitimate_offset_address_p): Likewise.
9143 (rs6000_legitimize_address): Likewise.
9144 (rs6000_emit_set_const): Likewise.
9145 (rs6000_const_vec): Likewise.
9146 (rs6000_emit_move): Likewise.
9147 (rs6000_darwin64_record_arg_recurse): Likewise.
9148 (swap_selector_for_mode): Likewise.
9149 (paired_init_builtins): Likewise.
9150 (altivec_init_builtins): Likewise.
9151 (rs6000_expand_float128_convert): Likewise.
9152 (emit_load_locked): Likewise.
9153 (emit_store_conditional): Likewise.
9154 (rs6000_output_function_epilogue): Likewise.
9155 (rs6000_handle_altivec_attribute): Likewise.
9156 (rs6000_function_value): Likewise.
9157 (emit_fusion_gpr_load): Likewise.
9158 (emit_fusion_p9_load): Likewise.
9159 (emit_fusion_p9_store): Likewise.
9160 * config/rx/rx.c (rx_gen_move_template): Likewise.
9161 (flags_from_mode): Likewise.
9162 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9163 (s390_slb_comparison): Likewise.
9164 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9165 (s390_vector_mode_supported_p): Likewise.
9166 (s390_cc_modes_compatible): Likewise.
9167 (s390_match_ccmode_set): Likewise.
9168 (s390_canonicalize_comparison): Likewise.
9169 (s390_emit_compare_and_swap): Likewise.
9170 (s390_branch_condition_mask): Likewise.
9171 (s390_rtx_costs): Likewise.
9172 (s390_secondary_reload): Likewise.
9173 (__SECONDARY_RELOAD_CASE): Likewise.
9174 (s390_expand_cs): Likewise.
9175 (s390_preferred_simd_mode): Likewise.
9176 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9177 * config/sh/sh.c (sh_print_operand): Likewise.
9178 (dump_table): Likewise.
9179 (sh_secondary_reload): Likewise.
9180 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9181 * config/sh/sh.md (casesi_worker_1): Likewise.
9182 (casesi_worker_2): Likewise.
9183 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9184 (fcc_comparison_operator): Likewise.
9185 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9186 (emit_soft_tfmode_cvt): Likewise.
9187 (sparc_preferred_simd_mode): Likewise.
9188 (output_cbranch): Likewise.
9189 (sparc_print_operand): Likewise.
9190 (sparc_expand_vec_perm_bmask): Likewise.
9191 (vector_init_bshuffle): Likewise.
9192 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9193 (spu_vector_mode_supported_p): Likewise.
9194 (spu_expand_insv): Likewise.
9195 (spu_emit_branch_or_set): Likewise.
9196 (spu_handle_vector_attribute): Likewise.
9197 (spu_builtin_splats): Likewise.
9198 (spu_builtin_extract): Likewise.
9199 (spu_builtin_promote): Likewise.
9200 (spu_expand_sign_extend): Likewise.
9201 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9202 (tilegx_simd_int): Likewise.
9203 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9204 (tilepro_simd_int): Likewise.
9205 * config/v850/v850.c (const_double_split): Likewise.
9206 (v850_print_operand): Likewise.
9207 (ep_memory_offset): Likewise.
9208 * config/vax/vax.c (vax_rtx_costs): Likewise.
9209 (vax_output_int_move): Likewise.
9210 (vax_output_int_add): Likewise.
9211 (vax_output_int_subtract): Likewise.
9212 * config/visium/predicates.md (visium_branch_operator): Likewise.
9213 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9214 (visium_print_operand_address): Likewise.
9215 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9216 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9217 (xtensa_expand_conditional_branch): Likewise.
9218 (xtensa_copy_incoming_a7): Likewise.
9219 (xtensa_output_literal): Likewise.
9220 * dfp.c (decimal_real_maxval): Likewise.
9221 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9224 Alan Hayward <alan.hayward@arm.com>
9225 David Sherwood <david.sherwood@arm.com>
9227 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9228 (mode_nunits_inline): Likewise.
9229 (mode_inner_inline): Likewise.
9230 (mode_unit_size_inline): Likewise.
9231 (mode_unit_precision_inline): Likewise.
9232 (emit_insn_modes_h): Likewise. Also emit a #define of the
9234 (emit_mode_wider): Add an E_ prefix to mode names.
9235 (emit_mode_complex): Likewise.
9236 (emit_mode_inner): Likewise.
9237 (emit_mode_adjustments): Likewise.
9238 (emit_mode_int_n): Likewise.
9239 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9240 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9241 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9242 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9243 (CRC32_BUILTIN, ENTRY): Likewise.
9244 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9245 (aarch64_pop_regs): Likewise.
9246 (aarch64_process_components): Likewise.
9247 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9248 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9249 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9250 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9251 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9252 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9254 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9255 (pcmpistr): Likewise.
9256 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9257 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9258 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9260 * config/rl78/rl78.c (mduc_regs): Likewise.
9261 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9262 (htm_expand_builtin): Likewise.
9263 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9264 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9265 * config/xtensa/xtensa.c (print_operand): Likewise.
9266 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9267 (NUM_MODE_VECTOR_INT): Likewise.
9268 * genoutput.c (null_operand): Likewise.
9269 (output_operand_data): Likewise.
9270 * genrecog.c (print_parameter_value): Likewise.
9271 * lra.c (debug_operand_data): Likewise.
9273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9275 * dwarf2out.c (scompare_loc_descriptor_wide)
9276 (scompare_loc_descriptor_narrow): New functions, split out from...
9277 (scompare_loc_descriptor): ...here.
9278 * expmed.c (emit_store_flag_int): New function, split out from...
9279 (emit_store_flag): ...here.
9281 2017-08-30 Richard Biener <rguenther@suse.de>
9283 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9284 (dwarf2out_early_finish): Move setting of AT_pubnames from
9285 early debug output to early finish.
9287 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9289 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9290 and -mdata-region to the assembler.
9292 2017-08-30 Richard Biener <rguenther@suse.de>
9294 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9296 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9297 (dwarf2out_early_global_decl): Only generate a DIE for the
9298 abstract origin if it doesn't already exist or is a declaration DIE.
9299 (resolve_addr): Do not add the linkage name twice when
9300 generating a stub DIE for the DW_TAG_GNU_call_site target.
9302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9304 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9305 Use machine_mode rather than int for arg1_mode.
9307 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9310 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9311 that the second argument of the built-in functions to unpack
9312 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9313 switch statement instead a lot of if statements.
9314 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9315 Allow 64-bit values to be in Altivec registers as well as
9316 traditional floating point registers.
9317 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9319 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9321 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9322 MAX_REGS_PER_ADDRESS == 1.
9324 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9326 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9327 * config/i386/i386.c (ix86_option_override_internal): Simplify
9328 setting of opts->x_flag_entry.
9330 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9331 Jakub Jelinek <jakub@redhat.com>
9332 Richard Biener <rguenther@suse.de>
9334 PR tree-optimization/81503
9335 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9336 folded constant fits in the target type; reorder tests for clarity.
9338 2017-08-29 Martin Liska <mliska@suse.cz>
9340 * passes.def: Include pass_lower_switch.
9341 * stmt.c (dump_case_nodes): Remove and move to
9342 tree-switch-conversion.
9343 (case_values_threshold): Likewise.
9344 (expand_switch_as_decision_tree_p): Likewise.
9345 (emit_case_decision_tree): Likewise.
9346 (expand_case): Likewise.
9347 (balance_case_nodes): Likewise.
9348 (node_has_low_bound): Likewise.
9349 (node_has_high_bound): Likewise.
9350 (node_is_bounded): Likewise.
9351 (emit_case_nodes): Likewise.
9352 (struct simple_case_node): New struct.
9353 (add_case_node): Remove.
9354 (emit_case_dispatch_table): Use vector instead of case_list.
9355 (reset_out_edges_aux): Remove.
9356 (compute_cases_per_edge): Likewise.
9357 (expand_case): Build list of simple_case_node.
9358 (expand_sjlj_dispatch_table): Use it.
9359 * tree-switch-conversion.c (struct case_node): Moved from
9360 stmt.c and adjusted.
9361 (emit_case_nodes): Likewise.
9362 (node_has_low_bound): Likewise.
9363 (node_has_high_bound): Likewise.
9364 (node_is_bounded): Likewise.
9365 (case_values_threshold): Likewise.
9366 (reset_out_edges_aux): Likewise.
9367 (compute_cases_per_edge): Likewise.
9368 (add_case_node): Likewise.
9369 (dump_case_nodes): Likewise.
9370 (balance_case_nodes): Likewise.
9371 (expand_switch_as_decision_tree_p): Likewise.
9372 (emit_jump): Likewise.
9373 (emit_case_decision_tree): Likewise.
9374 (try_switch_expansion): Likewise.
9375 (do_jump_if_equal): Likewise.
9376 (emit_cmp_and_jump_insns): Likewise.
9377 (fix_phi_operands_for_edge): New function.
9378 (record_phi_operand_mapping): Likewise.
9379 (class pass_lower_switch): New pass.
9380 (pass_lower_switch::execute): New function.
9381 (make_pass_lower_switch): Likewise.
9382 (conditional_probability):
9383 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9384 * tree-pass.h: Add make_pass_lower_switch.
9386 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9389 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9392 2017-08-29 Richard Biener <rguenther@suse.de>
9394 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9395 we do not add a DW_AT_inline attribute twice.
9396 (gen_subprogram_die): Remove code setting DW_AT_inline on
9397 DECL_ABSTRACT_P nodes.
9399 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9401 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9402 calls to internal functions.
9403 (gimplify_modify_expr): Likewise.
9404 * tree-call-cdce.c (use_internal_fn): Likewise.
9405 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9406 (convert_to_divmod): Set the nothrow flag.
9407 * tree-if-conv.c (predicate_mem_writes): Likewise.
9408 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9409 (vectorizable_call): Likewise.
9410 (vectorizable_store): Likewise.
9411 (vectorizable_load): Likewise.
9412 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9413 (vect_recog_mask_conversion_pattern): Likewise.
9415 2017-08-29 Martin Liska <mliska@suse.cz>
9418 * gcc.c (driver_handle_option): Add new argument.
9419 * opts-common.c (handle_option): Pass
9420 target_option_override_hook.
9421 * opts-global.c (lang_handle_option): Add new option.
9422 (set_default_handlers): Add new argument.
9423 (decode_options): Likewise.
9424 * opts.c (target_handle_option): Likewise.
9425 (common_handle_option): Call target_option_override_hook.
9426 * opts.h (struct cl_option_handler_func): Add hook for
9427 target option override.
9428 (struct cl_option_handlers): Likewise.
9429 (set_default_handlers): Add new argument.
9430 (decode_options): Likewise.
9431 (common_handle_option): Likewise.
9432 (target_handle_option): Likewise.
9433 * toplev.c (toplev::main): Pass targetm.target_option.override
9436 2017-08-29 Richard Biener <rguenther@suse.de>
9437 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9439 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9440 life to the active subtree.
9442 2017-08-28 Jeff Law <law@redhat.com>
9444 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9445 derive_equivalences.
9446 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9448 (edge_info::derive_equivalences): New private member function
9450 * tree-ssa-dom.c (class edge_info): Changed from a struct
9451 to a class. Add ctor/dtor, methods and data members.
9452 (edge_info::edge_info): Renamed from allocate_edge_info.
9453 Initialize additional members.
9454 (edge_info::~edge_info): New.
9455 (free_dom_edge_info): Delete the edge info.
9456 (record_edge_info): Use new class & associated member functions.
9457 Tighten forms for testing for edge equivalences.
9458 (record_temporary_equivalences): Iterate over the simple
9459 equivalences rather than assuming there's only one per edge.
9460 (cprop_into_successor_phis): Iterate over the simple
9461 equivalences rather than assuming there's only one per edge.
9462 (optimize_stmt): Use operand_equal_p rather than pointer
9463 equality for mini-DSE code.
9465 2017-08-28 Nathan Sidwell <nathan@acm.org>
9467 * gcc.c (execute): Fold SIGPIPE handling into switch
9468 statement. Adjust internal error message.
9470 2017-08-28 Richard Biener <rguenther@suse.de>
9473 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9474 Do nothing for removed DIEs.
9476 2017-08-28 Richard Biener <rguenther@suse.de>
9478 PR tree-optimization/81977
9479 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9482 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9485 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9486 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9487 using targetm.gen_mem_thread_fence.
9489 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9492 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9493 predicate to register_operand. Reorder operands.
9494 (*btr<mode>): Ditto.
9495 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9496 (*btr<mode>_mask): Ditto.
9498 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9500 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9501 * config/rs6000/xmmintrin.h: New file.
9502 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9504 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9507 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9508 parameter and_insn and return it.
9509 (recombine_lvx_pattern): Insert a copy to ensure availability of
9510 the base register of the copied masking operation at the point of
9511 the instruction replacement.
9512 (recombine_stvx_pattern): Likewise.
9514 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9516 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9517 undocumented switches.
9518 (-mpower9-dform-vector): Likewise.
9519 (-mpower9-dform): Likewise.
9520 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9521 comments to delete references to -mpower9-dform* switches.
9522 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9523 Delete reference to -mpower9-dform* switches, test for
9524 -mpower9-vector instead.
9525 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9526 (OTHER_P9_VECTOR_MASKS): Likewise.
9527 (POWERPC_MASKS): Likewise.
9528 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9529 tests against -mpower9-dform* to -mpower9-vector. Delete code
9530 that checked for -mpower9-dform* consistancy with other options.
9531 Add test for -mpower9-misc to enable other power9 switches.
9532 (rs6000_init_hard_regno_mode_ok): Likewise.
9533 (rs6000_option_override_internal): Likewise.
9534 (rs6000_emit_prologue): Likewise.
9535 (rs6000_emit_epilogue): Likewise.
9536 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9537 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9538 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9540 (emit_fusion_p9_store): Likewise.
9541 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9542 resetting these macros if the assembler does not support ISA 3.0
9544 (TARGET_P9_DFORM_VECTOR): Likewise.
9545 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9546 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9548 2017-08-25 Alan Modra <amodra@gmail.com>
9550 PR rtl-optimization/81747
9551 * cse.c (cse_extended_basic_block): Don't attempt to record
9552 equivalences for degenerate conditional branches that branch
9553 to their fall-through.
9555 2017-08-24 Martin Sebor <msebor@redhat.com>
9558 * gimple-fold.c (size_must_be_zero_p): New function.
9559 (gimple_fold_builtin_memory_op): Call it.
9561 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9563 * config/rs6000/mm_malloc.h: New file.
9565 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9567 PR tree-optimization/81913
9568 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9569 analysis when either IVs in condition can wrap.
9571 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9573 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9574 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9576 2017-08-24 Richard Biener <rguenther@suse.de>
9579 * targhooks.c (default_target_can_inline_p): Properly
9580 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9581 is present and always compare.
9582 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9583 infer -mfpmath=sse from TARGET_SSE_P.
9584 (ix86_can_inline_p): Properly use target_option_default_node when
9585 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9587 2017-08-24 Richard Biener <rguenther@suse.de>
9590 * dwarf2out.c (output_die): Handle flag_generate_offload like
9592 (output_comp_unit): Likewise.
9593 (gen_array_type_die): Likewise.
9594 (dwarf2out_early_finish): Likewise.
9595 (note_variable_value_in_expr): Likewise.
9596 (dwarf2out_finish): Likewise. Adjust assert.
9597 * cgraphunit.c (symbol_table::compile): Move setting of
9598 flag_generate_offload earlier ...
9599 (symbol_table::finalize_compilation_unit): ... here, before
9600 early debug finalization.
9602 2017-08-24 Richard Biener <rguenther@suse.de>
9604 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9605 and ipa-fnsummary.h.
9606 (ix86_can_inline_p): When ix86_fpmath flags do not match
9607 check whether the callee uses FP math at all.
9609 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9612 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9613 instead of TYPE_PRECISION.
9615 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9616 Alan Hayward <alan.hayward@arm.com>
9617 David Sherwood <david.sherwood@arm.com>
9619 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9621 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9623 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9625 (ix86_frame::stack_realign_allocate): New field.
9626 (struct machine_frame_state): Modify comments.
9627 (machine_frame_state::sp_realigned_fp_end): New field.
9628 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9630 (sp_valid_at): Add assertion to assure no attempt to access invalid
9631 offset of a realigned stack.
9632 (fp_valid_at): Likewise.
9633 (choose_baseaddr): Modify comments.
9634 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9635 ix86_expand_prologue.
9636 (ix86_expand_prologue): Modify stack realignment and allocation.
9637 (ix86_expand_epilogue): Modify comments.
9638 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9639 avx2_runtime, avx512f, and avx512f_runtime.
9641 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9643 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9644 (mstackrealign): Do not init to -1.
9645 * config/i386/i386.c (ix86_option_override_internal):
9646 Check opts_set, not opts when setting default value of
9647 opts->x_ix86_force_align_arg_pointer.
9649 2017-08-23 Richard Biener <rguenther@suse.de>
9651 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9652 lang_hooks.decl_printable_name.
9653 * print-rtl-function.c (print_rtx_function): Likewise.
9654 * tree-pretty-print.c (dump_function_header): Likewise.
9656 2017-08-23 Richard Biener <rguenther@suse.de>
9659 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9660 -g0 at compile-time.
9662 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9665 * doc/sourcebuild.texi (Other hardware attributes):
9668 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9671 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9672 Fix single-use check.
9674 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9676 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9678 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9680 * doc/install.texi: Modify to add more details on running selected
9683 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9685 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9686 is combined with -mabi=ms.
9687 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9690 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9692 PR tree-optimization/81488
9693 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9694 and cached_basis fields.
9695 (MAX_SPREAD): New constant.
9696 (alloc_cand_and_find_basis): Initialize new fields.
9697 (clear_visited): New function.
9698 (create_phi_basis_1): Rename from create_phi_basis, set visited
9699 and cached_basis fields.
9700 (create_phi_basis): New wrapper function.
9701 (phi_add_costs_1): Rename from phi_add_costs, add spread
9702 parameter, set visited field, short-circuit when limits reached.
9703 (phi_add_costs): New wrapper function.
9704 (record_phi_increments_1): Rename from record_phi_increments, set
9706 (record_phi_increments): New wrapper function.
9707 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9708 (phi_incr_cost): New wrapper function.
9709 (all_phi_incrs_profitable_1): Rename from
9710 all_phi_incrs_profitable, set visited field.
9711 (all_phi_incrs_profitable): New wrapper function.
9713 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9714 Alan Hayward <alan.hayward@arm.com>
9715 David Sherwood <david.sherwood@arm.com>
9717 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9718 that takes the outer and inner modes.
9719 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9720 comparison as the canonical test for a paradoxical subreg.
9721 * combine.c (simplify_set): Use paradoxical_subreg_p.
9722 (make_extraction): Likewise.
9723 (force_to_mode): Likewise.
9724 (rtx_equal_for_field_assignment_p): Likewise.
9725 (gen_lowpart_for_combine): Likewise.
9726 (simplify_comparison): Likewise.
9727 * cse.c (equiv_constant): Likewise.
9728 * expmed.c (store_bit_field_1): Likewise.
9729 * final.c (alter_subreg): Likewise.
9730 * fwprop.c (propagate_rtx): Likewise.
9731 (forward_propagate_subreg): Likewise.
9732 * ira-conflicts.c (ira_build_conflicts): Likewise.
9733 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9734 * lra-constraints.c (curr_insn_transform): Likewise.
9735 (split_reg): Likewise.
9736 * lra-eliminations.c (move_plus_up): Likewise.
9737 (lra_eliminate_regs_1): Likewise.
9738 * recog.c (general_operand): Likewise.
9739 * ree.c (combine_reaching_defs): Likewise.
9740 * reload.c (push_reload): Likewise.
9741 (find_reloads): Likewise.
9742 * reload1.c (elimination_effects): Likewise.
9743 (compute_reload_subreg_offset): Likewise.
9744 (choose_reload_regs): Likewise.
9745 * rtlanal.c (subreg_lsb_1): Likewise.
9746 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9747 (simplify_subreg): Likewise.
9748 * var-tracking.c (track_loc_p): Likewise.
9749 * emit-rtl.c (byte_lowpart_offset): Likewise.
9750 (paradoxical_subreg_p): Delete out-of-line definition.
9752 2017-08-22 Jeff Law <law@redhat.com>
9754 PR tree-optimization/81741
9755 PR tree-optimization/71947
9756 * tree-ssa-dom.c: Include tree-inline.h.
9757 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9758 equivalences if one is more expensive to compute than the other.
9759 * tree-ssa-scopedtables.h (class const_or_copies): Make
9760 record_const_or_copy_raw method private.
9761 (class avail_exprs_stack): New method simplify_binary_operation.
9762 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9763 avail_exprs_stack::simplify_binary_operation as needed.
9764 (avail_exprs_stack::simplify_binary_operation): New function.
9766 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9768 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9769 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9770 (DOT_SYMBOLS): Likewise.
9771 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9772 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9773 (RS6000_ABI_NAME): Likewise.
9774 (TARGET_CMODEL): Likewise.
9775 (TOC_SECTION_ASM_OP): Likewise.
9776 (SET_CMODEL): New macro.
9777 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9779 2017-08-22 Richard Biener <rguenther@suse.de>
9781 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9782 to allow for free-lang-data replacements similar to verify_type_variant.
9784 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9785 Alan Hayward <alan.hayward@arm.com>
9786 David Sherwood <david.sherwood@arm.com>
9788 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9789 VOIDmode for the LABEL_REF.
9791 2017-08-22 Richard Biener <rguenther@suse.de>
9793 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9795 2017-08-22 Richard Biener <rguenther@suse.de>
9797 * common.opt (feliminate-dwarf2-dups): Ignore.
9798 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9799 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9800 same_die_p_wrap, compute_section_prefix,
9801 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9802 (comdat_symbol_id, comdat_symbol_number): Likewise.
9803 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9805 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9806 (output_die): Mark unreachable path unreachable.
9807 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9808 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9809 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9810 (dwarf2out_early_finish): Likewise.
9812 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9814 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9816 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9819 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9820 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9821 (avr_attribute_table) <io, io_low, address>: Initialize
9822 .decl_required with true.
9824 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9826 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9827 undocumented debugging options.
9828 (-mvsx-scalar-double): Likewise.
9829 (-mallow-df-permute): Likewise.
9830 (-mvectorize-builtins): Likewise.
9831 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9832 (rs6000_builtin_vectorized_function): Likewise.
9833 (rs6000_builtin_md_vectorized_function): Likewise.
9834 (rs6000_opt_vars): Likewise.
9836 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9839 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9840 (*btrq_imm): Rename from *btrq.
9841 (*btcq_imm): Rename from *btcq.
9842 (btsc): New code attribute.
9843 (*<btsc><mode>): New insn pattern.
9844 (*btr<mode>): Ditto.
9845 (*<btsc><mode>_mask): New insn_and_split pattern.
9846 (*btr<mode>_mask): Ditto.
9848 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9849 Alan Hayward <alan.hayward@arm.com>
9850 David Sherwood <david.sherwood@arm.com>
9852 * function.c (pad_below): Simplify padding calculation.
9854 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9855 Alan Hayward <alan.hayward@arm.com>
9856 David Sherwood <david.sherwood@arm.com>
9858 * target.def (function_prologue): Remove frame size argument.
9859 (function_epilogue): Likewise.
9860 * doc/tm.texi: Regenerate.
9861 * final.c (final_start_function): Update call to function_prologue.
9862 (final_end_function): Update call to function_epilogue.
9863 (default_function_pro_epilogue): Remove frame size argument.
9864 * output.h (default_function_pro_epilogue): Likewise.
9865 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9866 (arm_output_function_prologue): Likewise.
9867 * config/frv/frv.c (frv_function_prologue): Likewise.
9868 (frv_function_epilogue): Likewise.
9869 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9870 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9871 (ia64_output_function_epilogue): Likewise.
9872 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9873 (m32r_output_function_epilogue): Likewise.
9874 * config/microblaze/microblaze.c (microblaze_function_prologue)
9875 (microblaze_function_epilogue): Likewise.
9876 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9877 (mips_output_function_epilogue): Likewise.
9878 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9879 (mmix_target_asm_function_epilogue): Likewise.
9880 * config/msp430/msp430.c (msp430_start_function): Likewise.
9881 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9882 (nds32_asm_function_epilogue): Likewise.
9883 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9884 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9885 (pa_output_function_epilogue): Likewise.
9886 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9887 (rs6000_output_function_epilogue): Likewise.
9888 * config/rl78/rl78.c (rl78_start_function): Likewise.
9889 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9890 (rs6000_output_function_epilogue): Likewise.
9891 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9892 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9893 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9894 (sparc_asm_function_epilogue): Likewise.
9896 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9898 * tree.h (type_has_mode_precision_p): New function.
9899 * convert.c (convert_to_integer_1): Use it.
9900 * expr.c (expand_expr_real_2): Likewise.
9901 (expand_expr_real_1): Likewise.
9902 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9903 * match.pd: Likewise.
9904 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9905 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9906 * tree-tailcall.c (process_assignment): Likewise.
9907 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9908 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9909 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9910 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9911 (vectorizable_assignment): Likewise.
9912 (vectorizable_shift): Likewise.
9913 (vectorizable_operation): Likewise.
9914 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9916 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9918 * match.pd: Add pow (C, x) simplification.
9920 2017-08-21 Richard Biener <rguenther@suse.de>
9922 PR tree-optimization/81900
9923 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9924 for blocks with abnormal predecessors.
9925 (compute_antic): Do not set visited flag prematurely.
9927 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9930 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9932 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9934 * stor-layout.h (vector_type_mode): Move to...
9935 * tree.h (vector_type_mode): ...here.
9936 * stor-layout.c (vector_type_mode): Move to...
9937 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9939 2017-08-21 Richard Biener <rguenther@suse.de>
9941 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9942 register_external_die hooks.
9943 (debug_false_tree_charstarstar_uhwistar): Declare.
9944 (debug_nothing_tree_charstar_uhwi): Likewise.
9945 * debug.c (do_nothing_debug_hooks): Adjust.
9946 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9947 (debug_nothing_tree_charstar_uhwi): Likewise.
9948 * dbxout.c (dbx_debug_hooks): Adjust.
9949 (xcoff_debug_hooks): Likewise.
9950 * sdbout.c (sdb_debug_hooks): Likewise.
9951 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9952 * dwarf2out.c (macinfo_label_base): New global.
9953 (dwarf2out_register_external_die): New function for the
9954 register_external_die hook.
9955 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9956 (dwarf2_debug_hooks): Use them.
9957 (dwarf2_lineno_debug_hooks): Adjust.
9958 (struct die_struct): Add with_offset flag.
9959 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9960 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9961 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9962 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9963 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9964 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9965 defining section names for the early LTO debug variants.
9966 (reset_indirect_string): New helper.
9967 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9968 (print_dw_val): Add support for offsetted symbol references.
9969 (get_ultimate_context): Split out from is_cxx.
9970 (is_cxx): Use get_ultimate_context.
9971 (is_fortran): Add decl overload.
9972 (compute_comp_unit_symbol): Split out worker from
9973 compute_section_prefix.
9974 (compute_section_prefix): Call compute_comp_unit_symbol and
9975 set comdat_type_p here.
9976 (output_die): Skip DIE symbol output for the LTO added one.
9977 Handle DIE symbol references with offset.
9978 (output_comp_unit): Guard section name mangling properly.
9979 For LTO debug sections emit a symbol at the section beginning
9980 which we use to refer to its DIEs.
9981 (add_abstract_origin_attribute): For DIEs registered via
9982 dwarf2out_register_external_die directly refer to the early
9983 DIE rather than indirectly through the shadow one we created.
9984 Remove obsolete call to dwarf2out_abstract_function for
9985 non-function/block origins.
9986 (gen_array_type_die): When generating early LTO debug do
9987 not emit DW_AT_string_length.
9988 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9989 late when in LTO. As suggested place a gcc_unreachable for
9990 the DECL_ABSTRACT_P case.
9991 (gen_subprogram_die): Avoid another specification DIE
9992 for early built declarations/definitions for the late LTO case.
9993 (gen_variable_die): Add type references for late duplicated VLA dies
9995 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9996 we have the abstract instance already.
9997 (process_scope_var): Adjust decl DIE contexts in LTO which
9998 first puts them in limbo.
9999 (gen_decl_die): Do not generate type DIEs late apart from
10000 types for VLAs or for decls we do not yet have a DIE. Do not
10001 call dwarf2out_abstract_function late.
10002 (dwarf2out_early_global_decl): Make sure to create DIEs
10003 for abstract instances of a decl first.
10004 (dwarf2out_late_global_decl): Adjust comment.
10005 (output_macinfo_op): With multiple macro sections use
10006 macinfo_label_base to distinguish labels.
10007 (output_macinfo): Likewise. Update macinfo_label_base.
10008 Pass in the line info label.
10009 (note_variable_value_in_expr): When generating LTO resolve
10010 all variable values here by generating DIEs as needed.
10011 (init_sections_and_labels): Add early LTO debug flag parameter
10012 and generate different sections and names if set. Add generation
10013 counter for the labels so we can have multiple of them.
10014 (reset_dies): Helper to allow DIEs to be output multiple times.
10015 (dwarf2out_finish): When outputting DIEs to the fat part of an
10016 LTO object first reset DIEs.
10017 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10018 (modified_type_die): Check for decl_ultimate_origin being self
10020 (gen_type_die_with_usage): Likewise.
10021 (gen_typedef_die): Allow decl_ultimate_origin being self.
10022 (set_decl_abstract_flags): Remove.
10023 (set_block_abstract_flags): Likewise.
10024 (dwarf2out_abstract_function): Treat the early generated DIEs
10025 as the abstract copy and only add DW_AT_inline and
10026 DW_AT_artificial here and call set_decl_origin_self.
10027 If the DIE has an abstract origin don't do anything.
10028 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10029 if we have none yet (Go fails to build one, PR78628).
10030 (variably_modified_type_p): Prevent endless recursion for Ada
10031 cyclic pointer types.
10032 * lto-streamer-in.c: Include debug.h.
10033 (dref_queue): New global.
10034 (lto_read_tree_1): Stream in DIE references.
10035 (lto_input_tree): Register DIE references.
10036 (input_function): Stream DECL_DEBUG_ARGS.
10037 * lto-streamer-out.c: Include debug.h.
10038 (lto_write_tree_1): Output DIE references.
10039 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10040 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10041 (output_function): Stream DECL_DEBUG_ARGS.
10042 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10043 Stream DECL_ABSTRACT_ORIGIN.
10044 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10045 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10046 DECL_CONTEXT for file-scope decls.
10047 * lto-streamer.h (struct dref_entry): Declare.
10048 (dref_queue): Likewise.
10049 * cfgexpand.c (pass_expand::execute): Do not call the
10050 outlining_inline_function hook here.
10051 * lto-wrapper.c (debug_obj): New global.
10052 (tool_cleanup): Unlink it if required.
10053 (debug_objcopy): New function.
10054 (run_gcc): Handle early debug sections in the IL files by
10055 extracting them to separate files, partially linkin them and
10056 feeding the result back as result to the linker.
10057 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10058 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10059 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10060 sections into a separate segment.
10061 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10063 (darwin_asm_dwarf_section): Likewise.
10064 (darwin_asm_output_dwarf_offset): Likewise.
10065 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10067 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10068 Alan Hayward <alan.hayward@arm.com>
10069 David Sherwood <david.sherwood@arm.com>
10071 * read-md.h (md_reader::record_potential_iterator_use): Replace
10072 pointer argument with an rtx and an index.
10073 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10074 (apply_mode_iterator): Likewise.
10075 (apply_code_iterator): Likewise.
10076 (apply_int_iterator): Likewise.
10077 (apply_subst_iterator): Likewise.
10078 (record_iterator_use): Likewise.
10079 (record_attribute_use): Likewise.
10080 (md_reader::record_potential_iterator_use): Likewise. Update calls
10081 to record_iterator_use and apply_iterator.
10082 (iterator_use): Replace ptr with x and index.
10083 (attribute_use): Likewise.
10084 (apply_attribute_uses): Update calls to apply_iterator.
10085 (apply_iterators): Likewise. Update initialization of iterator_use.
10086 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10087 and record_potential_iterator_use.
10088 (rtx_reader::read_rtx_operand): Likewise.
10090 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10091 Alan Hayward <alan.hayward@arm.com>
10092 David Sherwood <david.sherwood@arm.com>
10094 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10097 2017-08-21 Richard Biener <rguenther@suse.de>
10099 PR middle-end/81884
10100 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10101 at struct end conservatively when comparing common bases.
10103 2017-08-21 Richard Biener <rguenther@suse.de>
10105 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10106 (mem_ref_in_stmt): Remove.
10107 (determine_max_movement): Use ref index to get at the reference.
10108 (invariantness_dom_walker::before_dom_children): Deal with
10109 lim data already initialized.
10110 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10112 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10114 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10115 (TARGET_ISA_ROUND): Ditto.
10116 (TARGET_ROUND): Ditto.
10117 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10118 * config/i386/i386.md: Ditto.
10119 * config/i386/sse.md: Ditto.
10120 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10121 with OPTION_MASK_ISA_SSE4_1.
10123 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10126 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10127 __builtin_ia32_lzcnt_u16.
10129 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10132 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10133 (rs6000_set_current_function): Rewrite function to use it.
10135 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10138 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10139 and TYPE_WARN_IF_NOT_ALIGN.
10140 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10141 (handle_warn_if_not_align): New.
10142 (place_union_field): Call handle_warn_if_not_align.
10143 (place_field): Call handle_warn_if_not_align.
10144 Copy TYPE_WARN_IF_NOT_ALIGN.
10145 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10146 (layout_type): Likewise.
10147 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10149 (tree_decl_common): Add warn_if_not_align.
10150 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10151 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10152 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10153 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10154 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10155 * doc/extend.texi: Document warn_if_not_aligned attribute.
10156 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10158 2017-08-17 Martin Liska <mliska@suse.cz>
10161 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10162 (get_data_dependence): Use it as pointer type.
10163 (distribute_loop): Likewise.
10165 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10167 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10168 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10169 (p8_vmrgow_<mode>_direct): New define_insn.
10170 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10171 handle endianness for vmrgew and vmrgow permute patterns.
10173 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10175 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10176 * config/rs6000/rs6000-cpus.def: Remove comment.
10177 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10178 (POWERPC_MASKS): Likewise.
10179 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10180 use of TARGET_VSX_TIMODE.
10181 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10182 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10183 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10184 (rs6000_option_override_internal): Remove dead code.
10185 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10186 (rs6000_legitimize_reload_address): Likewise.
10187 (rs6000_legitimate_address_p): Likewise.
10188 (rs6000_opt_masks): Delete "vsx-timode".
10189 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10190 from function comment.
10191 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10192 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10193 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10195 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10196 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10197 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10198 (VSX_TI): Likewise.
10200 (define_peephole2): Likewise.
10202 2017-08-17 Martin Sebor <msebor@redhat.com>
10205 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10206 past the end of an array.
10207 (test_pp_format): Add test cases.
10209 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10211 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10212 missing ECF_NOTHROW flags.
10214 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10217 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10218 operands residing in integer registers.
10219 (*vsx_le_perm_load_<mode>): Likewise.
10220 (*vsx_le_perm_store_<mode>): Likewise.
10221 (define_peephole2): Add peepholes to optimize the above.
10223 2017-08-17 Marek Polacek <polacek@redhat.com>
10225 PR middle-end/81814
10226 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10227 to mimic what shorten_compare did. Change the return type to bool.
10228 (fold_cond_expr_with_comparison): Update call to
10229 operand_equal_for_comparison_p.
10230 (fold_ternary_loc): Likewise.
10232 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10234 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10236 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10238 2017-08-17 Richard Biener <rguenther@suse.de>
10240 * tree-ssa-structalias.c (solve_graph): When propagating
10241 to successors update the graphs succ edges and avoid duplicate work.
10243 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10246 * config/i386/i386.c (ix86_option_override_internal): Save target
10247 specific options after ix86_stack_protector_guard_reg was changed.
10249 2017-08-17 Richard Biener <rguenther@suse.de>
10251 PR tree-optimization/81827
10252 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10253 (new_var_info): Initialize it conservatively.
10254 (get_call_vi): Mark register vars.
10255 (new_scalar_tmp_constraint_exp): Likewise.
10256 (handle_rhs_call): Likewise.
10257 (handle_const_call): Likewise.
10258 (create_function_info_for): Likewise.
10259 (solve_constraints): Sort varinfos to separate register from
10260 non-register vars to pack points-to solution bitmaps during
10263 2017-08-17 Marek Polacek <polacek@redhat.com>
10265 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10267 2017-08-17 Richard Biener <rguenther@suse.de>
10269 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10270 to true when overflow is undefined and we saturated the result.
10272 2017-08-17 Alan Modra <amodra@gmail.com>
10275 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10276 Don't use store multiple if only one reg needs saving.
10277 (interesting_frame_related_regno): New function.
10278 (rs6000_frame_related): Don't emit frame info for regs that
10280 (rs6000_emit_epilogue): Likewise.
10282 2017-08-16 Nathan Sidwell <nathan@acm.org>
10284 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10285 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10286 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10287 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10288 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10289 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10290 (verify_type): Adjust for TYPE_BINFO move.
10291 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10292 process TYPE_BINFO directly.
10293 (hash_tree): Likewise.
10294 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10296 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10299 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10301 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10303 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10306 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10307 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10308 Add ix86_binary_operator_ok to insn constraint.
10309 (*iordi_1_bts): Ditto.
10310 (*xordi_1_btc): Ditto.
10311 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10312 Update corresponding peephole2 pattern.
10316 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10318 PR tree-optimization/81832
10319 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10320 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10322 2017-08-16 Marek Polacek <polacek@redhat.com>
10325 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10326 regarding size_zero_node.
10328 2017-08-16 Martin Liska <mliska@suse.cz>
10331 * config.gcc: Respect previously set extra_objs in case
10334 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10336 PR tree-optimization/81835
10337 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10338 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10339 not depend on the phi.
10341 2017-08-16 Alan Modra <amodra@gmail.com>
10343 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10346 2017-08-16 Alan Modra <amodra@gmail.com>
10348 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10349 (save_reg_p): ..into this. Update all callers.
10350 (first_reg_to_save): Simplify.
10352 2017-08-16 Alan Modra <amodra@gmail.com>
10354 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10357 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10361 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10362 constant count if that count is less than 32.
10364 2017-08-15 Nathan Sidwell <nathan@acm.org>
10366 * gcc.c (execute): Emit friendlier message if inferior is killed
10367 by an external cause.
10369 2017-08-15 Richard Biener <rguenther@suse.de>
10371 PR tree-optimization/81790
10372 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10373 CONSTRUCTORs from simplifying and VN.
10375 2017-08-14 Martin Sebor <msebor@redhat.com>
10377 * builtin-attrs.def: Add comments.
10379 2017-08-14 Martin Sebor <msebor@redhat.com>
10382 * doc/extend.texi (attribute nonstring): Document new attribute.
10384 2017-08-14 Martin Sebor <msebor@redhat.com>
10387 * tree-diagnostic.c (default_tree_printer): Handle %G.
10388 * gimple-pretty-print.h (percent_G_format): Declare new function.
10389 * gimple-pretty-print.c (percent_G_format): Define.
10390 * tree-pretty-print.c (percent_K_format): Add argument.
10392 2017-08-14 Martin Sebor <msebor@redhat.com>
10394 PR translation/79998
10395 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10396 Remove a stray space.
10398 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10401 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10402 (*iordi_1_bts): Ditto.
10403 (*xordi_1_btc): Ditto.
10405 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10408 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10411 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10412 quoted strings, and make more translator-friendly.
10413 (darwin_rs6000_override_options): Likewise.
10414 (rs6000_option_override_internal): Likewise.
10415 (rs6000_return_in_memory): Fix overlong line.
10416 (init_cmulative_args): Use quoted strings, and make more
10417 translator-friendly.
10418 (rs6000_pass_by_reference): Fix overlong line.
10419 (def_builtin): Use quoted strings.
10420 (altivec_expand_predicate_builtin): Use quoted strings, and make
10421 more translator-friendly.
10422 (htm_expand_builtin): Use quoted strings.
10423 (cpu_expand_builtin): Use quoted strings, and make more
10424 translator-friendly.
10425 (altivec_expand_builtin): Likewise.
10426 (paired_expand_predicate_builtin): Likewise.
10427 (rs6000_invalid_builtin): Likewise.
10428 (builtin_function_type): Use quoted strings.
10429 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10430 more translator-friendly.
10431 (rs6000_trampoline_init): Likewise.
10432 (rs6000_handle_altivec_attribute): Likewise.
10433 (rs6000_inner_target_options): Use quoted strings.
10434 (rs6000_disable_incompatible_switches): Likewise.
10435 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10436 strings, and make more translator-friendly.
10437 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10439 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10441 PR tree-optimization/81799
10442 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10443 cond_expr to simple gimple operand.
10445 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10447 PR middle-end/46932
10448 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10450 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10454 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10455 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10456 TARGET_GASISR_PROLOGUES.
10457 * config/avr/avr.c (avr_option_override): Same.
10458 (avr_pass_pre_proep::execute): Same.
10460 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10463 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10464 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10466 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10468 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10469 define_insn_and_split. Split to a memory load from 0 in
10470 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10471 using PTR mode iterator.
10472 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10473 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10474 (*add_tp_<mode>): Redefine as define_insn_and_split.
10475 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10476 address space. Merge with *add_tp_x32 using PTR mode iterator.
10477 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10478 Split to an add with a memory load from 0 in
10479 DEFAULT_TLS_SEG_REG address space.
10481 2017-08-12 Andrew Pinski <apinski@cavium.com>
10483 * config/aarch64/aarch64-option-extensions.def (rdma):
10484 Fix feature string to what Linux prints out in /proc/cpuinfo.
10486 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10489 * dwarf2out.c (modified_type_die): For C typedef types that have
10490 an ultimate origin, process the ultimate origin instead of the
10492 (gen_typedef_die): Assert that input DECLs have no ultimate
10494 (gen_type_die_with_usage): For typedef variants that have an
10495 ultimate origin, just call gen_decl_die on the original DECL.
10496 (process_scope_var): Avoid creating DIEs for local typedefs and
10497 concrete static variables.
10499 2017-08-12 Alan Modra <amodra@gmail.com>
10503 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10504 match gnu-user.h startfile.
10505 (ENDFILE_LINUX_SPEC): Similarly.
10507 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10510 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10512 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10514 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10515 * config/aarch64/aarch64.md (mov<mode>): Change.
10516 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10517 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10518 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10520 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10522 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10523 for storage order barriers.
10525 2017-08-11 Martin Liska <mliska@suse.cz>
10528 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10529 variables of void type.
10531 2017-08-11 Martin Liska <mliska@suse.cz>
10533 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10534 TARGET_SUPPORTS_ALIASES.
10535 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10536 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10537 (optimize_weakref): Likewise.
10538 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10539 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10540 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10542 2017-08-11 Martin Liska <mliska@suse.cz>
10545 * config/i386/i386.c (make_resolver_func): Do complete
10546 refactoring of the function.
10548 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10551 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10552 * config/i386/i386.c (ix86_stack_protect_guard): Use
10553 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10554 * doc/invoke.texi (x86 Options): Document
10555 -mstack-protector-guard-symbol= option.
10557 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10559 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10560 * config/i386/i386.c (ix86_split_stack_guard): New function.
10561 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10562 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10563 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10564 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10565 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10566 (split_stack_space_check): Call ix86_split_stack_guard.
10568 2017-08-10 Martin Sebor <msebor@redhat.com>
10570 * print-tree.c (print_node): Print location using the established
10572 Replace spaces with colons.
10573 (debug_raw, debug): Ditto.
10575 2017-08-10 Martin Sebor <msebor@redhat.com>
10578 * pretty-print.c (pp_format): Correct the handling of %s precision.
10580 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10583 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10585 (ix86_finalize_stack_frame_flags): This. Also clear
10586 frame_pointer_needed if -fno-omit-frame-pointer is used without
10588 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10589 with ix86_finalize_stack_frame_flags.
10590 (ix86_expand_epilogue): Likewise.
10591 (ix86_expand_split_stack_prologue): Likewise.
10592 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10594 2017-08-10 Martin Liska <mliska@suse.cz>
10597 * c-attribs.c (handle_target_attribute):
10598 Report warning for an empty string argument of target attribute.
10600 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10603 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10605 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10606 or DECL_NONLOCAL labels.
10607 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10608 or DECL_NONLOCAL labels here.
10610 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10612 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10613 to indicate when early gimple folding has been disabled.
10614 (rs6000_gimple_fold_builtin): Add debug content.
10615 (rs6000_invalid_builtin): Fix whitespace.
10616 (rs6000_expand_builtin): Fix whitespace.
10617 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10619 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10622 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10623 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10625 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10627 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10628 (qdf24xx): Likewise.
10629 * config/aarch64/aarch64.md: Include falkor.md.
10630 * config/aarch64/falkor.md: New.
10632 2017-08-09 Marek Polacek <polacek@redhat.com>
10635 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10636 * diagnostic.c (emit_diagnostic): Add a comment.
10637 (emit_diagnostic_valist): New function.
10639 2017-08-09 Marek Polacek <polacek@redhat.com>
10642 * input.c (make_location): New overload.
10643 * input.h (make_location): Declare.
10645 2017-08-08 Alan Modra <amodra@gmail.com>
10646 H.J. Lu <hongjiu.lu@intel.com>
10649 * gcc.c (NO_PIE_SPEC): Delete.
10650 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10652 (LINK_PIE_SPEC): ..to here.
10653 (LINK_COMMAND_SPEC): Support -no-pie.
10654 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10655 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10657 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10658 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10659 (ENDFILE_CRTEND_SPEC): Similarly.
10661 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10664 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10665 (mstack-protector-guard-offset=): Ditto.
10666 * config/i386/i386.c (ix86_option_override): Handle
10667 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10669 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10670 ix86_stack_protect_guard_offset variables.
10671 (TARGET_STACK_PROTECT_GUARD): Always define.
10672 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10673 and -mstack-protector-guard-offset= options.
10675 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10677 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10678 boundary case for the last candidate.
10680 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10682 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10683 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10685 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10687 PR middle-end/19706
10688 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10689 * config/aarch64/aarch64-builtins.c
10690 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10691 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10692 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10694 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10695 Andrew Pinski <pinskia@gmail.com>
10697 PR middle-end/19706
10698 * internal-fn.def (XORSIGN): New.
10699 * optabs.def (xorsign_optab): New.
10700 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10701 (convert_expand_mult_copysign): New.
10702 (pass_optimize_widening_mul::execute): Call
10703 convert_expand_mult_copysign.
10705 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10707 PR tree-optimization/81354
10708 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10709 Insert on edges rather than explicitly creating landing pads.
10710 (analyze_candidates_and_replace): Commit edge inserts.
10712 2017-08-08 Richard Biener <rguenther@suse.de>
10714 PR middle-end/81719
10715 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10716 (expand_simple_operations): Also look through ADDR_EXPRs with
10717 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10719 2017-08-08 Richard Biener <rguenther@suse.de>
10721 PR tree-optimization/81723
10722 * tree-vect-slp.c (struct bst_traits): New hash traits.
10723 (bst_fail): New global.
10724 (vect_build_slp_tree_2): New worker, split out from ...
10725 (vect_build_slp_tree): ... this now wrapping it with using
10726 bst_fail set to cache SLP tree build fails. Properly handle
10728 (vect_analyze_slp_instance): Allocate and free bst_fail.
10730 2017-08-08 Martin Liska <mliska@suse.cz>
10733 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10734 LABEL_DECLs that can be from a different function.
10736 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10738 PR tree-optimization/81744
10739 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10740 loop's number of iterations.
10742 2017-08-08 Martin Liska <mliska@suse.cz>
10744 * asan.c: Include header files.
10745 * attribs.c (build_decl_attribute_variant): New function moved
10747 (build_type_attribute_qual_variant): Likewise.
10748 (cmp_attrib_identifiers): Likewise.
10749 (simple_cst_list_equal): Likewise.
10750 (omp_declare_simd_clauses_equal): Likewise.
10751 (attribute_value_equal): Likewise.
10752 (comp_type_attributes): Likewise.
10753 (build_type_attribute_variant): Likewise.
10754 (lookup_ident_attribute): Likewise.
10755 (remove_attribute): Likewise.
10756 (merge_attributes): Likewise.
10757 (merge_type_attributes): Likewise.
10758 (merge_decl_attributes): Likewise.
10759 (merge_dllimport_decl_attributes): Likewise.
10760 (handle_dll_attribute): Likewise.
10761 (attribute_list_equal): Likewise.
10762 (attribute_list_contained): Likewise.
10763 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10764 (lookup_attribute_by_prefix): Likewise.
10765 * bb-reorder.c: Include header files.
10766 * builtins.c: Likewise.
10767 * calls.c: Likewise.
10768 * cfgexpand.c: Likewise.
10769 * cgraph.c: Likewise.
10770 * cgraphunit.c: Likewise.
10771 * convert.c: Likewise.
10772 * dwarf2out.c: Likewise.
10773 * final.c: Likewise.
10774 * fold-const.c: Likewise.
10775 * function.c: Likewise.
10776 * gimple-expr.c: Likewise.
10777 * gimple-fold.c: Likewise.
10778 * gimple-pretty-print.c: Likewise.
10779 * gimple.c: Likewise.
10780 * gimplify.c: Likewise.
10781 * hsa-common.c: Likewise.
10782 * hsa-gen.c: Likewise.
10783 * internal-fn.c: Likewise.
10784 * ipa-chkp.c: Likewise.
10785 * ipa-cp.c: Likewise.
10786 * ipa-devirt.c: Likewise.
10787 * ipa-fnsummary.c: Likewise.
10788 * ipa-inline.c: Likewise.
10789 * ipa-visibility.c: Likewise.
10791 * lto-cgraph.c: Likewise.
10792 * omp-expand.c: Likewise.
10793 * omp-general.c: Likewise.
10794 * omp-low.c: Likewise.
10795 * omp-offload.c: Likewise.
10796 * omp-simd-clone.c: Likewise.
10797 * opts-global.c: Likewise.
10798 * passes.c: Likewise.
10799 * predict.c: Likewise.
10800 * sancov.c: Likewise.
10801 * sanopt.c: Likewise.
10802 * symtab.c: Likewise.
10803 * toplev.c: Likewise.
10804 * trans-mem.c: Likewise.
10805 * tree-chkp.c: Likewise.
10806 * tree-eh.c: Likewise.
10807 * tree-into-ssa.c: Likewise.
10808 * tree-object-size.c: Likewise.
10809 * tree-parloops.c: Likewise.
10810 * tree-profile.c: Likewise.
10811 * tree-ssa-ccp.c: Likewise.
10812 * tree-ssa-live.c: Likewise.
10813 * tree-ssa-loop.c: Likewise.
10814 * tree-ssa-sccvn.c: Likewise.
10815 * tree-ssa-structalias.c: Likewise.
10816 * tree-ssa.c: Likewise.
10817 * tree-streamer-in.c: Likewise.
10818 * tree-vectorizer.c: Likewise.
10819 * tree-vrp.c: Likewise.
10820 * tsan.c: Likewise.
10821 * ubsan.c: Likewise.
10822 * varasm.c: Likewise.
10823 * varpool.c: Likewise.
10824 * tree.c: Remove functions moved to attribs.[ch].
10825 * tree.h: Likewise.
10826 * config/aarch64/aarch64.c: Add attrs.h header file.
10827 * config/alpha/alpha.c: Likewise.
10828 * config/arc/arc.c: Likewise.
10829 * config/arm/arm.c: Likewise.
10830 * config/avr/avr.c: Likewise.
10831 * config/bfin/bfin.c: Likewise.
10832 * config/c6x/c6x.c: Likewise.
10833 * config/cr16/cr16.c: Likewise.
10834 * config/cris/cris.c: Likewise.
10835 * config/darwin.c: Likewise.
10836 * config/epiphany/epiphany.c: Likewise.
10837 * config/fr30/fr30.c: Likewise.
10838 * config/frv/frv.c: Likewise.
10839 * config/ft32/ft32.c: Likewise.
10840 * config/h8300/h8300.c: Likewise.
10841 * config/i386/winnt.c: Likewise.
10842 * config/ia64/ia64.c: Likewise.
10843 * config/iq2000/iq2000.c: Likewise.
10844 * config/lm32/lm32.c: Likewise.
10845 * config/m32c/m32c.c: Likewise.
10846 * config/m32r/m32r.c: Likewise.
10847 * config/m68k/m68k.c: Likewise.
10848 * config/mcore/mcore.c: Likewise.
10849 * config/microblaze/microblaze.c: Likewise.
10850 * config/mips/mips.c: Likewise.
10851 * config/mmix/mmix.c: Likewise.
10852 * config/mn10300/mn10300.c: Likewise.
10853 * config/moxie/moxie.c: Likewise.
10854 * config/msp430/msp430.c: Likewise.
10855 * config/nds32/nds32-isr.c: Likewise.
10856 * config/nds32/nds32.c: Likewise.
10857 * config/nios2/nios2.c: Likewise.
10858 * config/nvptx/nvptx.c: Likewise.
10859 * config/pa/pa.c: Likewise.
10860 * config/pdp11/pdp11.c: Likewise.
10861 * config/powerpcspe/powerpcspe.c: Likewise.
10862 * config/riscv/riscv.c: Likewise.
10863 * config/rl78/rl78.c: Likewise.
10864 * config/rx/rx.c: Likewise.
10865 * config/s390/s390.c: Likewise.
10866 * config/sh/sh.c: Likewise.
10867 * config/sol2.c: Likewise.
10868 * config/sparc/sparc.c: Likewise.
10869 * config/spu/spu.c: Likewise.
10870 * config/stormy16/stormy16.c: Likewise.
10871 * config/tilegx/tilegx.c: Likewise.
10872 * config/tilepro/tilepro.c: Likewise.
10873 * config/v850/v850.c: Likewise.
10874 * config/vax/vax.c: Likewise.
10875 * config/visium/visium.c: Likewise.
10876 * config/xtensa/xtensa.c: Likewise.
10878 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10881 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10882 constraints since the -mupper-regs-* switches have been
10884 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10885 into a vector from a double word element that was extracted from
10886 another vector, and eliminate extra XXPERMDI instructions.
10887 (vsx_concat_<mode>_2): Likewise.
10888 (vsx_concat_<mode>_3): Likewise.
10889 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10890 concat to allow optimizing inserts from previous extracts.
10892 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10894 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10895 memory reference to a SSP offset in TLS address space.
10896 (ix86_print_operand) <case '@'>: Remove.
10897 (ix86_print_operand_punct_valid_p): Remove '@' code.
10898 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10899 UNSPEC_SP_TLS_TEST.
10900 (stack_tls_protect_set_<mode>): Remove.
10901 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10902 (stack_tls_protect_test_<mode>): Remove.
10903 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10905 2017-08-07 Olivier Hainque <hainque@adacore.com>
10908 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10910 2017-08-07 Douglas Rupp <rupp@adacore.com>
10912 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10913 variable was referenced as multidir in command.
10915 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10918 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10921 2017-08-07 Martin Liska <mliska@suse.cz>
10923 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10924 * config/rl78/rl78.c: Add include of attribs.h.
10925 * config/sh/sh.c: Likewise.
10926 * config/v850/v850.c: Likewise.
10928 2017-08-07 Tom de Vries <tom@codesourcery.com>
10930 PR middle-end/78266
10931 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10933 2017-08-07 Martin Liska <mliska@suse.cz>
10935 * config/mips/mips.c: Include attribs.h.
10937 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10940 * doc/invoke.texi: Document change in behvaior for -Ofast for
10943 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10945 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10947 (aarch64_pop_regs): Likewise.
10948 (aarch64_gen_load_pair): Likewise.
10949 (aarch64_save_callee_saves): Likewise.
10950 (aarch64_restore_callee_saves): Likewise.
10952 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10954 * config/i386/i386.c: Revert the last change.
10956 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10959 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10961 (ix86_finalize_stack_frame_flags): This. Also clear
10962 frame_pointer_needed if -fno-omit-frame-pointer is used without
10964 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10965 with ix86_finalize_stack_frame_flags.
10966 (ix86_expand_epilogue): Likewise.
10967 (ix86_expand_split_stack_prologue): Likewise.
10969 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10972 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10973 to P_AES for Westmere.
10975 2017-08-07 Jonathan Yong <10walls@gmail.com>
10977 * config/i386/mingw.opt (fset-stack-executable): Removed.
10978 * config/i386/cygming.opt (fset-stack-executable): Moved
10980 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10982 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10984 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10986 2017-08-07 Marek Polacek <polacek@redhat.com>
10988 PR middle-end/81737
10989 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10991 2017-08-07 Martin Liska <mliska@suse.cz>
10993 * attribs.h (canonicalize_attr_name): New function.
10994 (cmp_attribs): Move from c-format.c and adjusted.
10995 (is_attribute_p): Moved from tree.h.
10996 * tree-inline.c: Add new includes.
10997 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10998 (private_is_attribute_p): Remove.
10999 (private_lookup_attribute): Likewise.
11000 (private_lookup_attribute_by_prefix): Simplify.
11001 (remove_attribute): Use is_attribute_p.
11002 * tree.h: Remove removed declarations.
11004 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11006 PR middle-end/81698
11007 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11008 instead of computing it in the function. Formatting fix.
11009 (expand_case): Don't rely on default_edge being the first edge,
11010 clear it if removing it, pass default_edge to
11011 emit_case_dispatch_table.
11012 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11015 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11017 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11018 insn in the function, emit NOP after the insn.
11020 2017-08-06 Tom de Vries <tom@codesourcery.com>
11022 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11025 2017-08-06 Tom de Vries <tom@codesourcery.com>
11027 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11030 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11032 PR tree-optimization/57371
11033 * match.pd: New pattern.
11035 2017-08-04 Marek Polacek <polacek@redhat.com>
11037 PR middle-end/81695
11038 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11039 perform the computation in offset_int.
11041 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11043 PR tree-optimization/81136
11044 * tree-vectorizer.h: Include tree-hash-traits.h.
11045 (vec_base_alignments): New typedef.
11046 (vec_info): Add a base_alignments field.
11047 (vect_record_base_alignments): Declare.
11048 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11050 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11051 (create_data_ref): Add an is_conditional_in_stmt argument.
11052 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11053 the is_conditional_in_stmt field.
11054 (data_ref_loc): Add an is_conditional_in_stmt field.
11055 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11056 (find_data_references_in_stmt): Update call to create_data_ref.
11057 (graphite_find_data_references_in_stmt): Likewise.
11058 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11059 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11060 (vect_record_base_alignment): New function.
11061 (vect_record_base_alignments): Likewise.
11062 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11063 for nested statements even if we fail to compute a misalignment.
11064 Use pooled base alignments for unconditional references.
11065 (vect_find_same_alignment_drs): Compare base addresses instead
11067 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11068 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11070 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11072 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11073 Add an explicit name for the enum. Use auto_vec for slp_instances
11074 and grouped_stores.
11075 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11077 (_bb_vec_info): Add a constructor and destructor.
11078 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11079 (destroy_loop_vec_info): Delete.
11080 (vect_destroy_datarefs): Likewise.
11081 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11082 (vec_info::vec_info): New function.
11083 (vec_info::~vec_info): Likewise.
11084 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11085 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11086 destroy_loop_vec_info.
11087 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11088 (_loop_vec_info::_loop_vec_info): ...this.
11089 (destroy_loop_vec_info): Replace with...
11090 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11091 the stmt_vec_infos. Leave handling of vec_info information to its
11092 destructor. Remove explicit vector releases.
11093 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11094 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11095 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11096 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11097 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11098 (destroy_bb_vec_info): Replace with...
11099 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11100 information to its destructor.
11101 (vect_slp_analyze_bb_1): Use new and delete instead of
11102 new_bb_vec_info and destroy_bb_vec_info.
11103 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11106 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11108 * tree-data-ref.h (subscript): Add access_fn field.
11109 (data_dependence_relation): Add could_be_independent_p.
11110 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11111 (same_access_functions): Move to tree-data-ref.c.
11112 * tree-data-ref.c (ref_contains_union_access_p): New function.
11113 (access_fn_component_p): Likewise.
11114 (access_fn_components_comparable_p): Likewise.
11115 (dr_analyze_indices): Add a reference to access_fn_component_p.
11116 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11118 (constant_access_functions): Likewise.
11119 (add_other_self_distances): Likewise.
11120 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11121 (initialize_data_dependence_relation): Use XCNEW and remove
11122 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11123 of access functions that have the same type. Allow the
11124 subsequence to end with different bases in some circumstances.
11125 Record the chosen access functions in SUB_ACCESS_FN.
11126 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11127 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11128 (subscript_dependence_tester_1): Likewise dra and drb.
11129 (build_classic_dist_vector): Update calls accordingly.
11130 (subscript_dependence_tester): Likewise.
11131 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11132 DDR_COULD_BE_INDEPENDENT_P.
11133 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11134 comp_alias_ddrs instead of may_alias_ddrs.
11135 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11137 (vect_analyze_data_ref_dependence): Use it if
11138 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11139 distance vectors if that fails.
11140 (dependence_distance_ge_vf): New function.
11141 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11142 LOOP_VINFO_MAY_ALIAS_DDRS.
11144 2017-08-04 Richard Biener <rguenther@suse.de>
11146 PR middle-end/81705
11147 * fold-const.c (fold_binary_loc): Properly restrict
11148 minus_var0 && minus_var1 case when associating undefined overflow
11151 2017-08-04 Tom de Vries <tom@codesourcery.com>
11153 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11155 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11157 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11158 Don't start diagnostic messages with a capital letter.
11159 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11161 (rs6000_invalid_builtin): Likewise.
11162 (rs6000_trampoline_init): Likewise.
11164 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11167 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11168 after setting changeable df flags.
11170 2017-08-03 Richard Biener <rguenther@suse.de>
11172 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11173 up if the use is in USE - X.
11175 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11177 * toplev.c (dumpfile.h): New include.
11178 (internal_error_reentered): New static function. Use it...
11179 (internal_error_function): ...here to handle reentered internal_error.
11181 2017-08-03 Richard Biener <rguenther@suse.de>
11183 PR middle-end/81148
11184 * fold-const.c (split_tree): Add minus_var and minus_con
11185 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11186 here but always use minus_*.
11187 (associate_trees): Assert we never associate with MINUS_EXPR
11188 and NULL first operand. Do not recurse for PLUS_EXPR operands
11189 when associating as MINUS_EXPR either.
11190 (fold_binary_loc): Track minus_var and minus_con.
11192 2017-08-03 Tom de Vries <tom@codesourcery.com>
11195 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11196 ACCEL_COMPILER, apply finish_options on
11197 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11199 2017-08-03 Tom de Vries <tom@codesourcery.com>
11202 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11203 function_entry_patch_area_size > 0.
11205 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11208 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11209 instead of 10??LU, perform unit multiplication in wide_int,
11210 don't change alloc_object_size_limit if the limit is larger
11213 PR tree-optimization/81655
11214 PR tree-optimization/81588
11215 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11216 the case when ranges[i].low and high are 1 for unsigned type with
11219 PR middle-end/81052
11220 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11221 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11223 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11225 * tree-vrp.h: Add include guard.
11227 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11230 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11231 (ud2): New insn pattern.
11232 * config/i386/i386.c (ix86_expand_epilogue):
11233 For naked functions, generate ud2 instead of trap insn.
11235 2017-08-02 Marek Polacek <polacek@redhat.com>
11238 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11240 2017-08-02 Tom de Vries <tom@codesourcery.com>
11241 Cesar Philippidis <cesar@codesourcery.com>
11243 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11244 Add missing edge probabilities.
11246 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11248 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11249 Correct endianness.
11251 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11253 PR middle-end/79499
11254 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11255 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11256 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11257 prologue_seq sequences - if any.
11259 2017-08-02 Richard Biener <rguenther@suse.de>
11261 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11262 via vectors if supported, integer extracts via punning if supported
11263 or otherwise vector extracts.
11265 2017-08-02 Richard Biener <rguenther@suse.de>
11267 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11269 (bitmap_insert_into_set): ... this.
11271 2017-08-02 Richard Biener <rguenther@suse.de>
11273 PR tree-optimization/81633
11275 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11277 PR tree-optimization/71752
11278 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11280 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11282 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11283 (machine_function::call_ms2sysv_pad_out): Remove field.
11284 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11285 (ix86_compute_frame_layout): Likewise.
11287 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11290 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11291 attribute with interrupt attribute.
11293 2017-08-01 Andrew Pinski <apinski@cavium.com>
11295 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11296 BIT_INSERT_EXPR's operand 1
11297 to see if the types precision matches.
11299 2017-08-01 Martin Liska <mliska@suse.cz>
11301 PR middle-end/70140
11302 * builtins.c (expand_builtin_memcpy_args): Remove.
11303 (expand_builtin_memcpy): Call newly added function
11304 expand_builtin_memory_copy_args.
11305 (expand_builtin_memcpy_with_bounds): Likewise.
11306 (expand_builtin_mempcpy): Remove last argument.
11307 (expand_builtin_mempcpy_with_bounds): Likewise.
11308 (expand_builtin_memory_copy_args): New function created from
11309 expand_builtin_mempcpy_args with small modifications.
11310 (expand_builtin_mempcpy_args): Remove.
11311 (expand_builtin_stpcpy): Remove unused argument.
11312 (expand_builtin): Likewise.
11313 (expand_builtin_with_bounds): Likewise.
11315 2017-08-01 Martin Liska <mliska@suse.cz>
11318 Make mempcpy more optimal (PR middle-end/70140).
11320 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11324 __builtin_vec_cmpne verify both arguments are compatible vectors
11325 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11326 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11327 move computation of aligned to after checking the argument types.
11331 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11332 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11335 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11336 Doug Rupp <rupp@adacore.com>
11337 Olivier Hainque <hainque@adacore.com>
11339 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11340 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11342 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11343 for TARGET_OS_CPP_BUILTIN.
11344 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11345 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11347 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11348 passing required abi options to the assembler for EABI configurations.
11349 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11350 of .text.hot and .text.unlikely sections for kernel modules when
11351 using ARM style exceptions.
11352 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11353 options. Add EXTRA_CC1_SPEC.
11354 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11356 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11358 (ARM_TARGET2_DWARF_FORMAT): Define.
11359 * config/arm/t-vxworks: Adjust multilib control to removal of the
11360 Diab command line options.
11362 2017-08-01 Martin Liska <mliska@suse.cz>
11364 PR gcov-profile/81561
11365 * gcov.c (unblock): Make unblocking safe as we need to preserve
11366 index correspondence of blocks and block_lists.
11368 2017-08-01 Richard Biener <rguenther@suse.de>
11370 PR tree-optimization/81181
11371 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11372 (compute_antic): ... end of iteration here.
11374 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11376 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11377 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11378 (ftree-slp-vectorize): Likewise.
11379 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11380 can no longer be set independent of flag_tree_loop_vectorize.
11381 * omp-general.c (emp_max_vf): Likewise.
11382 * opts.c (enable_fdo_optimizations): Remove references to
11383 flag_tree_vectorize, these are now implicit.
11384 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11385 and leave it for the options machinery.
11387 2017-08-01 Martin Liska <mliska@suse.cz>
11389 PR middle-end/70140
11390 * builtins.c (expand_builtin_memcpy_args): Remove.
11391 (expand_builtin_memcpy): Call newly added function
11392 expand_builtin_memory_copy_args.
11393 (expand_builtin_memcpy_with_bounds): Likewise.
11394 (expand_builtin_mempcpy): Remove last argument.
11395 (expand_builtin_mempcpy_with_bounds): Likewise.
11396 (expand_builtin_memory_copy_args): New function created from
11397 expand_builtin_mempcpy_args with small modifications.
11398 (expand_builtin_mempcpy_args): Remove.
11399 (expand_builtin_stpcpy): Remove unused argument.
11400 (expand_builtin): Likewise.
11401 (expand_builtin_with_bounds): Likewise.
11403 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11406 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11407 print "ds:" only for immediates in generic address space.
11409 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11412 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11413 (ix86_function_ok_for_sibcall): Return false for naked functions.
11415 2017-08-01 Richard Biener <rguenther@suse.de>
11417 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11418 (compute_antic): Seed worklist with exit block predecessors.
11419 * cfganal.c (dfs_find_deadend): For a cycle return the source
11420 of the edge closing it.
11422 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11424 * config/aarch64/aarch64.c
11425 (aarch64_can_const_movi_rtx_p): Move 0 check.
11427 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11429 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11430 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11432 * match.pd: Ditto in address comparison pattern.
11434 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11436 PR tree-optimization/81627
11437 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11438 closed ssa form for store-store chain.
11440 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11442 PR tree-optimization/81620
11443 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11444 for store-store chain.
11446 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11448 PR tree-optimization/81588
11449 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11450 ranges[i].in_p, invert comparison code ccode. For >/>=,
11451 swap rhs1 and rhs2 and comparison code unconditionally,
11452 for </<= don't do that. Don't swap rhs1/rhs2 again if
11453 ranges[i].in_p, instead invert comparison code ccode if
11454 opcode or oe->rank is BIT_IOR_EXPR.
11457 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11458 a direct optab to conversion optab.
11459 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11460 with GET_MODE_INNER as last argument instead of optab_handler.
11461 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11462 vector extraction if possible and optab is available.
11463 * expr.c (store_constructor): Use convert_optab_handler instead
11464 of optab_handler. Use vector initialization from smaller
11465 vectors if possible and optab is available.
11466 * tree-vect-stmts.c (vectorizable_load): Likewise.
11467 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11468 now have two modes.
11469 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11470 of vec_init from half-sized vectors with the same element mode.
11471 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11472 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11473 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11474 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11475 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11476 after mode in gen_vec_extract* calls.
11477 (vec_extract<mode>): Renamed to ...
11478 (vec_extract<mode><ssescalarmodelower>): ... this.
11479 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11480 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11481 element mode after mode in gen_vec_init* calls.
11482 (VEC_INIT_HALF_MODE): New mode iterator.
11483 (vec_init<mode>): Renamed to ...
11484 (vec_init<mode><ssescalarmodelower>): ... this.
11485 (vec_init<mode><ssehalfvecmodelower>): New expander.
11486 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11487 (vec_extractv2sfsf): ... this.
11488 (vec_initv2sf): Renamed to ...
11489 (vec_initv2sfsf): ... this.
11490 (vec_extractv2si): Renamed to ...
11491 (vec_extractv2sisi): ... this.
11492 (vec_initv2si): Renamed to ...
11493 (vec_initv2sisi): ... this.
11494 (vec_extractv4hi): Renamed to ...
11495 (vec_extractv4hihi): ... this.
11496 (vec_initv4hi): Renamed to ...
11497 (vec_initv4hihi): ... this.
11498 (vec_extractv8qi): Renamed to ...
11499 (vec_extractv8qiqi): ... this.
11500 (vec_initv8qi): Renamed to ...
11501 (vec_initv8qiqi): ... this.
11502 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11503 (vec_init<mode>): Renamed to ...
11504 (vec_init<mode><VEC_base_l>): ... this.
11505 (vec_extract<mode>): Renamed to ...
11506 (vec_extract<mode><VEC_base_l>): ... this.
11507 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11508 (vec_initv2sfsf): ... this.
11509 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11510 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11511 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11512 element mode after mode in gen_vec_init* calls.
11513 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11514 (vec_init<mode><Vel>): ... this.
11515 (vec_extract<mode>): Renamed to ...
11516 (vec_extract<mode><Vel>): ... this.
11517 * config/aarch64/iterators.md (Vel): New mode attribute.
11518 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11519 Add element mode after mode in gen_vec_extract* calls.
11520 * config/s390/vector.md (non_vec_l): New mode attribute.
11521 (vec_extract<mode>): Renamed to ...
11522 (vec_extract<mode><non_vec_l>): ... this.
11523 (vec_init<mode>): Renamed to ...
11524 (vec_init<mode><non_vec_l>): ... this.
11525 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11526 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11528 * config/arm/iterators.md (V_elem_l): New mode attribute.
11529 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11530 (vec_extract<mode><V_elem_l>): ... this.
11531 (vec_extractv2di): Renamed to ...
11532 (vec_extractv2didi): ... this.
11533 (vec_init<mode>): Renamed to ...
11534 (vec_init<mode><V_elem_l>): ... this.
11535 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11536 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11537 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11538 Add element mode after gen_vec_extract* calls.
11539 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11540 (vec_init<mode><unitmode>): ... this.
11541 (vec_extract<mode>): Renamed to ...
11542 (vec_extract<mode><unitmode>): ... this.
11543 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11544 (vec_init<mode><unitmode>): ... this.
11545 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11546 (vec_initv2sfsf): ... this.
11547 (vec_extractv2sf): Renamed to ...
11548 (vec_extractv2sfsf): ... this.
11549 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11550 Add element mode after gen_vec_extract* calls.
11551 * config/mips/mips.md (unitmode): New mode iterator.
11552 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11553 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11554 * config/spu/spu.md (inner_l): New mode attribute.
11555 (vec_init<mode>): Renamed to ...
11556 (vec_init<mode><inner_l>): ... this.
11557 (vec_extract<mode>): Renamed to ...
11558 (vec_extract<mode><inner_l>): ... this.
11559 * config/sparc/sparc.md (veltmode): New mode iterator.
11560 (vec_init<VMALL:mode>): Renamed to ...
11561 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11562 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11563 (vec_initv2sisi): ... this.
11564 (vec_initv2sf): Renamed to ...
11565 (vec_initv2sfsf): ... this.
11566 (vec_extractv2sf): Renamed to ...
11567 (vec_extractv2sfsf): ... this.
11568 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11569 (vec_init<mode>): Renamed to ...
11570 (vec_init<mode><VEC_base_l>): ... this.
11571 (vec_extract<mode>): Renamed to ...
11572 (vec_extract<mode><VEC_base_l>): ... this.
11573 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11574 (vec_initv2sfsf): ... this.
11575 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11576 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11577 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11578 gen_vec_init* calls.
11580 2017-08-01 Richard Biener <rguenther@suse.de>
11582 PR tree-optimization/81297
11583 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11584 TREE_OVERFLOW from INTEGER_CSTs.
11586 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11588 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11590 2017-07-31 Carl Love <cel@us.ibm.com>
11592 * config/rs6000/rs6000-c: Add support for built-in functions
11593 vector signed char vec_xl_be (signed long long, signed char *);
11594 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11595 vector signed int vec_xl_be (signed long long, signed int *);
11596 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11597 vector signed long long vec_xl_be (signed long long, signed long long *);
11598 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11599 vector signed short vec_xl_be (signed long long, signed short *);
11600 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11601 vector double vec_xl_be (signed long long, double *);
11602 vector float vec_xl_be (signed long long, float *);
11603 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11604 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11605 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11607 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11608 (altivec_expand_builtin): Add switch statement to call
11609 altivec_expand_xl_be for each builtin.
11610 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11611 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11612 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11613 __builtin_vsx_le_be_v16qi.
11614 * doc/extend.texi: Update the built-in documentation file for the
11615 new built-in functions.
11617 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11620 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11622 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11624 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11626 * config.gcc: Add z14.
11627 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11628 CPU model numbers for z13s and z14.
11629 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11631 * config/s390/s390-opts.h (enum processor_type): Rename
11632 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11633 * config/s390/s390.c (processor_table): Add field for CPU name to
11634 be passed to Binutils.
11635 (s390_asm_output_machine_for_arch): Use the new field in
11636 processor_table for Binutils.
11637 (s390_expand_builtin): Replace arch12 with z14.
11638 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11639 (s390_get_sched_attrmask): Likewise.
11640 (s390_get_unit_mask): Likewise.
11641 * config/s390/s390.opt: Add z14 to processor_type enum.
11643 2017-07-31 Martin Jambor <mjambor@suse.cz>
11646 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11647 regardless of optimization level.
11649 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11650 Martin Liska <mliska@suse.cz>
11652 * predict.def: Remove old comment and adjust probability.
11653 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11654 PREDICT statements.
11656 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11659 * config/i386/i386.c (ix86_function_naked): New function.
11660 (ix86_can_use_return_insn_p): Return false for naked functions.
11661 (ix86_expand_prologue): Skip prologue for naked functions.
11662 (ix86_expand_epilogue): Skip epilogue for naked functions
11663 and emit trap instruction.
11664 (ix86_warn_func_return): New function.
11665 (ix86_attribute_table): Add "naked" attribute specification.
11666 (TARGET_WARN_FUNC_RETURN): Define.
11667 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11669 2017-07-31 Martin Liska <mliska@suse.cz>
11671 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11672 (dump_gimple_bb_header): Always dump BB info.
11673 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11675 2017-07-31 Martin Liska <mliska@suse.cz>
11678 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11679 also with current_function_decl non-null equality.
11681 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11684 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11685 change type to the element type, instead add eltype variable and
11686 use it where we are interested in the element type.
11688 PR tree-optimization/81603
11689 * ipa-polymorphic-call.c
11690 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11691 offset arithmetic in offset_int, bail out if the resulting bit offset
11692 doesn't fit into shwi.
11694 2017-07-31 Martin Liska <mliska@suse.cz>
11696 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11697 (gimplify_save_expr): Fix comment.
11699 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11702 * config/i386/i386.c (ix86_function_arg): Update arguments for
11704 (ix86_compute_frame_layout): Set the initial stack offset to
11705 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11706 INCOMING_FRAME_SP_OFFSET.
11707 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11708 stack before exception handler returns.
11709 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11710 the 'ERROR_CODE' for exception handler.
11712 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11714 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11715 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11716 (ASM_OUTPUT_REG_POP): Ditto.
11717 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11718 instead of asm_fprintf to output pure string.
11720 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11722 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11723 to imported_module_or_decl hook.
11724 (debug_nothing_tree_tree_tree_bool): Remove.
11725 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11726 * debug.c (do_nothing_debug_hooks): Use
11727 debug_nothing_tree_tree_tree_bool_bool instead of
11728 debug_nothing_tree_tree_tree_bool.
11729 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11730 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11731 * sdbout.c (sdb_debug_hooks): Likewise.
11732 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11733 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11735 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11736 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11737 attribute, don't add anything.
11739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11741 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11742 (fold_build2_stat_loc): Likewise.
11743 (fold_build3_stat_loc): Likewise.
11744 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11745 (fold_build1_loc): Remove macro.
11746 (fold_build2_loc): Likewise.
11747 (fold_build3_loc): Likewise.
11749 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11751 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11752 (gimple_build_debug_bind_source_stat): Likewise.
11753 * gimple.h (gimple_build_debug_bind): Remove macro.
11754 (gimple_build_debug_bind_source): Likewise.
11756 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11758 * bitmap.c (bitmap_alloc): Adjust.
11759 (bitmap_gc_alloc): Likewise.
11760 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11762 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11764 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11765 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11766 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11767 (bitmap_gc_alloc_stat): Likewise.
11768 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11772 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11773 * rtl.h (shallow_copy_rtx): Remove macro.
11775 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11777 * emit-rtl.c (gen_raw_REG): Adjust.
11778 * gengenrtl.c (gendef): Likewise.
11779 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11780 * rtl.h (rtx_alloc): Remove macro.
11782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11784 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11785 (build_tree_list_stat): Likewise.
11786 * tree.h (build_tree_list): Remove macro.
11787 (build_tree_list_vec): Likewise.
11789 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11791 * tree.c (make_vector_stat): Remove _stat from name.
11792 (build_vector_stat): Likewise.
11793 * tree.h (make_vector_stat): Remove macro.
11794 (build_vector_stat): Likewise.
11796 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11798 * tree.h (build_var_debug_value): Remove prototype.
11800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11802 * tree.c (tree_cons_stat): Remove _stat from name.
11803 * tree.h (tree_cons): Remove macro.
11805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11807 * tree.c (build_vl_exp_stat): Remove _stat from name.
11808 * tree.h (build_vl_exp): Remove macro.
11810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11812 * tree.c (build_decl_stat): Remove _stat from name.
11813 * tree.h (build_decl): Remove macro.
11815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11817 * gimple.c (gimple_build_with_ops_stat): Adjust.
11818 (gimple_alloc_stat): Remove _stat from name.
11819 * gimple.h (gimple_alloc): Remove macro.
11821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11823 * tree.c (make_tree_vec_stat): Remove _stat from name.
11824 (grow_tree_vec_stat): Likewise.
11825 * tree.h (make_tree_vec_stat): Adjust prototype.
11826 (grow_tree_vec_stat): Likewise.
11827 (make_tree_vec): Remove macro.
11828 (grow_tree_vec): Likewise.
11830 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11832 * fold-const.c (fold_build1_stat_loc): Adjust.
11833 (fold_build2_stat_loc): Likewise.
11834 (fold_build3_stat_loc): Likewise.
11835 * tree.c (build0_stat): Remove _stat from name.
11836 (build1_stat): Likewise.
11837 (build2_stat): Likewise.
11838 (build3_stat): Likewise.
11839 (build4_stat): Likewise.
11840 (build5_stat): Likewise.
11841 * tree.h (build1_loc): Remove macro, and rename _stat function
11843 (build2_loc): Likewise.
11844 (build3_loc): Likewise.
11845 (build4_loc): Likewise.
11846 (build5_loc): Likewise.
11848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11850 * tree.c (make_int_cst_stat): Remove _stat from name.
11851 * tree.h (make_int_cst_stat): Adjust prototype.
11852 (make_int_cst): Remove macro.
11854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11856 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11857 * tree.h (make_tree_binfo_stat): Adjust prototype.
11858 (make_tree_binfo): Remove.
11860 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11862 * tree.c (copy_node_stat): Rename to copy_node.
11863 (build_distinct_type_copy): Adjust.
11864 * tree.h (copy_node_stat): Adjust prototype.
11865 (copy_node): Remove macro.
11867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11869 * tree.c (make_node_stat): rename to make_node.
11870 (build_tree_list_stat): Adjust.
11871 (build0_stat): Likewise.
11872 (build2_stat): Likewise.
11873 (build3_stat): Likewise.
11874 (build4_stat): Likewise.
11875 (build5_stat): Likewise.
11876 (build_decl_stat): Likewise.
11877 * tree.h (make_node_stat): Adjust prototype.
11878 (make_node): remove macro.
11880 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11882 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11883 (PPC_FEATURE2_SCV): Likewise.
11884 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11886 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11888 * config/aarch64/aarch64.c
11889 (aarch64_internal_mov_immediate): Add new special pattern.
11890 * config/aarch64/aarch64.md (*movdi_aarch64):
11891 Add reg/32bit const mov case.
11893 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11894 Richard Sandiford <richard.sandiford@linaro.org>
11896 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11897 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11898 Add integer and movi cases.
11899 (movi-split-hf-df-sf split, fp16): New.
11900 (enabled): Added TARGET_FP_F16INST.
11901 * config/aarch64/iterators.md (GPF_HF): New.
11902 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11904 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11906 * config/aarch64/aarch64.c
11907 (aarch64_simd_container_mode): Add prototype.
11908 (aarch64_expand_mov_immediate): Add HI support.
11909 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11910 (aarch64_can_const_movi_rtx_p): New.
11911 (aarch64_preferred_reload_class):
11912 Remove restrictions of using FP registers for certain SIMD operations.
11913 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11914 (aarch64_valid_floating_const): Add integer move validation.
11915 (aarch64_simd_imm_scalar_p): Remove.
11916 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11917 (aarch64_legitimate_constant_p): Expand list of supported cases.
11918 * config/aarch64/aarch64-protos.h
11919 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11920 (aarch64_reinterpret_float_as_int): New.
11921 (aarch64_simd_imm_scalar_p): Remove.
11922 * config/aarch64/constraints.md (Uvi): New.
11923 (Dd): Split into Ds and new Dd.
11924 * config/aarch64/aarch64.md (*movsi_aarch64):
11926 (*movdi_aarch64): Add SIMD mov case.
11928 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11930 * tree-predcom.c: (struct chain): Handle store-store chain in which
11931 stores for elimination only store loop invariant values.
11932 (execute_pred_commoning_chain): Ditto.
11933 (prepare_initializers_chain_store_elim): Ditto.
11934 (prepare_finalizers): Ditto.
11935 (is_inv_store_elimination_chain): New function.
11936 (initialize_root_vars_store_elim_1): New function.
11938 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11940 * tree-predcom.c: Revise general description of the pass.
11941 (enum chain_type): New enum type for store elimination.
11942 (struct chain): New field supporting store elimination.
11943 (struct component): Ditto.
11944 (dump_chain): Dump store-stores chain.
11945 (release_chain): Release resources.
11946 (split_data_refs_to_components): Compute and create component
11947 contains only stores for elimination.
11948 (get_chain_last_ref_at): New function.
11949 (make_invariant_chain): Initialization.
11950 (make_rooted_chain): Specify chain type in parameter and record it.
11951 (add_looparound_copies): Skip for store-stores chain.
11952 (determine_roots_comp): Compute type of chain and pass it to
11954 (initialize_root_vars_store_elim_2): New function.
11955 (finalize_eliminated_stores): New function.
11956 (remove_stmt): Handle store for elimination.
11957 (execute_pred_commoning_chain): Execute predictive commoning on
11958 store-store chains.
11959 (determine_unroll_factor): Skip unroll for store-stores chain.
11960 (prepare_initializers_chain_store_elim): New function.
11961 (prepare_initializers_chain): Hanlde store-store chain.
11962 (prepare_finalizers_chain, prepare_finalizers): New function.
11963 (tree_predictive_commoning_loop): Return integer value indicating
11964 if loop is unrolled or lcssa form is corrupted.
11965 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11967 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11969 * tree-predcom.c (initialize_root): Delete.
11970 (execute_pred_commoning_chain): Initialize root vars and replace
11971 reference of non-combined chain directly, rather than call above
11974 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11976 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11977 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11979 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11981 * tree-predcom.c (struct chain): New field init_seq.
11982 (release_chain): Release init_seq.
11983 (prepare_initializers_chain): Record intialization stmts in above
11985 (insert_init_seqs): New function.
11986 (tree_predictive_commoning_loop): Call insert_init_seqs.
11988 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11990 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11992 2017-07-28 Richard Biener <rguenther@suse.de>
11994 * match.pd: Remove superfluous :c.
11995 * genmatch.c (simplify::id): Add member.
11996 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11998 (current_id): New global.
11999 (dt_node::parent): Move from ...
12000 (dt_operand::parent): ... here. Add for_id member.
12001 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12002 (decision_tree::find_node): Relax order requirement when
12003 merging DT_TRUE nodes to ones inbetween the current simplify
12004 and the one we try to merge with. Add diagnostic whenever
12005 we need to enforce pattern order by not merging.
12006 (decision_tree::insert): Set current_id.
12007 (decision_tree::print_node): Dump parent node and for_id.
12008 (parser::last_id): Add member.
12009 (parser::push_simplify): Assign unique id.
12010 (parser::parser): Initialize last_id.
12012 2017-07-28 Martin Liska <mliska@suse.cz>
12015 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12016 gimple_build_debug_bind.
12018 2017-07-28 Richard Biener <rguenther@suse.de>
12020 PR tree-optimization/81502
12021 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12023 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12025 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12026 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12027 for BIT_FIELD_REF args.
12028 * fold-const.c (make_bit_field_ref): Likewise.
12029 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12031 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12034 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12035 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12036 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12037 Or it into SANITIZER_UNDEFINED.
12038 * ubsan.c: Include gimple-fold.h and varasm.h.
12039 (ubsan_expand_ptr_ifn): New function.
12040 (instrument_pointer_overflow): New function.
12041 (maybe_instrument_pointer_overflow): New function.
12042 (instrument_object_size): Formatting fix.
12043 (pass_ubsan::execute): Call instrument_pointer_overflow
12044 and maybe_instrument_pointer_overflow.
12045 * internal-fn.c (expand_UBSAN_PTR): New function.
12046 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12047 * sanitizer.def (__ubsan_handle_pointer_overflow,
12048 __ubsan_handle_pointer_overflow_abort): New builtins.
12049 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12050 * internal-fn.def (UBSAN_PTR): New internal function.
12051 * opts.c (sanitizer_opts): Add pointer-overflow.
12052 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12053 * fold-const.c (build_range_check): Compute pointer range check in
12054 integral type if pointer arithmetics would be needed. Formatting
12057 2017-07-28 Martin Liska <mliska@suse.cz>
12060 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12061 parameters that are of a variable-length.
12063 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12065 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12067 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12068 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12069 (CRT_CALL_STATIC_FUNCTION): Likewise.
12070 (ASM_DEFAULT_SPEC): New define.
12071 (ASM_SPEC32): Likewise.
12072 (ASM_SPEC64): Likewise.
12073 (ASM_SPEC_COMMON): Likewise.
12074 (ASM_SPEC): Likewise.
12075 (INVALID_64BIT): Likewise.
12076 (LINK_OS_DEFAULT_SPEC): Likewise.
12077 (LINK_OS_SPEC32): Likewise.
12078 (LINK_OS_SPEC64): Likewise.
12079 (POWERPC_LINUX): Likewise.
12080 (PTRDIFF_TYPE): Likewise.
12081 (RESTORE_FP_PREFIX): Likewise.
12082 (RESTORE_FP_SUFFIX): Likewise.
12083 (SAVE_FP_PREFIX): Likewise.
12084 (SAVE_FP_SUFFIX): Likewise.
12085 (SIZE_TYPE): Likewise.
12086 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12087 (TARGET_64BIT): Likewise.
12088 (TARGET_64BIT): Likewise.
12089 (TARGET_AIX): Likewise.
12090 (WCHAR_TYPE_SIZE): Likewise.
12091 (WCHAR_TYPE): Undefine.
12092 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12093 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12094 (CPP_OS_RTEMS_SPEC): Delete.
12095 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12096 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12098 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12100 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12102 PR tree-optimization/81578
12103 * tree-parloops.c (build_new_reduction): Bail out if
12104 reduction_code isn't one of the standard OpenMP reductions.
12105 Move the details printing after that decision.
12107 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12109 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12110 related to reload_in_progress.
12111 (splat_input_operand): Likewise.
12112 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12114 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12116 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12117 (TARGET_INSTANTIATE_DECLS): Likewise.
12118 (legitimate_indexed_address_p): Delete reload_in_progress code.
12119 (rs6000_debug_legitimate_address_p): Likewise.
12120 (rs6000_eliminate_indexed_memrefs): Likewise.
12121 (rs6000_emit_le_vsx_store): Likewise.
12122 (rs6000_emit_move_si_sf_subreg): Likewise.
12123 (rs6000_emit_move): Likewise.
12124 (register_to_reg_type): Likewise.
12125 (rs6000_pre_atomic_barrier): Likewise.
12126 (rs6000_machopic_legitimize_pic_address): Likewise.
12127 (rs6000_allocate_stack_temp): Likewise.
12128 (rs6000_address_for_fpconvert): Likewise.
12129 (rs6000_address_for_altivec): Likewise.
12130 (rs6000_secondary_memory_needed_rtx): Delete function.
12131 (rs6000_check_sdmode): Likewise.
12132 (rs6000_alloc_sdmode_stack_slot): Likewise.
12133 (rs6000_instantiate_decls): Likewise.
12134 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12135 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12136 Delete reload_in_progress.
12137 (*vec_reload_and_plus_<mptrsize>): Likewise.
12138 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12139 (vsx_div_v2di): Likewise.
12140 (vsx_udiv_v2di): Likewise.
12142 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12144 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12145 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12146 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12147 (rs6000_debug_reg_global): Delete print of LRA status.
12148 (rs6000_option_override_internal): Delete dead LRA related code.
12149 (rs6000_lra_p): Delete function.
12150 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12152 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12154 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12155 * config/riscv/rtems.h: New file.
12157 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12158 Sudakshina Das <sudi.das@arm.com>
12160 * config/aarch64/aarch64.md
12161 (define_split for and<mode>3nr_compare): Move
12162 non aarch64_logical_operand to a register.
12163 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12164 register immediate operand to a register.
12165 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12167 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12169 PR middle-end/81564
12170 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12172 2017-07-27 Richard Biener <rguenther@suse.de>
12174 PR tree-optimization/81573
12175 PR tree-optimization/81494
12176 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12177 multi defuse cycle case.
12179 2017-07-27 Richard Biener <rguenther@suse.de>
12181 PR tree-optimization/81571
12182 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12185 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12187 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12188 earlier and only if MASK_FPU is set. Adjust formatting.
12190 2017-07-27 Martin Liska <mliska@suse.cz>
12192 * opt-functions.awk: Add validation of value of Init.
12193 * optc-gen.awk: Pass new argument.
12195 2017-07-27 Martin Liska <mliska@suse.cz>
12197 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12198 Fix wrong condition.
12200 2017-07-27 Martin Liska <mliska@suse.cz>
12202 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12203 BBs and edges seen by autoFDO.
12205 2017-07-27 Richard Biener <rguenther@suse.de>
12207 PR tree-optimization/81502
12208 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12209 with incompatible but same sized type.
12210 (execute_update_addresses_taken): Likewise.
12212 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12214 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12215 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12217 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12220 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12221 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12222 Change s_operand to memory_operand.
12224 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12226 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12227 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12228 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12229 Emit instructions rather than returning an expression. Handle TFmode
12230 and KFmode by casting to TImode.
12231 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12232 (rs6000_emit_le_vsx_store): Likewise.
12233 * config/rs6000/vsx.md (VSX_TI): New iterator.
12234 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12235 (*vsx_le_undo_permute_<mode>): Likewise.
12236 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12237 emit the split sequence.
12238 (*vsx_le_perm_store_<mode>): Likewise.
12240 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12242 PR tree-optimization/81555
12243 PR tree-optimization/81556
12244 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12245 if true, force CHANGED for the recursive invocation.
12246 (reassociate_bb): Remember original length of ops array, pass
12247 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12249 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12250 attributes for noipa attribute. For naked attribute use
12251 lookup_attribute first before lookup_attribute_spec.
12252 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12254 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12255 for functions with noipa attribute.
12256 (cgraph_externally_visible_p): Return true for functions with noipa
12258 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12259 for functions with noipa attribute.
12260 * doc/extend.texi: Document noipa function attribute.
12261 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12262 also for functions with noipa attribute.
12263 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12265 2017-07-26 Andrew Pinski <apinski@cavium.com>
12267 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12268 vec_unalign_load_cost and vec_unalign_store_cost.
12270 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12272 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12273 -mvsx-small-integer option.
12274 (ISA_3_0_MASKS_IEEE): Likewise.
12275 (OTHER_VSX_VECTOR_MASKS): Likewise.
12276 (POWERPC_MASKS): Likewise.
12277 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12278 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12279 code, only testing for DImode being allowed in non-VSX floating
12281 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12282 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12284 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12285 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12286 TARGET_P8_VECTOR test.
12287 (rs6000_secondary_reload_simple_move): Likewise.
12288 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12289 since TARGET_P9_VECTOR was already tested.
12290 (rs6000_opt_masks): Remove -mvsx-small-integer.
12291 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12292 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12294 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12295 test for TARGET_VEXTRACTUB was used, and that uses
12297 (p9 extract splitter): Likewise.
12298 (vsx_extract_<mode>_di_p9): Likewise.
12299 (vsx_extract_<mode>_store_p9): Likewise.
12300 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12301 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12302 the elimination of TARGET_VSX_SMALL_INTEGER.
12303 (vsx_extract_<mode>_p8): Likewise.
12304 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12305 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12306 (vsx_set_<mode>_p9): Likewise.
12307 (vsx_set_v4sf_p9): Likewise.
12308 (vsx_set_v4sf_p9_zero): Likewise.
12309 (vsx_insert_extract_v4sf_p9): Likewise.
12310 (vsx_insert_extract_v4sf_p9_2): Likewise.
12311 * config/rs6000/rs6000.md (sign extend splitter): Change
12312 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12313 (floatsi<mode>2_lfiwax_mem): Likewise.
12314 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12315 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12316 since a test for TARGET_P9_VECTOR was used.
12317 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12318 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12319 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12320 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12321 TARGET_P8_VECTOR test.
12322 (fix_trunc<mode>si2_stfiwx): Likewise.
12323 (fix_trunc<mode>si2_internal): Likewise.
12324 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12325 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12327 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12328 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12329 TARGET_P8_VECTOR test.
12330 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12331 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12332 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12334 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12335 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12336 since a test for TARGET_P9_VECTOR was used.
12337 (splitter for loading small constants): Likewise.
12339 2017-07-26 Andrew Pinski <apinski@cavium.com>
12341 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12344 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12347 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12348 (fp_valid_at): Likewise.
12350 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12352 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12353 (qdf24xx_addrcost_table): Likewise.
12354 (cortexa57_tunings): Update to use generic_branch_cost.
12355 (cortexa72_tunings): Likewise.
12356 (cortexa73_tunings): Likewise.
12357 (qdf24xx_tunings): Likewise.
12359 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12361 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12362 (thunderx2t99_branch_cost): Likewise.
12363 (cortexa35_tunings): Update to use generic_branch_cost.
12364 (cortexa53_tunings): Likewise.
12365 (cortexa57_tunings): Likewise.
12366 (cortexa72_tunings): Likewise.
12367 (cortexa73_tunings): Likewise.
12368 (thunderx2t99_tunings): Likewise.
12370 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12372 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12373 (sparc_option_override): Honour MASK_FSMULD.
12374 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12375 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12376 * config/sparc/sparc.opt (mfsmuld): New option.
12377 * doc/invoke.texi (mfsmuld): Document option.
12379 2017-07-26 Marek Polacek <polacek@redhat.com>
12381 PR middle-end/70992
12382 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12384 2017-07-26 Richard Biener <rguenther@suse.de>
12386 * gimple-match-head.c (do_valueize): Return OP if valueize
12388 (get_def): New helper to get at the def stmt of a SSA name
12389 if valueize allows.
12390 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12391 do_valueize to get at the def stmt.
12392 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12394 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12396 PR middle-end/46932
12397 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12399 2017-07-26 Martin Liska <mliska@suse.cz>
12402 * function.c (expand_function_start): Make expansion of
12403 nonlocal_goto_save_area after parm_birth_insn.
12405 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12407 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12408 from all CPU target flags enable members.
12410 2017-07-26 Richard Biener <rguenther@suse.de>
12412 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12413 (decision_tree::gen): Make 'type' const.
12414 (write_predicate): Likewise.
12416 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12418 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12419 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12420 (rs6000_option_override_internal): Likewise.
12421 (rs6000_expand_vector_set): Likewise.
12422 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12423 (TARGET_UPPER_REGS_SF): Likewise.
12424 (TARGET_UPPER_REGS_DI): Likewise.
12425 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12426 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12427 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12428 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12429 (Splitters for DI constants in Altivec registers): Likewise.
12430 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12431 (vsx_set_v4sf_p9): Likewise.
12432 (vsx_set_v4sf_p9_zero): Likewise.
12433 (vsx_insert_extract_v4sf_p9): Likewise.
12434 (vsx_insert_extract_v4sf_p9_2): Likewise.
12436 2017-07-25 Carl Love <cel@us.ibm.com>
12438 * doc/extend.texi: Update the built-in documentation file for the
12439 existing built-in functions
12440 vector signed char vec_cnttz (vector signed char);
12441 vector unsigned char vec_cnttz (vector unsigned char);
12442 vector signed short vec_cnttz (vector signed short);
12443 vector unsigned short vec_cnttz (vector unsigned short);
12444 vector signed int vec_cnttz (vector signed int);
12445 vector unsigned int vec_cnttz (vector unsigned int);
12446 vector signed long long vec_cnttz (vector signed long long);
12447 vector unsigned long long vec_cnttz (vector unsigned long long);
12449 2017-07-25 Andrew Pinski <apinski@cavium.com>
12451 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12452 accesses where the use is for the first operand of a BIT_INSERT.
12454 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12457 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12458 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12460 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12462 * config/i386/gstabs.h: Delete.
12463 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12465 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12467 * config/i386/i386.c (ix86_decompose_address): Do not check for
12468 register RTX when looking at index_reg or base_reg.
12469 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12471 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12473 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12474 to update EH info here.
12476 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12478 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12480 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12482 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12484 2017-07-25 Torsten Duwe <duwe@suse.de>
12486 * common.opt: Introduce -fpatchable-function-entry
12487 command line option, and its variables function_entry_patch_area_size
12488 and function_entry_patch_area_start.
12489 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12490 including a two-value parser.
12491 * target.def (print_patchable_function_entry): New target hook.
12492 * targhooks.h (default_print_patchable_function_entry): New function.
12493 * targhooks.c (default_print_patchable_function_entry): Likewise.
12494 * toplev.c (process_options): Switch off IPA-RA if
12495 patchable function entries are being generated.
12496 * varasm.c (assemble_start_function): Look at the
12497 patchable-function-entry command line switch and current
12498 function attributes and maybe generate NOP instructions by
12499 calling the print_patchable_function_entry hook.
12500 * doc/extend.texi: Document patchable_function_entry attribute.
12501 * doc/invoke.texi: Document -fpatchable_function_entry
12502 command line option.
12503 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12505 * doc/tm.texi: Re-generate.
12507 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12510 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12511 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12513 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12515 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12517 2017-07-25 Richard Biener <rguenther@suse.de>
12519 PR tree-optimization/81455
12520 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12521 not walk in cycles when looking for guards.
12523 2017-07-25 Richard Biener <rguenther@suse.de>
12525 PR tree-optimization/81529
12526 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12527 when optimizing backedge uses.
12529 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12531 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12533 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12534 to dl_section_ref. On AIX, append an expression to subtract
12535 the size of the section length to dl_section_ref.
12537 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12539 * configure.ac: If any of the config.* scripts fail, exit 1.
12540 * configure: Regenerate.
12542 2017-07-25 Richard Biener <rguenther@suse.de>
12544 PR middle-end/81546
12545 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12546 of immediate uses, be more verbose on errors.
12548 2017-07-25 Richard Biener <rguenther@suse.de>
12550 PR tree-optimization/81510
12551 * tree-vect-loop.c (vect_is_simple_reduction): When the
12552 reduction stmt is not inside the loop bail out.
12554 2017-07-25 Richard Biener <rguenther@suse.de>
12556 PR tree-optimization/81303
12557 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12558 profitability check against LOOP_VINFO_NITERSM1.
12560 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12562 * domwalk.c (cmp_bb_postorder): Simplify.
12563 (sort_bbs_postorder): New function. Use it...
12564 (dom_walker::walk): ...here to optimize common cases.
12566 2017-07-25 Martin Liska <mliska@suse.cz>
12569 * ipa-visibility.c (function_and_variable_visibility): Make the
12570 redirection just on target that supports aliasing.
12571 Fix GNU coding style.
12573 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12576 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12578 * config/arm/rtems.h: Likewise.
12579 * config/bfin/rtems.h: Likewise.
12580 * config/i386/rtemself.h: Likewise.
12581 * config/lm32/rtems.h: Likewise.
12582 * config/m32c/rtems.h: Likewise.
12583 * config/m68k/rtemself.h: Likewise.
12584 * config/microblaze/rtems.h: Likewise.
12585 * config/mips/rtems.h: Likewise.
12586 * config/moxie/rtems.h: Likewise.
12587 * config/nios2/rtems.h: Likewise.
12588 * config/powerpcspe/rtems.h: Likewise.
12589 * config/rs6000/rtems.h: Likewise.
12590 * config/rtems.h: Likewise.
12591 * config/sh/rtems.h: Likewise.
12592 * config/sh/rtemself.h: Likewise.
12593 * config/sparc/rtemself.h: Likewise.
12595 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12598 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12599 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12600 * tree-ssa-structalias.c (alias_get_name): Same.
12602 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12605 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12606 instructions if no du chain is found.
12608 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12610 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12612 2017-07-25 Richard Biener <rguenther@suse.de>
12614 PR middle-end/81505
12615 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12618 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12620 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12621 upper-regs options.
12622 (ISA_2_7_MASKS_SERVER): Likewise.
12623 (ISA_3_0_MASKS_IEEE): Likewise.
12624 (OTHER_P8_VECTOR_MASKS): Likewise.
12625 (OTHER_VSX_VECTOR_MASKS): Likewise.
12626 (POWERPC_MASKS): Likewise.
12627 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12628 duplicate list of options.
12629 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12630 explicit -mupper-regs options.
12631 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12632 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12633 alias for -mupper-regs-df.
12634 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12635 (rs6000_init_hard_regno_mode_ok): Likewise.
12636 (rs6000_option_override_internal): Likewise.
12637 (rs6000_opt_masks): Likewise.
12638 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12639 options in terms of whether -mvsx or -mpower8-vector was used.
12640 (TARGET_UPPER_REGS_DI): Likewise.
12641 (TARGET_UPPER_REGS_SF): Likewise.
12642 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12643 -mupper-regs-* options.
12645 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12647 * passes.c (emergency_dump_function): Print some empty lines and a
12648 header before the RTL dump.
12650 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12652 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12654 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12657 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12658 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12660 2017-07-24 Carl Love <cel@us.ibm.com>
12662 * config/rs6000/rs6000-c.c: Add support for built-in functions
12663 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12664 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12665 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12666 vec_extract_fp_from_shortl): Add defines for the two builtins.
12667 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12668 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12670 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12671 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12672 * doc/extend.texi: Update the built-in documentation file for the
12673 new built-in function.
12675 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12678 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12680 * doc/generic.texi: Likewise.
12681 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12682 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12684 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12686 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12687 (aarch64_mls_elt_merge<mode>): Likewise.
12689 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12691 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12692 having __cxa_atexit.
12694 2017-07-23 Michael Collison <michael.collison@arm.com>
12696 * config/arm/arm.c (arm_option_override): Deprecate
12697 use of -mstructure-size-boundary.
12698 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12699 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12701 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12704 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12705 Reduce cost estimate for direct moves.
12707 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12710 * config/i386/i386.c (ix86_option_override_internal): Disable
12711 BMI, BMI2 and TBM instructions for -m16.
12713 2017-07-21 Carl Love <cel@us.ibm.com>
12715 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12716 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12717 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12718 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12719 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12720 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12721 VMULOSW): New enum "unspec" values.
12722 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12723 altivec_vmulosw): New patterns.
12724 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12725 VMULOSW): Add definitions.
12727 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12729 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12730 (qdf24xx): Likewise.
12731 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12732 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12733 (AARCH64_FL_V8_1): Renumber.
12734 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12735 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12736 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12737 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12738 rdma to feature modifiers list.
12740 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12742 PR middle-end/56727
12743 * ipa-visibility (function_and_variable_visibility): Convert
12744 recursive PLT call to direct call if appropriate.
12746 2017-07-21 Andrew Pinski <apinski@cavium.com>
12748 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12749 operand 1 to see if the types precision matches.
12750 * fold-const.c (operand_equal_p): Likewise.
12752 2017-07-21 Richard Biener <rguenther@suse.de>
12754 PR tree-optimization/81303
12755 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12756 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12757 (vect_peeling_hash_get_lowest_cost): Adjust.
12758 (vect_enhance_data_refs_alignment): Likewise. Use
12759 vect_get_peeling_costs_all_drs to compute the penalty for no
12760 peeling to match up costs.
12762 2017-07-21 Richard Biener <rguenther@suse.de>
12764 PR tree-optimization/81500
12765 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12766 we didn't identify a reduction path.
12768 2017-07-21 Tom de Vries <tom@codesourcery.com>
12769 Cesar Philippidis <cesar@codesourcery.com>
12771 PR gcov-profile/81442
12772 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12775 2017-07-21 Tom de Vries <tom@codesourcery.com>
12778 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12780 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12781 nvptx_override_options_after_change.
12783 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12785 * dwarf2out.c (output_file_names): Avoid double testing for
12786 dwarf_version >= 5.
12788 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12790 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12792 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12794 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12796 (try_crossjump_to_edge): Do not punt on partitioned functions.
12798 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12800 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12801 Put all BBs reachable only via paths crossing cold region to cold
12803 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12805 2016-07-21 Richard Biener <rguenther@suse.de>
12807 PR tree-optimization/81303
12808 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12809 into account prologue and epilogue iterations when raising
12810 min_profitable_iters to sth at least covering one vector iteration.
12812 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12814 * config/arm/arm.c (arm_test_cpu_arch_dat):
12817 2017-07-20 Nathan Sidwell <nathan@acm.org>
12819 Remove TYPE_METHODS.
12820 * tree.h (TYPE_METHODS): Delete.
12821 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12822 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12823 (dbxout_type_methods): Scan TYPE_FIELDS.
12824 (dbxout_type): Don't check TYPE_METHODS here.
12825 * function.c (use_register_for_decl): Always ignore register for
12826 class types when not optimizing.
12827 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12828 * tree.c (free_lang_data_in_type): Stitch out member functions and
12829 templates from TYPE_FIELDS.
12830 (build_distinct_type_copy, verify_type_variant,
12831 verify_type): Member fns are on TYPE_FIELDS.
12832 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12833 * tree-pretty-print.c (dump_generic_node): Likewise.
12835 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12838 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12839 V2TImode and V4TImode.
12840 (ix86_expand_vector_extract): Likewise.
12841 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12842 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12843 (ssescalarmode): Handle V4TImode and V2TImode.
12844 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12845 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12846 (VEXTRACTI128_MODE): New mode iterator.
12847 (splitter for *vec_extractv?ti first element): New.
12848 (VEC_INIT_MODE): New mode iterator.
12849 (vec_init<mode>): Consolidate 3 expanders into one using
12850 VEC_INIT_MODE mode iterator.
12852 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12854 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12855 non_spilled_static_chain_regno_p.
12857 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12859 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12861 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12863 * bb-reorder.c (connect_traces): Allow copying of blocks within
12866 2017-07-20 Richard Biener <rguenther@suse.de>
12868 * gimple.h (gimple_phi_result): Add gphi * overload.
12869 (gimple_phi_result_ptr): Likewise.
12870 (gimple_phi_arg): Likewise. Adjust index assert to only
12871 allow actual argument accesses rather than all slots available
12873 (gimple_phi_arg_def): Add gphi * overload.
12874 * tree-phinodes.c (make_phi_node): Initialize only actual
12876 (resize_phi_node): Clear memory not covered by old node,
12877 do not initialize excess argument slots.
12878 (reserve_phi_args_for_new_edge): Initialize new argument slot
12881 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12883 PR tree-optimization/81388
12885 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12887 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12888 by removing computation of may_be_zero.
12890 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12891 Tom de Vries <tom@codesourcery.com>
12893 PR middle-end/81030
12894 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12895 when gimple level profile disagrees with what RTL expander did.
12897 2017-07-20 Richard Biener <rguenther@suse.de>
12899 PR tree-optimization/61171
12900 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12901 (vect_analyze_stmt): Add slp instance parameter.
12902 (vectorizable_reduction): Likewise.
12903 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12904 (vect_is_simple_reduction): Deal with chains not detected
12905 as SLP reduction chain, specifically not properly associated
12906 chains containing a mix of plus/minus.
12907 (get_reduction_op): Remove.
12908 (get_initial_defs_for_reduction): Simplify, pass in whether
12909 this is a reduction chain, pass in the SLP node for the PHIs.
12910 (vect_create_epilog_for_reduction): Get the SLP instance as
12912 (vectorizable_reduction): Get the SLP instance as arg.
12913 During analysis remember the SLP node with the PHIs in the
12914 instance. Simplify getting at the vectorized reduction PHIs.
12915 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12916 through SLP instance.
12917 (vect_slp_analyze_operations): Likewise.
12918 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12919 (vect_transform_stmt): Likewise.
12921 2017-07-20 Tom de Vries <tom@codesourcery.com>
12923 PR tree-optimization/81489
12924 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12925 read of phi arg location to before loop that modifies phi.
12927 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12929 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12932 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12934 PR middle-end/81331
12935 * except.c (execute): Fix ordering issue.
12937 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12939 PR rtl-optimization/81423
12940 * combine.c (make_compound_operation_int): Don't try to optimize
12941 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12943 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12945 PR rtl-optimization/81423
12946 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12947 with a constant that is -1 in the truncated to mode.
12949 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12951 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12952 (determine_unlikely_bbs): ... here.
12953 * predict.h (propagate_unlikely_bbs_forward): Declare.
12954 * cfgexpand.c (pass_expand::execute): Use it.
12955 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12957 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12958 propagate_unlikely_bbs_forward.
12960 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12962 PR middle-end/81331
12963 * except.c (maybe_add_nop_after_section_switch): New function.
12966 2017-07-19 Tom de Vries <tom@codesourcery.com>
12968 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12970 2017-07-19 Tom de Vries <tom@codesourcery.com>
12972 * gimple.h (gimple_phi_arg): Make assert more strict.
12974 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12976 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12977 * config/rs6000/mmintrin.h: New file.
12978 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12980 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12982 PR tree-optimization/81346
12983 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12985 2017-07-19 Tom de Vries <tom@codesourcery.com>
12987 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12989 2017-07-19 Tom de Vries <tom@codesourcery.com>
12991 * config/nvptx/nvptx-modes.def: Add V2DImode.
12992 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12993 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12994 (nvptx_output_mov_insn): Handle lack of mov.b128.
12995 (nvptx_print_operand): Handle 'H' and 'L' codes.
12996 (nvptx_vector_mode_supported): Allow V2DImode.
12997 (nvptx_preferred_simd_mode): New function.
12998 (nvptx_data_alignment): New function.
12999 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13000 nvptx_preferred_simd_mode.
13001 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13003 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13005 2017-07-19 Tom de Vries <tom@codesourcery.com>
13007 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13008 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13009 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13010 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13011 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13012 (mov<VECIM>_insn): New define_insn.
13013 (define_expand "mov<VECIM>): New define_expand.
13015 2017-07-19 Tom de Vries <tom@codesourcery.com>
13017 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13019 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13021 PR tree-optimization/81346
13022 * fold-const.h (fold_div_compare, range_check_type): Declare.
13023 * fold-const.c (range_check_type): New function.
13024 (build_range_check): Use range_check_type.
13025 (fold_div_compare): No longer static, rewritten into
13026 a match.pd helper function.
13027 (fold_comparison): Don't call fold_div_compare here.
13028 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13029 as helper function.
13031 2017-07-19 Nathan Sidwell <nathan@acm.org>
13033 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13034 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13035 * tree.c (find_decls_types_r, verify_type): Use
13036 TYPE_{MIN,MAX}_VALUE_RAW.
13037 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13038 (hash_tree): Likewise.
13039 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13041 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13044 2017-07-18 Tom de Vries <tom@codesourcery.com>
13046 PR middle-end/81464
13047 * omp-expand.c (expand_omp_for_static_chunk): Handle
13048 equal-argument loop exit phi.
13050 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13053 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13054 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13055 operand 2 predicate.
13056 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13057 operand 2 predicate.
13058 (ror,rol -> rorx splitters): Use const_int_operand as
13059 operand 2 predicate.
13061 2017-06-18 Richard Biener <rguenther@suse.de>
13063 PR tree-optimization/81410
13064 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13065 the gap in the ! slp_perm SLP case after each group.
13067 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13069 PR middle-end/81463
13070 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13073 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13075 PR middle-end/81462
13076 * predict.c (set_even_probabilities): Cleanup; do not affect
13077 probabilities that are already known.
13078 (combine_predictions_for_bb): Call even when count is set.
13080 2017-07-18 Nathan Sidwell <nathan@acm.org>
13082 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13085 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13088 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13089 optimization for loop niter analysis.
13091 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13094 * config/avr/avr.c (avr_optimize_casesi): Don't use
13095 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13097 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13099 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13100 body_cost_vec from _vect_peel_extended_info.
13101 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13102 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13105 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13107 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13109 2017-07-18 Richard Biener <rguenther@suse.de>
13111 PR tree-optimization/80620
13112 PR tree-optimization/81403
13113 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13114 info when re-using a VN table entry.
13116 2017-07-18 Richard Biener <rguenther@suse.de>
13118 PR tree-optimization/81418
13119 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13120 vectype_in. Verify that with lane-reducing reduction operations
13121 we have a single def-use cycle.
13123 2017-07-17 Carl Love <cel@us.ibm.com>
13125 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13127 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13128 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13129 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13130 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13131 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13132 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13133 VMULOSW): New enum "unspec" values.
13134 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13135 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13136 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13137 altivec_vmulosw): New patterns.
13138 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13139 VMULOSW): Add definitions.
13140 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13142 * config/alpha/alpha.c: Include predict.h.
13144 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13146 * tree-vrp.c (compare_assert_loc): Fix comparison function
13147 to return predictable results.
13149 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13151 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13153 (subdi3): Likewise.
13154 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13155 * doc/invoke.texi (mexpand-adddi): Update text.
13157 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13159 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13160 that also clobbers the CC register. The old expand code is moved
13162 (*arc_clzsi2): ... here.
13163 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13164 the CC register. The old expand code is moved to ...
13165 (arc_ctzsi2): ... here.
13167 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13169 * config/arc/arc.opt (mindexed-loads): Use initial value
13170 TARGET_INDEXED_LOADS_DEFAULT.
13171 (mauto-modify-reg): Use initial value
13172 TARGET_AUTO_MODIFY_REG_DEFAULT.
13173 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13174 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13175 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13176 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13178 2017-07-17 Martin Liska <mliska@suse.cz>
13181 * opts.c (finish_options): Do not allow -fgnu-tm
13182 w/ -fsanitize={kernel-,}address. Say sorry.
13184 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13187 * tree-loop-distribution.c (classify_partition): Only assert on
13188 numer of iterations.
13189 (merge_dep_scc_partitions): Delete prameter. Update function call.
13190 (distribute_loop): Remove code handling loop with unknown niters.
13191 (pass_loop_distribution::execute): Skip loop with unknown niters.
13193 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13196 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13197 function sort_partitions_by_post_order.
13199 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13201 PR tree-optimization/81374
13202 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13203 the max index of basic blocks, rather than number of basic blocks.
13205 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13207 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13209 (arc_legitimate_pic_operand_p): Likewise.
13210 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13212 (arc_needs_pcl_p): Likewise.
13213 (arc_legitimate_pc_offset_p): Likewise.
13214 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13215 function is also used in constrains.md.
13216 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13217 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13218 PLUS. Only return true/false in known cases, otherwise assert.
13219 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13220 is already called in arc_legitimate_constant_p.
13221 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13223 (LEGITIMATE_PIC_OPERAND_P): Use
13224 arc_raw_symbolic_reference_mentioned_p function.
13225 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13230 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13231 Andrew Burgess <andrew.burgess@embecosm.com>
13233 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13234 (arc_return_address_register): New function.
13235 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13236 (arc_handle_fndecl_attribute): Add naked attribute.
13237 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13238 (TARGET_WARN_FUNC_RETURN): Likewise.
13239 (arc_allocate_stack_slots_for_args): New function.
13240 (arc_warn_func_return): Likewise.
13241 (machine_function): Change type fn_type.
13242 (arc_compute_function_type): Consider new naked function type,
13243 change function return type.
13244 (arc_must_save_register): Adapt to handle new
13245 arc_compute_function_type's return type.
13246 (arc_expand_prologue): Likewise.
13247 (arc_expand_epilogue): Likewise.
13248 (arc_return_address_regs): Delete.
13249 (arc_return_address_register): New function.
13250 (arc_epilogue_uses): Use above function.
13251 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13252 (arc_function_type): Change encoding, add naked type.
13253 (ARC_INTERRUPT_P): Change to handle the new encoding.
13254 (ARC_FAST_INTERRUPT_P): Likewise.
13255 (ARC_NORMAL_P): Define.
13256 (ARC_NAKED_P): Likewise.
13257 (arc_compute_function_type): Delete prototype.
13258 * config/arc/arc.md (in_ret_delay_slot): Use
13259 arc_return_address_register function.
13260 (simple_return): Likewise.
13261 (p_return_i): Likewise.
13263 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13265 PR tree-optimization/81428
13266 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13267 can't be built for those types.
13269 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13271 Remove stuff dead since r239246.
13273 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13274 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13275 (avr_inform_devices): Remove dead stuff.
13277 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13279 * config/arm/arm_neon.h: Fix softp typo.
13281 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13283 PR tree-optimization/81365
13284 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13285 aggregate moves onto bb predecessor edges, make sure there are no
13286 loads that could alias the lhs in between the start of bb and the
13289 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13292 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13293 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13294 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13296 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13298 PR tree-optimization/81396
13299 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13300 (init_symbolic_number): Initialize it to 1.
13301 (perform_symbolic_merge): Add n_ops from both operands into the new
13303 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13304 without base_addr as useless if they need more than one operation.
13305 (bswap_replace): Handle !bswap case for NULL base_addr.
13307 2017-07-17 Tom de Vries <tom@codesourcery.com>
13310 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13313 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13315 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13316 conditional builtin define __FIX_LEON3FT_B2BST.
13318 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13320 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13321 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13324 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13326 PR rtl-optimization/81424
13327 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13328 to remove potential trapping from operands if -fnon-call-exceptions.
13330 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13333 profile_proability for scalling.
13334 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13336 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13338 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13340 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13342 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13343 fixpoint arithmetics.
13345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13347 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13348 fixpoint arithmetics.
13350 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13352 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13353 fixpoint arithmetics.
13355 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13357 * profile-count.h (profile_probability::from_reg_br_prob_note,
13358 profile_probability::to_reg_br_prob_note): New functions.
13359 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13360 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13361 * predict.c (probability_reliable_p): Update.
13362 (edge_probability_reliable_p): Update.
13363 (br_prob_note_reliable_p): Update.
13364 (invert_br_probabilities): Update.
13365 (add_reg_br_prob_note): New function.
13366 (combine_predictions_for_insn): Update.
13367 * asan.c (asan_clear_shadow): Update.
13368 * cfgbuild.c (compute_outgoing_frequencies): Update.
13369 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13370 (update_br_prob_note): Update.
13371 (rtl_verify_edges): Update.
13372 (purge_dead_edges): Update.
13373 (fixup_reorder_chain): Update.
13374 * emit-rtl.c (try_split): Update.
13375 * ifcvt.c (cond_exec_process_insns): Update.
13376 (cond_exec_process_if_block): Update.
13377 (dead_or_predicable): Update.
13378 * internal-fn.c (expand_addsub_overflow): Update.
13379 (expand_neg_overflow): Update.
13380 (expand_mul_overflow): Update.
13381 * loop-doloop.c (doloop_modify): Update.
13382 * loop-unroll.c (compare_and_jump_seq): Update.
13383 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13384 * predict.h: Update.
13385 * reorg.c (mostly_true_jump): Update.
13387 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13388 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13389 * config/arc/arc.c: (emit_unlikely_jump): Update.
13390 * config/arm/arm.c: (emit_unlikely_jump): Update.
13391 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13392 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13393 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13394 (ix86_print_operand): Update.
13395 (ix86_split_fp_branch): Update.
13396 (predict_jump): Update.
13397 * config/ia64/ia64.c (ia64_print_operand): Update.
13398 * config/mmix/mmix.c (mmix_print_operand): Update.
13399 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13400 (rs6000_expand_split_stack_prologue): Update.
13401 * config/rs6000/rs6000.c: Update.
13402 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13403 (s390_expand_vec_movstr): Update.
13404 (s390_expand_cs_tdsi): Update.
13405 (s390_expand_split_stack_prologue): Update.
13406 * config/sh/sh.c (sh_print_operand): Update.
13407 (expand_cbranchsi4): Update.
13408 (expand_cbranchdi4): Update.
13409 * config/sparc/sparc.c (output_v9branch): Update.
13410 * config/spu/spu.c (get_branch_target): Update.
13411 (ea_load_store_inline): Update.
13412 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13413 * config/tilepro/tilepro.c: Update.
13415 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13417 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13418 (gimplify_save_expr): Likewise.
13420 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13422 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13424 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13426 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13428 * timevar.def (TV_IPA_FNSUMMARY): Define.
13430 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13432 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13433 to back store errata sensitive sequence from being generated.
13434 (sqrtdf2_fix): Likewise.
13436 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13438 * tree-ssa-threadupdate.c (compute_path_counts,
13439 update_joiner_offpath_counts): Use profile_probability.
13441 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13444 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13446 * config/arm/arm-c.c (arm_cpu_builtins): Define
13447 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13449 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13452 array entries to represent __ieee128 versions of the
13453 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13454 scalar_extract_sig, and scalar_insert_exp built-in functions.
13455 (altivec_resolve_overloaded_builtin): Add special case handling
13456 for the __builtin_scalar_insert_exp function, as represented by
13457 the P9V_BUILTIN_VEC_VSIEDP constant.
13458 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13459 exponent support for __ieee128 argument.
13460 (VSESQP): Add scalar extract signature support for __ieee128
13462 (VSTDCNQP): Add scalar test negative support for __ieee128
13464 (VSIEQP): Add scalar insert exponent support for __int128 argument
13465 with __ieee128 result.
13466 (VSIEQPF): Add scalar insert exponent support for __ieee128
13467 argument with __ieee128 result.
13468 (VSTDCQP): Add scalar test data class support for __ieee128
13470 (VSTDCNQP): Add overload support for scalar test negative with
13471 __ieee128 argument.
13472 (VSTDCQP): Add overload support for scalar test data class
13473 __ieee128 argument.
13474 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13475 UNSPEC_VSX_SXSIGDP.
13476 (UNSPEC_VSX_SIEXPQP): New constant.
13477 (xsxexpqp): New insn for VSX scalar extract exponent quad
13479 (xsxsigqp): New insn for VSX scalar extract significand quad
13481 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13482 precision with floating point argument.
13483 (xststdcqp): New expand for VSX scalar test data class quad
13485 (xststdcnegqp): New expand for VSX scalar test negative quad
13487 (xststdcqp): New insn to match expansions for VSX scalar test data
13488 class quad precision and VSX scalar test negative quad precision.
13489 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13490 special case operand checking to enforce that second operand of
13491 VSX scalar test data class with quad precision argument is a 7-bit
13493 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13494 prototypes and descriptions of __ieee128 versions of
13495 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13496 scalar_test_data_class, and scalar_test_neg built-in functions.
13498 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13500 PR tree-optimization/81162
13501 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13502 replace a negate with an add.
13504 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13506 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13508 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13510 * config/arm/arm-c.c (arm_cpu_builtins): Define
13511 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13513 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13515 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13516 (armv8-r): Set ARM Cortex-R52 as default CPU.
13517 * config/arm/arm-tables.opt: Regenerate.
13518 * config/arm/arm-tune.md: Regenerate.
13519 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13521 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13522 extension for -mcpu=cortex-r52.
13524 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13526 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13527 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13528 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13529 (fp-armv8): Define it as FP_ARMv8 only.
13530 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13531 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13533 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13534 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13535 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13536 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13537 than TARGET_FPU_ARMV8.
13538 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13539 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13540 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13542 * config/arm/neon.md (neon_vrint): Likewise.
13543 (neon_vcvt): Likewise.
13544 (neon_<fmaxmin_op><mode>): Likewise.
13545 (<fmaxmin><mode>3): Likewise.
13546 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13547 * config/arm/predicates.md (arm_cond_move_operator): Check against
13548 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13550 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13552 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13553 to top of function.
13555 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13557 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13558 loop in comment with memset.
13560 2017-07-14 Martin Liska <mliska@suse.cz>
13562 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13563 * dwarf2out.c (is_java): Remove the function.
13564 (output_pubname): Remove usage of the function.
13565 (lower_bound_default): Remove usage of DW_LANG_Java.
13566 (gen_compile_unit_die): Likewise.
13567 * gcc.c: Remove compiler defaults for .java and .zip files.
13568 * gimple-expr.c (remove_suffix): Change as there's no longer
13569 extension than 4-letter one.
13570 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13571 (gimplify_save_expr): Likewise.
13572 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13573 as it's possible even for other languages than Java.
13574 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13575 * lto-opts.c (lto_write_options): Remove reference to Java.
13576 * opts.c (strip_off_ending): Update file extension handling.
13577 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13578 * tree-eh.c (lower_resx): Likewise.
13579 * tree.c (free_lang_data_in_type): Remove dead code.
13580 (find_decls_types_r): Likewise.
13581 (build_common_builtin_nodes): Remove Java from a comment.
13582 (verify_type): Remove dead code.
13583 * varasm.c (assemble_external): Remove Java from a comment.
13585 2017-07-14 Martin Liska <mliska@suse.cz>
13587 * opts.c (finish_options): Add quotes.
13588 (common_handle_option): Likewise.
13590 2017-07-14 Martin Liska <mliska@suse.cz>
13592 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13593 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13594 Remove N_SO_PASCAL.
13595 * dwarf2out.c (lower_bound_default): Do not handle
13597 (gen_compile_unit_die): Likewise.
13598 * gcc.c: Remove default extension binding for GNU Pascal.
13599 * stmt.c: Remove Pascal language from a comment.
13600 * xcoffout.c: Likewise.
13602 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13605 * diagnostic-show-locus.c (fixit_cmp): New function.
13606 (layout::layout): Sort m_fixit_hints.
13607 (column_range::column_range): Assert that the values are valid.
13608 (struct char_span): New struct.
13609 (correction::overwrite): New method.
13610 (struct source_line): New struct.
13611 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13612 calls in terms of classes source_line and char_span, and
13613 correction::overwrite.
13614 (selftest::test_overlapped_fixit_printing_2): New function.
13615 (selftest::diagnostic_show_locus_c_tests): Call it.
13617 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13619 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13620 early if there is no lhs.
13622 2017-07-13 Martin Liska <mliska@suse.cz>
13624 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13625 (gen_reference_type_die): Likewise.
13626 * stor-layout.c: Remove Pascal-related comment.
13628 2017-07-13 Martin Liska <mliska@suse.cz>
13630 * opts.c (finish_options): Add quotes to error messages.
13631 (parse_sanitizer_options): Likewise.
13633 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13635 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13637 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13639 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13641 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13643 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13645 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13647 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13650 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13651 provides the hardware capability bits, define the macro
13652 __BUILTIN_CPU_SUPPORTS__.
13653 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13654 if GLIBC does not provide the hardware capability bits. Add a
13655 gcc_unreachable call if the built-in cpu function is neither
13656 __builtin_cpu_is nor __builtin_cpu_supports.
13657 (rs6000_get_function_versions_dispatcher): Change the warning
13658 that an old GLIBC is used which does not export the capability
13659 bits to be an error.
13660 * doc/extend.texi (target_clones attribute): Document the
13661 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13662 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13663 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13664 the macros defined by GCC if the newer GLIBC is available.
13666 2017-07-12 Jeff Law <law@redhat.com>
13668 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13669 remaining includes slightly.
13670 * config/riscv/riscv-builtins.c: Include profile-count.h.
13672 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13675 * config/avr/avr.c (avr_set_current_function): In diagnostic
13676 messages: Quote keywords and (parts of) identifiers.
13677 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13680 2017-07-12 Carl Love <cel@us.ibm.com>
13682 * config/rs6000/rs6000-c.c: Add support for built-in functions
13683 vector bool char vec_revb (vector bool char);
13684 vector bool short vec_revb (vector short char);
13685 vector bool int vec_revb (vector bool int);
13686 vector bool long long vec_revb (vector bool long long);
13687 * doc/extend.texi: Update the built-in documentation file for the
13688 new built-in functions.
13690 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13692 * config/s390/s390.md: Remove movcc splitter.
13694 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13696 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13697 load/store on condition.
13699 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13702 * config/avr/avr.c (avr_encode_section_info)
13703 [progmem && !TREE_READONLY]: Error if progmem object needs
13706 2017-07-11 Michael Collison <michael.collison@arm.com>
13708 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13711 2017-07-11 Carl Love <cel@us.ibm.com>
13713 * config/rs6000/rs6000-c.c: Add support for builtins
13714 vector unsigned int vec_parity_lsbb (vector signed int);
13715 vector unsigned int vec_parity_lsbb (vector unsigned int);
13716 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13717 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13718 vector unsigned long long vec_parity_lsbb (vector signed long long);
13719 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13720 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13721 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13722 * doc/extend.texi: Update the built-in documentation file for the
13723 new built-in functions.
13725 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13727 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13728 (layout::m_primary_loc): New field.
13729 (layout::layout): Initialize new field. Move location filtering
13730 logic from here to...
13731 (layout::maybe_add_location_range): ...this new method. Add
13732 support for filtering to just the lines already specified by other
13734 (layout::will_show_line_p): New method.
13735 (gcc_rich_location::add_location_if_nearby): New method.
13736 (selftest::test_add_location_if_nearby): New test function.
13737 (selftest::diagnostic_show_locus_c_tests): Call it.
13738 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13741 2017-07-11 Tom de Vries <tom@codesourcery.com>
13743 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13744 (bb_first_real_insn): New function.
13745 (nvptx_single): Add extra initialization of broadcasted condition
13748 2017-07-11 Nathan Sidwell <nathan@acm.org>
13750 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13752 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13754 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13755 Binutils doc as TEXI will mess them up.
13756 * doc/invoke.texi (AVR Options): Same here.
13758 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13760 * config/sparc/sparc.opt (mfix-ut700): New option.
13761 (mfix-gr712rc): Likewise.
13762 (sparc_fix_b2bst): New variable.
13763 * doc/invoke.texi (SPARC options): Document them.
13764 (ARM options): Fix warnings.
13765 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13766 instructions to prevent sequences that can trigger the store-store
13767 errata for certain LEON3FT processors.
13768 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13769 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13770 * config/sparc/sparc.md (fix_b2bst): New attribute.
13771 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13773 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13776 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13778 (*rsqrtsf2_sse): Ditto.
13780 (div<mode>3): Macroize insn from divdf3 and divsf3
13781 using MODEF mode iterator.
13783 2017-07-10 Martin Sebor <msebor@redhat.com>
13785 PR tree-optimization/80397
13786 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13787 instead of testing for equality to INTEGER_TYPE.
13789 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13791 * config.gcc: Remove uclibc from arc target spec.
13793 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13795 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13797 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13800 * lto-wrapper.c (remove_option): New function.
13801 (merge_and_complain): Merge PIC/PIE options more realistically.
13803 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13805 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13809 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13810 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13811 * config.in: Regenerate.
13812 * configure: Regenerate.
13813 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13814 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13815 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13816 (TARGET_GASISR_PROLOGUES): ...target mask.
13817 * common/config/avr/avr-common.c
13818 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13819 Set -mgas-isr-prologues.
13820 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13821 INSERT_PASS_BEFORE for it.
13822 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13823 * config/avr/avr.c (avr_option_override)
13824 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13825 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13826 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13827 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13828 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13830 (make_avr_pass_pre_proep): New function.
13831 (emit_push_sfr) <treg>: Add argument to function and use it
13832 instead of TMP_REG.
13833 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13834 and set machine->gasisr.yes.
13835 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13836 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13837 __gcc_isr.n_pushed to .L__stack_usage.
13838 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13839 (avr_asm_final_postscan_insn): ...this new static function.
13840 * config/avr/avr.h (machine_function)
13841 <is_no_gccisr, use_L__stack_usage>: New fields.
13842 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13843 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13844 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13845 (gasisr, *gasisr): New expander and insn.
13846 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13847 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13848 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13850 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13852 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13855 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13857 Move jump-tables out of .text again.
13860 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13861 (ASM_OUTPUT_ADDR_VEC): New function.
13862 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13863 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13864 INSN_ADDRESSes as asm comment.
13865 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13866 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13867 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13868 * config/avr/avr.md (*tablejump): Adjust comment.
13869 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13870 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13872 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13873 (avr_output_addr_vec): New proto.
13874 (avr_log_t) <insn_addresses>: New field.
13876 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13879 * config/i386/i386.c (ix86_function_arg_advance): Set
13880 outgoing_args_on_stack to true if there are outgoing arguments
13882 (ix86_function_arg): Likewise.
13883 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13884 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13885 * config/i386/i386.h (machine_function): Add
13886 outgoing_args_on_stack.
13888 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13890 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13891 supporting pthreds.
13892 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13894 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13896 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13897 (REAL_H): Remove $(MACHMODE_H).
13898 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13900 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13901 $(MACHMODE_H) and double-int.h.
13902 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13904 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13907 2017-07-07 Andrew Pinski <apinski@cavium.com>
13909 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13910 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13912 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13914 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13915 Add warning if GCC was not configured to link against a GLIBC that
13916 exports the hardware capability bits.
13917 (make_resolver_func): Make resolver function private and not a
13918 COMDAT function. Create the name with clone_function_name instead
13919 of make_unique_name.
13922 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13923 correct operand in doing the split.
13925 2017-07-07 Carl Love <cel@us.ibm.com>
13927 * config/rs6000/rs6000-c: Add support for built-in function
13928 vector unsigned short vec_pack_to_short_fp32 (vector float,
13930 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13931 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13932 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13933 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13934 (convert_4f32_8i16): Add define_expand.
13935 * doc/extend.texi: Update the built-in documentation file for the
13936 new built-in function.
13938 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13940 * config/sparc/m8.md: New file.
13941 * config/sparc/sparc.md: Include m8.md.
13943 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13945 * config/sparc/sparc.opt: New option -mvis4b.
13946 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13947 (sparc_option_override): Handle VIS4B.
13948 (enum sparc_builtins): Define
13949 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13950 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13951 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13952 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13953 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13954 (check_constant_argument): New function.
13955 (sparc_vis_init_builtins): Define builtins
13956 __builtin_vis_dictunpack{8,16,32},
13957 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13958 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13959 __builtin_vis_fpcmpde{8,16,32}shl and
13960 __builtin_vis_fpcmpur{8,16,32}shl.
13961 (sparc_expand_builtin): Check that the constant operands to
13962 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13963 constant and in range.
13964 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13966 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13967 (SPARC_IMM5_P): Likewise.
13968 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13969 (enabled): Handle vis4b.
13970 (UNSPEC_DICTUNPACK): New unspec.
13971 (UNSPEC_FPCMPSHL): Likewise.
13972 (UNSPEC_FPUCMPSHL): Likewise.
13973 (UNSPEC_FPCMPDESHL): Likewise.
13974 (UNSPEC_FPCMPURSHL): Likewise.
13975 (cpu_feature): New CPU feature `vis4b'.
13976 (dictunpack{8,16,32}): New insns.
13977 (FPCSMODE): New mode iterator.
13978 (fpcscond): New code iterator.
13979 (fpcsucond): Likewise.
13980 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13981 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13982 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13983 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13984 * config/sparc/constraints.md: Define constraints `q' for unsigned
13985 2-bit integer constants and `t' for unsigned 5-bit integer
13987 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13989 (imm5_operand_dictunpack16): Likewise.
13990 (imm5_operand_dictunpack32): Likewise.
13991 (imm2_operand): Likewise.
13992 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13993 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13994 ditunpack* and fpcmp*shl builtins.
13996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13998 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13999 * config.in: Add HAVE_AS_SPARC6 define.
14000 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14002 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14004 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14005 (CPP_CPU_SPEC): Handle m8.
14006 (ASM_CPU_SPEC): Likewise.
14007 * config/sparc/sparc-opts.h (enum processor_type): Add
14009 * config/sparc/sparc.c (m8_costs): New struct.
14010 (sparc_option_override): Handle TARGET_CPU_m8.
14011 (sparc32_initialize_trampoline): Likewise.
14012 (sparc64_initialize_trampoline): Likewise.
14013 (sparc_issue_rate): Likewise.
14014 (sparc_register_move_cost): Likewise.
14015 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14016 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14017 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14018 (CPP_CPU_SPEC): Handle M8.
14019 (ASM_CPU_SPEC): Likewise.
14020 (AS_M8_FLAG): Define.
14021 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14022 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14023 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14025 * configure: Regenerate.
14026 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14029 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14031 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14033 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14034 ("*movdi_insn_sp32"): Do not set v3pipe.
14035 ("*movsi_insn"): Likewise.
14036 ("*movdi_insn_sp64"): Likewise.
14037 ("*movsf_insn"): Likewise.
14038 ("*movdf_insn_sp32"): Likewise.
14039 ("*movdf_insn_sp64"): Likewise.
14040 ("*zero_extendsidi2_insn_sp64"): Likewise.
14041 ("*sign_extendsidi2_insn"): Likewise.
14042 ("*mov<VM32:mode>_insn"): Likewise.
14043 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14044 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14045 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14046 ("<vlop:code><VL:mode>3"): Likewise.
14047 ("*not_<vlop:code><VL:mode>3"): Likewise.
14048 ("*nand<VL:mode>_vis"): Likewise.
14049 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14050 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14051 ("one_cmpl<VL:mode>2"): Likewise.
14052 ("faligndata<VM64:mode>_vis"): Likewise.
14053 ("alignaddrsi_vis"): Likewise.
14054 ("alignaddrdi_vis"): Likweise.
14055 ("alignaddrlsi_vis"): Likewise.
14056 ("alignaddrldi_vis"): Likewise.
14057 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14058 ("bmaskdi_vis"): Likewise.
14059 ("bmasksi_vis"): Likewise.
14060 ("bshuffle<VM64:mode>_vis"): Likewise.
14061 ("cmask8<P:mode>_vis"): Likewise.
14062 ("cmask16<P:mode>_vis"): Likewise.
14063 ("cmask32<P:mode>_vis"): Likewise.
14064 ("pdistn<P:mode>_vis"): Likewise.
14065 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14067 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14069 * config/sparc/sparc.md ("subtype"): New insn attribute.
14070 ("*wrgsr_sp64"): Set insn subtype.
14071 ("*rdgsr_sp64"): Likewise.
14072 ("alignaddrsi_vis"): Likewise.
14073 ("alignaddrdi_vis"): Likewise.
14074 ("alignaddrlsi_vis"): Likewise.
14075 ("alignaddrldi_vis"): Likewise.
14076 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14077 ("fexpand_vis"): Likewise.
14078 ("fpmerge_vis"): Likewise.
14079 ("faligndata<VM64:mode>_vis"): Likewise.
14080 ("bshuffle<VM64:mode>_vis"): Likewise.
14081 ("cmask8<P:mode>_vis"): Likewise.
14082 ("cmask16<P:mode>_vis"): Likewise.
14083 ("cmask32<P:mode>_vis"): Likewise.
14084 ("fchksm16_vis"): Likewise.
14085 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14086 ("fmean16_vis"): Likewise.
14087 ("fp<plusminus_insn>64_vis"): Likewise.
14088 ("<plusminus_insn>v8qi3"): Likewise.
14089 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14090 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14091 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14092 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14093 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14094 ("*movqi_insn"): Likewise.
14095 ("*movhi_insn"): Likewise.
14096 ("*movsi_insn"): Likewise.
14097 ("movsi_pic_gotdata_op"): Likewise.
14098 ("*movdi_insn_sp32"): Likewise.
14099 ("*movdi_insn_sp64"): Likewise.
14100 ("movdi_pic_gotdata_op"): Likewise.
14101 ("*movsf_insn"): Likewise.
14102 ("*movdf_insn_sp32"): Likewise.
14103 ("*movdf_insn_sp64"): Likewise.
14104 ("*zero_extendhisi2_insn"): Likewise.
14105 ("*zero_extendqihi2_insn"): Likewise.
14106 ("*zero_extendqisi2_insn"): Likewise.
14107 ("*zero_extendqidi2_insn"): Likewise.
14108 ("*zero_extendhidi2_insn"): Likewise.
14109 ("*zero_extendsidi2_insn_sp64"): Likewise.
14110 ("ldfsr"): Likewise.
14111 ("prefetch_64"): Likewise.
14112 ("prefetch_32"): Likewise.
14113 ("tie_ld32"): Likewise.
14114 ("tie_ld64"): Likewise.
14115 ("*tldo_ldub_sp32"): Likewise.
14116 ("*tldo_ldub1_sp32"): Likewise.
14117 ("*tldo_ldub2_sp32"): Likewise.
14118 ("*tldo_ldub_sp64"): Likewise.
14119 ("*tldo_ldub1_sp64"): Likewise.
14120 ("*tldo_ldub2_sp64"): Likewise.
14121 ("*tldo_ldub3_sp64"): Likewise.
14122 ("*tldo_lduh_sp32"): Likewise.
14123 ("*tldo_lduh1_sp32"): Likewise.
14124 ("*tldo_lduh_sp64"): Likewise.
14125 ("*tldo_lduh1_sp64"): Likewise.
14126 ("*tldo_lduh2_sp64"): Likewise.
14127 ("*tldo_lduw_sp32"): Likewise.
14128 ("*tldo_lduw_sp64"): Likewise.
14129 ("*tldo_lduw1_sp64"): Likewise.
14130 ("*tldo_ldx_sp64"): Likewise.
14131 ("*mov<VM32:mode>_insn"): Likewise.
14132 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14133 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14135 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14137 * config/sparc/sparc.md ("type"): New insn type viscmp.
14138 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14140 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14141 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14142 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14143 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14145 ("n7_vis_logical_11cycle"): Likewise.
14146 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14147 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14148 * config/sparc/niagara.md ("niag_vis"): Likewise.
14149 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14150 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14152 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14154 * config/sparc/sparc.md: New instruction type `bmask'.
14155 (bmaskdi_vis): Use the `bmask' type.
14156 (bmasksi_vis): Likewise.
14157 * config/sparc/ultra3.md (us3_array): Likewise.
14158 * config/sparc/niagara7.md (n7_array): Likewise.
14159 * config/sparc/niagara4.md (n4_array): Likewise.
14160 * config/sparc/niagara2.md (niag2_vis): Likewise.
14161 (niag3_vis): Likewise.
14162 * config/sparc/niagara.md (niag_vis): Likewise.
14164 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14166 * ipa-comdats.c: Remove optimize check from gate.
14167 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14168 for functions not optimized.
14169 (ipa_fn_summary_read): Skip optimize check.
14170 (ipa_fn_summary_write): Likewise.
14171 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14173 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14175 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14176 (check_callers): Give up on caller which is not optimized.
14177 (inline_small_functions): Likewise.
14178 (ipa_inline): Do not give up when not optimizing.
14179 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14180 away unoptimizes cdtors.
14181 (whole_program_function_and_variable_visibility): Do
14182 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14183 * ipa.c (process_references): Do not check optimize.
14184 (symbol_table::remove_unreachable_nodes): Update optimize check.
14185 (set_writeonly_bit): Update optimize check.
14186 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14187 (pass_ipa_single_use::gate): Remove.
14189 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14191 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14192 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14193 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14194 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14195 permute_load, permute_store, adjust_extract, adjust_splat,
14196 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14197 replace_swap_with_copy, dump_swap_insn_table,
14198 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14199 recombine_lvx_pattern, recombine_stvx_pattern,
14200 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14201 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14202 to file rs6000-p8swap.c.
14203 * config/rs6000/rs6000-p8swap.c: New file.
14204 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14205 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14206 and rs6000*-*-* targets.
14208 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14210 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14212 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14214 * lto-wrapper.c (merge_and_complain): Do not merge
14215 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14216 fsigned_zeros, ftrapping_math, fwrapv.
14217 (append_compiler_options): Do not track these options.
14218 (append_linker_options): Likewie
14220 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14222 * cgraphunit.c (cgraph_node::finalize_function): When
14223 !flag_toplevel_reorde set no_reorder flag.
14224 (varpool_node::finalize_decl): Likewise.
14225 (symbol_table::compile): Drop no toplevel reorder path.
14227 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14229 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14230 edges; zero probability is not better than uninitialized.
14232 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14234 * asan.h (asan_sanitize_allocas_p): Declare.
14235 * asan.c (asan_sanitize_allocas_p): New function.
14236 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14237 (handle_builtin_alloca): Likewise.
14238 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14239 if !asan_sanitize_allocas_p.
14240 * params.def (asan-instrument-allocas): Add new option.
14241 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14242 * opts.c (common_handle_option): Disable allocas sanitization for
14245 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14247 * asan.c: Include gimple-fold.h.
14248 (get_last_alloca_addr): New function.
14249 (handle_builtin_stackrestore): Likewise.
14250 (handle_builtin_alloca): Likewise.
14251 (asan_emit_allocas_unpoison): Likewise.
14252 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14253 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14254 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14255 (instrument_builtin_call): Pass gimple iterator to
14256 get_mem_refs_of_builtin_call.
14257 (last_alloca_addr): New global.
14258 * asan.h (asan_emit_allocas_unpoison): Declare.
14259 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14260 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14261 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14262 if function calls alloca.
14263 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14264 * gimple-fold.h (replace_call_with_value): Declare.
14265 * internal-fn.c: Include asan.h.
14266 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14267 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14269 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14271 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14272 (C_SELFTEST_FLAGS): New.
14273 (CPP_SELFTEST_FLAGS): New.
14274 (SELFTEST_DEPS): New, from deps of s-selftest.
14275 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14276 (CPP_SELFTEST_DEPS): New.
14277 (selftest): Add dependency on s-selftest-c++.
14278 (s-selftest): Rename to...
14279 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14280 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14281 than SELFTEST_FLAGS.
14282 (selftest-gdb): Rename to...
14283 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14285 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14286 (selftest-valgrind): Rename to...
14287 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14289 (selftest-valgrind): Reintroduce as an alias for
14290 selftest-c-valgrind.
14291 (s-selftest-c++): New.
14292 (selftest-c++-gdb): New.
14293 (selftest-c++-valgrind): New.
14295 2017-07-06 Olivier Hainque <hainque@adacore.com>
14297 * gcc.c (process_command): When deciding if undefined variables
14298 should be ignored when processing specs, accept "gcc -v" as well.
14300 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14302 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14303 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14305 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14307 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14308 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14309 * config/arm/arm-tables.opt: Regenerate.
14310 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14312 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14314 2017-07-06 Carl Love <cel@us.ibm.com>
14316 * ChangeLog: Clean up from mid air collision
14318 2017-07-06 Carl Love <cel@us.ibm.com>
14320 * config/rs6000/rs6000-c.c: Add support for built-in functions
14321 vector signed int vec_subc (vector signed int, vector signed int);
14322 vector signed __int128 vec_subc (vector signed __int128,
14323 vector signed __int128);
14324 vector unsigned __int128 vec_subc (vector unsigned __int128,
14325 vector unsigned __int128);
14326 vector signed int vec_sube (vector signed int, vector signed int,
14327 vector signed int);
14328 vector unsigned int vec_sube (vector unsigned int,
14329 vector unsigned int,
14330 vector unsigned int);
14331 vector signed __int128 vec_sube (vector signed __int128,
14332 vector signed __int128,
14333 vector signed__int128);
14334 vector unsigned __int128 vec_sube (vector unsigned __int128,
14335 vector unsigned __int128,
14336 vector unsigned __int128);
14337 vector signed int vec_subec (vector signed int, vector signed int,
14338 vector signed int);
14339 vector unsigned int vec_subec (vector unsigned int,
14340 vector unsigned int,
14341 vector unsigned int);
14342 vector signed __int128 vec_subec (vector signed __int128,
14343 vector signed __int128,
14344 vector signed__int128);
14345 vector unsigned __int128 vec_subec (vector unsigned __int128,
14346 vector unsigned __int128,
14347 vector unsigned __int128);
14348 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14349 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14350 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14351 BU_ALTIVEC_OVERLOAD_X definitions.
14352 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14353 * doc/extend.texi: Update the built-in documentation file for the new
14354 built-in functions.
14356 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14359 * diagnostic-show-locus.c (layout::layout): Use start and finish
14360 spelling location for the start and finish of each range.
14361 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14362 Add unused aspect param.
14363 * input.c (expand_location_1): Add "aspect" param, and use it
14364 to access the correct part of the location.
14365 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14367 (expand_location_to_spelling_point): Likewise.
14368 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14369 param, and pass it to expand_location_1.
14371 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14373 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14374 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14375 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14376 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14377 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14378 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14379 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14380 _mm_maskz_getmant_ss): New intrinsics.
14381 (__builtin_ia32_getexpss128_mask): Changed to ...
14382 __builtin_ia32_getexpss128_round ... this.
14383 (__builtin_ia32_getexpsd128_mask): Changed to ...
14384 __builtin_ia32_getexpsd128_round ... this.
14385 * config/i386/i386-builtin-types.def
14386 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14387 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14388 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14389 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14390 __builtin_ia32_getmantss_mask_round): New builtins.
14391 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14392 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14393 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14394 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14395 * config/i386/sse.md
14396 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14397 avx512f_sgetexp<mode><mask_scalar_name>
14398 <round_saeonly_scalar_name> ... this.
14399 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14400 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14401 vgetexp<ssescalarmodesuffix>
14402 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14403 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14404 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14405 avx512f_vgetmant<mode><mask_scalar_name>
14406 <round_saeonly_scalar_name> ... this.
14407 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14408 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14409 vgetmant<ssescalarmodesuffix>
14410 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14411 %0<mask_scalar_operand4>, %1, %2
14412 <round_saeonly_scalar_mask_op4>, %3} ... this.
14413 * config/i386/subst.md (mask_scalar_operand4,
14414 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14415 round_saeonly_scalar_nimm_predicate): New subst attributes.
14417 2017-07-06 Julia Koval <julia.koval@intel.com>
14419 * config/i386/i386.c (ix86_erase_embedded_rounding):
14420 Remove code for old rounding pattern.
14422 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14424 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14426 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14428 * doc/sourcebuild.texi (Test Directives, Variants of
14429 dg-require-support): Add documentation for dg-require-stack-check.
14431 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14433 * config/i386/subst.md (mask_scalar, round_scalar,
14434 round_saeonly_scalar): New meta-templates.
14435 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14436 round_scalar_mask_operand3, round_scalar_mask_op3,
14437 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14438 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14439 round_saeonly_scalar_constraint,
14440 round_saeonly_scalar_prefix): New subst attribute.
14441 * config/i386/sse.md
14442 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14443 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14444 <round_scalar_name> ... this.
14445 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14446 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14447 <round_scalar_name> ... this.
14448 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14449 <sse>_vm<code><mode>3<mask_scalar_name>
14450 <round_saeonly_scalar_name> ... this.
14451 (v<plusminus_mnemonic><ssescalarmodesuffix>
14452 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14453 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14454 v<plusminus_mnemonic><ssescalarmodesuffix>
14455 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14456 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14457 (v<multdiv_mnemonic><ssescalarmodesuffix>
14458 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14459 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14460 v<multdiv_mnemonic><ssescalarmodesuffix>
14461 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14462 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14463 (v<maxmin_float><ssescalarmodesuffix>
14464 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14465 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14466 v<maxmin_float><ssescalarmodesuffix>
14467 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14468 %0<mask_scalar_operand3>, %1, %<iptr>2
14469 <round_saeonly_scalar_mask_op3>} ... this.
14471 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14473 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14474 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14476 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14477 Alan Hayward <alan.hayward@arm.com>
14478 David Sherwood <david.sherwood@arm.com>
14480 * combine.c (simplify_if_then_else): Remove "enum" before
14482 * compare-elim.c (can_eliminate_compare): Likewise.
14483 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14485 (aarch64_lookup_simd_builtin_type): Likewise.
14486 (aarch64_simd_builtin_type): Likewise.
14487 (aarch64_init_simd_builtin_types): Likewise.
14488 (aarch64_simd_expand_args): Likewise.
14489 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14491 (aarch64_reverse_mask): Likewise.
14492 (aarch64_simd_emit_reg_reg_move): Likewise.
14493 (aarch64_gen_adjusted_ldpstp): Likewise.
14494 (aarch64_ccmp_mode_to_code): Likewise.
14495 (aarch64_operands_ok_for_ldpstp): Likewise.
14496 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14497 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14499 (aarch64_min_divisions_for_recip_mul): Likewise.
14500 (aarch64_reassociation_width): Likewise.
14501 (aarch64_get_condition_code_1): Likewise.
14502 (aarch64_simd_emit_reg_reg_move): Likewise.
14503 (aarch64_simd_attr_length_rglist): Likewise.
14504 (aarch64_reverse_mask): Likewise.
14505 (aarch64_operands_ok_for_ldpstp): Likewise.
14506 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14507 (aarch64_gen_adjusted_ldpstp): Likewise.
14508 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14510 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14511 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14512 (arm_lookup_simd_builtin_type): Likewise.
14513 (arm_simd_builtin_type): Likewise.
14514 (arm_init_simd_builtin_types): Likewise.
14515 (arm_expand_builtin_args): Likewise.
14516 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14517 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14518 (ft32_setup_incoming_varargs): Likewise.
14519 (ft32_function_arg): Likewise.
14520 (ft32_function_arg_advance): Likewise.
14521 (ft32_pass_by_reference): Likewise.
14522 (ft32_arg_partial_bytes): Likewise.
14523 (ft32_valid_pointer_mode): Likewise.
14524 (ft32_addr_space_pointer_mode): Likewise.
14525 (ft32_addr_space_legitimate_address_p): Likewise.
14526 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14528 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14529 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14530 (iamcu_alignment): Likewise.
14531 (canonicalize_vector_int_perm): Likewise.
14532 (ix86_noce_conversion_profitable_p): Likewise.
14533 (ix86_mpx_bound_mode): Likewise.
14534 (ix86_operands_ok_for_move_multiple): Likewise.
14535 * config/microblaze/microblaze-protos.h
14536 (microblaze_expand_conditional_branch_reg): Likewise.
14537 * config/microblaze/microblaze.c
14538 (microblaze_expand_conditional_branch_reg): Likewise.
14539 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14541 (rs6000_reassociation_width): Likewise.
14542 (rs6000_invalid_binary_op): Likewise.
14543 (fusion_p9_p): Likewise.
14544 (emit_fusion_p9_load): Likewise.
14545 (emit_fusion_p9_store): Likewise.
14546 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14548 (riscv_hard_regno_mode_ok_p): Likewise.
14549 (riscv_address_insns): Likewise.
14550 (riscv_split_symbol): Likewise.
14551 (riscv_legitimize_move): Likewise.
14552 (riscv_function_value): Likewise.
14553 (riscv_hard_regno_nregs): Likewise.
14554 (riscv_expand_builtin): Likewise.
14555 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14556 (riscv_build_integer): Likewise.
14557 (riscv_split_integer): Likewise.
14558 (riscv_legitimate_constant_p): Likewise.
14559 (riscv_cannot_force_const_mem): Likewise.
14560 (riscv_regno_mode_ok_for_base_p): Likewise.
14561 (riscv_valid_base_register_p): Likewise.
14562 (riscv_valid_offset_p): Likewise.
14563 (riscv_valid_lo_sum_p): Likewise.
14564 (riscv_classify_address): Likewise.
14565 (riscv_legitimate_address_p): Likewise.
14566 (riscv_address_insns): Likewise.
14567 (riscv_load_store_insns): Likewise.
14568 (riscv_force_binary): Likewise.
14569 (riscv_split_symbol): Likewise.
14570 (riscv_force_address): Likewise.
14571 (riscv_legitimize_address): Likewise.
14572 (riscv_move_integer): Likewise.
14573 (riscv_legitimize_const_move): Likewise.
14574 (riscv_legitimize_move): Likewise.
14575 (riscv_address_cost): Likewise.
14576 (riscv_subword): Likewise.
14577 (riscv_output_move): Likewise.
14578 (riscv_canonicalize_int_order_test): Likewise.
14579 (riscv_emit_int_order_test): Likewise.
14580 (riscv_function_arg_boundary): Likewise.
14581 (riscv_pass_mode_in_fpr_p): Likewise.
14582 (riscv_pass_fpr_single): Likewise.
14583 (riscv_pass_fpr_pair): Likewise.
14584 (riscv_get_arg_info): Likewise.
14585 (riscv_function_arg): Likewise.
14586 (riscv_function_arg_advance): Likewise.
14587 (riscv_arg_partial_bytes): Likewise.
14588 (riscv_function_value): Likewise.
14589 (riscv_pass_by_reference): Likewise.
14590 (riscv_setup_incoming_varargs): Likewise.
14591 (riscv_print_operand): Likewise.
14592 (riscv_elf_select_rtx_section): Likewise.
14593 (riscv_save_restore_reg): Likewise.
14594 (riscv_for_each_saved_reg): Likewise.
14595 (riscv_register_move_cost): Likewise.
14596 (riscv_hard_regno_mode_ok_p): Likewise.
14597 (riscv_hard_regno_nregs): Likewise.
14598 (riscv_class_max_nregs): Likewise.
14599 (riscv_memory_move_cost): Likewise.
14600 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14601 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14602 (rl78_addr_space_address_mode): Likewise.
14603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14606 (rs6000_reassociation_width): Likewise.
14607 (rs6000_invalid_binary_op): Likewise.
14608 (fusion_p9_p): Likewise.
14609 (emit_fusion_p9_load): Likewise.
14610 (emit_fusion_p9_store): Likewise.
14611 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14612 (ok_for_simple_move_operands): Likewise.
14613 (ok_for_simple_move_strict_operands): Likewise.
14614 (ok_for_simple_arith_logic_operands): Likewise.
14615 (visium_legitimize_reload_address): Likewise.
14616 (visium_select_cc_mode): Likewise.
14617 (output_cbranch): Likewise.
14618 (visium_split_double_move): Likewise.
14619 (visium_expand_copysign): Likewise.
14620 (visium_expand_int_cstore): Likewise.
14621 (visium_expand_fp_cstore): Likewise.
14622 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14623 (visium_function_arg): Likewise.
14624 (visium_function_arg_advance): Likewise.
14625 (visium_libcall_value): Likewise.
14626 (visium_setup_incoming_varargs): Likewise.
14627 (visium_legitimate_constant_p): Likewise.
14628 (visium_legitimate_address_p): Likewise.
14629 (visium_legitimize_address): Likewise.
14630 (visium_secondary_reload): Likewise.
14631 (visium_register_move_cost): Likewise.
14632 (visium_memory_move_cost): Likewise.
14633 (prepare_move_operands): Likewise.
14634 (ok_for_simple_move_operands): Likewise.
14635 (ok_for_simple_move_strict_operands): Likewise.
14636 (ok_for_simple_arith_logic_operands): Likewise.
14637 (visium_function_value_1): Likewise.
14638 (rtx_ok_for_offset_p): Likewise.
14639 (visium_legitimize_reload_address): Likewise.
14640 (visium_split_double_move): Likewise.
14641 (visium_expand_copysign): Likewise.
14642 (visium_expand_int_cstore): Likewise.
14643 (visium_expand_fp_cstore): Likewise.
14644 (visium_split_cstore): Likewise.
14645 (visium_select_cc_mode): Likewise.
14646 (visium_split_cbranch): Likewise.
14647 (output_cbranch): Likewise.
14648 (visium_print_operand_address): Likewise.
14649 * expmed.c (flip_storage_order): Likewise.
14650 * expmed.h (emit_cstore): Likewise.
14651 (flip_storage_order): Likewise.
14652 * genrecog.c (validate_pattern): Likewise.
14653 * hsa-gen.c (gen_hsa_addr): Likewise.
14654 * internal-fn.c (expand_arith_overflow): Likewise.
14655 * ira-color.c (allocno_copy_cost_saving): Likewise.
14656 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14657 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14658 (process_invariant_for_inheritance): Likewise.
14659 * lra-eliminations.c (move_plus_up): Likewise.
14660 * omp-low.c (lower_oacc_reductions): Likewise.
14661 * simplify-rtx.c (simplify_subreg): Likewise.
14662 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14663 (TARGET_CHKP_BOUND_MODE): Likewise..
14664 * targhooks.c (default_chkp_bound_mode): Likewise.
14665 (default_setup_incoming_vararg_bounds): Likewise.
14666 * targhooks.h (default_chkp_bound_mode): Likewise.
14667 (default_setup_incoming_vararg_bounds): Likewise.
14668 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14669 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14670 (have_whole_vector_shift): Likewise.
14671 * tree-vect-stmts.c (vectorizable_load): Likewise.
14672 * doc/tm.texi: Regenerate.
14674 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14676 Graceful degrade if Binutils PR21472 is not available.
14679 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14680 .rodata in flash test fails.
14681 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14682 * confgure: Regenerate.
14683 * config.in: Regenerate.
14684 * config/avr/avr.c (avr_asm_named_section)
14685 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14686 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14687 (avr_asm_init_sections): Same.
14689 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14691 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14692 (fma<VH:mode>4_intrinsic): Likewise.
14693 (*fmsub<VCVTF:mode>4): Likewise.
14694 (*fmsub<VH:mode>4_intrinsic): Likewise.
14696 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14699 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14700 Don't depend on "optimize > 0".
14701 (out_movhi_r_mr, out_movqi_mr_r): Same.
14702 (out_movhi_mr_r, out_movqi_r_mr): Same.
14703 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14704 io_address_operand on "optimize > 0".
14706 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14708 * tree-loop-distribution.c: Add general explanantion on the pass.
14709 (generate_loops_for_partition): Mark distributed loop.
14710 (pg_add_dependence_edges): New parameter. Handle alias data
14711 dependence specially and record it in the parameter if asked.
14712 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14713 (init_partition_graph_vertices, add_partition_graph_edge): New.
14714 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14715 (free_partition_graph_vdata, build_partition_graph): New.
14716 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14717 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14718 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14719 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14720 (version_for_distribution_p, finalize_partitions): New.
14721 (distribute_loop): Handle alias data dependence specially. Factor
14722 out loop fusion code as functions and call these functions.
14724 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14726 * tree-loop-distribution.c (classify_partition): New parameter and
14727 better handle reduction statement.
14728 (rdg_build_partitions): Revise comment.
14729 (distribute_loop): Compute statements in all partitions and pass it
14730 to classify_partition.
14732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14734 * tree-loop-distribution.c (enum partition_type): New.
14735 (struct partition): New field type.
14736 (partition_merge_into): Add parameter. Update partition type.
14737 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14738 (build_rdg_partition_for_vertex): Compute partition type.
14739 (rdg_build_partitions): Dump partition type.
14740 (distribute_loop): Update calls to partition_merge_into.
14742 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14744 * tree-loop-distribution.c (struct ddr_hasher): New.
14745 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14747 (classify_partition): Call get_data_dependence.
14748 (pg_add_dependence_edges): Ditto.
14749 (distribute_loop): Release data dependence hash table.
14751 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14753 * tree-loop-distribution.c (ref_base_address): Delete.
14754 (similar_memory_accesses): Rename ...
14755 (share_memory_accesses): ... to this. Check if partitions access
14756 the same memory reference.
14757 (distribute_loop): Call share_memory_accesses.
14759 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14761 * tree-loop-distribution.c (struct partition): New field recording
14762 its data reference.
14763 (partition_alloc, partition_free): Init and release data refs.
14764 (partition_merge_into): Merge data refs.
14765 (build_rdg_partition_for_vertex): Collect data refs for partition.
14766 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14768 (distribute_loop): Remve data refs from vertice data of partition
14771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14773 * tree-loop-distribution.c (params.h): Include header file.
14774 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14775 (datarefs_vec): New global var.
14776 (create_rdg_vertices): Use datarefs_vec directly.
14777 (free_rdg): Don't free data references.
14778 (build_rdg): Update use. Don't free data references.
14779 (distribute_loop): Compute global variable for data references.
14780 Bail out if there are too many data references.
14782 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14784 * tree-loop-distribution.c (loop_nest): New global var.
14785 (build_rdg): Use loop directly, rather than loop nest.
14786 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14788 (distribute_loop): Compute global variable loop nest. Update use.
14790 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14792 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14793 (partition_merge_into): New parameter. Dump reason for fusion.
14794 (distribute_loop): Update use of partition_merge_into.
14796 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14798 * tree-loop-distribution.c (bb_top_order_index): New.
14799 (bb_top_order_index_size, bb_top_order_cmp): New.
14800 (stmts_from_loop): Use topological order.
14801 (pass_loop_distribution::execute): Compute and release topological
14802 order for basic blocks.
14804 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14806 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14811 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14812 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14813 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14814 * internal-fn.def (LOOP_DIST_ALIAS): New.
14815 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14816 (fold_loop_internal_call): ... this.
14817 (vect_loop_dist_alias_call): New function.
14818 (set_uid_loop_bbs): Call fold_loop_internal_call.
14819 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14822 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14825 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14826 Require dead FLAGS_REG at the beginning of a peephole.
14828 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14831 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14832 arguments in the call to __builtin_ia32_sbb_u32.
14833 (_subborrow_u64): Swap _X and _Y arguments in the call to
14834 __builtin_ia32_sbb_u64.
14836 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14839 * tree-vrp.c (compare_assert_loc): Turn into a function template
14840 with stable template parameter. Only test if a->e is NULL,
14841 !a->e == !b->e has been verified already. Use e == NULL or
14842 e != NULL instead of e or ! e tests. If stable is true, don't use
14843 iterative_hash_expr, on the other side allow a or b or both NULL
14844 and sort the NULLs last.
14845 (process_assert_insertions): Sort using compare_assert_loc<false>
14846 instead of compare_assert_loc, later sort using
14847 compare_assert_loc<true> before calling process_assert_insertions_for
14848 in a loop. Use break instead of continue once seen NULL pointer.
14850 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14852 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14853 Cortex-R7 and Cortex-R8 processors.
14855 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14857 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14858 uninitialized while src is not.
14860 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14862 * common/config/arm/arm-common.c: Adjust include path for
14864 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14865 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14866 (arm-cpu-data.h): Likewise.
14867 (arm-cpu-cdata.h): Likewise.
14868 * config/arm/arm-cpu.h: Delete.
14869 * config/arm/arm-cpu-cdata.h: Delete.
14870 * config/arm/arm-cpu-data.h: Delete.
14872 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14874 * config/arm/arm-cpus.in (cortex-a55): New.
14875 (cortex-a75): Likewise.
14876 (cortex-a75.cortex-a55): Likewise.
14877 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14879 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14880 * config/arm/arm-cpu-cdata.h: Regenerate.
14881 * config/arm/arm-cpu-data.h: Regenerate.
14882 * config/arm/arm-cpu.h: Regenerate.
14883 * config/arm/arm-tables.opt: Regenerate.
14884 * config/arm/arm-tune.md: Regenerate.
14886 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14888 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14890 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14892 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14895 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14897 PR tree-optimization/81292
14898 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14899 full_string_p, also call adjust_related_strinfos if the adjustment
14900 is simple, otherwise invalidate related strinfos.
14902 2017-07-04 Martin Liska <mliska@suse.cz>
14905 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14906 newly created variable as DECL_IGNORED_P.
14908 2017-07-04 Martin Liska <mliska@suse.cz>
14911 * ipa-inline.c (inline_small_functions):
14912 Use xstrdup_for_dump.
14914 2017-07-04 Tom de Vries <tom@codesourcery.com>
14916 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14918 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14921 * config/darwin.c (darwin_function_switched_text_sections):
14924 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14926 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14928 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14930 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14932 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14934 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14935 min_profitable_iters, and th as inclusive lower bounds.
14936 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14937 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14938 for min_profitable_iters and min_profitable_estimate.
14939 (vect_transform_loop): Treat th as an inclusive lower bound.
14940 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14942 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14945 * config/darwin.c (darwin_function_switched_text_sections):
14946 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14947 in two pieces, and suppress the use of buf.
14949 2017-07-03 Nathan Sidwell <nathan@acm.org>
14951 * hash-table.h (hash_table_mod1): Fix indentation.
14953 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14955 PR middle-end/81290
14956 * predict.c (force_edge_cold): Be more careful about propagation
14958 * profile-count.h (profile_probability::guessed,
14959 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14961 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14963 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14965 * doc/invoke.texi (rcpc architecture extension): Document it.
14967 2017-07-03 Richard Biener <rguenther@suse.de>
14969 PR tree-optimization/60510
14970 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14971 the scalar reduction PHI and use it.
14972 (vectorizable_reduction): Properly guard the single_defuse_cycle
14973 path for non-SLP reduction chains where we cannot use it.
14974 Rework reduc_def/index and vector type deduction. Rework
14975 vector operand gathering during reduction op code-gen.
14976 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14977 chains dissolve the chain and leave it to non-SLP reduction
14980 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14982 * tree-data-ref.h (dr_alignment): Declare.
14983 * tree-data-ref.c (dr_alignment): New function.
14984 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14985 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14987 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14989 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14991 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14992 and base_misalignment fields.
14993 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14994 * tree-data-ref.c: Include builtins.h.
14995 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14996 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14997 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14998 * tree-vect-data-refs.c: Include tree-cfg.h.
14999 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15000 fields instead of calculating an alignment here.
15001 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15002 innermost_loop_behavior fields.
15004 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15006 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15008 (DR_STEP_ALIGNMENT): New macro.
15009 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15010 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15011 (create_data_ref): Print it.
15012 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15013 to tell whether the step preserves vector (mis)alignment.
15014 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15015 Move the check for an integer step and generalise to all INTEGER_CST.
15016 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15017 Print the outer step alignment.
15019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15021 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15022 with offset_alignment.
15023 (DR_ALIGNED_TO): Delete.
15024 (DR_OFFSET_ALIGNMENT): New macro.
15025 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15026 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15027 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15028 (create_data_ref): Likewise.
15029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15030 (vect_analyze_data_refs): Likewise.
15031 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15032 creating dummy innermost behavior.
15034 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15036 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15037 with a "innermost_loop_behavior *" and refeence tree.
15038 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15039 (create_data_ref): Update call accordingly.
15040 * tree-predcom.c (find_looparound_phi): Likewise.
15042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15044 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15045 fields with dr_wrt_vec_loop.
15046 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15047 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15048 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15049 (vect_dr_behavior): New function.
15050 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15051 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15052 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15053 track whether the step preserves the misalignment.
15054 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15055 Use vect_dr_behavior.
15056 (vect_setup_realignment): Update call accordingly.
15057 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15058 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15059 call to vect_create_addr_base_for_vector_ref.
15060 (vect_create_cond_for_align_checks): Likewise.
15061 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15062 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15063 (vect_recog_mask_conversion_pattern): Likewise.
15064 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15065 (new_stmt_vec_info): Remove redundant zeroing.
15067 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15069 * common/config/arm/arm-common.c (arm_be8_option): New function.
15070 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15071 (ISA_ARMv6): Add isa_bit_be8.
15072 * config/arm/arm.h (arm_be8_option): Add prototype.
15073 (BE8_SPEC_FUNCTION): New define.
15074 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15075 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15076 (mlittle-endian): Similarly.
15077 (mbe8, mbe32): New options.
15078 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15079 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15081 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15083 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15085 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15087 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15088 (cleanup_tree_cfg_bb): Use it.
15089 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15091 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15093 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15096 * loop-doloop.c (add_test): Update profile.
15098 2017-07-03 Martin Liska <mliska@suse.cz>
15101 * sanopt.c (rewrite_usage_of_param): New function.
15102 (sanitize_rewrite_addressable_params): Likewise.
15103 (pass_sanopt::execute): Call rewrite_usage_of_param.
15105 2017-07-03 Richard Biener <rguenther@suse.de>
15107 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15108 back to using VIEW_CONVERT_EXPR.
15110 2017-07-03 Martin Liska <mliska@suse.cz>
15113 * doc/extend.texi: Document when a resolver function is
15114 generated for target_clones.
15116 2017-07-03 Martin Liska <mliska@suse.cz>
15118 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15119 and shadow memory of auto variables which are subject of
15120 use-after-scope sanitization.
15121 (asan_expand_mark_ifn): Add do set only when is_poison.
15123 2016-07-03 Richard Biener <rguenther@suse.de>
15125 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15127 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15128 (vectorizable_reduction): Perform reduction PHI creation when
15129 visiting a reduction PHI and adjust and simplify code generation
15130 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15131 (vect_transform_loop): Visit reduction PHIs.
15132 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15133 defs into the SLP tree.
15134 (vect_build_slp_tree): Reduction defs terminate the recursion.
15135 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15137 (vect_get_vec_defs_for_stmt_copy): Export.
15138 (vect_get_vec_defs): Likewise.
15139 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15141 (vect_get_vec_defs_for_stmt_copy): Declare.
15142 (vect_get_vec_defs): Likewise.
15144 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15146 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15147 parameter with a "loop" parameter and use it instead of the
15148 loop containing DR_STMT. Don't check simple_iv when doing
15149 BB analysis. Describe the two analysis modes in the comment.
15151 2017-07-03 Tom de Vries <tom@codesourcery.com>
15153 PR tree-optimization/69468
15154 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15155 (find_same_succ_bb): Handle ignore_edge_flags.
15157 2017-07-03 Tom de Vries <tom@codesourcery.com>
15159 PR tree-optimization/81192
15160 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15162 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15164 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15166 2017-07-03 Tom de Vries <tom@codesourcery.com>
15168 PR tree-optimization/81192
15169 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15170 BB_SAME_SUCC (bb) == NULL.
15172 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15174 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15177 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15179 * dumpfile.c: Include profile-count.h
15180 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15182 (insert_cond_bb): Update profile.
15183 * tree-cfg.h (insert_cond_bb): Update prototype.
15184 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15185 * tree-dump.c: Do not include tree-cfg.
15187 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15189 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15191 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15193 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15196 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15198 * tree-complex.c (expand_complex_div_wide): update profile.
15200 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15201 Alan Hayward <alan.hayward@arm.com>
15202 David Sherwood <david.sherwood@arm.com>
15204 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15205 (CORETYPES_H): New define.
15206 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15207 (insn-modes-inline.h, s-modes-inline-h): New rules.
15208 (generated_files): Add insn-modes-inline.h.
15209 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15210 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15211 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15212 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15213 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15214 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15215 (build/gencodes.o, build/genconditions.o): Likewise.
15216 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15217 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15218 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15219 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15220 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15221 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15222 * coretypes.h: Include everything up to real.h for generators.
15223 Include insn-modes.h first. Include wide-int-print.h after
15224 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15225 * machmode.h: Don't include insn-modes.h here.
15226 * function-tests.c: Remove includes of signop.h, machmode.h,
15227 double-int.h and wide-int.h.
15229 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15231 * optc-save-gen.awk: Likewise.
15232 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15233 * godump.c: Remove include of wide-int-print.h.
15234 * pretty-print.h: Likewise.
15235 * wide-int-print.cc: Likewise.
15236 * wide-int.cc: Likewise.
15237 * hash-map-tests.c: Remove include of signop.h.
15238 * hash-set-tests.c: Likewise.
15239 * rtl-tests.c: Likewise.
15240 * mkconfig.sh: Remove include of machmode.h.
15241 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15243 (emit_insn_modes_inline_h): ...this new function. Emit the code
15244 into an insn-modes-inline.h header file, adding appropriate
15245 include guards and end comments.
15246 (emit_insn_modes_c_header): Remove include of machmode.h.
15247 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15249 (main): Handle -i flag and call emit_insn_modes_inline_h when
15252 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15254 * tree-ssa-strlen.c (strinfo): Rename the length field to
15255 nonzero_chars. Add a full_string_p field.
15256 (compare_nonzero_chars, zero_length_string_p): New functions.
15257 (get_addr_stridx): Add an offset_out parameter.
15258 Use compare_nonzero_chars.
15259 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15260 (new_strinfo): Update after above changes to strinfo.
15261 (set_endptr_and_length): Set full_string_p.
15262 (get_string_length): Update after above changes to strinfo.
15263 (unshare_strinfo): Update call to new_strinfo.
15264 (maybe_invalidate): Likewise.
15265 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15266 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15267 as a uhwi instead of an shwi. Update after above changes to
15268 strinfo and new_strinfo.
15269 (zero_length_string): Assert that chainsi contains full strings.
15270 Use zero_length_string_p. Update call to new_strinfo.
15271 (adjust_related_strinfos): Update after above changes to strinfo.
15272 Copy full_string_p from origsi.
15273 (adjust_last_stmt): Use zero_length_string_p.
15274 (handle_builtin_strlen): Update after above changes to strinfo and
15275 new_strinfo. Install the lhs as the string length if the previous
15276 entry didn't describe a full string.
15277 (handle_builtin_strchr): Update after above changes to strinfo
15279 (handle_builtin_strcpy): Likewise.
15280 (handle_builtin_strcat): Likewise.
15281 (handle_builtin_malloc): Likewise.
15282 (handle_pointer_plus): Likewise.
15283 (handle_builtin_memcpy): Likewise. Track nonzero characters
15284 that aren't necessarily followed by a nul terminator.
15285 (handle_char_store): Likewise.
15287 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15289 PR tree-optimization/80769
15290 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15291 for malloc and calloc. Document the new invariant that all related
15292 strinfos have delayed lengths or none do.
15293 (verify_related_strinfos): Move earlier in file.
15294 (set_endptr_and_length): New function, split out from...
15295 (get_string_length): ...here. Also set the lengths of related
15297 (zero_length_string): Assert that chainsi has known (rather than
15299 (adjust_related_strinfos): Likewise.
15301 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15303 PR tree-optimization/81136
15304 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15305 assert that two references with the same misalignment have the same
15306 compile-time misalignment if those compile-time misalignments
15309 2017-07-01 Andi Kleen <ak@linux.intel.com>
15311 * print-tree.c (print_node): Print all attributes.
15313 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15315 * cfg.c (scale_bbs_frequencies): New function.
15316 * cfg.h (scale_bbs_frequencies): Declare it.
15317 * cfgloopanal.c (single_likely_exit): Cleanup.
15318 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15320 (scale_loop_profile): Likewise.
15321 (loop_version): Likewise.
15322 (create_empty_loop_on_edge): Update.
15323 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15324 scale_loop_frequencies, scale_loop_profile, loopify,
15325 loop_version): Update prototypes.
15326 * modulo-sched.c (sms_schedule): Update.
15327 * predict.c (unlikely_executed_edge_p): Also check probability.
15328 (probably_never_executed_edge_p): Fix typo.
15329 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15330 * tree-parloops.c (gen_parallel_loop): Update.
15331 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15333 * tree-ssa-loop-split.c (split_loop): Update.
15334 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15335 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15336 (vect_loop_versioning): Update.
15337 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15339 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15341 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15343 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15345 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15346 to keep profile consistent.
15348 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15350 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15351 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15352 * profile-count.h (max_safe_multiplier): Make unsigned.
15353 (profile_count::guessed_zero): New.
15355 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15357 * bb-reorder.c (fix_up_crossing_landing_pad,
15358 fix_crossing_conditional_branches): Use make_single_succ_edge
15359 to keep profile consistent.
15361 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15363 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15366 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15369 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15370 the right scopes, make sure cond_jump isn't preserved between multiple
15371 iterations. Search for fallthru edge whenever there are 3+ edges and
15372 use find_fallthru_edge for it.
15374 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15376 Patch by Alexander Monakov <amonakov@ispras.ru>
15377 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15378 probabilities consistently.
15380 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15382 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15383 emit_cmp_and_jump_insns.
15385 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15388 * tree-inline.c (expand_call_inline): Combine profile statuses.
15390 2017-06-30 Andrew Pinski <apinski@cavium.com>
15392 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15393 fold_stmt returned true.
15395 2017-06-30 Nathan Sidwell <nathan@acm.org>
15397 * ggc.h (empty_string): Delete.
15398 * cfgexpand.c (expand_asm_stmt): Use plain "".
15399 * optabs.c (expand_asm_memory_barrier): Likewise.
15400 * stringpool.c (empty_string): Delete.
15401 (digit_vector, digit_string): Delete.
15402 (ggc_alloc_string): Use plain "", don't optimize single digit
15403 strings. Use ggc_alloc_atomic.
15405 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15407 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15408 comparison set and one other set, use the cost of the non-comparison
15411 2017-06-30 Nathan Sidwell <nathan@acm.org>
15413 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15416 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15418 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15419 loops. Remove now unneeded calls to gimple_switch_set_label() that
15420 just set removed labels to NULL_TREE.
15422 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15424 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15425 (set_range_info): ...here. Only call set_range_info_raw if domain
15427 (set_nonzero_bits): Call set_range_info_raw.
15428 * tree-ssanames.h (set_range_info_raw): New.
15430 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15433 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15434 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15435 of nonimmediate_operand and <store_mask_constraint> instead of m
15436 for the input operand. For V8FI iterator, always split if input
15437 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15438 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15439 <store_mask_predicate> instead of register_operand and
15440 <store_mask_constraint> instead of v for the input operand. Make
15441 sure both operands aren't MEMs for if not <mask_applied>.
15443 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15445 * lto-wrapper.c (copy_file) Close both file descriptors before
15448 2017-06-30 Martin Liska <mliska@suse.cz>
15451 * multiple_target.c (create_dispatcher_calls): Make ifunc
15452 also for function that don't have calls or are not referenced.
15454 2017-06-30 Richard Biener <rguenther@suse.de>
15456 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15457 analyze the first scalar stmt. Move vector type computation
15458 for the BB case here from ...
15459 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15460 live operation processing in the SLP case properly.
15462 2017-06-30 Richard Biener <rguenther@suse.de>
15464 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15466 2017-06-30 Martin Liska <mliska@suse.cz>
15469 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15470 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15472 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15474 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15477 2017-06-30 Richard Biener <rguenther@suse.de>
15479 PR tree-optimization/81249
15480 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15481 condition reduction result to original scalar type.
15483 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15485 * profile-count.h (enum profile_quality): Fix typos and whitespace
15488 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15490 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15491 type for branch probabilities.
15493 2017-06-29 Julian Brown <julian@codesourcery.com>
15494 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15496 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15497 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15498 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15499 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15501 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15503 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15504 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15505 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15506 CC usage from generic code to here.
15507 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15508 CC usage into the target macros.
15510 2017-06-29 Maya Rashish <coypu@sdf.org>
15512 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15515 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15517 * arm/arm-builtins.c: Include profile-count.h
15518 * except.c (sjlj_emit_function_enter): Use
15519 profile_probability::unlikely.
15521 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15523 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15524 and tocrel_offset be pointer args rather than implicitly using
15526 (legitimate_constant_pool_address_p, rs6000_emit_move,
15527 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15528 tocrel_offset and use in toc_relative_expr_p call.
15529 (print_operand, print_operand_address): Use static tocrel_base_oac
15530 and tocrel_offset_oac.
15531 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15534 2017-06-29 Maya Rashish <coypu@sdf.org>
15536 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15538 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15540 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15541 objects, take into account only the alignment of 'op0' and 'mode1' if
15544 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15546 * ccmp.c (ccmp_tree_comparison_p): New function.
15547 (ccmp_candidate_p): Update to use above function.
15548 (get_compare_parts): New function.
15549 (expand_ccmp_next): Update to use new functions.
15550 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15552 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15553 take mode as argument.
15554 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15555 * expr.c (expand_expr_real_1): Pass mode as argument.
15557 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15559 * combine.c (combine_instructions): Print insns to dump_file, together
15562 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15564 * asan.c (asan_emit_stack_protection): Update.
15565 (create_cond_insert_point): Update.
15566 * auto-profile.c (afdo_propagate_circuit): Update.
15567 * basic-block.h (struct edge_def): Turn probability to
15568 profile_probability.
15569 (EDGE_FREQUENCY): Update.
15570 * bb-reorder.c (find_traces_1_round): Update.
15571 (better_edge_p): Update.
15572 (sanitize_hot_paths): Update.
15573 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15574 (make_single_succ_edge): Update.
15575 (check_bb_profile): Update.
15576 (dump_edge_info): Update.
15577 (update_bb_profile_for_threading): Update.
15578 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15579 probabilitycount to 0.
15580 * cfgbuild.c (compute_outgoing_frequencies): Update.
15581 * cfgcleanup.c (try_forward_edges): Update.
15582 (outgoing_edges_match): Update.
15583 (try_crossjump_to_edge): Update.
15584 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15585 (expand_gimple_tailcall): Update.
15586 (construct_init_block): Use make_single_succ_edge.
15587 (construct_exit_block): Use make_single_succ_edge.
15588 * cfghooks.c (verify_flow_info): Update.
15589 (redirect_edge_succ_nodup): Update.
15590 (split_edge): Update.
15591 (account_profile_record): Update.
15592 * cfgloopanal.c (single_likely_exit): Update.
15593 * cfgloopmanip.c (scale_loop_profile): Update.
15594 (set_zero_probability): Remove.
15595 (duplicate_loop_to_header_edge): Update.
15596 * cfgloopmanip.h (loop_version): Update prototype.
15597 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15598 (force_nonfallthru_and_redirect): Update.
15599 (update_br_prob_note): Update.
15600 (rtl_verify_edges): Update.
15601 (purge_dead_edges): Update.
15602 (rtl_lv_add_condition_to_bb): Update.
15603 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15604 * cgraphunit.c (init_lowered_empty_function): Update.
15605 (cgraph_node::expand_thunk): Update.
15606 * cilk-common.c: Include profile-count.h
15607 * dojump.c (inv): Remove.
15608 (jumpifnot): Update.
15609 (jumpifnot_1): Update.
15610 (do_jump_1): Update.
15612 (do_jump_by_parts_greater_rtx): Update.
15613 (do_compare_rtx_and_jump): Update.
15614 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15615 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15616 * dwarf2cfi.c: Include profile-count.h
15617 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15618 (sjlj_emit_dispatch_table): Likewise.
15619 * explow.c: Include profile-count.h
15620 * expmed.c (emit_store_flag_force): Update.
15621 (do_cmp_and_jump): Update.
15622 * expr.c (compare_by_pieces_d::generate): Update.
15623 (compare_by_pieces_d::finish_mode): Update.
15624 (emit_block_move_via_loop): Update.
15625 (store_expr_with_bounds): Update.
15626 (store_constructor): Update.
15627 (expand_expr_real_2): Update.
15628 (expand_expr_real_1): Update.
15629 * expr.h (try_casesi, try_tablejump): Update prototypes.
15630 * gimple-pretty-print.c (dump_probability): Update.
15631 (dump_profile): New.
15632 (dump_gimple_label): Update.
15633 (dump_gimple_bb_header): Update.
15634 * graph.c (draw_cfg_node_succ_edges): Update.
15635 * hsa-gen.c (convert_switch_statements): Update.
15636 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15637 (find_if_case_1): Update.
15638 (find_if_case_2): Update.
15639 * internal-fn.c (expand_arith_overflow_result_store): Update.
15640 (expand_addsub_overflow): Update.
15641 (expand_neg_overflow): Update.
15642 (expand_mul_overflow): Update.
15643 (expand_vector_ubsan_overflow): Update.
15644 * ipa-cp.c (good_cloning_opportunity_p): Update.
15645 * ipa-split.c (split_function): Use make_single_succ_edge.
15646 * ipa-utils.c (ipa_merge_profiles): Update.
15647 * loop-doloop.c (add_test): Update.
15648 (doloop_modify): Update.
15649 * loop-unroll.c (compare_and_jump_seq): Update.
15650 (unroll_loop_runtime_iterations): Update.
15651 * lra-constraints.c (lra_inheritance): Update.
15652 * lto-streamer-in.c (input_cfg): Update.
15653 * lto-streamer-out.c (output_cfg): Update.
15654 * mcf.c (adjust_cfg_counts): Update.
15655 * modulo-sched.c (sms_schedule): Update.
15656 * omp-expand.c (expand_omp_for_init_counts): Update.
15657 (extract_omp_for_update_vars): Update.
15658 (expand_omp_ordered_sink): Update.
15659 (expand_omp_for_ordered_loops): Update.
15660 (expand_omp_for_generic): Update.
15661 (expand_omp_for_static_nochunk): Update.
15662 (expand_omp_for_static_chunk): Update.
15663 (expand_cilk_for): Update.
15664 (expand_omp_simd): Update.
15665 (expand_omp_taskloop_for_outer): Update.
15666 (expand_omp_taskloop_for_inner): Update.
15667 * omp-simd-clone.c (simd_clone_adjust): Update.
15668 * optabs.c (expand_doubleword_shift): Update.
15669 (expand_abs): Update.
15670 (emit_cmp_and_jump_insn_1): Update.
15671 (expand_compare_and_swap_loop): Update.
15672 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15673 * predict.c (predictable_edge_p): Update.
15674 (edge_probability_reliable_p): Update.
15675 (set_even_probabilities): Update.
15676 (combine_predictions_for_insn): Update.
15677 (combine_predictions_for_bb): Update.
15678 (propagate_freq): Update.
15679 (estimate_bb_frequencies): Update.
15680 (force_edge_cold): Update.
15681 * profile-count.c (profile_count::dump): Add missing space into dump.
15682 (profile_count::debug): Add newline.
15683 (profile_count::differs_from_p): Explicitly convert to unsigned.
15684 (profile_count::stream_in): Update.
15685 (profile_probability::dump): New member function.
15686 (profile_probability::debug): New member function.
15687 (profile_probability::differs_from_p): New member function.
15688 (profile_probability::differs_lot_from_p): New member function.
15689 (profile_probability::stream_in): New member function.
15690 (profile_probability::stream_out): New member function.
15691 * profile-count.h (profile_count_quality): Rename to ...
15692 (profile_quality): ... this one.
15693 (profile_probability): New.
15694 (profile_count): Update.
15695 * profile.c (compute_branch_probabilities): Update.
15696 * recog.c (peep2_attempt): Update.
15697 * sched-ebb.c (schedule_ebbs): Update.
15698 * sched-rgn.c (find_single_block_region): Update.
15699 (compute_dom_prob_ps): Update.
15700 (schedule_region): Update.
15701 * sel-sched-ir.c (compute_succs_info): Update.
15702 * stmt.c (struct case_node): Update.
15703 (do_jump_if_equal): Update.
15704 (get_outgoing_edge_probs): Update.
15705 (conditional_probability): Update.
15706 (emit_case_dispatch_table): Update.
15707 (expand_case): Update.
15708 (expand_sjlj_dispatch_table): Update.
15709 (emit_case_nodes): Update.
15710 * targhooks.c: Update.
15711 * tracer.c (better_p): Update.
15712 (find_best_successor): Update.
15713 * trans-mem.c (expand_transaction): Update.
15714 * tree-call-cdce.c: Update.
15715 * tree-cfg.c (gimple_split_edge): Upate.
15716 (move_sese_region_to_fn): Upate.
15717 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15718 * tree-eh.c (lower_resx): Upate.
15719 (cleanup_empty_eh_move_lp): Upate.
15720 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15721 * tree-inline.c (copy_edges_for_bb): Update.
15722 (copy_cfg_body): Update.
15723 * tree-parloops.c (gen_parallel_loop): Update.
15724 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15725 (gimple_gen_time_profiler): Update.
15726 * tree-ssa-dce.c (remove_dead_stmt): Update.
15727 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15728 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15729 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15730 (unloop_loops): Update.
15731 (try_peel_loop): Update.
15732 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15733 * tree-ssa-loop-split.c (connect_loops): Update.
15734 (split_loop): Update.
15735 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15736 (hoist_guard): Update.
15737 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15738 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15739 (value_replacement): Update.
15740 * tree-ssa-reassoc.c (branch_fixup): Update.
15741 * tree-ssa-tail-merge.c (replace_block_by): Update.
15742 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15743 (create_edge_and_update_destination_phis): Update.
15744 (compute_path_counts): Update.
15745 (recompute_probabilities): Update.
15746 (update_joiner_offpath_counts): Update.
15747 (freqs_to_counts_path): Update.
15748 (duplicate_thread_path): Update.
15749 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15750 (struct switch_conv_info): Update.
15751 (gen_inbound_check): Update.
15752 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15753 (vect_do_peeling): Update.
15754 (vect_loop_versioning): Update.
15755 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15756 (optimize_mask_stores): Update.
15757 * ubsan.c (ubsan_expand_null_ifn): Update.
15758 * value-prof.c (gimple_divmod_fixed_value): Update.
15759 (gimple_divmod_fixed_value_transform): Update.
15760 (gimple_mod_pow2): Update.
15761 (gimple_mod_pow2_value_transform): Update.
15762 (gimple_mod_subtract): Update.
15763 (gimple_mod_subtract_transform): Update.
15764 (gimple_ic): Update.
15765 (gimple_stringop_fixed_value): Update.
15766 (gimple_stringops_transform): Update.
15767 * value-prof.h: Update.
15769 2017-06-29 Carl Love <cel@us.ibm.com>
15771 * config/rs6000/rs6000-c.c: Add support for built-in functions
15772 vector signed int vec_signed (vector float);
15773 vector signed long long vec_signed (vector double);
15774 vector signed int vec_signed2 (vector double, vector double);
15775 vector signed int vec_signede (vector double);
15776 vector signed int vec_signedo (vector double);
15777 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15778 instruction generator.
15779 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15780 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15781 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15783 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15784 vunsignede_v2df): Add define_expands.
15785 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15786 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15787 VEC_UNSIGNEDO): Add definitions.
15788 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15789 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15790 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15791 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15792 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15793 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15794 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15795 vec_unsignede, vec_unsignedo): Add builtin defines.
15796 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15798 * doc/extend.texi: Update the built-in documentation file for the
15799 new built-in functions.
15801 2017-06-29 Richard Biener <rguenther@suse.de>
15803 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15804 reduction chains to LOOP_VINFO_REDUCTIONS.
15805 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15806 SLP reductions after processing reduction chains.
15808 2017-06-29 Nathan Sidwell <nathan@acm.org>
15810 * builtins.c (fold_builtin_FUNCTION): Use
15811 lang_hooks.decl_printable_name.
15813 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15815 PR middle-end/81194
15816 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15817 with only one label.
15818 * stmt.c (expand_case): Assert NCASES is greater than one.
15820 2017-06-29 Richard Biener <rguenther@suse.de>
15822 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15824 (group_case_labels): Likewise.
15825 (find_taken_edge): Push sanity checking on val to workers...
15826 (find_taken_edge_cond_expr): ... here
15827 (find_taken_edge_switch_expr): ... and here, handle cases
15828 with just a default label.
15829 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15830 (group_case_labels): Likewise.
15831 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15832 group_case_labels does anything cleanup the CFG again.
15834 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15836 PR tree-optimization/81196
15837 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15838 exit condition comparing two IVs.
15840 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15842 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15843 profile to the dummy entry at the end of the list of architectures.
15844 * config/arm/arm-cpu-cdata.h: Regenerated.
15846 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15847 Michael Collison <michael.collison@arm.com>
15850 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15852 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15853 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15854 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15855 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15856 of shift when the shift amount is masked with constant equal to
15857 the size of the mode.
15858 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15861 2017-06-29 Martin Liska <mliska@suse.cz>
15863 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15865 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15868 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15871 (edge_growth_cache_entry::edge_growth_cache_entry): New
15873 (reset_edge_growth_cache): Update to use constructor.
15875 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15877 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15878 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15879 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15881 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15883 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15884 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15886 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15888 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15889 (*-linux-uclibc*): Add t-uclibc tmake_file.
15890 * config/t-musl: New.
15891 * config/t-uclibc: New.
15893 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15895 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15897 (gen_comm_data): Emit architectural setting of arch_prof.
15898 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15900 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15901 (armv8-m.base, armv8-m.main): Likewise.
15902 * arm-protos.h (arm_build_target): Add profile field.
15903 (arch_option): Likewise.
15904 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15906 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15907 arm_active_target.profile.
15909 2017-06-28 Richard Biener <rguenther@suse.de>
15911 PR middle-end/81227
15912 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15913 TYPE_OVERFLOW_WRAPS.
15914 * match.pd (negate_expr_p): Likewise.
15915 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15916 fold_build2, not fold_binary.
15918 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15920 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15921 Convert memory address to Pmode.
15922 (aarch64_print_operand): Assert MEM operands are always Pmode.
15924 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15927 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15928 Remove redundant if.
15929 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15930 * config/arm/aarch-common-protos.h
15931 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15932 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15934 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15937 * multiple_target.c (create_dispatcher_calls): Set the default
15938 clone to be static, not public.
15940 2017-06-28 Richard Biener <rguenther@suse.de>
15942 * tree-vect-loop.c (vectorizable_reduction): Move special
15943 cond reduction IV var creation ...
15944 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15945 parameter. Use STMT_VINFO_VECTYPE.
15946 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15949 2017-06-28 Martin Liska <mliska@suse.cz>
15952 * ipa-visibility.c (non_local_p): Handle visibility.
15954 2017-06-28 Martin Liska <mliska@suse.cz>
15957 * common.opt: Add IntegerRange to various options.
15958 * opt-functions.awk (integer_range_info): New function.
15959 * optc-gen.awk: Add integer_range_info to cl_options struct.
15960 * opts-common.c (decode_cmdline_option): Handle
15961 CL_ERR_INT_RANGE_ARG.
15962 (cmdline_handle_error): Likewise.
15963 * opts.c (print_filtered_help): Show valid interval in
15964 when --help is provided.
15965 * opts.h (struct cl_option): Add range_min and range_max fields.
15966 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15968 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15970 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15971 (x * C EQ/NE y * C): New transformation.
15973 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15975 * genmultilib (combination_space): Accept '+' in option names.
15977 2017-06-28 Martin Liska <mliska@suse.cz>
15980 * asan.c (instrument_derefs): Bail out inner references
15981 that are hard register variables.
15983 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15986 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15987 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15989 2017-06-28 Richard Biener <rguenther@suse.de>
15991 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15992 (vect_get_slp_defs): Adjust.
15993 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15995 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15997 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15998 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15999 (vectorizable_reduction): Adjust.
16000 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16002 (vect_get_slp_defs): Likewise.
16003 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16004 (vectorizable_bswap): Adjust.
16005 (vectorizable_call): Likewise.
16006 (vectorizable_conversion): Likewise.
16007 (vectorizable_assignment): Likewise.
16008 (vectorizable_shift): Likewise.
16009 (vectorizable_operation): Likewise.
16010 (vectorizable_store): Likewise.
16011 (vectorizable_condition): Likewise.
16012 (vectorizable_comparison): Likewise.
16014 2017-06-28 Michael Collison <michael.collison@arm.com>
16017 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16019 (arm_gen_movmemqi): Removed unused variable 'i'.
16020 Convert 'for' loop into 'while' loop.
16021 (arm_expand_prologue): Remove last unnecessary set of insn.
16022 (thumb_pop): Remove unused variable 'pushed_words'.
16023 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16025 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16027 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16028 * config/s390/s390-protos.h: Add prototype of
16029 s390_rel_address_ok_p.
16030 * config/s390/s390.c (s390_got_symbol): New function.
16031 (s390_rel_address_ok_p): New function.
16032 (legitimize_pic_address): Use s390_rel_address_ok_p.
16033 (s390_load_got): Use s390_got_symbol.
16034 (s390_option_override): Issue error if
16035 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16036 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16038 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16040 2017-06-27 Andrew Pinski <apinski@cavium.com>
16042 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16043 (X * copysign (1.0, X)): New pattern.
16044 (X * copysign (1.0, -X)): New pattern.
16045 (copysign (-1.0, CST)): New pattern.
16047 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16049 * genmultilib (combination_space): Remove variable.
16050 Validate reuse rules against regular expression for any sequence
16051 of multilib options in any order.
16053 2017-06-27 Michael Collison <michael.collison@arm.com>
16055 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16056 call aarch64_split_simd_combine.
16057 * (aarch64_combine_internal<mode>): Delete pattern.
16058 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16059 Allow register and subreg operands.
16061 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16063 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16064 specific need, just fallback on defaults.
16065 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16067 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16068 Olivier Hainque <hainque@adacore.com>
16070 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16072 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16073 targets. Pick a default if no particular attempt applied.
16074 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16077 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16079 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16080 (x86_64-wrs-vxworks7): Likewise.
16082 2017-06-27 Marek Polacek <polacek@redhat.com>
16085 * ubsan.c (instrument_null): Check get_base_address's result for null.
16087 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16089 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16091 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16093 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16094 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16095 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16096 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16097 New function types.
16098 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16099 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16100 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16101 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16102 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16103 BUILT_IN_FEUPDATEENV): New builtins.
16104 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16105 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16106 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16107 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16109 (builtin_structptr_types): Adjust size.
16110 * tree.c (builtin_structptr_types): Add four entries.
16112 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16113 Olivier Hainque <hainque@adacore.com>
16115 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16116 (TLS_SYM): New local macro, forcing reference to __tls__ on
16117 link command lines for VxWorks 7 RTPs, triggering initialization
16119 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16120 OS features TLS support, true for RTPs on VxWorks 7.
16121 * config/vxworks.c (vxworks_override_options): Setup emutls
16124 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16126 * predict.c (test_prediction_value_range): Use -1U instead of -1
16127 to avoid narrowing conversion warning.
16128 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16129 to avoid narrowing conversion warning.
16130 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16132 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16134 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16136 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16137 64bit configurations.
16138 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16139 (SIZE_TYPE): Likewise.
16140 * config/vxworks.c (vxworks_emutls_var_fields): Use
16141 long_unsigned_type_node instead of unsigned_type_node as the offset
16142 field type, which is "pointer" mode in emutls.c.
16144 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16147 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16149 PR middle-end/81207
16150 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16151 gimple_vuse copying separately from gimple_vdef copying.
16153 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16155 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16158 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16159 Olivier Hainque <hainque@adacore.com>
16161 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16162 for all vxworks7 targets.
16163 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16164 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16165 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16166 variations for VX6/VX7 and 32/64bits later on in ...
16167 (VXWORKS_LIB_SPEC): Leverage new macros.
16168 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16169 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16171 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16173 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16174 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16176 2017-06-26 Carl Love <cel@us.ibm.com>
16178 * config/rs6000/rs6000-c.c: Add support for built-in functions
16179 vector bool char vec_reve (vector bool char);
16180 vector signed char vec_reve (vector signed char);
16181 vector unsigned char vec_reve (vector unsigned char);
16182 vector bool int vec_reve (vector bool int);
16183 vector signed int vec_reve (vector signed int);
16184 vector unsigned int vec_reve (vector unsigned int);
16185 vector bool long long vec_reve (vector bool long long);
16186 vector signed long long vec_reve (vector signed long long);
16187 vector unsigned long long vec_reve (vector unsigned long long);
16188 vector bool short vec_reve (vector bool short);
16189 vector signed short vec_reve (vector signed short);
16190 vector double vec_reve (vector double);
16191 vector float vec_reve (vector float);
16192 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16193 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16194 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16195 (altivec_vreve): New pattern.
16196 * config/rs6000/altivec.h (vec_reve): New define.
16197 * doc/extend.texi (vec_rev): Update the built-in documentation file
16198 for the new built-in functions.
16200 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16202 PR tree-optimization/71815
16203 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16205 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16207 (slsr_process_phi): Likewise.
16208 (replace_uncond_cands_and_profitable_phis): Don't replace a
16209 multiply candidate with a stride of 1 (copy or cast).
16210 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16212 (lowest_cost_path): Likewise.
16213 (total_savings): Likewise.
16215 2017-06-26 Richard Biener <rguenther@suse.de>
16218 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16219 Use def_builtin_pure for all gather builtins.
16221 2017-06-26 Richard Biener <rguenther@suse.de>
16223 PR tree-optimization/81203
16224 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16225 non-dominating BBs.
16227 2017-06-26 Marek Polacek <polacek@redhat.com>
16230 * doc/invoke.texi: Document -Wmultistatement-macros.
16232 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16234 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16235 arm_neon_ok_no_float_abi effective target.
16237 2017-06-26 Richard Biener <rguenther@suse.de>
16239 PR tree-optimization/80928
16240 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16241 (copy_bbs): Set BB_DUPLICATED flag early.
16242 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16244 (execute_on_shrinking_pred): Likewise.
16245 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16246 BB_DUPLICATED blocks.
16247 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16248 iterate over all PHIs considering removal of *gsi.
16250 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16252 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16255 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16257 * config/rs6000/rs6000-string.c: (expand_block_clear,
16258 do_load_for_compare, select_block_compare_mode,
16259 compute_current_alignment, expand_block_compare,
16260 expand_strncmp_align_check, expand_strn_compare,
16261 expand_block_move, rs6000_output_load_multiple)
16262 Move functions related to string/block move/compare
16263 to a separate file.
16264 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16265 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16266 for this function which is now used in two files.
16267 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16268 * config.gcc: Add rs6000-string.o to extra_objs for
16269 targets powerpc*-*-* and rs6000*-*-*.
16271 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16274 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16275 32-bit, since indexed is not valid for DImode.
16276 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16277 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16278 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16279 (define_peephole2 for Altivec d-form store): Likewise.
16282 * multiple_target.c (create_dispatcher_calls): Only create the
16283 dispatcher call if the function is the default clone of a
16284 versioned function.
16286 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16288 PR middle-end/80902
16289 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16290 a call, force the call to not be a tail call.
16292 2017-06-23 Jeff Law <law@redhat.com>
16294 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16297 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16299 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16300 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16301 handling for early expansion of vector shifts (sl,sr,sra,rl).
16302 (builtin_function_type): Add vector shift right instructions
16303 to the unsigned argument list.
16305 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16307 rtl-optimizatoin/79286
16308 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16309 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16310 trap. PIC register plus a const unspec without offset can never trap.
16312 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16314 * tree.h (builtin_structptr_type): New type.
16315 (builtin_structptr_types): Declare new array.
16316 * tree.c (builtin_structptr_types): New array.
16317 (free_lang_data, build_common_tree_nodes): Use it.
16319 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16322 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16325 2017-06-22 Matt Turner <mattst88@gmail.com>
16327 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16328 Lake models to skylake case. Assume skylake for unknown
16329 models with clflushopt.
16331 2017-06-22 Jeff Law <law@redhat.com>
16333 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16334 frame sizes that do not satisfy aarch64_uimm12_shift.
16336 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16338 * profile-count.h (apply_probability,
16339 apply_scale, probability_in): Fix checks for zero.
16341 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16343 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16344 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16346 2016-06-22 Richard Biener <rguenther@suse.de>
16348 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16349 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16350 REDUC_MAX_EXPR support.
16351 (vectorizable_reduction): Likewise.
16352 (vect_create_epilog_for_reduction): Likewise.
16354 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16356 * match.pd (A / (1 << B) -> A >> B): New.
16357 * generic-match-head.c: Include optabs-tree.h.
16358 * gimple-match-head.c: Likewise.
16359 * optabs-tree.h (target_supports_op_p): New.
16360 * optabs-tree.c (target_supports_op_p): New.
16362 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16364 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16365 $gcc_cv_ld --help output.
16366 (gcc_cv_ld_demangle): Likewise.
16367 (gcc_cv_ld_eh_frame_hdr): Likewise.
16368 (gcc_cv_ld_pie): Likewise.
16369 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16370 (gcc_cv_ld_buildid): Likewise.
16371 (gcc_cv_ld_sysroot): Likewise.
16372 (ld_bndplt_support): Likewise.
16373 (ld_pushpopstate_support): Likewise.
16374 * configure: Regenerate.
16375 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16377 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16380 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16381 operands indexes to avoid gap between operands and match_dups.
16383 2017-06-21 Andrew Pinski <apinski@cavium.com>
16385 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16386 Increment Arith_shift and Arith_shift_reg by 1.
16387 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16389 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16390 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16391 (aarch64_strip_extend): Add new argument and test for it.
16392 (aarch64_cheap_mult_shift_p): New function.
16393 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16394 add a cost if it is true.
16395 Update calls to aarch64_strip_extend.
16396 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16398 2017-06-21 Andrew Pinski <apinski@cavium.com>
16400 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16402 (thunderxt88): Likewise.
16403 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16404 (thunderx_prefetch_tune): New variable.
16405 (thunderx2t99_prefetch_tune): Update for the correct values.
16406 (thunderxt88_tunings): New variable.
16407 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16408 generic_prefetch_tune.
16409 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16411 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16413 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16414 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16415 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16416 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16417 (aarch64_atomic_cas<mode>, GPI): Likewise.
16419 2017-06-21 Martin Liska <mliska@suse.cz>
16421 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16422 statements on cold and hot labels.
16423 * predict.c (tree_estimate_probability_bb): Remove the
16424 prediction from this place.
16426 2017-06-21 Martin Liska <mliska@suse.cz>
16428 PR tree-optimization/79489
16429 * gimplify.c (maybe_add_early_return_predict_stmt): New
16431 (gimplify_return_expr): Call the function.
16432 * predict.c (tree_estimate_probability_bb): Remove handling
16434 * predict.def: Update comment about early return predictor.
16435 * gimple-predict.h (is_gimple_predict): New function.
16436 * predict.def: Change default value of early return to 66.
16437 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16439 * passes.def: Put pass_strip_predict_hints to the beginning of
16442 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16444 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16445 FUNCTION_DECL declarations.
16446 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16448 (dwaf2out_decl): Likewise.
16449 * godump.c (go_early_global_decl): Skip call to the real debug hook
16450 for FUNCTION_DECL declarations.
16451 * passes.c (rest_of_decl_compilation): Skip call to the
16452 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16453 -fdump-go-spec is passed.
16455 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16457 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16458 Reorder for compactness.
16459 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16460 (def_builtin_pure, def_builtin_pure2): New functions.
16461 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16463 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16465 * match.pd (nop_convert): New predicate.
16466 ((A +- CST1) +- CST2): Allow some NOP conversions.
16468 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16471 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16472 with ctors/dtors if GOVD_SHARED is set.
16474 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16476 * config/aarch64/aarch64.md (movti_aarch64):
16477 Emit mov rather than orr.
16478 (movtf_aarch64): Likewise.
16479 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16480 Emit mov rather than orr.
16482 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16484 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16485 Swap alternatives, make integer dup more expensive.
16487 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16489 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16490 Return true for non-tls symbols.
16492 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16494 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16495 (cortex-a75): Likewise.
16496 (cortex-a75.cortex-a55): Likewise.
16497 * config/aarch64/aarch64-tune.md: Regenerate.
16498 * doc/invoke.texi (-mtune): Document new values for -mtune.
16500 2017-06-21 Tom de Vries <tom@codesourcery.com>
16502 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16503 stack_size feature.
16504 (Effective-Target Keywords, Other attributes): Suggest using
16505 dg-add-options stack_size feature to get stack limit in stack_size
16506 effective target documentation.
16508 2017-06-21 Julian Brown <julian@codesourcery.com>
16509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16511 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16512 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16513 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16515 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16516 attribute type list for neon_multiply.
16517 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16518 attribute type list for neon_multiply.
16519 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16520 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16521 attribute type list for neon_multiply.
16522 * config/arm/types.md (crypto_pmull): Add.
16523 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16524 attribute type list.
16526 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16528 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16531 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16533 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16534 to make sure not to dereference a NULL cost_classes_ptr pointer.
16536 2017-06-20 Carl Love <cel@us.ibm.com>
16538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16539 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16540 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16541 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16542 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16543 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16544 VMULOSW): New enum "unspec" values.
16545 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16546 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16547 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16548 altivec_vmulosw): New patterns.
16549 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16550 VMULOSW): Add definitions.
16552 2017-06-20 Julia Koval <julia.koval@intel.com>
16554 * config/i386/i386.c: Fix rounding expand for new pattern.
16555 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16557 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16559 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16560 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16562 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16564 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16567 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16569 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16570 architecture, then by implementer ID.
16571 * config/aarch64/aarch64-tune.md: Regenerate.
16573 2017-06-20 Richard Biener <rguenther@suse.de>
16575 PR middle-end/81097
16576 * fold-const.c (split_tree): Fold to type before negating.
16578 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16580 * diagnostic-show-locus.c
16581 (selftest::test_fixit_deletion_affecting_newline): New function.
16582 (selftest::diagnostic_show_locus_c_tests): Call it.
16584 2017-06-20 Andreas Schwab <schwab@suse.de>
16587 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16590 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16592 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16593 __ARM_FEATURE_COPROC according to support.
16595 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16597 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16598 Rewritten to avoid overflow for > 32-bit pointers.
16601 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16602 by removing enum keyword.
16603 (ubsan_type_descriptor): Likewise. Formatting fix.
16606 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16607 splitter): Require TARGET_SSE2 in the condition.
16609 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16612 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16613 for doing vector set of SFmode on ISA 3.0.
16614 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16615 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16617 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16618 SFmode value into a V4SF variable that was extracted from another
16619 V4SF variable without converting the element to double precision
16620 and back to single precision vector format.
16621 (vsx_insert_extract_v4sf_p9_2): Likewise.
16623 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16625 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16626 in UWHI to avoid undefined overflow.
16629 * ubsan.h (enum ubsan_encode_value_phase): New.
16630 (ubsan_encode_value): Change second argument to
16631 enum ubsan_encode_value_phase with default value of
16632 UBSAN_ENCODE_VALUE_GENERIC.
16633 * ubsan.c (ubsan_encode_value): Change second argument to
16634 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16635 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16636 create_tmp_var_raw instead of create_tmp_var and use a
16638 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16639 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16640 ubsan_encode_value callers.
16643 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16644 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16645 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16647 2017-06-19 Richard Biener <rguenther@suse.de>
16649 PR middle-end/81118
16650 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16651 estimates if we changed anything.
16653 2017-06-19 Richard Biener <rguenther@suse.de>
16655 PR tree-optimization/80887
16656 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16657 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16658 simplified lookups, then reset mprts_hook.
16659 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16661 (try_to_simplify): Likewise.
16663 2017-06-19 Martin Liska <mliska@suse.cz>
16666 * gimplify.c (gimplify_switch_expr):
16667 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16669 2017-06-19 Martin Liska <mliska@suse.cz>
16671 * doc/install.texi: Document that PGO runs in 4 stages.
16673 2017-06-19 Martin Liska <mliska@suse.cz>
16676 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16677 to dispatcher function name.
16678 * multiple_target.c (replace_function_decl): New function.
16679 (create_dispatcher_calls): Redirect both edges and references.
16681 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16683 * profile-count.c (profile_count::dump): Dump quality.
16684 (profile_count::differs_from_p): Update for unsigned val.
16685 * profile-count.h (profile_count_quality): New enum.
16686 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16688 2017-06-19 Richard Biener <rguenther@suse.de>
16690 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16691 struct function as arg.
16692 (estimate_numbers_of_iterations): Export overload with loop arg.
16693 (free_numbers_of_iterations_estimates_loop): Use an overload of
16694 free_numbers_of_iterations_estimates instead.
16695 * tree-cfg.c (remove_bb): Adjust.
16696 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16697 * tree-parloops.c (gen_parallel_loop): Likewise.
16698 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16700 (tree_unroll_loops_completely): Likewise.
16701 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16702 Use an overload instead and export.
16703 (estimated_loop_iterations): Adjust.
16704 (max_loop_iterations): Likewise.
16705 (likely_max_loop_iterations): Likewise.
16706 (estimate_numbers_of_iterations): Take struct function as arg
16708 (loop_exits_before_overflow): Adjust.
16709 (free_numbers_of_iterations_estimates_loop): Use an overload.
16710 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16711 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16713 2017-06-19 Richard Biener <rguenther@suse.de>
16716 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16717 RANGE_EXPR conservatively.
16719 2017-06-16 Carl Love <cel@us.ibm.com>
16721 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16722 definitions for vec_float, vec_float2, vec_floato,
16723 vec_floate built-ins.
16724 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16725 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16727 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16728 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16729 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16730 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16731 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16732 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16733 vec_floato): Add builtin defines.
16734 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16735 Update the built-in documentation file for the new built-in
16738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16740 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16741 (mthumb): Mark as the negative of -marm.
16743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16745 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16747 (ARM Options, -mtune): Document that this accepts the same
16748 extension options as -mcpu.
16749 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16753 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16754 permitted extensions.
16756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16758 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16759 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16760 (armv8-m.main): Add option +nodsp.
16761 * config/arm/arm-cpu-cdata.h: Regenerated.
16763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16765 * config/arm/t-fuchsia: New file.
16766 * config.gcc (arm*-*-fuchsia*): Use it.
16768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16770 * config/arm/t-symbian: Rewrite for new option infrastructure.
16772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16774 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16775 (MULTILIB_REQUIRED): Likewise.
16777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16779 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16780 (MULTILIB_RESUE): Likewise.
16781 (MULTILIB_MATCHES): Likewise.
16782 (MULTLIB_REQUIRED): Likewise.
16784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16786 * config/arm/t-rtems: Rewrite for new option framework.
16788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16790 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16791 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16792 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16793 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16794 * config/arm/t-multilib: ... here.
16795 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16796 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16797 armv7-a and armv8*-a when A-profile libraries have not been built.
16798 * config/arm/t-rmprofile: Rewrite.
16800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16802 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16803 with a backslash. Remove the backslash after substituting unescaped
16805 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16809 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16810 the driver to pass through the default setting of -mfloat-abi.
16811 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16813 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16814 (all_feat_combs): New rule.
16815 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16817 * config/arm/t-aprofile: Rewrite.
16819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16821 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16822 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16823 fpu is not specified by the user/command-line.
16824 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16825 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16826 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16827 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16828 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16829 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16833 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16834 * config/arm/t-arm-elf: Rewritten.
16836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16838 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16839 have some floating-point instructions.
16840 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16841 (TARGET_MAYBE_HARD_FLOAT): New macro.
16842 * config/arm/arm-builtins.c (arm_init_builtins): Use
16843 TARGET_MAYBE_HARD_FLOAT.
16844 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16848 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16849 (configargs.h): Include it.
16850 (arm_print_hint_for_fpu_option): New function.
16851 (arm_parse_fpu_option): New function.
16852 (candidate_extension): New class.
16853 (arm_canon_for_multilib): New function.
16854 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16855 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16856 (ARCH_CANONICAL_SPECS): New macro.
16857 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16861 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16862 are set after handling multilib fragments. Set target_cpu_default2
16865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16867 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16869 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16870 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16872 (check_arch): Likewise.
16873 * config/arm/arm.c (arm_configure_build_target): Handle
16874 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16875 options in the default.
16877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16879 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16880 when an option is an alias of another.
16881 * config/arm/parsecpu.awk (optalias): New parser token.
16882 (gen_comm_data): Mark non-alias options as such. Emit entries
16883 for extension aliases.
16884 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16885 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16886 (armv6kz, armv6zk, armv6t2): Likewise.
16887 (armv7): Make vfpv3-d16 an alias.
16888 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16890 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16891 Sort in canonical order.
16892 (armv8-a): Sort in canonical order.
16893 (armv8.1-a, armv8.2-a): Likewise.
16894 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16896 (cortex-a9): Sort in canonical order.
16897 * config/arm/arm.c (selftests.h): Include it.
16898 (arm_test_cpu_arch_data): New function.
16899 (arm_run_self_tests): New function.
16900 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16901 (targetm): Move declaration to the end of the file.
16902 * arm-cpu-cdata.h: Regenerated.
16904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16906 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16907 call to target_mode_check describing the type of option passed.
16908 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16909 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16910 arm_parse_cpu_option_name to match parameters against list of
16912 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16913 arm_arch_core_flags data structure.
16914 * config/arm/arm-cpu_cdata.h: Regenerated.
16916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16918 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16920 (arm_print_hint_for_cpu_option): Likewise.
16921 (arm_print_hint_for_arch_option): Likewise.
16922 (arm_parse_cpu_option_name): Likewise.
16923 (arm_parse_arch_option_name): Likewise.
16924 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16925 of entries in the all_fpus list.
16926 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16927 (arm_parse_cpu_option_name): Declare.
16928 (arm_parse_arch_option_name): Declare.
16929 (arm_parse_option_features): Declare.
16930 (arm_intialize_isa): Declare.
16931 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16933 (gen_comm_data): ... here. Make definitions non-static.
16934 * config/arm/arm-cpu-data.h: Regenerated.
16935 * config/arm/arm-cpu-cdata.h: Regenerated.
16937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16939 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16940 (cpu_arch_extension): New structure.
16941 (cpu_arch_option, arch_option, cpu_option): New structures.
16942 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16943 architecture types.
16944 (gen_data): Generate new format data tables.
16945 * config/arm/arm.c (cpu_tune): New structure.
16946 (cpu_option, processors): Delete.
16947 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16948 (arm_print_hint_for_cpu_option): ... this and ...
16949 (arm_print_hint_for_arch_option): ... this.
16950 (arm_parse_arch_cpu_name): Delete. Replace with ...
16951 (arm_parse_cpu_option_name): ... this and ...
16952 (arm_parse_arch_option_name): ... this.
16953 (arm_unrecognized_feature): Change type of target parameter to
16955 (arm_parse_arch_cpu_features): Delete. Replace with ...
16956 (arm_parse_option_features): ... this.
16957 (arm_configure_build_target): Rework to use new configuration data
16959 (arm_print_tune_info): Rework for new configuration data tables.
16960 * config/arm/arm-cpu-data.h: Regenerated.
16961 * config/arm/arm-cpu.h: Regenerated.
16963 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16965 * Makefile.in (OBJS): Move sbitmap.o from here ...
16966 (OBJS-libcommon): ... to here.
16968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16970 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16971 (ISA_ALL_CRYPTO): New macro.
16972 (ISA_ALL_SIMD): New macro
16973 (ISA_ALL_FP): New macro.
16974 * config/arm/arm.c (fpu_bitlist): Update initializer.
16975 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16977 (arm9e): Add fpu. Add option for nofp
16978 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16979 (arm926ej-s, arm1026ej-s): Likewise.
16980 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16981 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16982 neon-fp16, neon-vfpv4, nofp and nosimd.
16983 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16984 (cortex-a8): Add fpu. Add option for nofp.
16985 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16986 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16987 (cortex-r4f): Add fpu.
16988 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16989 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16991 (cortex-r8): Likewise.
16992 (cortex-m4): Add fpu. Add option for nofp.
16993 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16994 (cortex-a17.cortex-a7): Likewise.
16995 (cortex-a32): Add fpu. Add options for crypto and nofp.
16996 (cortex-a35, cortex-a53): Likewise.
16997 (cortex-a57): Add fpu. Add option for crypto.
16998 (cortex-a72, cortex-a73): Likewise.
16999 (exynos-m1): Likewise.
17000 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17001 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17002 (cortex-m33): Add fpu. Add option for nofp.
17003 * config/arm/arm-cpu-cdata.h: Regenerated
17004 * config/arm/arm-cpu-data.h: Regenerated.
17006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17008 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17009 (armv5te, armv5tej): Likewise.
17010 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17011 (armv7): Add options fp and vfpv3-d16.
17012 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17013 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17015 (armv7ve): Likewise.
17016 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17017 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17018 (armv8-a): Add nocrypto option.
17019 (armv8.1-a, armv8.2-a): Likewise.
17020 (armv8-m.main): add options fp, fp.dp and nofp.
17022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17024 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17026 (armv8-a+crc): Delete.
17027 (armv8.1-a): Add options simd, crypto and nofp.
17028 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17029 (armv8.2-a+fp16): Delete.
17030 (armv8-m.main): Add option dsp.
17031 (armv8-m.main+dsp): Delete.
17032 (cortex-a8): Add fpu. Add nofp option.
17033 (cortex-a9): Add fpu. Add nofp and nosimd options.
17034 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17035 link to main cpu and architecture data structures.
17036 (gen_comm_data): Only put isa attributes from the main architecture
17038 (option): New statement for architecture and CPU entries.
17039 * arm.c (struct cpu_option): New structure.
17040 (struct processors): Add entry for options.
17041 (arm_unrecognized_feature): New function.
17042 (arm_parse_arch_cpu_name): Ignore any characters after the first
17044 (arm_parse_arch_cpu_feature): New function.
17045 (arm_configure_build_target): Separate out any CPU and architecture
17046 features and parse separately. Don't error out if -mfpu=auto is
17047 used with only an architecture string.
17048 (arm_print_asm_arch_directives): New function.
17049 (arm_file_start): Call it.
17050 * config/arm/arm-cpu-cdata.h: Regenerated.
17051 * config/arm/arm-cpu-data.h: Likewise.
17052 * config/arm/arm-tables.opt: Likewise.
17054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17056 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17057 assembler when it is not -mfpu=auto.
17059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17061 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17062 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17063 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17064 (ASM_CPU_SPEC): Rewrite.
17065 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17066 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17067 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17068 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17069 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17070 copied string is NUL-terminated. Also strip any characters prefixed
17072 (arm_rewrite_selected_arch): New function.
17073 (arm_rewrite_march): New function.
17075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17077 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17078 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17079 (march, mcpu, mtune): Convert to string-based options.
17080 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17081 (arm_parse_arch_cpu_name): New function.
17082 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17083 identify selected architecture or CPU.
17084 (arm_option_save): New function.
17085 (TARGET_OPTION_SAVE): Redefine.
17086 (arm_option_restore): Restore string options.
17087 (arm_option_print): Print string options.
17089 2017-06-16 Martin Sebor <msebor@redhat.com>
17091 PR tree-optimization/80933
17092 PR tree-optimization/80934
17093 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17094 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17095 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17096 (gimple_fold_builtin): Call them.
17098 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17100 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17101 as unlikely; update profile.
17103 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17105 * predict.c (force_edge_cold): Handle declaring edges impossible
17108 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17110 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17112 (try_unroll_loop_completely): Fix reporting.
17114 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17116 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17118 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17121 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17122 if given a non-constant argument for an intrinsic which requires a
17125 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17127 * profile.c (compare_freqs): New function.
17128 (branch_prob): Sort edge list.
17129 (find_spanning_tree): Assume that the list is priority sorted.
17131 2017-06-16 Richard Biener <rguenther@suse.de>
17133 PR tree-optimization/81090
17134 * passes.def (pass_record_bounds): Remove.
17135 * tree-pass.h (make_pass_record_bounds): Likewise.
17136 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17137 make_pass_record_bounds): Likewise.
17138 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17139 not free niter estimates at the beginning but at the end.
17140 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17142 2017-06-16 Richard Biener <rguenther@suse.de>
17144 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17145 initializer to workaround ICE in host GCC 4.8.
17147 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17149 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17151 (clone_inlined_nodes): Update.
17153 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17155 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17156 prefetch settings, and enable prefetching by default at -O3.
17158 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17160 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17161 Set flag_prefetch_loop_arrays according to tuning data.
17163 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17165 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17166 New tune structure.
17167 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17168 [Unrelated to main purpose of the patch] Place the pointer field last
17169 to enable type checking errors when tune structure are wrongly merged.
17170 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17171 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17172 (thunderx2t99_prefetch_tune): New tune constants.
17173 (tune_params *_tunings): Update all tunings (no functional change).
17174 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17175 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17176 from tunings structures.
17178 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17181 * ubsan.c (instrument_null): Add T argument, use it instead
17182 of computing it based on IS_LHS.
17183 (instrument_object_size): Likewise.
17184 (pass_ubsan::execute): Adjust instrument_null and
17185 instrument_object_size callers to pass gimple_get_lhs or
17186 gimple_assign_rhs1 result to it. Use instrument_null instead of
17187 calling get_base_address and instrument_mem_ref. Handle
17188 aggregate call arguments for object-size sanitization.
17190 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17192 PR tree-optimization/81089
17193 * tree-vrp.c (is_masked_range_test): Validate operands of
17196 2017-06-15 Martin Sebor <msebor@redhat.com>
17199 * dumpfile.c (dump_register): Avoid calling memset to initialize
17200 a class with a default ctor.
17201 * gcc.c (struct compiler): Remove const qualification.
17202 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17203 * hash-table.h: Ditto.
17204 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17206 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17207 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17209 * params.h (struct param_info): Make struct members non-const.
17210 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17211 with default initialization.
17212 * vec.h (vec_copy_construct, vec_default_construct): New helper
17214 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17215 with vec_copy_construct.
17216 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17217 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17218 * doc/invoke.texi (-Wclass-memaccess): Document.
17220 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17222 * emit-rtl.h (is_leaf): Update comment about local
17223 register allocator.
17225 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17228 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17229 for a variable to have a section before checking if the section has a
17231 Set section to.persistent if persistent attribute is set.
17232 Warn if .persistent attribute is used on an automatic variable.
17234 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17236 PR rtl-optimization/80474
17237 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17239 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17241 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17244 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17246 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17247 (casesi): Emit bounds checking as RTL.
17248 (casesi_internal_mips16_<mode>): Remove bounds checking.
17250 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17252 * config/xtensa/xtensa.c (xtensa_option_override): Append
17253 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17254 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17255 xtensa_doloop_hooks): Define unconditionally.
17256 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17258 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17259 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17260 for it in xtensa_option_override.
17261 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17262 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17264 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17266 * doc/cppopts.texi: Document '-' special value to -MF.
17268 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17270 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17271 (cortex_a53_fconst): Likewise.
17272 (cortex_a53_fpmul): Likewise.
17273 (cortex_a53_f_load_64): Likewise.
17274 (cortex_a53_f_load_many): Likewise.
17275 (cortex_a53_advsimd_alu): Likewise.
17276 (cortex_a53_advsimd_alu_q): Likewise.
17277 (cortex_a53_advsimd_mul): Likewise.
17278 (cortex_a53_advsimd_mul_q): Likewise.
17279 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17280 Add missing fmul, r2f_cvt and fconst cases.
17282 2017-06-14 Richard Biener <rguenther@suse.de>
17284 PR middle-end/81088
17285 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17287 (fold_binary_loc): When associating do not treat pre-existing
17288 TREE_OVERFLOW on literal constants as a reason to allow
17289 TREE_OVERFLOW on associated literal constants.
17291 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17293 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17294 (MASK_FEATURES): New macro.
17295 * config/sparc/sparc.c (sparc_option_override): Remove the special
17296 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17298 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17300 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17301 a division of 0 if non-call exceptions are enabled.
17303 2017-06-14 Andrew Pinski <apinski@cavium.com>
17304 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17307 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17308 Improve vector initialization code gen for only variable case.
17310 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17312 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17314 2017-06-14 Richard Biener <rguenther@suse.de>
17316 PR tree-optimization/81083
17317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17320 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17322 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17323 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17324 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17325 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17326 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17327 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17329 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17331 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17332 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17336 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17340 * config/rs6000/t-rtems: Don't handle SPE.
17342 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17344 * config/rs6000/t-linux: Don't handle SPE.
17346 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17348 * config/rs6000/eabispe.h: Delete file.
17350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17352 * config/rs6000/t-spe: Delete file.
17354 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17356 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17357 (rs6000_legitimate_offset_address_p): Return false for anything in
17358 V2SImode or V2SFmode.
17360 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17362 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17363 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17364 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17366 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17367 (rs6000_legitimate_offset_address_p): Ditto.
17368 (rs6000_emit_move): Ditto.
17369 (rs6000_init_builtins): Remove V4HI_type_node.
17371 2017-06-13 Martin Liska <mliska@suse.cz>
17374 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17375 (gate_asan): Likewise.
17376 * asan.h (asan_no_sanitize_address_p): Remove the function.
17377 (sanitize_flags_p): New function.
17378 * builtins.def: Fix coding style.
17379 * common.opt: Use renamed enum value.
17380 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17381 * doc/extend.texi: Document no_sanitize attribute.
17382 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17383 to SANITIZE_UNDEFINED_NONDEFAULT.
17384 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17385 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17386 Use sanitize_flags_p.
17387 * gimplify.c (gimplify_function_tree): Likewise.
17388 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17389 * opts.c (parse_no_sanitize_attribute): New function.
17390 (common_handle_option): Use renamed enum value.
17391 * opts.h (parse_no_sanitize_attribute): Declare.
17392 * tree.c (sanitize_flags_p): New function.
17393 * tree.h: Declared here.
17394 * tsan.c: Use sanitize_flags_p.
17395 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17396 (instrument_mem_ref): Likewise.
17397 (instrument_bool_enum_load): Likewise.
17398 (do_ubsan_in_current_function): Remove the function.
17399 (pass_ubsan::execute): Use sanitize_flags_p.
17400 * ubsan.h: Remove do_ubsan_in_current_function
17401 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17402 (dump_function_to_file): Use it here.
17404 2017-06-13 Martin Jambor <mjambor@suse.cz>
17406 PR tree-optimization/80803
17407 PR tree-optimization/81063
17408 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17409 (propagate_subaccesses_across_link): Enqueue subtree whenever
17410 necessary instead of relying on the caller.
17412 2017-06-13 Martin Jambor <mjambor@suse.cz>
17414 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17415 that have a first_link.
17416 (sort_and_splice_var_accesses): Do not check first_link before
17418 (subtree_mark_written_and_enqueue): Likewise.
17419 (propagate_all_subaccesses): Likewise and do not stop at first
17420 parent with a first_link.
17422 2017-06-13 Martin Jambor <mjambor@suse.cz>
17424 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17427 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17429 * match.pd: New pattern.
17431 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17433 * tree-vrp.c (is_masked_range_test): New function.
17434 (register_edge_assert_for): Determine ranges for
17437 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17439 PR tree-optimization/67328
17440 * fold-const.c (maskable_range_p): New function.
17441 (build_range_check): Generate bittests if possible.
17443 2017-06-13 Martin Liska <mliska@suse.cz>
17445 * gimple-pretty-print.c (dump_probability): Add new argument.
17446 (dump_edge_probability): Dump both probability and count.
17447 (dump_gimple_label): Likewise.
17448 (dump_gimple_bb_header): Likewise.
17450 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17453 * config/avr/avr-devices.c: Fix indentation.
17454 * config/avr/gen-avr-mmcu-specs.c: Dito.
17456 2017-06-13 Richard Biener <rguenther@suse.de>
17458 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17459 instead get vector type from stmt_info.
17460 (vectorizable_reduction): Adjust. Remove dead code.
17462 2017-06-13 Richard Biener <rguenther@suse.de>
17464 PR middle-end/81065
17465 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17466 case of C * (x * C2 + C3).
17467 (fold_addr_of_array_ref_difference): Properly fold index difference.
17469 2017-06-12 David S. Miller <davem@davemloft.net>
17472 * config/sparc/sparc.md (return expander): Emit frame blockage if
17473 function uses alloca.
17475 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17477 * combine.c (make_field_assignment): Check len rather than the mode
17478 precision when calling force_to_mode.
17480 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17482 Support multilibs and devices that see flash in RAM address range.
17485 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17486 (avr_mcu_t) <flash_pm_offset>: New field.
17487 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17488 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17489 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17490 (AVR_TINY_PM_OFFSET): Remove macro.
17491 * config/avr/avr.opt (-mshort-calls): New option.
17492 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17493 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17494 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17495 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17496 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17497 instead of avr_arch->have_jmp_call.
17498 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17499 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17500 avr_arch->flash_pm_offset to define.
17501 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17502 new field flash_pm_offset. Add entry for avrxmega3.
17503 (avr_texinfo): Add entry for avrxmega3.
17504 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17505 attiny212, attiny214,
17506 attiny412, attiny414, attiny416, attiny417,
17507 attiny814, attiny816, attiny817,
17508 attiny1614, attiny1616, attiny1617,
17509 attiny3214, attiny3216, attiny3217.
17510 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17511 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17512 (avr_print_operand_address) [AVR_TINY]: Same.
17513 (avr_asm_init_sections) <readonly_data_section>: Only patch
17514 callback if avr_arch->flash_pm_offset = 0.
17515 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17516 for rodata if avr_arch->flash_pm_offset != 0.
17517 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17518 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17519 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17520 (m_options): Append opt_rcall.
17521 (m_dirnames): Append dir_rcall.
17522 * config/avr/t-multilib: Regenerate.
17524 * configure.ac [target=avr]: Check whether avrxmega3 default
17525 linker description file works as needed.
17526 * configure: Regenerate.
17527 * doc/avr-mmcu.texi: Regenerate.
17528 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17529 <__AVR_ARCH__>: Document avrxmega3 and 103.
17530 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17531 <__AVR_SHORT_CALLS__>: Document it.
17532 <__AVR_PM_BASE_ADDRESS__>: Document it.
17533 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17534 (AVR Variable Attributes) <progmem>: Document this is
17535 not needed for avrxmega3.
17536 (AVR Named Address Spaces) <__flash>: Dito.
17538 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17540 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17542 2017-06-12 Doug Rupp <rupp@adacore.com>
17544 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17545 Append vxworks-stdint.h to the tm_file list.
17546 * config/vxworks-stdint.h: New file.
17548 2017-06-12 Martin Liska <mliska@suse.cz>
17550 PR tree-optimization/81041
17551 * tree-profile.c (gimple_gen_ic_func_profiler):
17552 Create an extra BB in profile-generate
17553 (gimple_gen_time_profiler): Likewise.
17555 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17557 PR tree-optimization/81003
17558 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17559 (update_range_test): Use it instead of force_gimple_operand_gsi.
17561 2017-06-12 Richard Biener <rguenther@suse.de>
17563 PR tree-optimization/81053
17564 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17565 with backedge value not defined in loop. Simplify def stmt
17568 2017-06-11 Tom de Vries <tom@codesourcery.com>
17571 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17573 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17574 nvptx_cannot_force_const_mem.
17576 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17578 * opts.c (finish_options): Move test for flag_split_stack after
17579 it has been initialized.
17581 2017-06-11 Jason Merrill <jason@redhat.com>
17583 * tree.h (id_equal): New.
17584 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17585 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17586 instead of strcmp of IDENTIFIER_POINTER.
17588 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17590 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17591 (mark_all_inlined_calls_cdtor): Fix formating.
17592 (inline_transform): Rescale profile before inlining.
17594 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17596 * cgraph.h (cgraph_edge::clone): Update prototype.
17597 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17598 (cgraph_node::create_clone): Update.
17599 (cgraph_node::create_version_clone): Update.
17600 * tree-inline.c (copy_bb): Update.
17601 (expand_call_inline): Update.
17603 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17605 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17606 factored out from ...
17607 (rs6000_emit_prologue): ... here.
17609 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17611 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17612 factored out from ...
17613 (rs6000_emit_prologue): ... here.
17615 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17617 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17619 (handle_missing_profiles): Fix computation of tp_first_run.
17620 (counts_to_freqs): Do not touch freqs when count is 0.
17622 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17624 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17627 2017-06-10 Tom de Vries <tom@codesourcery.com>
17629 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17630 attributes): Document signal effective target.
17632 2017-06-10 Tom de Vries <tom@codesourcery.com>
17634 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17635 Document effective target stack_size.
17637 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17639 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17640 to the edit_context if they can be auto-applied.
17642 2017-06-9 Ian Lance Taylor <iant@golang.org>
17644 * opts.c (finish_options): If -fsplit-stack, disable implicit
17645 -forder-blocks-and-partition.
17646 * doc/invoke.texi (Optimize Options): Document that when using
17647 -fsplit-stack -forder-blocks-and-partition is not implicitly
17650 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17652 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17653 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17654 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17655 * builtins.def (abort, trap, unreachable): Declare cold.
17656 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17657 * tree-core.h (ECF_COLD): New.
17658 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17659 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17661 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17663 * predict.c (unlikely_executed_stmt_p): Cleanup.
17665 2017-06-09 Richard Biener <rguenther@suse.de>
17667 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17668 model if the ref is always written to.
17670 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17672 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17674 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17676 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17679 2017-06-09 Tom de Vries <tom@codesourcery.com>
17682 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17683 "target cannot support label values" when encountering LABEL_REF.
17685 2017-06-09 Martin Liska <mliska@suse.cz>
17687 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17688 (gimple_gen_ic_func_profiler): Emit direct comparison
17689 of __gcov_indirect_call_callee with NULL.
17690 (gimple_gen_time_profiler): Change probability from
17691 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17693 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17695 * profile.c (edge_gcov_counts): Turn to pointer.
17696 (compute_branch_probabilities, compute_branch_probabilities): Update.
17697 (branch_prob): Do not clear edge_gcov_count.
17698 * profile.h (edge_gcov_counts): Turn to pointer.
17699 (edge_gcov_count): Update.
17701 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17703 * gimple.h (gimple_check_failed): Mark cold.
17705 2017-06-09 Richard Biener <rguenther@suse.de>
17707 PR tree-optimization/66623
17708 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17709 refactor check_reduction into two parts, properly computing
17710 whether we have to check reduction validity for outer loop
17713 2017-06-09 Richard Biener <rguenther@suse.de>
17715 PR tree-optimization/79483
17716 * graphite-scop-detection.c (order): New global.
17717 (get_order): Compute bb to order mapping that satisfies code
17718 generation constraints.
17719 (cmp_pbbs): New helper.
17720 (build_scops): Start domwalk at entry block, sort generated
17723 2017-06-09 Richard Biener <rguenther@suse.de>
17725 PR middle-end/81007
17726 * ipa-polymorphic-call.c
17727 (ipa_polymorphic_call_context::restrict_to_inner_class):
17728 Skip FIELD_DECLs with error_mark_node type.
17729 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17732 2017-06-09 Martin Liska <mliska@suse.cz>
17734 * predict.c (struct branch_predictor): New struct.
17735 (test_prediction_value_range): New test.
17736 (predict_c_tests): New function.
17737 * selftest-run-tests.c (selftest::run_tests): Run the function.
17738 * selftest.h: Declare new tests.
17740 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17743 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17744 gen_add3_insn did not fail.
17745 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17746 r0, construct that number in a temporary reg and add that reg to r0.
17747 If asked to put the result in r0 as well, fail.
17749 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17751 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17752 for early expansion of vec_eqv.
17754 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17756 PR middle-end/81005
17757 * ubsan.c (instrument_null): Avoid pointless code temporary.
17758 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17760 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17764 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17767 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17770 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17772 * predict.c (unlikely_executed_edge_p): Move ahead.
17773 (probably_never_executed_edge_p): Use it.
17775 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17777 PR middle-end/79988
17778 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17779 gimple_call_builtin_p call.
17781 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17783 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17784 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17785 rtl_check_failed_type2, rtl_check_failed_code1,
17786 rtl_check_failed_code2, rtl_check_failed_code_mode,
17787 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17788 rtvec_check_failed_bounds, rtl_check_failed_flag,
17789 _fatal_insn_not_found, _fatal_insn): Likewise.
17790 * tree.h (tree_contains_struct_check_failed,
17791 tree_check_failed, tree_not_check_failed,
17792 tree_class_check_failed, tree_range_check_failed,
17793 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17794 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17795 tree_operand_check_failed, omp_clause_check_failed,
17796 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17799 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17801 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17802 flag_branch_probabilities.
17803 * ipa-inline.c (edge_badness): Likewise.
17804 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17805 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17806 * predict.c (maybe_hot_frequency_p): Likewise.
17807 (probably_never_executed): Likewise.
17808 * sched-ebb.c (schedule_ebbs): Likewise.
17809 * sched-rgn.c (find_single_block_region): Likewise.
17810 * tracer.c (tail_duplicate): Likewise.
17812 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17814 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17815 longer requires x_flag_profile_use.
17817 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17819 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17820 instead of flag_reorder_blocks_and_partition.
17821 * dbxout.c (dbxout_function_end): Likewise.
17822 * dwarf2out.c (gen_subprogram_die): Likewise.
17823 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17824 * hw-doloop.c (reorg_loops): Likewise.
17825 * varasm.c (assemble_start_function,
17826 assemble_end_function): Likewise.
17827 (decide_function_section): Do not check for
17828 flag_reorder_blocks_and_partition.
17830 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17832 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17834 (chkp_get_hard_register_fake_addr_expr): Ditto.
17835 (chkp_build_addr_expr): Add check for hard reg case.
17836 (chkp_parse_array_and_component_ref): Ditto.
17837 (chkp_find_bounds_1): Ditto.
17838 (chkp_process_stmt): Don't generate bounds store for
17841 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17843 * predict.c (maybe_hot_bb_p): Do not check profile status.
17844 (maybe_hot_edge_p): Likewise.
17845 (probably_never_executed): Check for zero counts even if profile
17847 (unlikely_executed_edge_p): New function.
17848 (unlikely_executed_stmt_p): New function.
17849 (unlikely_executed_bb_p): New function.
17850 (set_even_probabilities): Use unlikely predicates.
17851 (combine_predictions_for_bb): Likewise.
17852 (predict_paths_for_bb): Likewise.
17853 (predict_paths_leading_to_edge): Likewise.
17854 (determine_unlikely_bbs): New function.
17855 (estimate_bb_frequencies): Use it.
17856 (compute_function_frequency): Use zero counts even if profile is
17858 * profile-count.h: Fix typo.
17860 2017-08-08 Julia Koval <julia.koval@intel.com>
17862 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17863 _mm512_mask_cvtsepi16_storeu_epi8,
17864 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17865 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17866 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17867 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17868 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17869 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17870 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17871 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17872 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17873 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17874 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17875 __builtin_ia32_pmovuswb256mem_mask,
17876 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17877 __builtin_ia32_pmovwb512mem_mask): New builtins.
17879 2017-08-08 Julia Koval <julia.koval@intel.com>
17881 PR target/73350,80862
17882 * config/i386/subst.md (round): Fix round pattern.
17883 * config/i386/i386.c (ix86_erase_embedded_rounding):
17884 Fix erasing rounding for the fixed pattern.
17886 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17888 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17890 2017-06-08 Martin Liska <mliska@suse.cz>
17892 PR gcov-profile/80911
17893 * gcov.c (block_info::block_info): New constructor.
17895 2017-06-07 Carl Love <cel@us.ibm.com>
17897 * config/rs6000/rs6000-c: The return type of the following
17898 built-in functions was implemented as int not long long. Fix sign
17899 of return value for the unsigned version of vec_mulo and vec_mule.
17900 vector unsigned long long vec_bperm (vector unsigned long long,
17901 vector unsigned char)
17902 vector signed long long vec_mule (vector signed int,
17904 vector unsigned long long vec_mule (vector unsigned int,
17905 vector unsigned int)
17906 vector signed long long vec_mulo (vector signed int,
17908 vector unsigned long long vec_mulo (vector unsigned int,
17909 vector unsigned int)
17910 * doc/extend.texi: Fix the documentation for the built-in
17913 2017-06-07 Carl Love <cel@us.ibm.com>
17916 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17919 2017-06-07 Carl Love <cel@us.ibm.com>
17921 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17922 support, Generate doublehv for signed int/float for BE case only.
17924 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17926 * doc/invoke.texi (mcx16): Rewrite.
17928 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17930 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17931 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17932 *mov<mode>_softfloat, and an anonymous splitter): Use
17933 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17935 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17937 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17939 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17940 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17941 (rs6000_debug_reg_global): Adjust.
17942 (rs6000_init_hard_regno_mode_ok): Adjust.
17943 (rs6000_dbx_register_number): Adjust.
17944 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17945 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17946 Remove SPE_ACC and SPEFSCR.
17947 (REG_ALLOC_ORDER): Ditto.
17948 (FRAME_POINTER_REGNUM): Change to 111.
17949 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17950 (REG_CLASS_NAMES): Ditto.
17951 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17952 (REGISTER_NAMES): Ditto.
17953 (ADDITIONAL_REG_NAMES): Ditto.
17954 (rs6000_reg_names): Ditto.
17955 * config/rs6000/rs6000.md: Renumber some register number
17958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17960 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17962 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17963 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17965 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17966 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17967 Delete the SPE high registers.
17968 (REG_ALLOC_ORDER): Ditto.
17969 (enum reg_class): Remove SPE_HIGH_REGS.
17970 (REG_CLASS_NAMES): Ditto.
17971 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17972 (REGISTER_NAMES): Ditto.
17973 (rs6000_reg_names): Ditto.
17974 * doc/tm.texi.in: Remove SPE as example.
17975 * doc/tm.texi: Regenerate.
17977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17979 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17980 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17981 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17982 * config/rs6000/rs6000.md (type): Remove "brinc".
17984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17986 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17987 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17988 * config/rs6000/linuxspe.h: Delete file.
17989 * config/rs6000/rs6000.md: Don't include spe.md.
17990 * config/rs6000/spe.h: Delete file.
17991 * config/rs6000/spe.md: Delete file.
17992 * config/rs6000/t-rs6000: Remove spe.md.
17994 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17996 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17997 (reg_or_none500mem_operand): Delete.
17998 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17999 instead of reg_or_none500mem_operand.
18001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18003 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18004 handling of SPE flags.
18005 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18009 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18011 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18012 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18013 paired_divv2sf3): Similar.
18014 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18015 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18016 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18018 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18019 Rename the paired_* instruction patterns.
18020 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18022 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18023 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18024 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18025 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18026 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18027 PAIRED_VECTOR_MODE.
18028 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18029 (spe_func_has_64bit_regs_p): Delete.
18030 (spe_expand_predicate_builtin): Delete.
18031 (spe_expand_evsel_builtin): Delete.
18032 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18033 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18034 (invalid_e500_subreg): Delete.
18035 (rs6000_legitimize_address): Always force_reg op2 as well, for
18036 paired single memory accesses.
18037 (rs6000_member_type_forces_blk): Delete.
18038 (rs6000_spe_function_arg): Delete.
18039 (rs6000_expand_unop_builtin): Delete SPE handling.
18040 (rs6000_expand_binop_builtin): Ditto.
18041 (spe_expand_stv_builtin): Delete.
18042 (bdesc_2arg_spe): Delete.
18043 (spe_expand_builtin): Delete.
18044 (spe_expand_predicate_builtin): Delete.
18045 (spe_expand_evsel_builtin): Delete.
18046 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18047 (spe_init_builtins): Delete.
18048 (spe_func_has_64bit_regs_p): Delete.
18049 (savres_routine_name): Delete "info" parameter. Adjust callers.
18050 (rs6000_emit_stack_reset): Ditto.
18051 (rs6000_dwarf_register_span): Delete.
18052 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18053 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18054 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18056 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18058 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18059 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18060 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18061 mulv2sf3, divv2sf3): Delete expanders.
18063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18065 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18067 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18069 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18070 * config/rs6000/rs6000.c: Ditto.
18072 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18074 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18075 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18076 comparison_operator.
18078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18080 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18081 * config/rs6000/rs6000.opt: Ditto.
18082 * config/rs6000/t-rtems: Ditto.
18084 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18086 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18087 TARGET_E500_SINGLE by 0, simplify.
18088 * config/rs6000/rs6000.c: Ditto.
18089 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18090 (spe_build_register_parallel): Delete.
18091 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18092 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18093 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18094 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18095 (E500_CONVERT): Delete.
18096 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18100 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18101 * config/rs6000/dfp.md: Ditto.
18102 (negdd2, *negdd2_fpr): Merge.
18103 (absdd2, *absdd2_fpr): Merge.
18104 (negtd2, *negtd2_fpr): Merge.
18105 (abstd2, *abstd2_fpr): Merge.
18106 * config/rs6000/e500.h: Delete file.
18107 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18108 TARGET_FPRS by 1 and simplify.
18109 * config/rs6000/rs6000-c.c: Ditto.
18110 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18111 TARGET_DF_SPE by 0.
18112 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18114 * config/rs6000/rs6000.md: Ditto.
18115 (floatdidf2, *floatdidf2_fpr): Merge.
18116 (move_from_CR_gt_bit): Delete.
18117 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18118 (E500_CR_IOR_COMPARE): Delete.
18119 (All patterns that require !TARGET_FPRS): Delete.
18120 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18124 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18128 * graphds.c (add_edge): Intitialize edge's attached data.
18129 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18130 pointer parameter. Call pointed function on each edge during
18131 graph traversing. Skip traversing the edge when the function
18133 (graphds_dfs, graphds_scc): Ditto.
18134 (for_each_edge): New parameter. Pass the new parameter to callback
18136 * graphds.h (skip_edge_callback): New function pointer type.
18137 (graphds_dfs, graphds_scc): New function pointer parameter.
18138 (graphds_edge_callback, for_each_edge): New parameter.
18140 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18142 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18143 out code checking if runtime alias check is possible to below ...
18144 Call the new function.
18145 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18146 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18148 2017-06-07 Marek Polacek <polacek@redhat.com>
18151 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18152 TYPE_OVERFLOW_WRAPS check.
18154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18156 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18157 if versioning is required.
18158 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18159 peeling with the check for versioning.
18161 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18163 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18164 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18165 Set true to new parameter if new ssa variable is defined.
18166 (vect_gen_vector_loop_niters): Refactor. Set range information
18167 for the new vector loop bound variable.
18168 (vect_do_peeling): Ditto.
18170 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18172 * tree-affine.c (ssa.h): Include header file.
18173 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18174 has wrapping overflow behavior.
18176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18178 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18182 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18183 (tree_to_aff_combination): ... here.
18185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18187 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18188 reg_pressure model function.
18189 (ivopts_global_cost_for_size): Delete.
18190 (determine_set_costs, iv_ca_recount_cost): Call new model function
18191 ivopts_estimate_reg_pressure.
18193 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18195 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18196 expensive than udiv. Remove floating point cases from mod.
18198 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18200 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18201 Increase idiv cost.
18203 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18205 * config/aarch64/aarch64.md
18206 (copysignsf3): Fix mask generation.
18208 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18210 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18212 (class dump_manager): Add register_dumps method.
18213 * dumpfile.c: Include langhooks.h.
18214 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18215 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18216 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18217 (dump_manager::dump_register): Start with 512 entries instead of 32.
18218 (dump_manager::register_dumps): New method.
18219 * toplev.c (general_init): Instead of invoking register_dumps
18220 langhook, invoke register_dumps method on the dump manager.
18221 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18224 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18226 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18227 Say that # requires an associated define_split to exist, and that
18228 the define_split must be suitable for use after register allocation.
18230 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18232 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18233 (compute_outgoing_frequencies): Also initialize zero counts.
18234 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18235 around loops; preserve more of profile when nothing changes.
18237 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18239 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18241 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18242 * config/arm/arm-cpu-cdata.h: Regenerate.
18243 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18244 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18245 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18246 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18247 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18249 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18251 * config/arm/t-rmprofile: Likewise.
18252 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18254 2017-06-06 David S. Miller <davem@davemloft.net>
18257 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18258 blockage if function uses alloca.
18260 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18262 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18263 New "uid" fields to hold pretty-print IDs of group and ref.
18264 Memory references are now identified as <group_id>:<ref_id>
18265 instead of using [random] addresses.
18266 (dump_mem_details): Simplify, no functional change.
18267 (dump_mem_ref): Simplify and make output more concise.
18268 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18269 (find_or_create_group): Initialize group uid.
18270 (record_ref): Initialize ref uid. Improve debug output.
18271 (prune_group_by_reuse, should_issue_prefetch_p,)
18272 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18273 (mark_nontemporal_store, determine_loop_nest_reuse):
18274 Improve debug output.
18276 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18278 * dbgcnt.def (prefetch): New debug counter.
18279 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18280 (schedule_prefetches): Stop issueing prefetches if debug counter
18283 2017-06-06 Tom de Vries <tom@codesourcery.com>
18285 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18286 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18288 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18290 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18291 Use aarch64_reg_or_zero predicate for operand 4.
18292 (aarch64_compare_and_swap<mode> define_insn_and_split):
18293 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18294 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18296 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18298 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18299 (arm_compute_save_core_reg_mask): This.
18300 (thumb1_compute_save_reg_mask): Rename into ...
18301 (thumb1_compute_save_core_reg_mask): This.
18302 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18303 (arm_compute_frame_layout): Likewise.
18305 2017-06-06 Richard Biener <rguenther@suse.de>
18307 PR tree-optimization/80974
18308 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18309 keep or clear leaders SSA info.
18311 2017-06-06 Tom de Vries <tom@codesourcery.com>
18313 * config/nvptx/nvptx.c (split_mode_p): New function.
18314 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18316 2017-06-06 Tom de Vries <tom@codesourcery.com>
18318 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18320 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18323 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18326 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18328 * shrink-wrap.c (handle_simple_exit): Update profile.
18329 (try_shrink_wrapping): Upate profile.
18331 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18333 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18334 (tree_guess_outgoing_edge_probabilities): New.
18335 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18336 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18338 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18340 * ipa-split.c (split_function): Initialize return bb profile.
18342 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18344 * profile.c (compute_branch_probabilities): Also initialize
18345 EXIT_BLOCK profile.
18347 2017-06-06 Richard Biener <rguenther@suse.de>
18349 PR tree-optimization/80928
18350 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18351 (vect_analyze_loop_operations): Properly guard analysis for
18353 (vect_transform_loop): Likewise.
18354 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18355 (vect_model_induction_cost): Do not cost for pure SLP.
18356 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18357 of induction in inner loop vectorization.
18358 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18359 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18360 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18362 (vect_analyze_slp_cost_1): Cost induction.
18363 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18364 (vect_get_slp_vect_defs): Likewise.
18365 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18366 (vect_transform_stmt): Handle SLP reductions.
18367 * tree-vectorizer.h (vectorizable_induction): Adjust.
18369 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18371 * config/rs6000/rs6000.c (make_resolver_func): Update
18372 init_lowered_empty_function call.
18374 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18376 * doc/invoke.texi: Document the -fprofile-abs-path option.
18377 * common.opt (fprofile-abs-path): New option.
18378 * gcov-io.h (gcov_write_filename): Declare.
18379 * gcov-io.c (gcov_write_filename): New function.
18380 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18381 * profile.c (output_location): Likewise.
18383 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18385 * shring-wrap.c: Revert accidental commit.
18387 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18389 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18391 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18393 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18395 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18396 profile in callgraph edge.
18397 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18398 (apply_scale): Likewise.
18399 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18400 Also scale profile when inlining function with zero profile.
18401 (initialize_cfun): Update exit block profile even when it is zero.
18402 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18403 when profile is read.
18405 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18407 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18408 (CLONE_*): New constants to define the processors we can generate
18409 code for with the target_clone attribute.
18410 (rs6000_clone_map): New array to identify which clone processors
18411 the current program is running on.
18412 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18413 target_clone attribute.
18414 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18415 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18416 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18417 (cpu_expand_builtin): Add support for target_clone attribute.
18418 (rs6000_valid_attribute_p): Allow "default" attribute.
18419 (get_decl_name): New debug function to simplify printing the
18420 current function name in debugging statements.
18421 (rs6000_clone_priority): New functions to support the target_clone
18422 attribute, and be able to generate code to switch between ISA 2.05
18423 through ISA 3.0 (power6 through power9).
18424 (rs6000_compare_version_priority): Likewise.
18425 (rs6000_get_function_versions_dispatcher): Likewise.
18426 (make_resolver_func): Likewise.
18427 (add_condition_to_bb): Likewise.
18428 (dispatch_function_versions): Likewise.
18429 (rs6000_generate_version_dispatcher_body): Likewise.
18430 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18431 (fusion_gpr_load_p): Fix a spacing issue.
18432 * doc/extend.texi (Common Function Attributes): Document that the
18433 PowerPC supports the target_clone attribute.
18435 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18437 * config/arm/arm.h: explain F symbol found in description of ARM
18438 register allocation in its legend.
18440 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18442 * config/mips/frame-header-opt.c: Include profile-count.h.
18443 * config/riscv/riscv.c: Include profile-count.h
18445 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18447 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18449 (sm_set_flag_if_changed): Add bbs field.
18450 (execute_sm_if_changed_flag_set): Pass BBS.
18451 (execute_sm): Update.
18453 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18455 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18458 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18460 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18461 (peephole2): New peephole2 to emit the above.
18462 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18464 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18466 * config/aarch64/aarch64.c (define_peephole2 above
18467 *sub_<shift>_<mode>): New peephole.
18469 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18471 * config/i386/i386.c (make_resolver_func): Update.
18472 * Makefile.in: Add profile-count.h and profile-count.o
18473 * auto-profile.c (afdo_indirect_call): Update to new API.
18474 (afdo_set_bb_count): Update.
18475 (afdo_propagate_edge): Update.
18476 (afdo_propagate_circuit): Update.
18477 (afdo_calculate_branch_prob): Update.
18478 (afdo_annotate_cfg): Update.
18479 * basic-block.h: Include profile-count.h
18480 (struct edge_def): Turn count to profile_count.
18481 (struct basic_block_def): Likewie.
18482 (REG_BR_PROB_BASE): Move to profile-count.h
18483 (RDIV): Move to profile-count.h
18484 * bb-reorder.c (max_entry_count): Turn to profile_count.
18485 (find_traces): Update.
18486 (rotate_loop):Update.
18487 (connect_traces):Update.
18488 (sanitize_hot_paths):Update.
18489 * bt-load.c (migrate_btr_defs): Update.
18490 * cfg.c (RDIV): Remove.
18491 (init_flow): Use alloc_block.
18492 (alloc_block): Uninitialize count.
18493 (unchecked_make_edge): Uninitialize count.
18494 (check_bb_profile): Update.
18495 (dump_edge_info): Update.
18496 (dump_bb_info): Update.
18497 (update_bb_profile_for_threading): Update.
18498 (scale_bbs_frequencies_int): Update.
18499 (scale_bbs_frequencies_gcov_type): Update.
18500 (scale_bbs_frequencies_profile_count): New.
18501 * cfg.h (update_bb_profile_for_threading): Update.
18502 (scale_bbs_frequencies_profile_count): Declare.
18503 * cfgbuild.c (compute_outgoing_frequencies): Update.
18504 (find_many_sub_basic_blocks): Update.
18505 * cfgcleanup.c (try_forward_edges): Update.
18506 (try_crossjump_to_edge): Update.
18507 * cfgexpand.c (expand_gimple_tailcall): Update.
18508 (construct_exit_block): Update.
18509 * cfghooks.c (verify_flow_info): Update.
18510 (dump_bb_for_graph): Update.
18511 (split_edge): Update.
18512 (make_forwarder_block): Update.
18513 (duplicate_block): Update.
18514 (account_profile_record): Update.
18515 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18516 (get_estimated_loop_iterations): Update.
18517 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18518 (single_likely_exit): Update.
18519 * cfgloopmanip.c (scale_loop_profile): Update.
18521 (set_zero_probability): Update.
18522 (lv_adjust_loop_entry_edge): Update.
18523 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18524 (purge_dead_edges): Update.
18525 (rtl_account_profile_record): Update.
18526 * cgraph.c (cgraph_node::create): Uninitialize count.
18527 (symbol_table::create_edge): Uninitialize count.
18528 (cgraph_update_edges_for_call_stmt_node): Update.
18529 (cgraph_edge::dump_edge_flags): Update.
18530 (cgraph_node::dump): Update.
18531 (cgraph_edge::maybe_hot_p): Update.
18532 * cgraph.h: Include profile-count.h
18533 (create_clone), create_edge, create_indirect_edge): Update.
18534 (cgraph_node): Turn count to profile_count.
18535 (cgraph_edge0: Likewise.
18536 (make_speculative, clone): Update.
18537 (create_edge): Update.
18538 (init_lowered_empty_function): Update.
18539 * cgraphclones.c (cgraph_edge::clone): Update.
18540 (duplicate_thunk_for_node): Update.
18541 (cgraph_node::create_clone): Update.
18542 * cgraphunit.c (cgraph_node::analyze): Update.
18543 (cgraph_node::expand_thunk): Update.
18544 * final.c (dump_basic_block_info): Update.
18545 * gimple-streamer-in.c (input_bb): Update.
18546 * gimple-streamer-out.c (output_bb): Update.
18547 * graphite.c (print_global_statistics): Update.
18548 (print_graphite_scop_statistics): Update.
18549 * hsa-brig.c: Include basic-block.h.
18550 * hsa-dump.c: Include basic-block.h.
18551 * hsa-gen.c (T sum_slice): Update.
18552 (convert_switch_statements):Update.
18553 * hsa-regalloc.c: Include basic-block.h.
18554 * ipa-chkp.c (chkp_produce_thunks): Update.
18555 * ipa-cp.c (struct caller_statistics): Update.
18556 (init_caller_stats): Update.
18557 (gather_caller_stats): Update.
18558 (ipcp_cloning_candidate_p): Update.
18559 (good_cloning_opportunity_p): Update.
18560 (get_info_about_necessary_edges): Update.
18561 (dump_profile_updates): Update.
18562 (update_profiling_info): Update.
18563 (update_specialized_profile): Update.
18564 (perhaps_add_new_callers): Update.
18565 (decide_about_value): Update.
18566 (ipa_cp_c_finalize): Update.
18567 * ipa-devirt.c (struct odr_type_warn_count): Update.
18568 (struct decl_warn_count): Update.
18569 (struct final_warning_record): Update.
18570 (possible_polymorphic_call_targets): Update.
18571 (ipa_devirt): Update.
18572 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18573 * ipa-icf.c (sem_function::merge): Update.
18574 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18575 * ipa-inline.c (compute_uninlined_call_time): Update.
18576 (compute_inlined_call_time): Update.
18577 (want_inline_small_function_p): Update.
18578 (want_inline_self_recursive_call_p): Update.
18579 (edge_badness): Update.
18580 (lookup_recursive_calls): Update.
18581 (recursive_inlining): Update.
18582 (inline_small_functions): Update.
18583 (dump_overall_stats): Update.
18584 (dump_inline_stats): Update.
18585 * ipa-profile.c (ipa_profile_generate_summary): Update.
18586 (ipa_propagate_frequency): Update.
18587 (ipa_profile): Update.
18588 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18589 * ipa-utils.c (ipa_merge_profiles): Update.
18590 * loop-doloop.c (doloop_modify): Update.
18591 * loop-unroll.c (report_unroll): Update.
18592 (unroll_loop_runtime_iterations): Update.
18593 * lto-cgraph.c (lto_output_edge): Update.
18594 (lto_output_node): Update.
18595 (input_node): Update.
18596 (input_edge): Update.
18597 (merge_profile_summaries): Update.
18598 * lto-streamer-in.c (input_cfg): Update.
18599 * lto-streamer-out.c (output_cfg): Update.
18600 * mcf.c (create_fixup_graph): Update.
18601 (adjust_cfg_counts): Update.
18602 (sum_edge_counts): Update.
18603 * modulo-sched.c (sms_schedule): Update.
18604 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18605 * predict.c (maybe_hot_count_p): Update.
18606 (probably_never_executed): Update.
18607 (dump_prediction): Update.
18608 (combine_predictions_for_bb): Update.
18609 (propagate_freq): Update.
18610 (handle_missing_profiles): Update.
18611 (counts_to_freqs): Update.
18612 (rebuild_frequencies): Update.
18613 (force_edge_cold): Update.
18614 * predict.h: Include profile-count.h
18615 (maybe_hot_count_p, counts_to_freqs): UPdate.
18616 * print-rtl-function.c: Do not include cfg.h
18617 * print-rtl.c: Include basic-block.h
18618 * profile-count.c: New file.
18619 * profile-count.h: New file.
18620 * profile.c (is_edge_inconsistent): Update.
18621 (correct_negative_edge_counts): Update.
18622 (is_inconsistent): Update.
18623 (set_bb_counts): Update.
18624 (read_profile_edge_counts): Update.
18625 (compute_frequency_overlap): Update.
18626 (compute_branch_probabilities): Update; Initialize and deinitialize
18628 (branch_prob): Update.
18629 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18630 (edge_gcov_count): New.
18631 (bb_gcov_count): New.
18632 * shrink-wrap.c (try_shrink_wrapping): Update.
18633 * tracer.c (better_p): Update.
18634 * trans-mem.c (expand_transaction): Update.
18635 (ipa_tm_insert_irr_call): Update.
18636 (ipa_tm_insert_gettmclone_call): Update.
18637 * tree-call-cdce.c: Update.
18638 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18639 (gimple_duplicate_sese_tail): Update.
18640 (gimple_account_profile_record): Update.
18641 (execute_fixup_cfg): Update.
18642 * tree-inline.c (copy_bb): Update.
18643 (copy_edges_for_bb): Update.
18644 (initialize_cfun): Update.
18645 (freqs_to_counts): Update.
18646 (copy_cfg_body): Update.
18647 (expand_call_inline): Update.
18648 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18649 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18650 (try_unroll_loop_completely): Update.
18651 (try_peel_loop): Update.
18652 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18653 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18654 * tree-ssa-loop-split.c (connect_loops): Update.
18655 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18656 * tree-ssa-reassoc.c (branch_fixup): Update.
18657 * tree-ssa-tail-merge.c (replace_block_by): Update.
18658 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18659 (compute_path_counts): Update.
18660 (update_profile): Update.
18661 (recompute_probabilities): Update.
18662 (update_joiner_offpath_counts): Update.
18663 (estimated_freqs_path): Update.
18664 (freqs_to_counts_path): Update.
18665 (clear_counts_path): Update.
18666 (ssa_fix_duplicate_block_edges): Update.
18667 (duplicate_thread_path): Update.
18668 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18669 (struct switch_conv_info): Update.
18670 * tree-tailcall.c (decrease_profile): Update.
18671 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18672 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18673 * value-prof.c (check_counter): Update.
18674 (gimple_divmod_fixed_value): Update.
18675 (gimple_mod_pow2): Update.
18676 (gimple_mod_subtract): Update.
18677 (gimple_ic_transform): Update.
18678 (gimple_stringop_fixed_value): Update.
18679 * value-prof.h (gimple_ic): Update.
18681 2017-06-02 Carl Love <cel@us.ibm.com>
18683 * config/rs6000/rs6000-c: Add support for built-in functions
18684 vector double vec_doublee (vector signed int);
18685 vector double vec_doublee (vector unsigned int);
18686 vector double vec_doublee (vector float);
18687 vector double vec_doubleh (vector signed int);
18688 vector double vec_doubleh (vector unsigned int);
18689 vector double vec_doubleh (vector float);
18690 vector double vec_doublel (vector signed int);
18691 vector double vec_doublel (vector unsigned int);
18692 vector double vec_doublel (vector float);
18693 vector double vec_doubleo (vector signed int);
18694 vector double vec_doubleo (vector unsigned int);
18695 vector double vec_doubleo (vector float);.
18696 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18697 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18699 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18700 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18701 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18703 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18704 vec_doublel, vec_doubleh.
18705 * doc/extend.texi: Update the built-in documentation file for the
18706 new built-in functions.
18708 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18711 * ipa-inline-analysis.c (free_growth_caches): Set
18712 edge_removal_hook_holder to NULL after removing it.
18714 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18716 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18717 comparision with zero.
18719 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18721 for early expansion of vec_min and vec_max builtins.
18722 (builtin_function_type): Add min/max unsigned variants to those
18723 identified as having unsigned arguments.
18725 2017-06-02 Olivier Hainque <hainque@adacore.com>
18727 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18729 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18731 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18732 Use VALL_F16 iterator rather than VALL.
18734 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18736 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18737 Emit CBNZ inside loop when doing a strong exchange and comparing
18738 against zero. Generate the CC flags after the loop.
18740 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18742 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18743 (dl_section_ref): New.
18744 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18745 On AIX, append an expression to subtract the size of the
18746 section length to dl_section_ref.
18748 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18751 for early expansion of vector absolute builtins.
18753 2017-06-02 Richard Biener <rguenther@suse.de>
18755 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18756 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18758 2017-06-02 Richard Biener <rguenther@suse.de>
18760 PR tree-optimization/80948
18761 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18762 stmt order as well.
18764 2017-06-02 Richard Biener <rguenther@suse.de>
18766 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18768 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18769 for inductions as relevant.
18771 2017-06-02 Richard Biener <rguenther@suse.de>
18773 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18774 (vectorizable_induction): ... this. Remove dead code.
18776 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18778 * builtins. (expand_builtin_alloca): Remove second parameter and
18779 infer its value from the first parameter instead.
18780 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18782 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18784 PR rtl-optimization/80903
18785 * loop-doloop.c (add_test): Unshare sequence.
18787 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18789 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18791 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18793 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18795 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18796 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18797 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18798 (xlogue_layout::get_stub_rtx): Make static.
18799 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18800 (xlogue_layout::compute_stub_managed_regs): Rename to...
18801 (xlogue_layout::count_stub_managed_regs): ...this.
18802 (xlogue_layout::is_stub_managed_reg): New function.
18803 (xlogue_layout::m_stub_names): Rename to...
18804 (xlogue_layout::s_stub_names): ...this, make static.
18805 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18806 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18807 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18808 xlogue_layout::s_stub_names): Instantiate statics.
18809 (stub_managed_regs): Remove.
18810 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18811 (disable_call_ms2sysv_xlogues): Rename to...
18812 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18813 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18815 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18816 change after reload_completed.
18817 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18819 (ix86_expand_prologue): Likewise.
18820 (ix86_expand_epilogue): Likewise.
18821 (ix86_expand_split_stack_prologue): Likewise.
18822 (ix86_compute_frame_layout): Remove frame parameter ...
18823 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18824 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18826 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18827 (ix86_frame): Move from here ...
18828 * config/i386/i386.h (ix86_frame): ... to here.
18829 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18830 complete ix86_frame data structure instead. Remove some_ld_name.
18832 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18834 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18835 symbols that hold a DECL_VALUE_EXPR.
18837 2017-06-01 Martin Jambor <mjambor@suse.cz>
18839 PR tree-optimization/80898
18840 * tree-sra.c (process_subtree_disqualification): Removed.
18841 (disqualify_candidate): Do not acll
18842 process_subtree_disqualification.
18843 (subtree_mark_written_and_enqueue): New function.
18844 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18845 RHS has been disqualified and re-queue LHS if necessary. Apart
18846 from that, ignore disqualified RHS.
18848 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18850 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18851 address fetch for z10 or later.
18853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18855 * config/arc/arc.md (tst_movb): Add guard when splitting.
18857 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18859 * config/arc/arc.c (arc_can_eliminate): Test against
18860 arc_frame_pointer_needed.
18862 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18864 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18865 to prevent store reordering.
18866 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18867 (type): Add block type.
18868 (stack_tie): Define special instruction to be used in
18871 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18873 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18874 constraint. It is not valid for the pattern.
18875 (noncommutative_binary_comparison): Likewise.
18877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18879 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18882 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18884 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18885 be used by the reg-alloc.
18887 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18889 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18890 reg-alloc when having mul64 or mul32x16 instructions.
18891 (mulsidi3): Likewise.
18892 (umulsidi3): Likewise.
18893 (mulsi32x16): New pattern.
18894 (mulsi64): Likewise.
18895 (mulsidi64): Likewise.
18896 (umulsidi64): Likewise.
18897 (MUL32x16_REG): Define.
18898 (mul64_600): Use MUL32x16_REG.
18899 (mac64_600): Likewise.
18900 (umul64_600): Likewise.
18901 (umac64_600): Likewise.
18903 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18905 * config/arc/arc.md (mulsi3_700): Make it commutative.
18907 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18909 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18911 (*sign_extendsidi2_insn): Likewise for movstosw.
18913 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18915 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18916 the type of the input discriminant value. Convert the
18917 discriminant value of signedness vary.
18919 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18921 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18922 Add to -Wall section.
18924 2017-06-01 Richard Biener <rguenther@suse.de>
18926 PR middle-end/66313
18927 * fold-const.c (fold_plusminus_mult_expr): If the factored
18928 factor may be zero use a wrapping type for the inner operation.
18929 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18930 and handle moved defs.
18931 (process_assignment): Properly guard the unary op case. Return a
18932 tri-state indicating that moving the stmt before the call may allow
18933 to continue. Pass through to_move.
18934 (find_tail_calls): Handle moving unrelated defs before
18937 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18940 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18941 splitter result in the canonical way.
18943 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18945 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18946 also for 32bit target. Update insn attributes.
18947 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18949 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18951 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18952 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18953 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18954 (_mm_maskz_min_ss): New intrinsics.
18956 2017-05-31 Martin Liska <mliska@suse.cz>
18958 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18959 Change comment style to one we normally use.
18960 (vectorizable_reduction): Likewise.
18961 (vectorizable_induction): Likewise.
18962 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18963 (vectorizable_call): Likewise.
18964 (vectorizable_simd_clone_call): Likewise.
18965 (vectorizable_conversion): Likewise.
18966 (vectorizable_assignment): Likewise.
18967 (vectorizable_shift): Likewise.
18968 (vectorizable_operation): Likewise.
18969 (vectorizable_store): Likewise.
18970 (vectorizable_load): Likewise.
18971 * tree-vectorizer.h: Likewise.
18973 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18975 * passes.c (emergency_dump_function): New.
18976 * tree-pass.h (emergency_dump_function): Declare.
18977 * plugin.c (plugins_internal_error_function): Remove.
18978 * plugin.h (plugins_internal_error_function): Remove declaration.
18979 * toplev.c (internal_error_function): New static function. Use it...
18980 (general_init): ...here.
18982 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18984 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18985 (arc_rtx_costs): Add costs for new patterns.
18986 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18987 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18989 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18991 * tree-ssa-strlen.c (get_next_strinfo): New function.
18992 (get_stridx_plus_constant): Use it.
18993 (zero_length_string): Likewise.
18994 (adjust_related_strinfos): Likewise.
18995 (adjust_last_stmt): Likewise.
18997 2017-05-31 Richard Biener <rguenther@suse.de>
19000 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19001 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19003 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19005 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19006 loop_vinfo argument and use of dependence distance vectors.
19007 Check instead whether the two references differ only in their
19008 initial value and assume that they have the same alignment if the
19009 difference is a multiple of the vector alignment.
19010 (vect_analyze_data_refs_alignment): Update call accordingly.
19012 2017-05-31 Martin Liska <mliska@suse.cz>
19015 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19017 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19019 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19020 (create_intersect_range_checks): Move from ...
19021 * tree-data-ref.c (create_intersect_range_checks_index)
19022 (create_intersect_range_checks): ... to here.
19023 (create_runtime_alias_checks): New function factored from ...
19024 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19025 here. Call above function.
19026 * tree-data-ref.h (create_runtime_alias_checks): New function.
19028 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19030 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19031 segment length for dr_b and compute it in wide_int.
19033 2017-05-31 Richard Biener <rguenther@suse.de>
19035 PR tree-optimization/80906
19036 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19037 and pass through iv_map.
19038 (copy_bb_and_scalar_dependences): Adjust.
19039 (translate_pending_phi_nodes): Likewise.
19040 (copy_loop_close_phi_args): Handle code-generating IVs instead
19043 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19045 * diagnostic-color.c (color_dict): Add "type-diff".
19046 (parse_gcc_colors): Update comment.
19047 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19048 -fdiagnostics-show-template-tree and -fno-elide-type.
19049 (GCC_COLORS): Add type-diff to example.
19051 (-fdiagnostics-show-template-tree): New.
19052 (-fno-elide-type): New.
19053 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19054 the pp_format_decoder callback. Call any m_format_postprocessor's
19056 (pretty_printer::pretty_printer): Initialize
19057 m_format_postprocessor.
19058 (pretty_printer::~pretty_printer): Delete any
19059 m_format_postprocessor.
19060 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19061 (class format_postprocessor): New class.
19062 (struct pretty_printer::format_decoder): Document the new parameters.
19063 (struct pretty_printer::m_format_postprocessor): New field.
19064 * tree-diagnostic.c (default_tree_printer): Update for new
19065 bool and const char ** params.
19066 * tree-diagnostic.h (default_tree_printer): Likewise.
19068 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19070 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19071 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19073 rs6000_gen_cell_microcode code.
19074 (rs6000_final_prescan_insn): Delete.
19075 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19076 "warn-cell-microcode" entries.
19077 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19078 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19079 throughout. Change cc_reg_not_micro_cr0_operand to
19080 cc_reg_not_cr0_operand throughout.
19081 (*extendhi<mode>2_noload): Delete.
19082 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19083 (mwarn-cell-microcode): Delete.
19084 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19085 -mgen-cell-microcode and -mwarn-cell-microcode.
19087 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19090 * config/i386/constraints.md (Yd): New constraint.
19092 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19093 and (?Yd, r) alternatives. Update insn attributes.
19094 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19095 and (?*Yd, r) alternatives. Update insn attributes.
19096 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19098 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19100 * gimplify.c (gimplify_modify_expr): Don't create a
19101 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19104 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19106 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19108 2017-05-30 Richard Biener <rguenther@suse.de>
19110 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19111 and reduc_def fields.
19112 (STMT_VINFO_REDUC_TYPE): New define.
19113 (STMT_VINFO_REDUC_DEF): Likewise.
19114 (vect_force_simple_reduction): Adjust prototype.
19115 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19116 (vect_is_simple_reduction): Remove check_reduction argument.
19117 (vect_force_simple_reduction): Adjust and set
19118 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19119 (vectorizable_reduction): Do not re-do reduction analysis
19120 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19121 * tree-parloops.c (gather_scalar_reductions): Adjust.
19123 2017-05-30 Richard Biener <rguenther@suse.de>
19125 PR middle-end/80901
19126 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19129 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19131 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19132 Introduce unknown_misalignment parameter and remove vf.
19133 (vect_peeling_hash_get_lowest_cost):
19134 Pass unknown_misalignment parameter.
19135 (vect_enhance_data_refs_alignment):
19136 Fix unsupportable data ref treatment.
19138 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19140 * tree-vect-data-refs.c (vect_get_data_access_cost):
19141 Workaround for SLP handling.
19142 (vect_enhance_data_refs_alignment):
19143 Compute costs for doing no peeling at all, compare to the best
19144 peeling costs so far and avoid peeling if cheaper.
19146 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19148 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19149 Return peeling info and set costs to zero for unlimited cost
19151 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19152 with unknown misalignment. Compute and costs for unknown
19153 misalignment, compare them to the costs for known misalignment
19154 and choose the cheapest for peeling.
19156 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19158 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19159 (vect_get_peeling_costs_all_drs): Create function.
19160 (vect_peeling_hash_get_lowest_cost):
19161 Use vect_get_peeling_costs_all_drs.
19162 (vect_peeling_supportable): Create function.
19163 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19165 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19168 DR_HAS_NEGATIVE_STEP.
19169 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19170 (vect_enhance_data_refs_alignment): Use.
19171 (vect_duplicate_ssa_name_ptr_info): Use.
19172 * tree-vectorizer.h (dr_misalignment): Use.
19173 (known_alignment_for_access_p): Use.
19175 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19178 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19180 (has_section_name): New function.
19182 2017-05-30 Martin Liska <mliska@suse.cz>
19185 * auto-profile.c (get_function_decl_from_block): Fix
19188 2017-05-30 Richard Biener <rguenther@suse.de>
19190 PR middle-end/80876
19191 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19193 2017-05-30 Martin Liska <mliska@suse.cz>
19195 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19196 * dumpfile.h (struct dump_file_info): Remove ctors.
19198 2017-05-30 Martin Liska <mliska@suse.cz>
19200 * predict.def: Fix GNU coding style.
19202 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19204 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19205 Mark 'to' argument with ATTRIBUTE_UNUSED.
19207 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19209 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19210 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19211 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19214 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19216 * doc/install.texi (Options specification): Restore entry of
19217 --enable-sjlj-exceptions.
19219 2017-05-27 Michael Eager <eager@eagercon.com>
19222 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19224 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19226 * config/microblaze/microblaze.h
19227 (FIXED_REGISTERS): Update in macro.
19228 (CALL_USED_REGISTERS): Update in macro.
19230 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19232 * doc/install.texi: Add links to macOS binary distributions.
19234 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19238 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19240 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19242 2017-05-26 Martin Liska <mliska@suse.cz>
19244 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19246 2017-05-26 Martin Liska <mliska@suse.cz>
19248 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19249 always leading ';; '.
19250 (dump_bb_info): Likewise.
19251 (brief_dump_cfg): Likewise.
19252 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19253 * dumpfile.c: Remove usage of TDF_VERBOSE.
19254 * dumpfile.h (enum dump_kind): Likewise.
19255 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19256 * print-tree.c (debug_verbose): Remove.
19257 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19258 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19259 * tree-diagnostic.c (default_tree_printer): Replace
19260 TDF_DIAGNOSTIC with TDF_SLIM.
19262 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19264 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19265 in parameter loop, rather than loop_vinfo.
19266 (create_intersect_range_checks): Ditto.
19267 (vect_create_cond_for_alias_checks): Update call to above functions.
19269 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19271 PR tree-optimization/80815
19272 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19273 for merging runtime alias checks. Handle negative DR_STEPs.
19275 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19277 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19279 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19280 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19281 out code pruning runtime alias checks.
19282 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19283 factored out from above.
19284 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19286 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19288 (prune_runtime_alias_test_list): New decalaration.
19290 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19292 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19293 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19294 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19295 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19296 (operator==, comp_dr_with_seg_len_pair): Ditto.
19297 (vect_prune_runtime_alias_test_list): Ditto.
19299 2017-05-26 Martin Liska <mliska@suse.cz>
19302 * params.def: Bound partial-inlining-entry-probability param.
19304 2017-05-26 Marek Polacek <polacek@redhat.com>
19307 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19310 2017-05-26 Richard Biener <rguenther@suse.de>
19312 PR tree-optimization/80842
19313 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19316 2017-05-26 Richard Biener <rguenther@suse.de>
19318 PR tree-optimization/80844
19319 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19321 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19323 * doc/md.texi (Machine Constraints): Update x86 family
19324 machine constraints section to match 'config/i386/constraints.md'.
19326 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19328 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19330 2017-05-25 Nathan Sidwell <nathan@acm.org>
19332 * doc/invoke.texi (--enable-languages): Update documentation.
19334 2017-05-25 Martin Liska <mliska@suse.cz>
19336 * dumpfile.c: Add TDF_FOLDING.
19337 * dumpfile.h (enum dump_kind): Likewise.
19338 * genmatch.c (dt_simplify::gen_1): Use it.
19340 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19342 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19344 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19346 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19347 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19349 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19351 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19352 * match.pd (X == C): Rewrite it here.
19353 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19354 with_certain_nonzero_bits2): New predicates.
19355 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19357 2017-05-24 Nathan Sidwell <nathan@acm.org>
19359 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19362 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19365 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19367 * config/powerpcspe: New port. Files are copied from the rs6000
19368 port, with "rs6000" in filenames replaced by "powerpcspe".
19370 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19372 PR rtl-optimization/80754
19373 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19375 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19377 * config/sparc/sparc.md (length): Return the correct value for -mflat
19378 sibcalls to match output_sibcall.
19380 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19384 * config/rs6000/rs6000.c (struct machine_function): Add new field
19386 (rs6000_get_separate_components): Init that field, use it.
19387 (rs6000_components_for_bb): Use the field.
19389 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19391 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19393 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19395 PR middle-end/80823
19396 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19398 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19401 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19402 address against address_operand predicate.
19403 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19405 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19407 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19408 parameters passed indirectly.
19410 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19412 * config/i386/i386.md (*movdi_internal): Remove SSE4
19413 alternative 18 (?r, *v). Update insn attributes.
19414 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19415 Update insn attributes.
19416 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19417 Update insn attributes.
19418 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19419 alternative 1 (r, v). Remove isa attribute.
19420 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19421 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19422 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19424 2017-05-23 Tom de Vries <tom@codesourcery.com>
19426 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19429 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19431 * cgraphunit.c (symbol_table::process_new_functions): Update.
19432 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19433 (inline_generate_summary): Rename to ...
19434 (ipa_fn_summary_generate): ... this one.
19435 (inline_read_summary): Rename to ...
19436 (ipa_fn_summary_read): ... this one.
19437 (inline_write_summary): Rename to ...
19438 (ipa_fn_summary_write): ... this one.
19439 (inline_free_summary): Rename to ...
19440 (ipa_free_fn_summary): ... this one.
19441 (pass_data_local_fn_summary, pass_local_fn_summary,
19442 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19443 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19444 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19445 make_pass_ipa_fn_summary): New.
19446 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19447 inline_write_summary, inline_free_summary): Remove.
19448 (ipa_free_fn_summary) : New.
19449 * ipa-inline.c (ipa_inline): Update.
19450 (pass_ipa_inline): Do not generate summaries.
19451 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19453 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19454 and add pass_ipa_fn_summary.
19455 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19457 (make_pass_inline_parameters): Remove.
19459 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19461 * omp-low.c (struct omp_context): Remove "default_kind" member.
19464 * omp-offload.c (execute_oacc_device_lower): Remove the
19465 parallelism dimensions function attributes for unparallelized
19466 OpenACC kernels constructs.
19468 2017-05-23 Martin Liska <mliska@suse.cz>
19470 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19472 (cgraph_edge::make_speculative): Likewise.
19473 (cgraph_edge::resolve_speculation): Likewise.
19474 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19475 (cgraph_node::dump): Likewise.
19476 * cgraph.h: Likewise.
19477 * cgraphunit.c (analyze_functions): Likewise.
19478 (symbol_table::compile): Likewise.
19479 * ipa-cp.c (print_all_lattices): Likewise.
19480 (determine_versionability): Likewise.
19481 (initialize_node_lattices): Likewise.
19482 (ipcp_verify_propagated_values): Likewise.
19483 (estimate_local_effects): Likewise.
19484 (update_profiling_info): Likewise.
19485 (create_specialized_node): Likewise.
19486 (perhaps_add_new_callers): Likewise.
19487 (decide_about_value): Likewise.
19488 (decide_whether_version_node): Likewise.
19489 (identify_dead_nodes): Likewise.
19490 (ipcp_store_bits_results): Likewise.
19491 * ipa-devirt.c (dump_targets): Likewise.
19492 (ipa_devirt): Likewise.
19493 * ipa-icf.c (sem_item::dump): Likewise.
19494 (sem_function::equals): Likewise.
19495 (sem_variable::equals): Likewise.
19496 (sem_item_optimizer::read_section): Likewise.
19497 (sem_item_optimizer::execute): Likewise.
19498 (congruence_class::dump): Likewise.
19499 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19500 (dump_inline_summary): Likewise.
19501 (estimate_node_size_and_time): Likewise.
19502 (inline_analyze_function): Likewise.
19503 * ipa-inline-transform.c (inline_call): Likewise.
19504 * ipa-inline.c (report_inline_failed_reason): Likewise.
19505 (want_early_inline_function_p): Likewise.
19506 (edge_badness): Likewise.
19507 (update_edge_key): Likewise.
19508 (inline_small_functions): Likewise.
19509 * ipa-profile.c (ipa_profile): Likewise.
19510 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19511 (ipa_make_edge_direct_to_target): Likewise.
19512 (remove_described_reference): Likewise.
19513 (ipa_impossible_devirt_target): Likewise.
19514 (propagate_controlled_uses): Likewise.
19515 (ipa_print_node_params): Likewise.
19516 (ipcp_transform_function): Likewise.
19517 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19518 (propagate_pure_const): Likewise.
19519 * ipa-reference.c (generate_summary): Likewise.
19520 (read_write_all_from_decl): Likewise.
19521 (propagate): Likewise.
19522 (ipa_reference_read_optimization_summary): Likewise.
19523 * ipa-utils.c (ipa_merge_profiles): Likewise.
19524 * ipa.c (walk_polymorphic_call_targets): Likewise.
19525 (symbol_table::remove_unreachable_nodes): Likewise.
19526 (ipa_single_use): Likewise.
19527 * passes.c (execute_todo): Likewise.
19528 * predict.c (drop_profile): Likewise.
19529 * symtab.c (symtab_node::get_dump_name): New function.
19530 (symtab_node::dump_name): Likewise.
19531 (symtab_node::dump_asm_name): Likewise.
19532 (symtab_node::dump_references): Likewise.
19533 (symtab_node::dump_referring): Likewise.
19534 (symtab_node::dump_base): Likewise.
19535 (symtab_node::debug_symtab): Likewise.
19536 * tree-sra.c (convert_callers_for_node): Likewise.
19537 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19538 * value-prof.c (init_node_map): Likewise.
19540 2017-05-23 Martin Liska <mliska@suse.cz>
19542 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19543 and symtab_node::debug_symtab to symbol_table::debug.
19544 * cgraphunit.c (analyze_functions): Use the renamed function.
19545 (symbol_table::compile): Likewise.
19546 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19547 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19548 * passes.c (execute_todo): Likewise.
19549 * symtab.c (symbol_table::dump): New function.
19550 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19552 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19554 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19555 that nonconst implies exec.
19557 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19559 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19560 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19561 (inline_edge_summary_vec): Turn into ...
19562 (ipa_call_summaries): ... this one.
19563 (redirect_to_unreachable, edge_set_predicate,
19564 evaluate_properties_for_edge, inline_summary_alloc,
19565 reset_ipa_call_summary, reset_inline_summary,
19566 inline_summary_t::duplicate): Update.
19567 (inline_edge_duplication_hook): Turn to ...
19568 (ipa_call_summary_t::duplicate): ... this one.
19569 (inline_edge_removal_hook): Turn to ...
19570 (ipa_call_summary_t::remove): ... this one.
19571 (dump_inline_edge_summary): Turn to ...
19572 (dump_ipa_call_summary): ... this one.
19573 (estimate_function_body_sizes): Update.
19574 (inline_update_callee_summaries): Update.
19575 (remap_edge_change_prob): Update.
19576 (remap_edge_summaries): Update.
19577 (inline_merge_summary): Update.
19578 (do_estimate_edge_time): Update.
19579 (inline_generate_summary): Update.
19580 (inline_read_section): Update.
19581 (inline_read_summary): Update.
19582 (inline_free_summary): Update.
19583 * ipa-inline.c (can_inline_edge_p): Update.
19584 (compute_inlined_call_time): Update.
19585 (want_inline_small_function_p): Update.
19586 (edge_badness): Update.
19587 (early_inliner): Update.
19588 * ipa-inline.h (inline_edge_summary): Turn to ...
19589 (ipa_call_summary): ... this one.
19590 (ipa_call_summary_t): New class.
19591 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19592 (ipa_call_summaries): New.
19593 (inline_edge_summary): Remove.
19594 (estimate_edge_growth): Update.
19595 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19596 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19597 * ipa-split.c (execute_split_functions): Update.
19598 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19600 2017-05-23 Tom de Vries <tom@codesourcery.com>
19602 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19603 attributes): Document rdrand effective target.
19605 2017-05-23 Tom de Vries <tom@codesourcery.com>
19607 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19608 attributes): Sort alphabetically.
19610 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19612 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19614 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19617 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19618 V2DF/V2DI splat into two separate patterns, one that handles
19619 registers, and the other that only handles memory. Drop support
19620 for splatting from a GPR on ISA 2.07 and then splitting the
19621 splat into direct move and splat.
19622 (vsx_splat_<mode>_reg): Likewise.
19623 (vsx_splat_<mode>_mem): Likewise.
19625 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19627 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19629 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19631 PR middle-end/80809
19632 * omp-low.c (finish_taskreg_remap): New function.
19633 (finish_taskreg_scan): If unit size of ctx->record_type
19634 is non-constant, unshare the size expression and replace
19635 decls in it with possible outer var refs.
19637 PR middle-end/80809
19638 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19639 GOVD_SHARED rather than GOVD_PRIVATE with it.
19640 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19641 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19643 PR middle-end/80853
19644 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19645 as last argument to build_outer_var_ref for pointer bases of array
19646 section reductions.
19648 2017-05-19 Martin Sebor <msebor@redhat.com>
19650 * print-tree.c (print_node): Print DECL_READ_P flag.
19652 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19654 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19655 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19656 * cgraph.c: Likewise.
19657 * cgraphunit.c: Likewise.
19658 * gengtype.c: Likewise.
19659 * ipa-cp.c: Likewise.
19660 * ipa-devirt.c: Likewise.
19661 * ipa-icf.c: Likewise.
19662 * ipa-predicate.c: Likewise.
19663 * ipa-profile.c: Likewise.
19664 * ipa-prop.c: Likewise.
19665 * ipa-split.c: Likewise.
19667 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19668 edge_predicate_pool, dump_inline_hints,
19669 inline_summary::account_size_time, redirect_to_unreachable,
19670 edge_set_predicate, set_hint_predicate,
19671 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19672 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19673 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19674 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19675 ipa_call_summary_t::remove, initialize_growth_caches,
19676 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19677 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19678 mark_modified, unmodified_parm_1, unmodified_parm,
19679 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19680 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19681 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19682 will_be_nonconstant_predicate, record_modified_bb_info,
19683 get_minimal_bb, record_modified, param_change_prob,
19684 phi_result_unknown_predicate, predicate_for_phi_result,
19685 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19686 estimate_function_body_sizes, compute_inline_parameters,
19687 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19688 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19689 inline_update_callee_summaries, remap_edge_change_prob,
19690 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19691 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19692 inline_analyze_function, inline_summary_t::insert,
19693 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19694 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19695 inline_free_summary): Move to ipa-fnsummary.h
19696 (predicate_t): Remove.
19697 * ipa-fnsummary.c: New file.
19698 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19699 (enum inline_hints_vals, inline_hints, agg_position_info,
19700 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19701 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19702 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19703 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19704 inline_read_summary, inline_write_summary, inline_free_summary,
19705 inline_analyze_function, initialize_inline_failed,
19706 inline_merge_summary, inline_update_overall_summary,
19707 compute_inline_parameters): Move to ipa-fnsummary.h
19708 * ipa-fnsummary.h: New file.
19709 * ipa-inline-transform.h: Include ipa-inline.h.
19710 * ipa-inline.c: LIkewise.
19712 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19714 * ipa-inline.c (edge_badness): Use inlined_time instead of
19715 inline_summaries->get.
19717 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19719 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19721 2017-05-22 Nathan Sidwell <nathan@acm.org>
19723 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19724 (fdump-lang): Document 'raw' option.
19725 * dumpfile.h (TDI_tu): Delete.
19726 * dumpfile.c (dump_files): Remove translation-unit.
19727 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19729 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19731 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19732 command option from $(AWK) call.
19733 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19734 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19735 [FORMAT]: Remove handling of variable.
19736 * config/avr/t-multilib: Regenerate.
19738 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19740 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19742 (dump_inline_summary): Do not print self_time.
19743 (estimate_function_body_sizes): Do not set self_time.
19744 (compute_inline_parameters): Likewise.
19745 (inline_read_section, inline_write_summary): Do not stream self_time.
19746 * ipa-inline.h (inline_summary): Drop self_time.
19748 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19750 * ipa-inline-analysis.c (account_size_time): Rename to ...
19751 (inline_summary::account_size_time): ... this one.
19752 (reset_ipa_call_summary): Turn to ...
19753 (ipa_call_summary::reset): ... this one.
19754 (reset_inline_summary): Turn to ...
19755 (inline_summary::reset): ... this one.
19756 (inline_summary_t::remove): Update.
19757 (inline_summary_t::duplicate): Update.
19758 (ipa_call_summary_t::remove): Update.
19759 (dump_inline_summary): Update.
19760 (estimate_function_body_sizes): Update.
19761 (compute_inline_parameters): Update.
19762 (estimate_node_size_and_time): Update.
19763 (inline_merge_summary): Update.
19764 (inline_update_overall_summary): Update.
19765 (inline_read_section): Update.
19766 (inline_write_summary): Update.
19767 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19768 add account_size_time and reset member functions.
19769 (ipa_call_summary): Add reset function.
19770 * ipa-predicate.h (predicate::operator &): Constify.
19772 2017-05-22 Richard Biener <rguenther@suse.de>
19774 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19776 2017-05-19 Jason Merrill <jason@redhat.com>
19778 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19780 2017-05-19 Marek Polacek <polacek@redhat.com>
19783 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19784 TYPE_OVERFLOW_WRAPS checks.
19786 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19788 * tree-core.h (enum omp_clause_default_kind): Add
19789 "OMP_CLAUSE_DEFAULT_PRESENT".
19790 * tree-pretty-print.c (dump_omp_clause): Handle it.
19791 * gimplify.c (enum gimplify_omp_var_data): Add
19792 "GOVD_MAP_FORCE_PRESENT".
19793 (gimplify_adjust_omp_clauses_1): Map it to
19794 "GOMP_MAP_FORCE_PRESENT".
19795 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19797 * gimplify.c (oacc_default_clause): Clarify.
19799 2017-05-19 Nathan Sidwell <nathan@acm.org>
19801 LANG_HOOK_REGISTER_DUMPS
19802 * toplev.c (general_init): Call register dump lang hook.
19803 * doc/invoke.texi: Document -fdump-lang option family.
19804 * dumpfile.c (dump_files): Remove class dump here.
19805 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19806 * dumpfile.h (tree_dump_index): Remove TDI_class.
19807 * langhooks-def.h (lhd_register_dumps): Declare.
19808 (LANG_HOOKS_REGISTER_DUMPS): Define.
19809 (LANG_HOOKS_INITIALIZER): Add it.
19810 * langhooks.c (lhd_register_dumps): Define.
19811 * langhooks.h (struct lang_hooks): Add register_dumps.
19813 2017-05-19 Nathan Sidwell <nathan@acm.org>
19815 * context.h (context::set_passes): New.
19816 * context.c (context::context): Do not create pass manager.
19817 * toplev.c (general_init): Create pass manager here.
19819 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19821 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19822 use this splitter if two add or or instructions would also work for
19823 the constant we want to generate.
19825 2017-05-19 Richard Biener <rguenther@suse.de>
19828 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19829 predicate evaluation.
19831 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19833 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19835 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19838 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19840 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19841 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19842 (inline_edge_summary_vec): Turn into ...
19843 (ipa_call_summaries): ... this one.
19844 (redirect_to_unreachable, edge_set_predicate,
19845 evaluate_properties_for_edge, inline_summary_alloc,
19846 reset_ipa_call_summary, reset_inline_summary,
19847 inline_summary_t::duplicate): Update.
19848 (inline_edge_duplication_hook): Turn to ...
19849 (ipa_call_summary_t::duplicate): ... this one.
19850 (inline_edge_removal_hook): Turn to ...
19851 (ipa_call_summary_t::remove): ... this one.
19852 (dump_inline_edge_summary): Turn to ...
19853 (dump_ipa_call_summary): ... this one.
19854 (estimate_function_body_sizes): Update.
19855 (inline_update_callee_summaries): Update.
19856 (remap_edge_change_prob): Update.
19857 (remap_edge_summaries): Update.
19858 (inline_merge_summary): Update.
19859 (do_estimate_edge_time): Update.
19860 (inline_generate_summary): Update.
19861 (inline_read_section): Update.
19862 (inline_read_summary): Update.
19863 (inline_free_summary): Update.
19864 * ipa-inline.c (can_inline_edge_p): Update.
19865 (compute_inlined_call_time): Update.
19866 (want_inline_small_function_p): Update.
19867 (edge_badness): Update.
19868 (early_inliner): Update.
19869 * ipa-inline.h (inline_edge_summary): Turn to ...
19870 (ipa_call_summary): ... this one.
19871 (ipa_call_summary_t): New class.
19872 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19873 (ipa_call_summaries): New.
19874 (inline_edge_summary): Remove.
19875 (estimate_edge_growth): Update.
19876 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19877 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19878 * ipa-split.c (execute_split_functions): Update.
19879 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19881 2017-05-19 Richard Biener <rguenther@suse.de>
19883 PR middle-end/80764
19884 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19886 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19888 * config/rs6000/rs6000.c (struct machine_function): Add field
19889 fpr_is_wrapped_separately.
19890 (rs6000_get_separate_components): Use 64 components. Handle the
19891 new FPR components.
19892 (rs6000_components_for_bb): Handle the FPR components.
19893 (rs6000_emit_prologue_components): Handle the FPR components.
19894 (rs6000_emit_epilogue_components): Handle the FPR components.
19895 (rs6000_set_handled_components): Handle the FPR components.
19896 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19897 that are already separately shrink-wrapped.
19898 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19899 that are already separately shrink-wrapped.
19901 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19904 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19907 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19908 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19909 cases where the register allocator uses a move and an offsettable
19910 memory operation to/from a FPR register on ISA 2.06/2.07.
19911 (define_peephole2 for Altivec d-form store): Likewise.
19913 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19916 * config/i386/mmx.md (*mov<mode>_internal): Enable
19917 alternatives 11, 12, 13 and 14 also for 32bit targets.
19918 Remove alternatives 15, 16, 17 and 18.
19919 * config/i386/sse.md (vec_concatv2di): Change
19920 alternative (!x, *y) to (x, ?!*Yn).
19922 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19924 * dumpfile.h (enum dump_kind): Remove stray comma.
19926 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19928 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19929 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19930 predicate::num_conditions
19931 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19932 (CHANGED): turn into predicate::changed.
19933 (agg_position_info): Move to ipa-predicate.h
19934 (add_condition, predicate::add_clause, predicate::operator &=,
19935 predicate::or_with, predicate::evaluate, predicate::probability,
19936 dump_condition, dump_clause, predicate::dump,
19937 predicate::remap_after_duplication, predicate::remap_after_inlining,
19938 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19939 (evaluate_conditions_for_known_args): Update.
19940 (set_cond_stmt_execution_predicate): Update.
19941 * ipa-inline.h: Include ipa-predicate.h
19942 (condition, inline_param_summary, conditions, agg_position_info,
19943 predicate): Move to ipa-predicate.h
19944 * ipa-predicate.c: New file.
19945 * ipa-predicate.h: New file.
19947 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19949 * final.c (leaf_function_p): Check we are not in a sequence.
19951 2017-05-18 Martin Liska <mliska@suse.cz>
19953 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19954 * dumpfile.c (dump_register): Use new enum dump_kind.
19955 (get_dump_file_name): Likewise.
19956 (dump_enable_all): Likewise.
19957 (dump_switch_p_1): Likewise.
19958 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19959 * dumpfile.h (enum dump_kind): New enum type.
19960 (struct dump_file_info): Create constructor and
19961 format fields and comments.
19962 * passes.c (pass_manager::register_one_dump_file):
19964 * statistics.c (statistics_early_init): Likewise.
19965 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19966 TDF_TREE with TDF_SLIM.
19967 (gather_memory_references_ref): Likewise.
19969 2017-05-18 Martin Liska <mliska@suse.cz>
19971 * vec.h (struct vnull): Use it.
19973 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19975 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19976 (true_predicate, false_predicate, true_predicate_p,
19977 false_predicate_p): Remove.
19978 (single_cond_predicate, not_inlined_predicate): Turn to member function
19980 (add_condition): Update.
19981 (add_clause): Turn to...
19982 (predicate::add_clause): ... this one; update; allow passing NULL
19984 (and_predicates): Turn to ...
19985 (predicate::operator &=): ... this one.
19986 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19987 (or_predicates): Turn to ...
19988 (predicate::or_with): ... this one.
19989 (evaluate_predicate): Turn to ...
19990 (predicate::evaluate): ... this one.
19991 (predicate_probability): Turn to ...
19992 (predicate::probability): ... this one.
19993 (dump_condition): Update.
19994 (dump_predicate): Turn to ...
19995 (predicate::dump): ... this one.
19996 (account_size_time): Update.
19997 (edge_set_predicate): Update.
19998 (set_hint_predicate): UPdate.
19999 (evaluate_conditions_for_known_args): Update.
20000 (evaluate_properties_for_edge): Update.
20001 (remap_predicate_after_duplication): Turn to...
20002 (predicate::remap_after_duplication): ... this one.
20003 (remap_hint_predicate_after_duplication): Update.
20004 (inline_summary_t::duplicate): UPdate.
20005 (dump_inline_edge_summary): Update.
20006 (dump_inline_summary): Update.
20007 (set_cond_stmt_execution_predicate): Update.
20008 (set_switch_stmt_execution_predicate): Update.
20009 (compute_bb_predicates): Update.
20010 (will_be_nonconstant_expr_predicate): Update.
20011 (will_be_nonconstant_predicate): Update.
20012 (phi_result_unknown_predicate): Update.
20013 (predicate_for_phi_result): Update.
20014 (array_index_predicate): Update.
20015 (estimate_function_body_sizes): Update.
20016 (estimate_node_size_and_time): Update.
20017 (estimate_ipcp_clone_size_and_time): Update.
20018 (remap_predicate): Rename to ...
20019 (predicate::remap_after_inlining): ... this one.
20020 (remap_hint_predicate): Update.
20021 (inline_merge_summary): Update.
20022 (inline_update_overall_summary): Update.
20023 (estimate_size_after_inlining): Update.
20024 (read_predicate): Rename to ...
20025 (predicate::stream_in): ... this one.
20026 (read_inline_edge_summary): Update.
20027 (write_predicate): Rename to ...
20028 (predicate::stream_out): ... this one.
20029 (write_inline_edge_summary): Update.
20030 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20031 (clause_t): Turn to uint32_t
20032 (predicate): Turn to class; implement constructor and operators
20034 (size_time_entry): Update.
20035 (inline_summary): Update.
20036 (inline_edge_summary): Update.
20038 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20040 * fold-const.c (fold_binary_loc): Move transformation...
20041 * match.pd (C - X CMP X): ... here.
20043 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20045 * config/sparc/sparc.c (sparc_option_override): Set function
20046 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20047 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20049 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20051 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20053 2017-05-18 Marek Polacek <polacek@redhat.com>
20056 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20057 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20058 gimple_assign_load_p.
20060 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20062 PR middle-end/80692
20063 * real.c (do_compare): Give decimal_do_compare preference over
20064 comparing just the signs.
20066 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20068 * doc/md.texi (Canonicalization of Instructions): Describe the
20069 canonical form of instructions that inherently set a condition
20072 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20074 PR middle-end/80775
20075 * tree-cfg.c: Move deletion of unreachable case statements to after
20076 the merging of consecutive case labels.
20078 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20080 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20081 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20082 restoring of callee-saved registers.
20084 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20086 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20087 * config/visium/visium.c (single_set_and_flags): Likewise.
20088 * config/visium/visium.md (Substitutions): Likewise.
20090 2017-05-17 Martin Liska <mliska@suse.cz>
20092 * cfg.c: Introduce dump_flags_t type and
20093 use it instead of int type.
20095 * cfghooks.c: Likewise.
20096 * cfghooks.h (struct cfg_hooks): Likewise.
20097 * cfgrtl.c: Likewise.
20098 * cfgrtl.h: Likewise.
20099 * cgraph.c (cgraph_node::get_body): Likewise.
20100 * coretypes.h: Likewise.
20101 * domwalk.c: Likewise.
20102 * domwalk.h: Likewise.
20103 * dumpfile.c (struct dump_option_value_info): Likewise.
20104 (dump_enable_all): Likewise.
20105 (dump_switch_p_1): Likewise.
20106 (opt_info_switch_p): Likewise.
20107 * dumpfile.h (enum tree_dump_index): Likewise.
20108 (struct dump_file_info): Likewise.
20109 * genemit.c: Likewise.
20110 * generic-match-head.c: Likewise.
20111 * gengtype.c (open_base_files): Likewise.
20112 * gimple-pretty-print.c: Likewise.
20113 * gimple-pretty-print.h: Likewise.
20114 * graph.c (print_graph_cfg): Likewise.
20115 * graphite-scop-detection.c (dot_all_sese): Likewise.
20116 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20117 * loop-unroll.c (report_unroll): Likewise.
20118 * passes.c (pass_manager::register_one_dump_file): Likewise.
20119 * print-tree.c: Likewise.
20120 * statistics.c: Likewise.
20121 * tree-cfg.c: Likewise.
20122 * tree-cfg.h: Likewise.
20123 * tree-dfa.c: Likewise.
20124 * tree-dfa.h: Likewise.
20125 * tree-dump.c (dump_function): Likewise.
20126 * tree-dump.h (struct dump_info): Likewise.
20127 * tree-pretty-print.c: Likewise.
20128 * tree-pretty-print.h: Likewise.
20129 * tree-ssa-live.c: Likewise.
20130 * tree-ssa-live.h: Likewise.
20131 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20132 * tree-vect-loop.c: Likewise.
20133 * tree-vect-slp.c: Likewise.
20135 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20136 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20138 PR tree-optimization/80457
20139 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20140 of all arguments to a statement as scalar_to_vec operations.
20141 (vectorizable_call): Adjust call to vect_model_simple_cost for
20143 (vectorizable_conversion): Likewise.
20144 (vectorizable_assignment): Likewise.
20145 (vectorizable_shift): Likewise.
20146 (vectorizable_operation): Likewise.
20147 (vectorizable_comparison): Likewise.
20148 (vect_is_simple_cond): Record the def types for operands.
20149 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20150 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20151 for statement argument count.
20153 2017-05-16 Carl Love <cel@us.ibm.com>
20155 * config/rs6000/rs6000-c: Add support for built-in functions
20156 vector unsigned long long vec_bperm (vector unsigned long long,
20157 vector unsigned char)
20158 vector signed long long vec_mule (vector signed int,
20160 vector unsigned long long vec_mule (vector unsigned int,
20161 vector unsigned int)
20162 vector signed long long vec_mulo (vector signed int,
20164 vector unsigned long long vec_mulo (vector unsigned int,
20165 vector unsigned int)
20166 vector signed char vec_sldw (vector signed char,
20167 vector signed char,
20169 vector unsigned char vec_sldw (vector unsigned char,
20170 vector unsigned char,
20172 vector signed short vec_sldw (vector signed short,
20173 vector signed short,
20175 vector unsigned short vec_sldw (vector unsigned short,
20176 vector unsigned short,
20178 vector signed int vec_sldw (vector signed int,
20181 vector unsigned int vec_sldw (vector unsigned int,
20182 vector unsigned int,
20184 vector signed long long vec_sldw (vector signed long long,
20185 vector signed long long,
20187 vector unsigned long long vec_sldw (vector unsigned long long,
20188 vector unsigned long long,
20190 * config/rs6000/rs6000-c: Add support for built-in functions
20191 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20192 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20193 * doc/extend.texi: Update the built-in documentation for the
20194 new built-in functions.
20196 2017-05-16 Marek Polacek <polacek@redhat.com>
20200 * tree.c (save_expr): Don't fold the expression.
20202 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20204 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20205 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20206 and (?*y,m). Update insn attributes.
20208 2017-05-16 Martin Liska <mliska@suse.cz>
20210 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20211 flags argument of print_gimple_stmt, print_gimple_expr,
20212 print_generic_stmt and print_generic_expr.
20213 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20214 * coretypes.h: Likewise.
20215 * except.c (dump_eh_tree): Likewise.
20216 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20217 * gimple-pretty-print.h: Likewise.
20218 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20219 (backprop::push_to_worklist): Likewise.
20220 (backprop::pop_from_worklist): Likewise.
20221 (backprop::process_use): Likewise.
20222 (backprop::intersect_uses): Likewise.
20223 (note_replacement): Likewise.
20224 * gimple-ssa-store-merging.c
20225 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20226 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20227 (pass_store_merging::execute): Likewise.
20228 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20229 (ssa_base_cand_dump_callback): Likewise.
20230 (dump_incr_vec): Likewise.
20231 (replace_refs): Likewise.
20232 (replace_mult_candidate): Likewise.
20233 (create_add_on_incoming_edge): Likewise.
20234 (create_phi_basis): Likewise.
20235 (insert_initializers): Likewise.
20236 (all_phi_incrs_profitable): Likewise.
20237 (introduce_cast_before_cand): Likewise.
20238 (replace_one_candidate): Likewise.
20239 * gimplify.c (gimplify_expr): Likewise.
20240 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20241 (set_rename): Likewise.
20242 (rename_uses): Likewise.
20243 (copy_loop_phi_nodes): Likewise.
20244 (add_close_phis_to_merge_points): Likewise.
20245 (copy_loop_close_phi_args): Likewise.
20246 (copy_cond_phi_args): Likewise.
20247 (graphite_copy_stmts_from_block): Likewise.
20248 (translate_pending_phi_nodes): Likewise.
20249 * graphite-poly.c (print_pdr): Likewise.
20250 (dump_gbb_cases): Likewise.
20251 (dump_gbb_conditions): Likewise.
20252 (print_scop_params): Likewise.
20253 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20254 (build_cross_bb_scalars_use): Likewise.
20255 (gather_bbs::before_dom_children): Likewise.
20256 * hsa-dump.c (dump_hsa_immed): Likewise.
20257 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20258 (get_replacement_map): Likewise.
20259 * ipa-inline-analysis.c (dump_condition): Likewise.
20260 (estimate_function_body_sizes): Likewise.
20261 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20262 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20263 * ipa-prop.c (ipa_dump_param): Likewise.
20264 (ipa_print_node_jump_functions_for_edge): Likewise.
20265 (ipa_modify_call_arguments): Likewise.
20266 (ipa_modify_expr): Likewise.
20267 (ipa_dump_param_adjustments): Likewise.
20268 (ipa_dump_agg_replacement_values): Likewise.
20269 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20270 * ipa-pure-const.c (check_stmt): Likewise.
20271 (pass_nothrow::execute): Likewise.
20272 * ipa-split.c (execute_split_functions): Likewise.
20273 * omp-offload.c (dump_oacc_loop_part): Likewise.
20274 (dump_oacc_loop): Likewise.
20275 * trans-mem.c (tm_log_emit): Likewise.
20276 (tm_memopt_accumulate_memops): Likewise.
20277 (dump_tm_memopt_set): Likewise.
20278 (dump_tm_memopt_transform): Likewise.
20279 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20280 (print_loop): Likewise.
20281 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20282 (chkp_gather_checks_info): Likewise.
20283 (chkp_get_check_result): Likewise.
20284 (chkp_remove_check_if_pass): Likewise.
20285 (chkp_use_outer_bounds_if_possible): Likewise.
20286 (chkp_reduce_bounds_lifetime): Likewise.
20287 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20288 (chkp_mark_completed_bounds): Likewise.
20289 (chkp_register_incomplete_bounds): Likewise.
20290 (chkp_mark_invalid_bounds): Likewise.
20291 (chkp_maybe_copy_and_register_bounds): Likewise.
20292 (chkp_build_returned_bound): Likewise.
20293 (chkp_get_bound_for_parm): Likewise.
20294 (chkp_build_bndldx): Likewise.
20295 (chkp_get_bounds_by_definition): Likewise.
20296 (chkp_generate_extern_var_bounds): Likewise.
20297 (chkp_get_bounds_for_decl_addr): Likewise.
20298 * tree-chrec.c (chrec_apply): Likewise.
20299 * tree-data-ref.c (dump_data_reference): Likewise.
20300 (dump_subscript): Likewise.
20301 (dump_data_dependence_relation): Likewise.
20302 (analyze_overlapping_iterations): Likewise.
20303 * tree-inline.c (expand_call_inline): Likewise.
20304 (tree_function_versioning): Likewise.
20305 * tree-into-ssa.c (dump_defs_stack): Likewise.
20306 (dump_currdefs): Likewise.
20307 (dump_names_replaced_by): Likewise.
20308 (dump_update_ssa): Likewise.
20309 (update_ssa): Likewise.
20310 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20311 * tree-parloops.c (build_new_reduction): Likewise.
20312 (try_create_reduction_list): Likewise.
20313 (ref_conflicts_with_region): Likewise.
20314 (oacc_entry_exit_ok_1): Likewise.
20315 (oacc_entry_exit_single_gang): Likewise.
20316 * tree-pretty-print.h: Likewise.
20317 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20318 (get_scalar_evolution): Likewise.
20319 (add_to_evolution): Likewise.
20320 (get_loop_exit_condition): Likewise.
20321 (analyze_evolution_in_loop): Likewise.
20322 (analyze_initial_condition): Likewise.
20323 (analyze_scalar_evolution): Likewise.
20324 (instantiate_scev): Likewise.
20325 (number_of_latch_executions): Likewise.
20326 (gather_chrec_stats): Likewise.
20327 (final_value_replacement_loop): Likewise.
20328 (scev_const_prop): Likewise.
20329 * tree-sra.c (dump_access): Likewise.
20330 (disqualify_candidate): Likewise.
20331 (create_access): Likewise.
20332 (reject): Likewise.
20333 (maybe_add_sra_candidate): Likewise.
20334 (create_access_replacement): Likewise.
20335 (analyze_access_subtree): Likewise.
20336 (analyze_all_variable_accesses): Likewise.
20337 (sra_modify_assign): Likewise.
20338 (initialize_constant_pool_replacements): Likewise.
20339 (find_param_candidates): Likewise.
20340 (decide_one_param_reduction): Likewise.
20341 (replace_removed_params_ssa_names): Likewise.
20342 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20343 * tree-ssa-copy.c (dump_copy_of): Likewise.
20344 (copy_prop_visit_cond_stmt): Likewise.
20345 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20346 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20347 (record_equivalences_from_stmt): Likewise.
20348 * tree-ssa-dse.c (compute_trims): Likewise.
20349 (delete_dead_call): Likewise.
20350 (delete_dead_assignment): Likewise.
20351 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20352 (forward_propagate_into_cond): Likewise.
20353 (pass_forwprop::execute): Likewise.
20354 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20355 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20357 (move_computations_worker): Likewise.
20358 (execute_sm): Likewise.
20359 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20360 (remove_exits_and_undefined_stmts): Likewise.
20361 (remove_redundant_iv_tests): Likewise.
20362 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20363 (adjust_iv_update_pos): Likewise.
20364 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20365 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20366 (value_replacement): Likewise.
20367 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20368 * tree-ssa-pre.c (print_pre_expr): Likewise.
20369 (get_representative_for): Likewise.
20370 (create_expression_by_pieces): Likewise.
20371 (insert_into_preds_of_block): Likewise.
20372 (eliminate_insert): Likewise.
20373 (eliminate_dom_walker::before_dom_children): Likewise.
20374 (eliminate): Likewise.
20375 (remove_dead_inserted_code): Likewise.
20376 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20377 * tree-ssa-reassoc.c (get_rank): Likewise.
20378 (eliminate_duplicate_pair): Likewise.
20379 (eliminate_plus_minus_pair): Likewise.
20380 (eliminate_not_pairs): Likewise.
20381 (undistribute_ops_list): Likewise.
20382 (eliminate_redundant_comparison): Likewise.
20383 (update_range_test): Likewise.
20384 (optimize_range_tests_var_bound): Likewise.
20385 (optimize_vec_cond_expr): Likewise.
20386 (rewrite_expr_tree): Likewise.
20387 (rewrite_expr_tree_parallel): Likewise.
20388 (linearize_expr): Likewise.
20389 (break_up_subtract): Likewise.
20390 (linearize_expr_tree): Likewise.
20391 (attempt_builtin_powi): Likewise.
20392 (attempt_builtin_copysign): Likewise.
20393 (transform_stmt_to_copy): Likewise.
20394 (transform_stmt_to_multiply): Likewise.
20395 (dump_ops_vector): Likewise.
20396 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20397 (print_scc): Likewise.
20398 (set_ssa_val_to): Likewise.
20399 (visit_reference_op_store): Likewise.
20400 (visit_use): Likewise.
20401 (sccvn_dom_walker::before_dom_children): Likewise.
20402 (run_scc_vn): Likewise.
20403 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20405 (expr_hash_elt::print): Likewise.
20406 (const_and_copies::pop_to_marker): Likewise.
20407 (const_and_copies::record_const_or_copy_raw): Likewise.
20408 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20409 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20410 (dump_predicates): Likewise.
20411 (find_uninit_use): Likewise.
20412 (warn_uninitialized_phi): Likewise.
20413 (pass_late_warn_uninitialized::execute): Likewise.
20414 * tree-ssa.c (verify_vssa): Likewise.
20415 (verify_ssa): Likewise.
20416 (maybe_optimize_var): Likewise.
20417 * tree-vrp.c (dump_value_range): Likewise.
20418 (dump_all_value_ranges): Likewise.
20419 (dump_asserts_for): Likewise.
20420 (register_edge_assert_for_2): Likewise.
20421 (vrp_visit_cond_stmt): Likewise.
20422 (vrp_visit_switch_stmt): Likewise.
20423 (vrp_visit_stmt): Likewise.
20424 (vrp_visit_phi_node): Likewise.
20425 (simplify_cond_using_ranges_1): Likewise.
20426 (fold_predicate_in): Likewise.
20427 (evrp_dom_walker::before_dom_children): Likewise.
20428 (evrp_dom_walker::push_value_range): Likewise.
20429 (evrp_dom_walker::pop_value_range): Likewise.
20430 (execute_early_vrp): Likewise.
20432 2017-05-16 Richard Biener <rguenther@suse.de>
20434 * dwarf2out.c (loc_list_from_tree_1): Do not create
20435 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20437 2017-05-16 Richard Biener <rguenther@suse.de>
20439 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20441 (note_variable_value_in_expr): If we resolved the decl ref
20442 do not push to the stack.
20444 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20446 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20447 operations in fast-math mode.
20448 (vaddq_f16): Likewise.
20449 (vmul_f16): Likewise.
20450 (vmulq_f16): Likewise.
20451 (vsub_f16): Likewise.
20452 (vsubq_f16): Likewise.
20453 * config/arm/neon.md (add<mode>3): New.
20455 (fma:<VH:mode>3): New. Also remove outdated comment.
20458 2017-05-16 Martin Liska <mliska@suse.cz>
20462 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20463 (odr_types_equivalent_p): Likewise.
20465 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20467 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20469 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20472 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20473 non-interunit SSE move alternatives with '?'.
20474 (zero-extendsidi peephole2): New peephole to skip intermediate
20475 general register in SSE zero-extend sequence.
20477 2017-05-15 Jeff Law <law@redhat.com>
20479 * reorg.c (relax_delay_slots): Create a new variable to hold
20480 the temporary target rather than clobbering TARGET_LABEL.
20482 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20483 missing argument to extract_bit_field call.
20484 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20486 2017-05-15 Martin Liska <mliska@suse.cz>
20489 * gcc.c (process_command): Do not allow empty argument of -o option.
20491 2017-05-15 Renlin Li <renlin.li@arm.com>
20493 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20494 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20495 * config/aarch64/constraints.md (Usf): Add long call check.
20496 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20497 (call_value): Likewise.
20498 (sibcall): Likewise.
20499 (sibcall_value): Likewise.
20501 (call_value_insn): New.
20502 (sibcall_insn): Update rtx pattern.
20503 (sibcall_value_insn): Likewise.
20504 (call_internal): Remove.
20505 (call_value_internal): Likewise.
20506 (sibcall_internal): Likewise.
20507 (sibcall_value_internal): Likewise.
20508 (call_reg): Likewise.
20509 (call_symbol): Likewise.
20510 (call_value_reg): Likewise.
20511 (call_value_symbol): Likewise.
20513 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20516 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20518 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20520 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20521 compatible with CCGOCmode and with CCZmode.
20523 2017-05-14 Martin Sebor <msebor@redhat.com>
20525 PR middle-end/77671
20526 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20527 (gimple_fold_builtin_snprintf): Same.
20528 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20529 (gimple_fold_builtin_snprintf): Same.
20530 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20531 of character types.
20532 (is_call_safe): New function.
20533 (try_substitute_return_value): Call it.
20534 (try_simplify_call): New function.
20535 (pass_sprintf_length::handle_gimple_call): Call it.
20537 2017-05-14 Martin Sebor <msebor@redhat.com>
20539 PR middle-end/80669
20540 * builtins.c (expand_builtin_stpncpy): Simplify.
20542 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20544 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20545 * config/i386/i386.h
20546 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20547 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20548 (struct machine_function): Add new members call_ms2sysv,
20549 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20550 (struct machine_frame_state): New fields sp_realigned and
20551 sp_realigned_offset.
20552 * config/i386/i386.c
20553 (enum xlogue_stub): New enum.
20554 (enum xlogue_stub_sets): New enum.
20555 (class xlogue_layout): New class.
20556 (struct ix86_frame): New fields stack_realign_allocate_offset,
20557 stack_realign_offset and outlined_save_offset. Modify comments to
20558 detail stack layout when using out-of-line stubs.
20559 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20560 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20561 -mcall-ms2sysv-xlogues.
20562 (stub_managed_regs): New static variable.
20563 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20564 registers managed by out-of-line stub.
20565 (disable_call_ms2sysv_xlogues): New function.
20566 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20567 m->call_ms2sysv when appropriate and compute frame layout for
20569 (sp_valid_at, fp_valid_at): New inline functions.
20570 (choose_basereg): New function.
20571 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20573 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20574 Use align parameter of choose_baseaddr to generated aligned SSE movs
20576 (pro_epilogue_adjust_stack): Modify to track
20577 machine_frame_state::sp_realigned.
20578 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20579 (ix86_nsaved_sseregs): Likewise.
20580 (ix86_emit_save_regs): Likewise.
20581 (ix86_emit_save_regs_using_mov): Likewise.
20582 (ix86_emit_save_sse_regs_using_mov): Likewise.
20583 (get_scratch_register_on_entry): Likewise.
20584 (gen_frame_set): New function.
20585 (gen_frame_load): Likewise.
20586 (gen_frame_store): Likewise.
20587 (emit_outlined_ms2sysv_save): Likewise.
20588 (emit_outlined_ms2sysv_restore): Likewise.
20589 (ix86_expand_prologue): Modify stack re-alignment code and call
20590 emit_outlined_ms2sysv_save when appropriate.
20591 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20592 parameter rtx_insn *insn, which allows the function to be used to only
20593 generate the notes.
20594 (ix86_expand_epilogue): Modify validity checks of frame and stack
20595 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20596 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20597 * config/i386/predicates.md
20598 (save_multiple): New predicate.
20599 (restore_multiple): Likewise.
20600 * config/i386/sse.md
20601 (save_multiple<mode>): New pattern.
20602 (save_multiple_realign<mode>): Likewise.
20603 (restore_multiple<mode>): Likewise.
20604 (restore_multiple_and_return<mode>): Likewise.
20605 (restore_multiple_leave_return<mode>): Likewise.
20606 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20608 2017-05-14 Julia Koval <julia.koval@intel.com>
20610 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20611 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20612 (__builtin_ia32_xsetbv): New builtins.
20613 * config/i386/i386.c (ix86_expand_special_args_builtin):
20615 (ix86_expand_builtin): Special expand for new intrinsics.
20616 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20617 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20618 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20620 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20622 * cfganal.c (inverted_post_order_compute): Change argument type
20624 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20625 * df-core.c (rest_of_handle_df_initialize): Adjust.
20626 (rest_of_handle_df_finish): Likewise.
20627 (df_analyze_1): Likewise.
20628 (df_analyze): Likewise.
20629 (loop_inverted_post_order_compute): Change argument to be a vec *.
20630 (df_analyze_loop): Adjust.
20631 (df_get_n_blocks): Likewise.
20632 (df_get_postorder): Likewise.
20633 * df.h (struct df_d): Change field to be a vec.
20634 * lcm.c (compute_laterin): Adjust.
20635 (compute_available): Likewise.
20636 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20637 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20638 * tree-ssa-pre.c (compute_antic): Likewise.
20640 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20642 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20643 (depth_first_search::depth_first_search): Change structure init
20644 function to this constructor.
20645 (depth_first_search::add_bb): Rename function to this member.
20646 (depth_first_search::execute): Likewise.
20647 (flow_dfs_compute_reverse_finish): Adjust.
20649 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20651 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20652 (longest_simple_path): Likewise.
20653 * shrink-wrap.c (spread_components): Likewise.
20654 (disqualify_problematic_components): Likewise.
20655 (emit_common_heads_for_components): Likewise.
20656 (emit_common_tails_for_components): Likewise.
20657 (insert_prologue_epilogue_for_components): Likewise.
20659 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20661 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20666 * df-core.c (df_set_blocks): Start using auto_bitmap.
20667 (df_compact_blocks): Likewise.
20668 * df-problems.c (df_rd_confluence_n): Likewise.
20669 * df-scan.c (df_insn_rescan_all): Likewise.
20670 (df_process_deferred_rescans): Likewise.
20671 (df_update_entry_block_defs): Likewise.
20672 (df_update_exit_block_uses): Likewise.
20673 (df_entry_block_bitmap_verify): Likewise.
20674 (df_exit_block_bitmap_verify): Likewise.
20675 (df_scan_verify): Likewise.
20676 * lra-constraints.c (lra_constraints): Likewise.
20677 (undo_optional_reloads): Likewise.
20678 (lra_undo_inheritance): Likewise.
20679 * lra-remat.c (calculate_gen_cands): Likewise.
20680 (do_remat): Likewise.
20681 * lra-spills.c (assign_spill_hard_regs): Likewise.
20682 (spill_pseudos): Likewise.
20683 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20684 (bitmap_set_subtract_values): Likewise.
20686 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20688 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20689 management with auto_bitmap.
20690 (fix_inter_tick): Likewise.
20691 (fix_recovery_deps): Likewise.
20692 * ira.c (add_store_equivs): Likewise.
20693 (find_moveable_pseudos): Likewise.
20694 (split_live_ranges_for_shrink_wrap): Likewise.
20695 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20696 (rtx_reuse_manager::seen_def_p): Likewise.
20697 (rtx_reuse_manager::set_seen_def): Likewise.
20698 * print-rtl.h (class rtx_reuse_manager): Likewise.
20700 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20702 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20704 (migrate_btr_def): Likewise.
20705 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20706 * df-core.c (loop_post_order_compute): Likewise.
20707 (loop_inverted_post_order_compute): Likewise.
20708 * hsa-common.h: Likewise.
20709 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20710 * init-regs.c (initialize_uninitialized_regs): Likewise.
20711 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20712 (inline_small_functions): Likewise.
20713 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20714 * ira.c (combine_and_move_insns): Likewise.
20715 (build_insn_chain): Likewise.
20716 * loop-invariant.c (find_invariants): Likewise.
20717 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20718 * predict.c (tree_predict_by_opcode): Likewise.
20719 (predict_paths_leading_to): Likewise.
20720 (predict_paths_leading_to_edge): Likewise.
20721 (estimate_loops_at_level): Likewise.
20722 (estimate_loops): Likewise.
20723 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20724 (spread_components): Likewise.
20725 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20726 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20727 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20728 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20729 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20730 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20731 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20732 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20733 (mark_threaded_blocks): Likewise.
20734 (thread_through_all_blocks): Likewise.
20735 * tree-ssa.c (verify_ssa): Likewise.
20736 (execute_update_addresses_taken): Likewise.
20737 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20739 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20741 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20743 (post_order_compute): Likewise.
20744 (inverted_post_order_compute): Likewise.
20745 (pre_and_rev_post_order_compute_fn): Likewise.
20747 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20749 * genrecog.c (int_set::int_set): Explicitly construct our
20750 auto_vec base class.
20751 * vec.h (auto_vec::auto_vec): New constructor.
20753 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20755 * bitmap.h (class auto_bitmap): New constructor taking
20756 bitmap_obstack * argument.
20758 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20760 * bitmap.h (class auto_bitmap): Change type of m_bits to
20761 bitmap_head, and adjust ctor / dtor and member operators.
20763 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20765 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20766 when returned register mode doesn't match original mode.
20768 2017-05-12 Jeff Law <law@redhat.com>
20769 Jakub Jelinek <jakub@redhat.com>
20771 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20772 we look for cc setter after the compare-elim changes.
20773 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20774 within the vector to match what compare-elim now expects.
20775 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20776 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20778 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20779 after the compare-elim changes.
20780 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20781 the vector to match what compare-elim now expects.
20782 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20783 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20784 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20785 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20786 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20788 * config/visium/visium.c (single_set_and_flags): Fix where
20789 we look for cc setter after the compare-elim changes.
20790 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20791 with the vector to match what compare-elim now expects.
20792 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20793 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20794 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20795 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20796 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20797 (neg<mode>2_insn_set_overflow): Likewise.
20799 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20801 PR middle-end/79794
20802 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20803 maybe_expand_insn call, set ops[0].target. If still set after call,
20804 set alt_rtl. Add extra arg to recursive calls.
20805 (extract_bit_field): Add alt_rtl argument. Pass to
20807 * expmed.h (extract_bit_field): Fix prototype.
20808 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20809 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20810 to extract_bit_field_calls.
20811 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20812 Pass alt_rtl to extract_bit_field calls.
20813 * calls.c (store_unaligned_arguments_into_psuedos)
20814 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20815 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20817 * optabs.h (struct expand_operand): Add target bitfield.
20819 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20821 * compare-elim.c (try_eliminate_compare): Canonicalize
20822 operation with embedded compare to
20823 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20824 (set (reg) (operation)].
20826 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20828 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20831 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20832 cost of adding a carry flag for ADC instruction.
20833 [case MINUS]: Ignore the cost of subtracting a carry flag
20834 for SBB instruction.
20836 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20838 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20840 * config/rs6000/bmiintrin.h: New file.
20841 * config/rs6000/bmi2intrin.h: New file.
20842 * config/rs6000/x86intrin.h: New file.
20844 2017-05-12 Jeff Law <law@redhat.com>
20846 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20849 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20851 PR middle-end/80707
20852 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20854 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20857 early expansion of vector divide builtins.
20858 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20859 builtins identified as having unsigned arguments.
20861 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20863 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20864 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20865 expansion of vector logical operations (and, andc, or, xor,
20868 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20870 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20871 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20873 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20875 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20876 early GIMPLE expansion of vector multiplies.
20878 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20880 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20881 TARGET_HAVE_MOVT conditional.
20882 (movt splitter): Likewise.
20884 2017-05-12 Richard Biener <rguenther@suse.de>
20886 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20887 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20888 Fold all stmts not inplace.
20890 2017-05-12 Richard Biener <rguenther@suse.de>
20892 PR tree-optimization/80713
20893 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20894 inserted_exprs bit for not removed stmts.
20896 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20898 PR middle-end/69921
20899 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20900 parallelized" attribute for parallelized OpenACC kernels.
20901 * omp-offload.c (execute_oacc_device_lower): Use it.
20903 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20904 Set "oacc kernels" attribute.
20905 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20906 parameter. Adjust all users.
20907 (oacc_fn_attrib_kernels_p): Remove function.
20908 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20909 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20910 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20911 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20912 assert "oacc kernels" attribute is set.
20914 2017-05-11 Carl Love <cel@us.ibm.com>
20916 * config/rs6000/rs6000-c: Add support for built-in functions
20917 vector unsigned char vec_popcnt (vector signed char)
20918 vector unsigned char vec_popcnt (vector unsigned char)
20919 vector unsigned short vec_popcnt (vector signed short)
20920 vector unsigned short vec_popcnt (vector unsigned short)
20921 vector unsigned int vec_popcnt (vector signed int)
20922 vector unsigned int vec_popcnt (vector unsigned int)
20923 vector unsigned long long vec_popcnt (vector signed long long)
20924 vector unsigned long long vec_popcnt (vector unsigned long long)
20925 vector signed long long vec_slo (vector signed long long,
20926 vector signed char)
20927 vector signed long long vec_slo (vector signed long long,
20928 vector unsigned char)
20929 vector unsigned long long vec_slo (vector unsigned long long,
20930 vector signed char)
20931 vector unsigned long long vec_slo (vector unsigned long long,
20932 vector unsigned char)
20933 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20934 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20935 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20936 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20937 * doc/extend.texi: Update the built-in documentation file for the
20938 new built-in functions.
20940 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20942 * attribs.h (sorted_attr_string): Move machine independent
20943 functions for target clone support from the i386 port to common
20944 code. Rename ix86_function_versions to common_function_versions.
20945 Rename make_name to make_unique_name.
20946 (common_function_versions): Likewise.
20947 (make_unique_name): Likewise.
20948 (make_dispatcher_decl): Likewise.
20949 (is_function_default_version): Likewise.
20950 * attribs.c (attr_strcmp): Likewise.
20951 (sorted_attr_string): Likewise.
20952 (common_function_versions): Likewise.
20953 (make_unique_name): Likewise.
20954 (make_dispatcher_decl): Likewise.
20955 (is_function_default_version): Likewise.
20956 * config/i386/i386.c (attr_strcmp): Likewise.
20957 (sorted_attr_string): Likewise.
20958 (ix86_function_versions): Likewise.
20959 (make_name): Likewise.
20960 (make_dispatcher_decl): Likewise.
20961 (is_function_default_version): Likewise.
20962 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20964 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20967 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20968 Account for direct move costs for vec_construct of integer
20971 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20974 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20975 (UNSPEC_STX_ATOMIC): Ditto.
20976 (loaddi_via_sse): New insn.
20977 (storedi_via_sse): Ditto.
20978 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20979 Update corresponding peephole2 patterns.
20980 (atomic_storedi_fpu): Ditto.
20982 2017-05-11 Julia Koval <julia.koval@intel.com>
20984 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20985 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20987 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20988 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20989 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20991 2017-05-11 Nathan Sidwell <nathan@acm.org>
20993 * graphite-poly.c: Include dumpfile.h.
20995 * dumpfle.h (dump_function): Declare here ...
20996 * tree-dump.h (dump_function): ... not here.
20997 * dumpfile.c: #include tree-cfg.h.
20998 (dump_function): Move here from ...
20999 * tree-dump.c (dump_function): ... here.
21000 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21001 * graphite-poly.c: Don't include tree-dump.h.
21002 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21003 * print-tree.c: Likewise.
21004 * stor-layout.c: Likewise.
21005 * tree-nested.c: Likewise.
21007 * dumpfile.c (dump_start): Use TDF_FLAGS.
21008 (dump_enable_all): Fix TDF_KIND check thinko.
21010 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21012 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21013 array entries to represent two legal parameterizations of the
21014 overloaded __builtin_cmpb function, as represented by the
21015 P6_OV_BUILTIN_CMPB constant.
21016 (altivec_resolve_overloaded_builtin): Add special case handling
21017 for the __builtin_cmpb function, as represented by the
21018 P6_OV_BUILTIN_CMPB constant.
21019 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21020 (BU_P6_64BIT_2): New macro.
21021 (BU_P6_OVERLOAD_2): New macro
21022 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21023 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21024 (CMPB): Add overload support to represent both 32-bit and 64-bit
21025 compare-bytes function.
21026 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21027 support for TARGET_CMPB.
21028 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21029 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21030 documentation of the __builtin_cmpb overloaded built-in function.
21032 2017-05-11 Richard Biener <rguenther@suse.de>
21034 PR tree-optimization/80705
21035 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21036 bases are not vectorizable.
21038 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21040 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21041 when counting register pressure.
21043 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21045 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21046 (iv_ca_more_deps): Renamed to ...
21047 (iv_ca_compare_deps): ... this.
21048 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21052 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21054 (determine_group_iv_costs): ... here.
21055 (find_inv_vars_cb): Record inv var if it's not recorded before.
21057 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21059 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21060 (get_shiftadd_cost): Ditto.
21062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21064 * tree-ssa-address.c: Include header file.
21065 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21067 (add_to_parts): Refactor.
21068 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21069 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21072 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21074 PR tree-optimization/53090
21075 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21077 (extract_cond_operands): Detect condition with IV on both sides
21078 and return COMP_IV_EXPR_2.
21079 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21080 (rewrite_use_compare): Simplify by removing call to function
21081 extract_cond_operands.
21083 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21085 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21086 (extract_cond_operands): Detect condition comparing against non-
21087 invariant bound and return appropriate enum value.
21088 (find_interesting_uses_cond): Update use of extract_cond_operands.
21089 Handle its return value accordingly.
21090 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21094 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21095 nonlinear iv_use computation in loop invariant sensitive way.
21097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21099 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21100 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21104 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21105 (dump_cand): Support iv_cand.inv_exprs.
21106 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21108 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21113 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21115 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21116 as local function. Include necessary header files.
21117 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21119 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21121 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21125 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21126 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21127 RSHIFT_EXPR and BIT_NOT_EXPR.
21129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21131 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21132 (adjust_setup_cost): New parameter supporting round up adjustment.
21133 (struct address_cost_data): Delete.
21134 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21135 (split_address_cost, ptr_difference_cost): Delete.
21136 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21137 (struct ainc_cost_data): New struct.
21138 (get_address_cost_ainc): New function.
21139 (get_address_cost, get_computation_cost): Reimplement.
21140 (determine_group_iv_cost_address): Record inv_expr for all uses of
21142 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21143 (iv_ca_has_deps): Reimplemented to ...
21144 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21146 (iv_ca_extend): Call iv_ca_more_deps.
21148 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21150 * tree-ssa-address.c (struct mem_address): Move to header file.
21151 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21152 * tree-ssa-address.h (struct mem_address): Move from C file.
21153 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21157 * tree-affine.h (aff_combination_type): New interface.
21158 (aff_combination_zero_p): Remove static.
21159 (aff_combination_const_p): New interface.
21160 (aff_combination_singleton_var_p): New interfaces.
21162 2017-05-11 Richard Biener <rguenther@suse.de>
21164 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21165 Skip unreachable blocks and destinations.
21166 (eliminate): Move stmt removal and fixup ...
21167 (fini_eliminate): ... here. Skip inserted exprs.
21168 (pass_pre::execute): Move fini_pre after fini_eliminate.
21169 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21170 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21171 PRE to get rid of dead code that has invalid SSA form and
21172 split critical edges again.
21174 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21176 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21178 2017-05-11 Richard Biener <rguenther@suse.de>
21180 * passes.c (execute_function_todo): Verify loops if they are
21181 said to be up-to-date.
21182 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21183 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21185 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21188 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21189 handle calling assemble_external ourself.
21192 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21193 modes with zero size. Enhance comment.
21195 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21197 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21198 built-ins for vec_xl and vec_xst with short and char pointer
21201 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21203 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21204 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21205 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21206 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21207 (_mm_maskz_min_round_ss): New intrinsics.
21208 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21209 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21210 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21211 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21212 (__builtin_ia32_minss_mask_round): New builtins.
21213 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21214 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21215 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21217 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21218 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21220 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21223 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21225 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21226 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21227 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21228 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21229 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21230 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21231 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21232 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21234 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21235 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21236 (__builtin_ia32_mulss_mask_round): New builtins.
21237 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21238 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21239 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21241 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21242 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21244 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21247 2017-05-10 Julia Koval <julia.koval@intel.com>
21249 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21250 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21251 (_mm256_setr_m128i): New intrinsics.
21253 2017-05-10 Julia Koval <julia.koval@intel.com>
21255 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21256 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21257 (_mm_maskz_rcp14_ss): New intrinsics.
21258 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21259 (__builtin_ia32_rcp14ss_mask): New builtins.
21260 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21262 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21264 PR tree-optimization/51513
21265 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21266 (assert_unreachable_fallthru_edge_p): Use it.
21267 (group_case_labels_stmt): Likewise.
21268 * tree-cfg.h: Prototype it.
21269 * stmt.c: Include cfghooks.h and tree-cfg.h.
21270 (emit_case_dispatch_table) <gap_label>: New local variable.
21271 Use it to fill dispatch table gaps.
21272 Test for default_label before updating probabilities.
21273 (expand_case) <default_label>: Remove unneeded initialization.
21274 Test for unreachable default case statement and remove its edge.
21275 Set default_label accordingly.
21276 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21278 2017-05-10 Carl Love <cel@us.ibm.com>
21280 * config/rs6000/rs6000-c: Add support for built-in functions
21281 vector signed char vec_neg (vector signed char)
21282 vector signed short int vec_neg (vector short int)
21283 vector signed int vec_neg (vector signed int)
21284 vector signed long long vec_neg (vector signed long long)
21285 vector float vec_neg (vector float)
21286 vector double vec_neg (vector double)
21287 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21289 * config/rs6000/altivec.h: Add define for vec_neg
21290 * doc/extend.texi: Update the built-in documentation for the
21291 new built-in functions.
21293 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21295 PR tree-optimization/77644
21296 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21298 2017-05-10 Nathan Sidwell <nathan@acm.org>
21300 * dumpfile.h (TDI_lang_all): New.
21301 (TDF_KIND): New. Renumber others
21302 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21304 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21306 (get_dump_file_name): Adjust suffix generation.
21307 (dump_enable_all): Use TDF_KIND.
21308 * doc/invoke.texi (-fdump-lang-all): Document.
21310 * dumpfile.h: Tabify.
21312 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21315 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21316 Move member access before delete.
21318 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21320 * tree-inline.c (expand_call_inline): Split block at stmt
21323 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21326 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21327 are now unused after splitting mov{sf,sd}_hardfloat.
21328 (f32_lr2): Likewise.
21329 (f32_lm): Likewise.
21330 (f32_lm2): Likewise.
21331 (f32_li): Likewise.
21332 (f32_li2): Likewise.
21333 (f32_lv): Likewise.
21334 (f32_sr): Likewise.
21335 (f32_sr2): Likewise.
21336 (f32_sm): Likewise.
21337 (f32_sm2): Likewise.
21338 (f32_si): Likewise.
21339 (f32_si2): Likewise.
21340 (f32_sv): Likewise.
21341 (f32_dm): Likewise.
21342 (f32_vsx): Likewise.
21343 (f32_av): Likewise.
21344 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21345 For movsf, order stores so the VSX stores occur before the GPR
21346 store which encourages the register allocator to use a traditional
21347 FPR instead of a GPR. For movsd, order the stores so that the GPR
21348 store comes before the VSX stores to allow the power6 to work.
21349 This is due to the power6 not having a 32-bit integer store
21350 instruction from a FPR.
21351 (movsf_hardfloat): Likewise.
21352 (movsd_hardfloat): Likewise.
21354 2017-05-09 Martin Sebor <msebor@redhat.com>
21356 PR translation/80280
21357 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21360 PR translation/80280
21361 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21362 data member added in r247778.
21363 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21365 2017-05-09 Nathan Sidwell <nathan@acm.org>
21367 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21369 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21372 2017-05-09 Marek Polacek <polacek@redhat.com>
21374 * doc/invoke.texi: Fix typo.
21376 2017-05-09 Richard Biener <rguenther@suse.de>
21378 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21379 (vrp_val_is_min): Likewise.
21380 (set_value_range_to_value): Likewise.
21381 (set_value_range_to_nonnegative): Likewise.
21382 (gimple_assign_nonzero_p): Likewise.
21383 (gimple_stmt_nonzero_p): Likewise.
21384 (vrp_int_const_binop): Likewise. Remove unreachable case.
21385 (adjust_range_with_scev): Adjust comments.
21386 (compare_range_with_value): Likewise.
21387 (extract_range_from_phi_node): Likewise.
21388 (test_for_singularity): Likewise.
21390 2017-05-09 Richard Biener <rguenther@suse.de>
21392 * tree-vrp.c (get_single_symbol): Add assert that we don't
21393 get overflowed constants as invariant part.
21394 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21395 checks. Use wi::cmp instead of recursing for integer constants.
21396 (compare_values): Just ignore whether we assumed undefined
21397 overflow instead of failing the compare.
21398 (extract_range_for_var_from_comparison_expr): Add comment before the
21399 TREE_NO_WARNING sets.
21400 (test_for_singularity): Likewise.
21401 (extract_range_from_comparison): Do not disable optimization
21402 when we assumed undefined overflow.
21403 (extract_range_basic): Remove init of unused var.
21405 2017-05-09 Richard Biener <rguenther@suse.de>
21407 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21408 (extract_range_from_multiplicative_op_1): Adjust.
21409 (extract_range_from_binary_expr_1): Use int_const_binop.
21411 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21414 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21415 rs6000_store_data_bypass_p in seven define_bypass directives and
21416 in several comments.
21417 * config/rs6000/rs6000-protos.h: Add prototype for
21418 rs6000_store_data_bypass_p function.
21419 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21420 function implements slightly different (rs6000-specific) semantics
21421 than store_data_bypass_p, returning false rather than aborting
21422 with assertion error when arguments do not satisfy the
21423 requirements of store data bypass.
21424 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21425 rs6000_store_data_bypass_p.
21427 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21429 * config/xtensa/xtensa-protos.h
21430 (xtensa_initial_elimination_offset): New declaration.
21431 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21432 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21433 macro definition, add case for FRAME_POINTER_REGNUM when
21434 FRAME_GROWS_DOWNWARD.
21435 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21436 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21437 xtensa_initial_elimination_offset.
21439 2017-05-08 Nathan Sidwell <nathan@acm.org>
21441 * doc/invoke.texi: Alphabetize -fdump options.
21443 2017-05-08 Martin Sebor <msebor@redhat.com>
21445 PR translation/80280
21446 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21448 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21450 * target.def (compute_frame_layout): New optional target hook.
21451 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21452 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21453 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21455 * reload1.c (verify_initial_elim_offsets): Likewise.
21456 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21457 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21458 (arm_get_frame_offsets): Split up into this ...
21459 (arm_compute_frame_layout): ... and this function.
21461 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21463 * config/aarch64/constraints.md (Usa): New constraint.
21464 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21466 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21468 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21469 with_multilib_list after it has been checked.
21471 2017-05-08 Richard Biener <rguenther@suse.de>
21473 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21474 (bitmap_set_subtract_values): Likewise.
21476 2017-05-08 Richard Biener <rguenther@suse.de>
21478 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21479 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21481 (gimple_stmt_nonzero_warnv_p): Rename to ...
21482 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21484 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21485 (extract_range_basic): Adjust, do not disable propagation on
21486 strict overflow sensitive simplification.
21487 (vrp_visit_cond_stmt): Likewise.
21489 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21491 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21492 body size unconditionally.
21494 2017-05-07 Jeff Law <law@redhat.com>
21497 2017-05-06 Jeff Law <law@redhat.com>
21498 PR tree-optimization/78496
21499 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21502 PR tree-optimization/78496
21503 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21504 (simplify_stmt_using_ranges): Call it.
21505 (vrp_dom_walker::before_dom_children): Extract equivalences
21506 from an ASSERT_EXPR with an equality comparison against a
21509 2017-05-06 Jeff Law <law@redhat.com>
21511 PR tree-optimization/78496
21512 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21515 PR tree-optimization/78496
21516 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21517 (simplify_stmt_using_ranges): Call it.
21518 (vrp_dom_walker::before_dom_children): Extract equivalences
21519 from an ASSERT_EXPR with an equality comparison against a
21522 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21524 * lra-constraints.c (lra_copy_reg_equiv): New function.
21525 (split_reg): Use it to copy equivalence information from the
21526 original register to the spill register.
21528 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21530 PR rtl-optimization/75964
21531 * simplify-rtx.c (simplify_const_relational_operation): Remove
21532 invalid handling of comparisons of integer ABS.
21534 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21536 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21537 initialize to zero.
21538 (init_regs): Remove declaration.
21539 (function_arg_advance_32): Initialize error_p as boolean variable.
21541 2017-05-05 Nathan Sidwell <nathan@acm.org>
21543 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21544 lines. Use for (;;).
21546 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21548 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21549 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21550 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21551 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21552 VF=2 that require versioning.
21554 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21556 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21559 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21561 * diagnostic.h (diagnostic_override_option_index): Convert from
21562 macro to inline function.
21564 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21566 * diagnostic.c (last_module_changed_p): New function.
21567 (set_last_module): New function.
21568 (diagnostic_report_current_module): Convert macro usage to
21569 the above functions.
21570 * diagnostic.h (diagnostic_context::last_module): Strengthen
21571 from const line_map * to const line_map_ordinary *.
21572 (diagnostic_last_module_changed): Delete macro.
21573 (diagnostic_set_last_module): Delete macro.
21575 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21577 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21578 with diagnostic_report_diagnostic.
21579 (diagnostic_n_impl_richloc): Likewise.
21580 * diagnostic.h (report_diagnostic): Delete macro.
21581 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21582 with diagnostic_report_diagnostic.
21583 * substring-locations.c (format_warning_va): Likewise.
21585 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21587 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21588 save/restor of format_spec. Move option-printing code to...
21589 (print_option_information): ...this new function, and
21590 reimplement by simply printing to the pretty_printer,
21591 rather than appending to the format string.
21593 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21595 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21596 handling logic into...
21597 (update_effective_level_from_pragmas): ...this new function.
21599 2017-05-04 Andrew Waterman <andrew@sifive.com>
21601 * config/riscv/riscv.opt (mstrict-align): New option.
21602 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21603 (SLOW_UNALIGNED_ACCESS): Define.
21604 (riscv_slow_unaligned_access): Declare.
21605 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21607 (riscv_slow_unaligned_access): New variable.
21608 (rocket_tune_info): Set slow_unaligned_access to true.
21609 (optimize_size_tune_info): Set slow_unaligned_access to false.
21610 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21611 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21612 (riscv_option_override): Set riscv_slow_unaligned_access.
21613 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21615 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21617 * config/riscv/riscv.md: Unify indentation.
21619 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21624 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21626 (extendsi<mode>2): Add support for doing sign extension via
21627 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21628 don't have ISA 3.0 instructions.
21629 (extendsi<mode>2 splitter): Likewise.
21630 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21631 generate the normal insns since SImode can now go in vector
21632 registers. Disallow the special UNSPECs needed for previous
21633 machines to hide SImode being used. Add new insns
21634 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21635 (fix_trunc<mode>si2_stfiwx): Likewise.
21636 (fix_trunc<mode>si2_internal): Likewise.
21637 (fixuns_trunc<mode>si2): Likewise.
21638 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21639 (fctiw<u>z_<mode>_smallint): Likewise.
21640 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21641 of floating point to 32-bit integer from doing a direct move to
21642 the GPR registers to do a store.
21643 (fctiwz_<mode>): Break long line.
21645 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21647 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21648 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21649 (addr_list, addr_offset_valid_p): New.
21650 (split_address_groups): Check offset validity with above function.
21651 (gt-tree-ssa-loop-ivopts.h): Include header file.
21653 2017-05-05 Nathan Sidwell <nathan@acm.org>
21655 * config.gcc (arm*-*-*): Add missing 'fi'.
21657 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21659 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21660 included in -fopt-info does not matter.
21661 * doc/optinfo.texi (-fopt-info): Fix description of default
21662 behavour. Explicitly say order of options included in -fopt-info
21665 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21667 * config.gcc: Allow combinations of aprofile and rmprofile values for
21668 --with-multilib-list.
21669 * config/arm/t-multilib: New file.
21670 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21671 variables. Remove setting of ISA and floating-point ABI in
21672 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21673 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21674 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21675 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21677 * config/arm/t-rmprofile: Likewise except for the matches changes.
21678 * doc/install.texi (--with-multilib-list): Document the combination of
21679 aprofile and rmprofile values and warn about pitfalls in doing that.
21681 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21683 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21684 (movdi_aarch64): Likewise.
21686 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21688 PR tree-optimization/80632
21689 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21691 (build_arrays): Initialize it for virtual phis.
21692 (fix_phi_nodes): Use it for virtual phis.
21694 PR tree-optimization/80558
21695 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21696 [x, y] op z into [x op, y op z] for op & or | if conditions
21699 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21700 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21703 * config/arm/arm.md (use_literal_pool): Remove.
21704 (64-bit immediate split): No longer takes cost into consideration
21705 if arm_disable_literal_pool is enabled.
21706 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21707 used when arm_disable_literal_pool is enabled.
21708 (arm_max_const_double_inline_cost): Remove use of
21709 arm_disable_literal_pool.
21710 (push_minipool_fix): Add assert.
21711 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21712 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21713 (no_literal_pool_sf_immediate): New.
21715 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21717 PR tree-optimization/80613
21718 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21719 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21721 2017-05-05 Richard Biener <rguenther@suse.de>
21723 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21725 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21727 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21728 of this flag from insn conditions due to removal from r247495.
21730 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21732 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21734 (arm_early_store_addr_dep_ptr): Likewise.
21735 * config/arm/aarch-common-protos.h
21736 (arm_early_load_addr_dep_ptr): Add prototype.
21737 (arm_early_store_addr_dep_ptr): Likewise.
21738 * config/arm/cortex-a53.md: Add new bypasses.
21740 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21742 * tree.c (next_type_uid): Change type to unsigned.
21743 (type_hash_canon): Decrement back next_type_uid if
21744 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21745 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21748 2017-05-04 Martin Sebor <msebor@redhat.com>
21750 * builtins.c: Fix a trivial typo in a comment.
21752 PR middle-end/79234
21753 * builtins.c (check_sizes): Adjust to handle reading past the end.
21754 Avoid printing excessive upper bound of ranges. Use %E to print
21755 tree nodes instead of converting them to %wu.
21756 (expand_builtin_memchr): New function.
21757 (compute_dest_size): Rename...
21758 (compute_objsize): ...to this.
21759 (expand_builtin_memcpy): Adjust.
21760 (expand_builtin_mempcpy): Adjust.
21761 (expand_builtin_strcat): Adjust.
21762 (expand_builtin_strcpy): Adjust.
21763 (check_strncat_sizes): Adjust.
21764 (expand_builtin_strncat): Adjust.
21765 (expand_builtin_strncpy): Adjust and simplify.
21766 (expand_builtin_memset): Adjust.
21767 (expand_builtin_bzero): Adjust.
21768 (expand_builtin_memcmp): Adjust.
21769 (expand_builtin): Handle memcmp.
21770 (maybe_emit_chk_warning): Check strncat just once.
21772 2017-05-04 Martin Sebor <msebor@redhat.com>
21774 PR preprocessor/79214
21775 PR middle-end/79222
21776 PR middle-end/79223
21777 * builtins.c (check_sizes): Add inlining context and issue
21778 warnings even when -Wno-system-headers is set.
21779 (check_strncat_sizes): Same.
21780 (expand_builtin_strncat): Same.
21781 (expand_builtin_memmove): New function.
21782 (expand_builtin_stpncpy): Same.
21783 (expand_builtin): Handle memmove and stpncpy.
21785 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21787 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21788 which is not used any more.
21790 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21792 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21794 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21796 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21797 (cortexa53_tunings): Likewise.
21798 (cortexa57_tunings): Likewise.
21799 (cortexa72_tunings): Likewise.
21800 (cortexa73_tunings): Likewise.
21802 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21804 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21805 Set loop alignment to 8.
21807 2017-05-04 Martin Sebor <msebor@redhat.com>
21809 PR translation/80280
21810 * builtins.c (expand_builtin_object_size): Add missing quoting to
21811 %D and like directives.
21812 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21813 (hsa_type_for_tree_type): Same.
21814 (verify_function_arguments): Same.
21815 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21816 * varasm.c (get_section): Same.
21819 2017-05-04 Martin Sebor <msebor@redhat.com>
21821 PR translation/80280
21822 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21824 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21826 * config/aarch64/aarch64.c (generic_addrcost_table):
21827 Change HI/TI mode setting.
21829 2017-05-04 Martin Jambor <mjambor@suse.cz>
21831 PR tree-optimization/80622
21832 * tree-sra.c (comes_initialized_p): New function.
21833 (build_accesses_from_assign): Only set write lazily when
21834 comes_initialized_p is false.
21835 (analyze_access_subtree): Use comes_initialized_p.
21836 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21837 instead of testing for PARM_DECL.
21839 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21841 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21842 constraint on operand 0 to allow more general addressing modes.
21843 Adjust output template.
21844 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21846 * config/aarch64/aarch64-protos.h
21847 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21848 * config/aarch64/constraints.md (Dp): New address constraint.
21849 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21852 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21854 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21855 update use of estimate_ipcp_clone_size_and_time.
21856 (estimate_local_effects): Update use of
21857 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21858 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21859 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21860 Return nonspecialized time.
21862 2017-05-04 Richard Biener <rguenther@suse.de>
21864 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21865 for the last VUSE which def dominates the PHI. Directly call
21867 (get_continuation_for_phi_1): Remove.
21869 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21871 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21872 to explain the use of truncating division. Cap the number of
21873 iterations to the maximum given by nb_iterations_upper_bound,
21876 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21878 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21879 * configure: Regenerate.
21880 * config.in: Regenerate.
21881 * config/i386/driver-mingw32.c: new file.
21882 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21883 * config.host: Link driver-mingw32.o on MinGW host.
21884 * doc/install.texi: Document new --enable-mingw-wildcard configure
21887 2017-05-04 Marek Polacek <polacek@redhat.com>
21889 PR tree-optimization/80612
21890 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21892 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21893 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21895 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21896 (movt splitter): Likewise.
21897 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21898 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21899 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21900 block for Thumb-1 with MOVT.
21901 (thumb2_legitimate_address_p): Move code block ...
21902 (can_avoid_literal_pool_for_label_p): ... into this new function.
21903 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21905 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21906 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21907 "M-profile targets with the MOVT instruction".
21909 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21911 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21912 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21913 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21915 2017-05-04 Martin Liska <mliska@suse.cz>
21917 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21918 variable cond_code.
21920 2017-05-04 Richard Biener <rguenther@suse.de>
21922 * tree.c (array_at_struct_end_p): Handle arrays at struct
21923 end with flexarrays more conservatively. Refactor and treat
21924 arrays of arrays or aggregates more strict. Fix
21925 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21926 * tree.c (array_at_struct_end_p): Adjust prototype.
21927 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21928 * gimple-fold.c (get_range_strlen): Likewise.
21929 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21931 2017-05-04 Richard Biener <rguenther@suse.de>
21933 PR tree-optimization/31130
21934 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21936 (supports_overflow_infinity): Likewise.
21937 (is_negative_overflow_infinity): Likewise.
21938 (is_positive_overflow_infinity): Likewise.
21939 (is_overflow_infinity): Likewise.
21940 (stmt_overflow_infinity): Likewise.
21941 (overflow_infinity_range_p): Likewise.
21942 (usable_range_p): Remove as always returning true.
21943 (make_overflow_infinity): Remove.
21944 (negative_overflow_infinity): Likewise.
21945 (positive_overflow_infinity): Likewise.
21946 (avoid_overflow_infinity): Likewise.
21947 (set_value_range): Adjust accordingly.
21948 (set_value_range_to_nonnegative): Likewise, remove now unused
21949 overflow_infinity arg.
21950 (vrp_operand_equal_p): Adjust.
21951 (update_value_range): Likewise.
21952 (range_int_cst_singleton_p): Likewise.
21953 (operand_less_p): Likewise.
21954 (compare_values_warnv): Likewise.
21955 (extract_range_for_var_from_comparison_expr): Likewise.
21956 (vrp_int_const_binop): Likewise.
21957 (zero_nonzero_bits_from_vr): Likewise.
21958 (extract_range_from_multiplicative_op_1): Likewise.
21959 (extract_range_from_binary_expr_1): Likewise.
21960 (extract_range_from_unary_expr): Likewise.
21961 (extract_range_from_comparison): Likewise.
21962 (extract_range_basic): Likewise.
21963 (adjust_range_with_scev): Likewise.
21964 (compare_ranges): Likewise.
21965 (compare_range_with_value): Likewise.
21966 (dump_value_range): Likewise.
21967 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21969 (simplify_cond_using_ranges): Adjust.
21971 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21973 * brig-builtins.def: Added a builtin for class_f64.
21974 * builtin-types.def: Added a builtin type needed by class_f64.
21976 2017-05-03 Jason Merrill <jason@redhat.com>
21978 * timevar.def: Add TV_CONSTEXPR.
21980 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21982 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21984 2017-05-03 Martin Jambor <mjambor@suse.cz>
21986 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21987 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21988 * ipa-cp.c (ipcp_propagate_stage): Do not call
21989 ipa_update_after_lto_read.
21990 * ipa-inline.c (ipa_inline): Likewise.
21992 2017-05-03 Martin Jambor <mjambor@suse.cz>
21994 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21995 tag. Added a default constructor and a destructor.
21996 (ipa_edge_args_sum_t): New class;
21997 (ipa_edge_args_sum): Declare.
21998 (ipa_edge_args_vector): Remove declaration.
21999 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22000 (ipa_free_edge_args_substructures): Remove declaration.
22001 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22002 (ipa_edge_args_info_available_for_edge_p): Likewise.
22003 * ipa-prop.c (ipa_edge_args_vector): Removed.
22004 (edge_removal_hook_holder): Likewise.
22005 (edge_duplication_hook_holder): Likewise.
22006 (ipa_edge_args_sum): New variable.
22007 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22008 ipa_edge_args_vector.
22009 (ipa_free_edge_args_substructures): Likewise.
22010 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22011 ipa_edge_args_vector.
22012 (ipa_edge_removal_hook): Turned into method
22013 ipa_edge_args_sum_t::remove.
22014 (ipa_edge_duplication_hook): Turned into method
22015 ipa_edge_args_sum_t::duplicate.
22016 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22017 registering edge hooks.
22018 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22019 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22020 ipa_edge_args_sum instead of ipa_edge_args_vector.
22021 * ipa-profile.c (ipa_profile): Likewise.
22023 2017-05-03 Martin Jambor <mjambor@suse.cz>
22025 * symbol-summary.h (function_summary): New method exists.
22026 (function_summary::symtab_removal): Deallocate through release.
22027 (call_summary): New class.
22028 (gt_ggc_mx): New overload.
22029 (gt_pch_nx): Likewise.
22030 (gt_pch_nx): Likewise.
22032 2017-05-03 Jeff Law <law@redhat.com>
22034 PR tree-optimization/78496
22035 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22036 from simplify_cond_using_ranges. Split off code to walk
22037 backwards through casts into ...
22038 (simplify_cond_using_ranges_2): New function.
22039 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22040 (execute_vrp): After identifying jump threads, call
22041 simplify_cond_using_ranges_2.
22043 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22046 * ipa-inline.h (inline_summary): Add ctor.
22047 (create_ggc): Do not use ggc_cleared_alloc.
22049 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22050 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22052 * gcc.c (handle_braces): Support escaping in switch matching
22054 * doc/invoke.texi (Spec Files): Document it.
22055 Remove superfluous @code markup in items.
22057 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22059 * diagnostic-show-locus.c (struct column_range): New struct.
22060 (get_affected_columns): New function.
22061 (get_printed_columns): New function.
22062 (struct correction): New struct.
22063 (correction::ensure_capacity): New function.
22064 (correction::ensure_terminated): New function.
22065 (struct line_corrections): New struct.
22066 (line_corrections::~line_corrections): New dtor.
22067 (line_corrections::add_hint): New function.
22068 (layout::print_trailing_fixits): Reimplement in terms of the new
22070 (selftest::test_overlapped_fixit_printing): New function.
22071 (selftest::diagnostic_show_locus_c_tests): Call it.
22073 2017-05-03 Nathan Sidwell <nathan@acm.org>
22075 Canonicalize canonical type hashing
22076 * tree.h (type_hash_canon_hash): Declare.
22077 * tree.c (type_hash_list, attribute_hash_list): Move into
22078 type_hash_canon_hash.
22079 (build_type_attribute_qual_variant): Break out hash code calc into
22080 type_hash_canon_hash.
22081 (type_hash_canon_hash): New. Generic type hash computation.
22082 (build_range_type_1, build_array_type_1, build_function_type,
22083 build_method_type_directly, build_offset_type, build_complex_type,
22084 make_vector_type): Call it.
22086 2017-05-03 Richard Biener <rguenther@suse.de>
22088 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22089 When all DRs have unknown misaligned do not always peel
22090 when there is a store but apply the same costing model as if
22091 there were only loads.
22093 2017-05-03 Richard Biener <rguenther@suse.de>
22096 PR tree-optimization/80492
22097 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22098 compare_base_decls returning dont-know properly.
22100 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22102 * config/arm/iterators.md (CCSI): New mode iterator.
22103 (arch): New mode attribute.
22104 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22105 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22106 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22107 code iterator for success result mode.
22108 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22109 the corresponding new insn generators.
22111 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22114 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22115 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22117 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22119 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22120 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22121 (DDR_A): Wrap DDR argument in brackets.
22122 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22123 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22124 (DDR_REVERSED_P): Likewise.
22126 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22128 PR tree-optimization/79472
22129 * tree-switch-conversion.c (struct switch_conv_info): Add
22130 contiguous_range and default_case_nonstandard fields.
22131 (collect_switch_conv_info): Compute contiguous_range and
22132 default_case_nonstandard fields, don't clear final_bb if
22133 contiguous_range and only the default case doesn't have the required
22135 (check_all_empty_except_final): Set default_case_nonstandard instead
22136 of failing if contiguous_range and the default case doesn't have empty
22138 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22139 and only the default case doesn't have the required constants. Skip
22141 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22142 if default_case_nonstandard.
22143 (build_constructors): Build constant 1 just once. Assert that default
22144 values aren't inserted in between cases if contiguous_range. Skip
22146 (build_arrays): Skip virtual phis.
22147 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22148 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22149 Handle virtual phis.
22150 (gen_inbound_check): Handle default_case_nonstandard case.
22151 (process_switch): Adjust check_final_bb caller. Call
22152 gather_default_values with the first non-default case instead of
22153 default case if default_case_nonstandard.
22155 2017-05-02 Nathan Sidwell <nathan@acm.org>
22157 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22158 check. Fix formatting.
22160 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22162 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22163 errors when comparing specialized and unspecialized times.
22165 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22167 * diagnostic-show-locus.c
22168 (layout::should_print_annotation_line_p): Make private.
22169 (layout::print_annotation_line): Make private.
22170 (layout::annotation_line_showed_range_p): Make private.
22171 (layout::show_ruler): Make private.
22172 (layout::print_source_line): Make private. Pass in line and
22173 line_width, rather than calling location_get_source_line. Drop
22175 (layout::print_leading_fixits): New method.
22176 (layout::print_any_fixits): Rename to...
22177 (layout::print_trailing_fixits): ...this, and make private.
22178 Don't print newline fixits.
22179 (diagnostic_show_locus): Move logic for printing one row into...
22180 (layout::print_line): ...this new function. Move the
22181 location_get_source_line call and error-handling from
22182 print_source_line to here. Call print_leading_fixits, and rename
22183 print_any_fixits to print_trailing_fixits.
22184 (selftest::test_fixit_insert_containing_newline): Update now that
22185 newlines are partially supported.
22186 (selftest::test_fixit_insert_containing_newline_2): New test.
22187 (selftest::test_fixit_replace_containing_newline): Update comments.
22188 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22189 * edit-context.c (class added_line): New class.
22190 (class edited_line): Describe newline handling in comment.
22191 (edited_line::actually_edited_p): New method.
22192 (edited_line::print_content): Delete redundant decl.
22193 (edited_line::m_predecessors): New field.
22194 (edited_file::print_content): Call edited_line::print_content.
22195 (edited_file::print_diff): Update to support newlines.
22196 (edited_file::print_diff_hunk): Likewise.
22197 (edited_file::print_run_of_changed_lines): New function.
22198 (edited_file::print_diff_line): Convert to...
22199 (print_diff_line): ...this.
22200 (edited_file::get_effective_line_count): New function.
22201 (edited_line::edited_line): Initialize new field m_predecessors.
22202 (edited_line::~edited_line): Clean up m_predecessors.
22203 (edited_line::apply_fixit): Handle newlines.
22204 (edited_line::get_effective_line_count): New function.
22205 (edited_line::print_content): New function.
22206 (edited_line::print_diff_lines): New function.
22207 (selftest::test_applying_fixits_insert_containing_newline): New
22209 (selftest::test_applying_fixits_replace_containing_newline): New
22211 (selftest::insert_line): New function.
22212 (selftest::test_applying_fixits_multiple_lines): Add example of
22214 (selftest::edit_context_c_tests): Call the new tests.
22216 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22218 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22219 parameter cand. Update dump information.
22220 (get_computation_cost): Update uses.
22222 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22224 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22225 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22226 (get_computation_at, rewrite_use_address): Update use of
22227 get_computation_aff.
22229 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22231 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22232 (get_computation): Delete.
22233 (get_computation_cost): Implement like get_computation_cost_at.
22234 Use get_computation_at.
22235 (get_computation_cost_at): Delete.
22236 (rewrite_use_nonlinear_expr): Use get_computation_at.
22237 (rewrite_use_compare, remove_unused_ivs): Ditto.
22239 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22241 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22243 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22245 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22246 (ivopts_global_cost_for_size): Rename parameter and update uses.
22247 (iv_ca_recount_cost): Update uses.
22248 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22249 candidates seperately in n_invs and n_cands.
22250 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22252 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22254 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22255 (find_inv_vars_cb): New.
22256 (find_depends): Renamed to ...
22257 (find_inv_vars): ... this.
22258 (add_candidate_1, force_var_cost): Call find_inv_vars.
22259 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22263 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22264 inv_vars. Add inv_exprs.
22265 (struct iv_cand): Rename depends_on to inv_vars.
22266 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22267 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22268 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22269 (dump_cand): Dump inv_vars.
22270 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22271 (record_invariant, find_depends, add_candidate_1): Ditto.
22272 (set_group_iv_cost, force_var_cost): Ditto.
22273 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22274 (get_computation_cost_at, get_computation_cost): Ditto.
22275 (determine_group_iv_cost_generic): Ditto.
22276 (determine_group_iv_cost_address): Ditto.
22277 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22278 (determine_group_iv_costs): Ditto.
22279 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22280 (iv_ca_set_remove_invariants): Renamed to ...
22281 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22282 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22283 (iv_ca_set_add_invariants): Renamed to ...
22284 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22285 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22286 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22287 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22288 (create_new_ivs): Remove useless dump.
22290 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22292 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22294 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22295 (iv_ca_set_no_cp, create_new_iv): Ditto.
22297 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22299 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22301 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22303 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22304 function tree_check2.
22306 2017-05-02 Martin Liska <mliska@suse.cz>
22308 * doc/gcov.texi: Add missing preposition.
22309 * gcov.c (function_info::function_info): Properly fill up
22310 all member variables.
22312 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22314 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22316 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22318 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22320 2017-05-02 Martin Liska <mliska@suse.cz>
22323 * lto-streamer-in.c (lto_read_tree_1): Remove
22324 LTO_STREAMER_DEBUG.
22325 * lto-streamer.c (struct tree_hash_entry): Likewise.
22326 (struct tree_entry_hasher): Likewise.
22327 (tree_entry_hasher::hash): Likewise.
22328 (tree_entry_hasher::equal): Likewise.
22329 (lto_streamer_init): Likewise.
22330 (lto_orig_address_map): Likewise.
22331 (lto_orig_address_get): Likewise.
22332 (lto_orig_address_remove): Likewise.
22333 * lto-streamer.h: Likewise.
22334 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22335 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22337 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22339 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22340 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22341 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22342 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22343 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22344 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22345 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22346 (mm_maskz_sub_ss): New intrinsics.
22347 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22348 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22349 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22350 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22351 (__builtin_ia32_subss_mask_round): New builtins.
22352 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22353 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22354 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22356 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22357 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22359 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22362 2017-05-02 Martin Jambor <mjambor@suse.cz>
22364 PR tree-optimization/78687
22365 * tree-sra.c (access): New field parent.
22366 (process_subtree_disqualification): New function.
22367 (disqualify_candidate): Call it.
22368 (build_accesses_from_assign): Reset write flag if creating an
22370 (build_access_subtree): Fill in parent field and also prpagate
22371 down grp_write flag.
22372 (create_artificial_child_access): New parameter set_grp_write, set
22373 grp_write to its value.
22374 (propagate_subaccesses_across_link): Also propagate grp_write flag
22376 (propagate_all_subaccesses): Push the closest parent back to work
22377 queue if add_access_to_work_queue returned true.
22379 2017-05-02 Richard Biener <rguenther@suse.de>
22381 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22382 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22383 -fstrict-overflow documentation.
22384 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22385 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22386 flag_strict_overflow.
22387 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22388 * lto-opts.c (lto_write_options): Do not stream it.
22389 * lto-wrapper.c (merge_and_complain): Do not handle it.
22390 * opts.c (default_options_table): Do not set -fstrict-overflow.
22391 (finish_options): Likewise do not clear it when sanitizing.
22392 * simplify-rtx.c (simplify_const_relational_operation): Do not
22393 test flag_strict_overflow.
22395 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22397 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22398 using enabled attribute.
22399 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22400 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22401 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22402 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22403 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22404 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22405 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22406 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22407 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22408 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22410 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22412 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22414 2017-05-02 Richard Biener <rguenther@suse.de>
22416 PR tree-optimization/80591
22418 2017-04-10 Richard Biener <rguenther@suse.de>
22420 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22423 2017-05-02 Richard Biener <rguenther@suse.de>
22425 PR tree-optimization/80549
22426 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22427 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22428 headers if they do not have a preheader.
22430 2017-05-02 Martin Liska <mliska@suse.cz>
22433 * common.opt: Fix typo.
22434 * doc/invoke.texi: Likewise.
22436 2017-05-01 Jan Beulich <jbeulich@suse.com>
22438 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22439 swapping, add (x,x,m,x,n) alternative.
22441 2017-05-01 Nathan Sidwell <nathan@acm.org>
22443 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22444 unnecessary unadjusted_alignment check.
22446 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22449 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22451 * gimplify.c (gimplify_cilk_detach): New function.
22452 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22453 * tree-core.h: Document EXPR_CILK_SPAWN.
22454 * tree.h (EXPR_CILK_SPAWN): Define.
22456 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22458 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22459 to use new fixit_hint representation, using the "replace" logic.
22460 (get_line_span_for_fixit_hint): Likewise.
22461 (layout::print_any_fixits): Likewise.
22462 (selftest::test_one_liner_many_fixits): Rename to...
22463 (selftest::test_one_liner_many_fixits_1): ...this, and update
22464 comment and expected output to reflect that the multiple fix-it
22465 hints are now consolidated into one insertion.
22466 (selftest::test_one_liner_many_fixits_2): New test.
22467 (selftest::test_diagnostic_show_locus_one_liner): Update for
22469 (selftest::test_fixit_consolidation): Update for fix-it API
22471 * diagnostic.c (print_parseable_fixits): Likewise.
22472 * edit-context.c (edited_line::m_line_events): Convert from
22473 auto_vec <line_event *> to auto_vec <line_event>.
22474 (class line_event): Convert from abstract base class to a concrete
22475 class, taking over the role of replace_event.
22476 (class insert_event): Delete.
22477 (class replace_event): Rename to class line_event. Convert to
22479 (edit_context::add_fixits): Reimplement.
22480 (edit_context::apply_insert): Delete.
22481 (edit_context::apply_replace): Rename to...
22482 (edit_context::apply_fixit): ...this. Convert to half-open range.
22483 (edited_file::apply_insert): Delete.
22484 (edited_file::apply_replace): Rename to...
22485 (edited_file::apply_fixit): ...this.
22486 (edited_line::~edited_line): Drop deletion of events.
22487 (edited_line::apply_insert): Delete.
22488 (edited_line::apply_replace): Rename to...
22489 (edited_line::apply_fixit): ...this. Convert to half-open range.
22490 Update for change to type of m_line_events.
22491 * edit-context.h (edit_context::apply_insert): Delete.
22492 (edit_context::apply_replace): Rename to...
22493 (edit_context::apply_fixit): ...this.
22495 2017-05-01 Martin Sebor <msebor@redhat.com>
22497 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22500 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22503 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22504 __get_cpuid_max returns 0.
22505 (__get_cpuid_count): Ditto.
22507 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22509 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22510 replacement expression is another instance of one of its arguments.
22512 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22515 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22516 check for stack push/pop autoinc.
22517 * config/i386/i386.c (ix86_agi_dependent): Return false
22518 if the only reason why modified_in_p returned true is that
22519 addr is SP based and set_insn is a push or pop.
22521 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22523 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22526 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22529 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22530 (account_size_time): Use two predicates - exec_pred and
22532 (evaluate_conditions_for_known_args): Compute both clause and
22534 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22535 (inline_summary_t::duplicate): Update.
22536 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22538 (compute_inline_parameters): Likewise.
22539 (estimate_edge_size_and_time): Update caluclation of time.
22540 (estimate_node_size_and_time): Compute both time and nonspecialized
22542 (estimate_ipcp_clone_size_and_time): Update.
22543 (inline_merge_summary): Update.
22544 (do_estimate_edge_time): Update.
22545 (do_estimate_edge_size): Update.
22546 (do_estimate_edge_hints): Update.
22547 (inline_read_section, inline_write_summary): Stream both new predicates.
22548 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22550 (compute_inlined_call_time): Cleanup.
22551 (big_speedup_p): Update.
22552 (edge_badness): Update.
22553 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22554 (size_time_entry): Replace predicate by exec_predicate and
22555 nonconst_predicate.
22556 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22557 (estimate_edge_time): Return also nonspec_time.
22558 (reset_edge_growth_cache): Update.
22560 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22562 PR rtl-optimization/80491
22563 * ifcvt.c (noce_process_if_block): When looking for x setter
22564 with missing else_bb, don't check only the insn right before
22565 cond_earliest, but look for the last insn that x is modified in
22566 within the same bb.
22568 PR rtl-optimization/80491
22569 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22571 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22573 PR tree-optimization/80487
22574 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22576 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22578 PR tree-optimization/79697
22579 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22580 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22581 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22583 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22584 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22586 2017-04-28 Martin Sebor <msebor@redhat.com>
22588 PR tree-optimization/80523
22589 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22590 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22592 (maybe_warn, format_directive, parse_directive): Use new functions.
22593 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22595 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22597 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22599 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22601 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22602 target_header_dir): Set correctly.
22603 * configure: Regenerated.
22604 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22605 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22606 instead of SYSTEM_HEADER_DIR.
22608 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22610 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22611 (estimate_local_effects): Likewise.
22612 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22613 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22614 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22615 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22616 do_estimate_edge_time, estimate_edge_time): Likewise.
22617 * ipa-inline-analysis.c (estimate_node_size_and_time,
22618 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22619 (estimate_time_after_inlining): Remove.
22621 2017-04-28 Martin Liska <mliska@suse.cz>
22623 * doc/gcov.texi: Enhance documentation of gcov.
22625 2017-04-28 Martin Liska <mliska@suse.cz>
22627 * doc/gcov.texi: Sort options in alphabetic order.
22628 * doc/gcov-dump.texi: Likewise.
22629 * doc/gcov-tool.texi: Likewise.
22630 * gcov.c (print_usage): Likewise.
22631 * gcov-dump.c (print_usage): Likewise.
22632 * gcov-tool.c (print_merge_usage_message): Likewise.
22633 (print_rewrite_usage_message): Likewise.
22634 (print_overlap_usage_message): Likewise.
22636 2017-04-28 Martin Liska <mliska@suse.cz>
22638 PR gcov-profile/53915
22639 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22641 2017-04-28 Martin Liska <mliska@suse.cz>
22643 PR gcov-profile/79891
22644 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22645 is marked by compiler as living on a line.
22646 (get_cycles_count): Remove usage of the union.
22647 (output_intermediate_file): Likewise.
22648 (find_source): Fix GNU coding style.
22649 (accumulate_line_counts): Remove old non-all block mode.
22650 (output_lines): Remove usage of the union.
22651 * profile.c (output_location): Include all BBs, even if
22652 belonging to a same line (and file) as a previous BB.
22654 2017-04-28 Martin Liska <mliska@suse.cz>
22656 * gcov.c (process_args): Handle new argument 'w'.
22657 (read_graph_file): Assign ID to BBs.
22658 (output_branch_count): Display BB # if verbose flag is set.
22659 (output_lines): Likewise for arcs.
22660 (print_usage): Add '--verbose' option help.
22661 * doc/gcov.texi: Document --verbose (-w) option.
22663 2017-04-28 Martin Liska <mliska@suse.cz>
22665 * gcov.c (struct block_location_info): New struct.
22666 (process_file): Fill up the new structure.
22667 (read_graph_file): Replace usage of encoding by the newly added
22669 (add_line_counts): Likewise.
22670 (accumulate_line_counts): Remove usage of the union.
22671 (function_info::function_info): New function.
22672 (function_info::~function_info): Likewise.
22673 (process_file): Call delete instead of release_function.
22674 (release_function): Release the function.
22675 (release_structures): Call delete instead of release_function.
22676 (solve_flow_graph): Replace usage of num_blocks.
22677 (find_exception_blocks): Likewise.
22678 (output_lines): Fix GNU coding style.
22680 2017-04-28 Martin Liska <mliska@suse.cz>
22683 * coverage.c (coverage_remove_note_file): New function.
22684 * coverage.h: Declare the function.
22685 * toplev.c (finalize): Clean if an error has been seen.
22687 2017-04-28 Martin Liska <mliska@suse.cz>
22689 PR gcov-profile/80031
22690 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22691 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22692 * gcov.c (read_graph_file): Read just number of blocks.
22693 * profile.c (branch_prob): Do not stream 0 flags per a basic
22696 2017-04-28 Martin Liska <mliska@suse.cz>
22698 * gcov-dump.c (tag_*): Add new argument to declarations.
22699 (dump_gcov_file): Likewise.
22700 (tag_blocks): Add and use new argument depth.
22701 (tag_arcs): Likewise.
22702 (tag_lines): Likewise.
22703 (tag_counters): Likewise.
22704 (tag_summary): Likewise.
22705 (dump_working_sets): Use depth to do a proper indentation.
22707 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22710 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22711 * symtab.c (symtab_node::debug_symtab): Move definition here.
22713 2017-04-28 Richard Biener <rguenther@suse.de>
22715 * lto-streamer.h (LTO_major_version): Bump to 7.
22717 2017-04-28 Richard Biener <rguenther@suse.de>
22719 * tree-vrp.c (assert_info): New struct.
22720 (add_assert_info): New helper.
22721 (register_edge_assert_for_2): Refactor to add asserts to a vector
22723 (register_edge_assert_for_1): Likewise.
22724 (register_edge_assert_for): Likewise.
22725 (finish_register_edge_assert_for): New helper actually registering
22726 asserts where live on edge.
22727 (find_conditional_asserts): Adjust.
22728 (find_switch_asserts): Likewise.
22729 (evrp_dom_walker::try_find_new_range): Generalize.
22730 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22732 2017-04-27 Marek Polacek <polacek@redhat.com>
22735 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22736 arg10 and arg11 to itype.
22738 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22740 * doc/extend.texi (Object Size Checking): Improve grammar.
22742 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22745 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22746 that the logic for permitting reciprocal estimates matches that
22749 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22752 * tree.c (type_cache_hasher::equal): Only compare
22753 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22754 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22755 non-aggregate element types.
22756 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22757 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22759 2017-04-27 Richard Biener <rguenther@suse.de>
22761 PR middle-end/80533
22762 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22763 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22764 keeping a reference to a trailing array.
22766 2017-04-27 Richard Biener <rguenther@suse.de>
22768 PR middle-end/80539
22769 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22770 being in loop-closed SSA form conservatively.
22771 (chrec_fold_multiply_poly_poly): Likewise.
22773 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22775 PR middle-end/79665
22776 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22777 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22779 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22782 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22783 (aarch64_function_arg_alignment): Return unsigned int again, but still
22784 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22785 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22786 Don't emit -Wpsabi note.
22787 (aarch64_function_arg_boundary): Likewise.
22788 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22791 2017-04-26 Nathan Sidwell <nathan@acm.org>
22793 * tree.h (crc32_unsigned_n): Declare.
22794 (crc32_unsigned, crc32_unsigned): Make inline.
22795 * tree.c (crc32_unsigned_bits): Replace with ...
22796 (crc32_unsigned_n): ... this.
22797 (crc32_unsigned, crc32_byte): Remove.
22798 (crc32_string): Remove unnecessary braces.
22800 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22802 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22803 * ipa-inline-analysis.c (MAX_TIME): Remove.
22804 (account_size_time): Use sreal for time.
22805 (dump_inline_summary): Update.
22806 (estimate_function_body_sizes): Update.
22807 (estimate_edge_size_and_time): Update.
22808 (estimate_calls_size_and_time): Update.
22809 (estimate_node_size_and_time): Update.
22810 (inline_merge_summary): Update.
22811 (inline_update_overall_summary): Update.
22812 (estimate_time_after_inlining): Update.
22813 (inline_read_section): Update.
22814 (inline_write_summary): Update.
22815 * ipa-inline.c (compute_uninlined_call_time): Update.
22816 (compute_inlined_call_time): Update.
22817 (recursive_inlining): Update.
22818 (inline_small_functions): Update.
22819 (dump_overall_stats): Update.
22820 * ipa-inline.h: Include sreal.h.
22821 (size_time_entry): Turn time to sreal.
22822 (inline_summary): Turn self_time nad time to sreal.
22824 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22826 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22828 (sreal::stream_out, sreal::stream_in): New.
22829 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22831 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22833 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22836 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22839 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22841 (dimode_scalar_chain::compute_convert_gain): Ditto.
22842 (dimode_scalar_chain::make_vector_copies): Ditto.
22843 (dimode_scalar_chain::convert_reg): Ditto.
22844 (dimode_scalar_chain::convert_insn): Ditto.
22845 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22846 (VI248_AVX512BW_1): New mode iterator.
22847 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22848 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22851 2017-04-25 Martin Sebor <msebor@redhat.com>
22853 PR tree-optimization/80497
22854 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22855 constants are representable in HOST_WIDE_INT.
22856 (parse_directive): Ditto.
22858 2017-04-25 Martin Sebor <msebor@redhat.com>
22861 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22862 (new_zero_array): Adjust signature.
22863 (dom_info::dom_init): Used unsigned rather that size_t.
22864 (dom_info::dom_info): Same.
22866 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22867 Jakub Jelinek <jakub@redhat.com>
22870 * config/arm/arm.c: Include gimple.h.
22871 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22872 returns negative, increment ncrn only if it returned positive.
22873 (arm_needs_doubleword_align): Return int instead of bool,
22874 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22875 members, but if there is any such non-FIELD_DECL
22876 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22877 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22878 returns negative, increment nregs only if it returned positive.
22879 (arm_setup_incoming_varargs): Likewise.
22880 (arm_function_arg_boundary): Emit -Wpsabi note if
22881 arm_needs_doubleword_align returns negative, return
22882 DOUBLEWORD_ALIGNMENT only if it returned positive.
22884 2017-04-25 Marek Polacek <polacek@redhat.com>
22887 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22888 first argument to type.
22890 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22893 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22894 type checks to test for compatibility instead of equality.
22896 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22897 Jakub Jelinek <jakub@redhat.com>
22900 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22902 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22903 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22904 the alignment computation, but return their maximum in warn_alignment.
22905 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22906 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22908 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22909 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22914 * config/arc/simdext.md (dmpyh): Fix typo.
22916 2017-04-25 Richard Biener <rguenther@suse.de>
22918 PR tree-optimization/80492
22919 * alias.c (compare_base_decls): Handle registers with asm
22920 specification conservatively.
22921 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22922 compare_base_decls returning dont-know properly.
22924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22926 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22927 (legitimate_offset_address_p): New function.
22928 (arc_legitimate_address_p): Use above function.
22930 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22932 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22934 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22936 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22937 ACCH registers whenever they are available.
22939 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22941 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22942 double regs fix when not used.
22944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22946 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22948 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22949 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22951 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22953 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22954 when emitting short ADD instructions.
22956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22958 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22960 (cmpsi_cc_c_insn): Likewise.
22961 (cbranchsi4_scratch): Compute proper instruction length using
22962 compact_hreg_operand.
22963 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22965 2017-04-25 Richard Biener <rguenther@suse.de>
22967 PR middle-end/80509
22968 * passes.c (pass_manager::pass_manager): Initialize
22969 m_name_to_pass_map.
22971 2017-04-25 Richard Biener <rguenther@suse.de>
22973 PR tree-optimization/79201
22974 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22976 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22979 * config/s390/vector.md: Split MEM->GPR vector moves for
22980 non-s_operand addresses.
22982 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22985 * config/s390/predicates.md (reload_const_wide_int_operand): New
22987 * config/s390/s390.md ("movti"): Remove d/P alternative.
22988 ("movti_bigconst"): New pattern definition.
22990 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22993 * s390-protos.h (s390_expand_cs_hqi): Removed.
22994 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22995 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22996 modes as well as CCZ1mode and CCZmode.
22997 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22998 signature of s390_emit_compare_and_swap.
22999 (s390_expand_cs_hqi): Likewise, make static.
23000 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23001 compare-and-swap, in some cases.
23002 (s390_expand_cs): Wrapper function.
23003 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23005 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23006 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23007 patterns for small and large integers. Forbid symref memory operands.
23008 Move expander to s390.c. Require cc register.
23009 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23010 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23011 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23012 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23013 symref memory operands. Remove CC mode and call s390_match_ccmode
23015 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23017 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23019 * config/s390/s390.md (define_peephole2): New peephole to help
23020 combining the load-and-test pattern with volatile memory.
23022 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23024 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23025 with CCZmode for TARGET_Z196.
23027 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23029 PR rtl-optimization/80501
23030 * combine.c (make_compound_operation_int): Set subreg_code to SET
23031 even for AND with mask of the sign bit of mode.
23033 PR rtl-optimization/80500
23034 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23035 sum's initial value.
23037 2017-04-25 Julian Brown <julian@codesourcery.com>
23038 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23040 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23042 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23044 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23046 2017-04-25 Julian Brown <julian@codesourcery.com>
23047 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23049 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23050 (thunderx2t99_sha): New Reservation.
23052 2017-04-25 Julian Brown <julian@codesourcery.com>
23053 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23055 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23056 type for 1-element load.
23058 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23060 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23062 2017-04-24 Martin Jambor <mjambor@suse.cz>
23064 PR tree-optimization/80293
23065 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23066 char arrays not totally scalarizable if it is false.
23067 (analyze_all_variable_accesses): Pass correct value in the new
23068 parameter. Add a statistics counter.
23070 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23072 PR middle-end/79931
23073 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23075 2017-04-24 Richard Biener <rguenther@suse.de>
23077 PR tree-optimization/80494
23078 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23079 out for complex types.
23081 2017-04-24 Richard Biener <rguenther@suse.de>
23083 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23084 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23085 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23086 (DFS): Adjust and never fail.
23087 (sccvn_dom_walker::fail): Remove.
23088 (sccvn_dom_walker::before_dom_children): Adjust.
23089 (run_scc_vn): Likewise and never fail.
23090 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23091 (pass_fre::execute): Likewise.
23093 2017-04-24 Richard Biener <rguenther@suse.de>
23095 PR tree-optimization/79725
23096 * tree-ssa-sink.c (statement_sink_location): Return whether
23097 failure reason was zero uses. Move that check later.
23098 (sink_code_in_bb): Deal with zero uses by removing the stmt
23101 2017-04-24 Richard Biener <rguenther@suse.de>
23104 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23105 pointer-based references.
23107 2017-04-24 Richard Biener <rguenther@suse.de>
23110 * pass_manager.h (pass_manager::operator new): Remove.
23111 (pass_manager::operator delete): Likewise.
23112 * passes.c (pass_manager::operator new): Remove.
23113 (pass_manager::operator delete): Likewise.
23114 (pass_manager::pass_manager): Zero individual pass members.
23116 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23119 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23120 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23121 Check "XEXP (src, 1)" operand here.
23122 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23123 Check "XEXP (src, 1)" operand here.
23124 (dimode_scalar_chain::make_vector_copies): Detect count register
23125 of a shift instruction. Zero extend count register from QImode
23126 to DImode to satisfy vector shift pattern count operand predicate.
23127 Substitute vector shift count operand with a DImode copy.
23128 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23131 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23133 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23134 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23135 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23136 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23137 (UNSPEC_NOREX_MEM): Remove definition.
23139 2017-04-21 Richard Biener <rguenther@suse.de>
23141 PR tree-optimization/79547
23142 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23143 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23144 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23145 without any constraints.
23147 2017-04-21 Richard Biener <rguenther@suse.de>
23149 PR tree-optimization/78847
23150 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23152 2017-04-21 Richard Biener <rguenther@suse.de>
23154 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23155 (build_distinct_type_copy): Likewise.
23156 (build_variant_type_copy): Likewise.
23157 * tree.c (build_qualified_type): Pass down mem-stat info.
23158 (build_distinct_type_copy): Likewise.
23159 (build_variant_type_copy): Likewise.
23161 2017-04-21 Richard Biener <rguenther@suse.de>
23163 PR tree-optimization/80237
23164 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23166 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23167 for a simplified result.
23169 2016-04-21 Richard Biener <rguenther@suse.de>
23171 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23172 sth as strict as a simple_iv but a chrec without symbols and an
23173 operand defined in the loop we are peeling (and not some subloop).
23174 (propagate_constants_for_unrolling): Propagate all constants.
23176 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23179 * config/i386/i386.c (print_reg): Remove assert for disalowed
23180 regno values, call output_operand_lossage instead.
23182 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23185 * config/i386/constraints.md (Yc): New register constraint.
23186 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23187 Use Yc constraint for alternative 2 of operand 0. Remove
23188 preferred_for_speed attribute.
23190 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23192 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23193 lastprivate clauses in SIMT case.
23195 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23197 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23199 2017-04-20 Richard Biener <rguenther@suse.de>
23201 PR tree-optimization/57796
23202 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23203 as N scalar stores.
23204 (vect_model_load_cost): Cost gathers as N scalar loads.
23206 2017-04-20 Richard Biener <rguenther@suse.de>
23208 * ggc-page.c (ggc_allocated_p): Rename to ...
23209 (safe_lookup_page_table_entry): ... this and return the lookup
23211 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23213 2017-04-20 Richard Biener <rguenther@suse.de>
23215 PR tree-optimization/80453
23216 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23217 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23218 from the conditions.
23219 (vn_phi_eq): Pass them down.
23220 (vn_phi_lookup): Record them.
23221 (vn_phi_insert): Likewise.
23223 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23225 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23226 uninitialized variable warning to avoid buffer overrun.
23228 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23231 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23232 is suppressed for '{ 0 }' in C.
23234 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23236 * BASE-VER: Set to 8.0.0.
23238 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23240 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23241 priority .init_array and .fini_array section with SECTION_NOTYPE
23244 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23246 PR middle-end/80423
23247 * tree.h (build_array_type): Add typeless_storage default argument.
23248 * tree.c (type_cache_hasher::equal): Also compare
23249 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23250 (build_array_type): Add typeless_storage argument, set
23251 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23253 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23254 (build_array_type): Likewise. Add typeless_storage argument.
23256 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23257 Jakub Jelinek <jakub@redhat.com>
23259 PR tree-optimization/80426
23260 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23261 operation on symbolic operands, also compute the overflow for the
23262 invariant part when the operation degenerates into a negation.
23264 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23267 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23268 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23271 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23273 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23276 * config/avr/avr.c (tree.h): Include it.
23277 (cgraph.h): Include it.
23278 (avr_encode_section_info): Don't warn for uninitialized progmem
23279 variable if it's just an alias.
23281 2017-04-19 Richard Biener <rguenther@suse.de>
23284 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23285 when needed by AutoPGO.
23287 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23290 * doc/lto.texi: Remove an extra 'that'.
23292 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23294 PR rtl-optimization/80429
23295 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23296 are only used in debug insns.
23298 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23299 Vladimir Makarov <vmakarov@redhat.com>
23301 * config/sparc/predicates.md (input_operand): Add comment. Return
23302 true for any memory operand when LRA is in progress.
23303 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23305 2017-04-18 Jeff Law <law@redhat.com>
23308 * mips.md ({return,simple_return}_internal): Do not overwrite
23311 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23313 PR tree-optimization/80443
23314 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23315 instead of adding 1, subtract -1 and similarly instead of subtracting
23318 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23320 PR rtl-optimization/80357
23321 * haifa-sched.c (tmp_bitmap): New variable.
23322 (model_recompute): Handle duplicate use records.
23323 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23324 (free_global_sched_pressure_data): Free it.
23326 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23329 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23330 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23331 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23332 instead of SYSTEM_HEADER_DIR.
23334 2017-04-18 Jeff Law <law@redhat.com>
23336 PR middle-end/80422
23337 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23338 predecessors after walking up the insn chain.
23340 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23343 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23344 sizetype type into debug info.
23346 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23349 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23350 unneeded test for TARGET_UPPER_REGS_SF.
23351 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23353 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23356 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23357 instead of gsi_after_labels.
23359 2017-04-18 Jeff Law <law@redhat.com>
23361 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23365 2017-04-13 Jeff Law <law@redhat.com>
23366 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23367 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23369 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23372 * config/avr/avr.c (intl.h): Include it.
23373 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23375 2017-04-18 Martin Liska <mliska@suse.cz>
23377 PR gcov-profile/78783
23378 * gcov-tool.c (gcov_output_files): Validate that destination
23379 file is either removed by the tool or by a user.
23381 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23382 Guy Benyei <guybe@mellanox.com>
23384 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23385 block, and do not negate it, the stored id is already negative.
23387 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23389 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23391 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23394 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23395 masks of options that should be turned off if the VSX vector
23396 options are turned off.
23397 (OTHER_P8_VECTOR_MASKS): Likewise.
23398 (OTHER_VSX_VECTOR_MASKS): Likewise.
23399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23400 rs6000_disable_incompatible_switches to validate no type switches
23402 (rs6000_incompatible_switch): New function to disallow turning on
23403 other vector options if -mno-vsx, -mno-power8-vector, or
23404 -mno-power9-vector are specified.
23406 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23408 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23410 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23412 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23413 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23414 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23415 (ARG_POINTER_CFA_OFFSET): Likewise.
23417 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23419 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23420 conditions to take advantage of various optimizations.
23422 2017-04-13 Jeff Law <law@redhat.com>
23424 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23425 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23426 (zero_extendsidi2_dext): Likewise.
23428 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23431 * fold-const.c (fold_ternary_loc): Revert
23432 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23435 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23437 PR rtl-optimization/80343
23438 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23439 new scratch pseudo.
23441 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23444 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23445 to ubsan_encode_value.
23447 2017-04-13 Jeff Law <law@redhat.com>
23449 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23450 appearing in DEBUG_INSNs.
23452 2017-04-13 Martin Liska <mliska@suse.cz>
23454 PR gcov-profile/80413
23455 * gcov-io.c (gcov_write_string): Copy to buffer just when
23456 allocated size is greater than zero.
23458 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23461 * dwarf2out.c (decls_for_scope): Ignore declarations of
23462 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23464 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23467 * ipa-visibility.c (non_local_p): Fix typos.
23468 (localize_node): When localizing symbol in same comdat group,
23469 dissolve the group only when we know external symbols are going
23471 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23473 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23475 PR tree-optimization/79390
23476 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23477 order does not result in usable sequence, retry with reversed operand
23483 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23484 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23485 op0 instead of fold_convert_loc (loc, type, arg0).
23487 2017-04-12 Jeff Law <law@redhat.com>
23489 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23490 has a delay slot in the generated code.
23492 * config/cris/cris.md (cris_preferred_reload_class): Return
23493 GENNONACR_REGS rather than GENERAL_REGS.
23495 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23498 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23499 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23500 signedness of the result type.
23502 2017-04-12 Richard Biener <rguenther@suse.de>
23503 Jeff Law <law@redhat.com>
23505 PR tree-optimization/80359
23506 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23507 trim stores to TARGET_MEM_REFs.
23509 2017-04-12 Richard Biener <rguenther@suse.de>
23511 PR tree-optimization/79390
23512 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23513 threading case even more.
23515 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23518 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23519 for quad_address_p for TImode, instead of just not indexed_address.
23521 2017-04-12 Richard Biener <rguenther@suse.de>
23522 Bernd Edlinger <bernd.edlinger@hotmail.de>
23524 PR middle-end/79671
23525 * alias.c (component_uses_parent_alias_set_from): Handle
23526 TYPE_TYPELESS_STORAGE.
23527 (get_alias_set): Likewise.
23528 * tree-core.h (tree_type_common): Add typeless_storage flag.
23529 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23530 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23531 for types containing members with TYPE_TYPELESS_STORAGE.
23532 (place_field): Likewise.
23533 (layout_type): Likewise for ARRAY_TYPE.
23534 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23535 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23536 TYPE_TYPELESS_STORAGE.
23537 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23539 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23542 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23543 first argument to type.
23545 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23549 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23550 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23551 (rs6000_expand_binop_builtin): Likewise.
23552 (rs6000_expand_ternop_builtin): Likewise; also add missing
23553 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23554 vshasigma built-ins.
23555 * doc/extend.texi: Document that vec_xxpermdi's third argument
23556 must be a constant.
23558 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23560 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23561 Use shift_const cost parameter when calculating gain of STV shifts.
23563 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23565 PR rtl-optimization/70478
23566 * lra-constraints.c (process_alt_operands): Check memory for
23567 disfavoring memory insn operand.
23569 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23571 PR middle-end/80100
23572 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23573 left shift in unsigned HOST_WIDE_INT type.
23575 PR rtl-optimization/80385
23576 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23577 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23580 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23581 if they have any depend clauses.
23583 2017-04-11 Martin Liska <mliska@suse.cz>
23586 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23587 * ipa-split.c (split_function): Create a local comdat symbol
23588 if caller is in a comdat group.
23590 2017-04-11 Martin Liska <mliska@suse.cz>
23593 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23596 2017-04-11 Martin Sebor <msebor@redhat.com>
23598 PR middle-end/80364
23599 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23600 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23602 (directive::set_width, directive::set_precision, format_character):
23604 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23607 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23610 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23611 conflict, set target->arch_name instead of target->cpu_name.
23613 2017-04-11 Richard Biener <rguenther@suse.de>
23615 PR tree-optimization/80374
23616 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23617 build_zero_cst, remove fold_convertible_p check again.
23619 2017-04-11 Martin Liska <mliska@suse.cz>
23622 * ubsan.c (instrument_object_size): Do not instrument register
23625 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23628 * config/i386/i386-builtin-types.def
23629 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23630 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23631 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23632 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23633 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23634 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23635 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23636 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23637 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23638 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23639 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23640 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23641 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23642 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23643 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23644 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23645 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23646 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23647 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23648 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23649 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23650 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23651 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23652 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23653 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23654 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23655 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23656 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23657 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23658 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23659 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23660 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23661 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23662 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23663 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23664 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23665 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23666 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23667 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23668 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23669 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23670 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23671 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23672 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23673 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23674 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23676 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23677 flag to second_arg_count, handle 4 argument function type _COUNT
23678 aliases, handle second_arg_count on second argument rather than last.
23680 2017-04-10 Jeff Law <law@redhat.com>
23682 PR tree-optimization/80374
23683 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23684 record anything if we can not convert integer_zero_node to the
23687 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23690 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23691 Enhance special handling given to the TARGET_P9_MINMAX option in
23692 relation to certain other options.
23694 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23696 PR tree-optimization/80153
23697 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23698 remove POINTER_PLUS_EXPR's base part directly, rather than through
23701 2017-04-10 Richard Biener <rguenther@suse.de>
23702 Bin Cheng <bin.cheng@arm.com>
23704 PR tree-optimization/80153
23705 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23706 the first element of pointer type aff_tree. Build result expr in
23708 (add_elt_to_tree): Convert to type unconditionally. Remove other
23709 fold_convert calls.
23710 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23711 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23713 2017-04-10 Richard Biener <rguenther@suse.de>
23715 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23718 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23720 PR rtl-optimization/70478
23721 * lra-constraints.c (curr_small_class_check): New.
23722 (update_and_check_small_class_inputs): New.
23723 (process_alt_operands): Update curr_small_class_check. Disfavor
23724 alternative insn memory operands. Check available regs for small
23727 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23730 * config/mips/mips.opt (-mvirt): Update description.
23731 * doc/invoke.texi (-mvirt): Likewise.
23733 2017-04-10 Richard Biener <rguenther@suse.de>
23735 PR middle-end/80362
23736 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23737 looking for NEGATE_EXPR in -A / -B to A / B folding.
23739 2017-04-10 Martin Liska <mliska@suse.cz>
23741 PR gcov-profile/80224
23742 * gcov.c (print_usage): Fix usage string.
23743 (get_gcov_intermediate_filename): Remove.
23744 (output_gcov_file): Use both for normal and intermediate format.
23745 (generate_results): Do not initialize special file for
23746 intermediate format.
23748 2017-04-10 Richard Biener <rguenther@suse.de>
23750 PR tree-optimization/80304
23751 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23754 2017-04-10 Nathan Sidwell <nathan@acm.org>
23757 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23758 (rs6000_init_builtins): Use it.
23760 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23762 * config/arm/arm.md (<mrc>): Add mode to SET source.
23763 (<mrrc>): Likewise.
23765 2017-04-10 Richard Biener <rguenther@suse.de>
23767 PR middle-end/80344
23768 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23770 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23773 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23774 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23775 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23776 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23777 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23778 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23779 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23780 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23781 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23782 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23783 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23784 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23785 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23786 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23787 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23788 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23789 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23790 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23791 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23792 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23793 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23794 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23795 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23797 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23799 PR rtl-optimization/70478
23800 * lra-constraints.c: Reverse the last patch.
23802 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23804 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23805 Add comment for WCHAR_T.
23807 2017-04-08 Martin Liska <mliska@suse.cz>
23810 2017-04-07 Martin Liska <mliska@suse.cz>
23813 * ipa-split.c (split_function): Add function part to a same comdat
23816 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23819 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23821 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23823 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23824 (vec_pairing): ...to this.
23825 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23826 (rs6000_sched_init): Adjust for name change.
23827 (struct rs6000_sched_context): Likewise.
23828 (rs6000_init_sched_context): Likewise.
23829 (rs6000_set_sched_context): Likewise.
23831 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23837 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23839 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23840 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23841 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23843 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23845 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23847 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23849 PR rtl-optimization/70703
23850 * ira-color.c (update_conflict_hard_regno_costs): Use
23851 int64_t instead of HOST_WIDE_INT.
23853 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23855 PR rtl-optimization/70478
23856 * lra-constraints.c (process_alt_operands): Disfavor alternative
23857 insn memory operands.
23859 2017-04-07 Jeff Law <law@redhat.com>
23861 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23862 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23864 2017-04-07 Martin Liska <mliska@suse.cz>
23867 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23868 Show error message instead of an ICE.
23870 2017-04-07 Martin Liska <mliska@suse.cz>
23873 * ipa-split.c (split_function): Add function part to a same comdat
23876 2017-04-07 Richard Biener <rguenther@suse.de>
23878 PR middle-end/80341
23879 * tree.c (get_unwidened): Also handle ! for_type case for
23881 * convert.c (do_narrow): Split out from ...
23882 (convert_to_integer_1): ... here. Do not pass final truncation
23883 type to get_unwidened for TRUNC_DIV_EXPR.
23885 2017-04-07 Richard Biener <rguenther@suse.de>
23887 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23889 (aff_combination_const): Adjust.
23890 (aff_combination_scale): Likewise.
23891 (aff_combination_add_elt): Likewise.
23892 (aff_combination_add_cst): Likewise.
23893 (aff_combination_convert): Likewise.
23894 (add_elt_to_tree): Likewise. Remove unused argument.
23895 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23897 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23899 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23901 * config/arm/arm.c (arm_default_short_enums): Use
23902 ARM_DEFAULT_SHORT_ENUMS.
23903 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23905 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23908 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23909 members with redundant out-of-class redeclaration.
23911 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23914 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23915 * config/i386/i386.md (*zero_extendsidi2):
23916 Add (?*x,*x) and (?*v,*v) alternatives.
23918 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23921 * config/i386/i386.c (ix86_expand_builtin)
23922 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23923 mode from insn data. Convert operands to insn operand mode.
23924 Copy operands that don't satisfy insn predicate to a register.
23926 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23928 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23931 2017-04-06 Richard Biener <rguenther@suse.de>
23933 PR tree-optimization/80334
23934 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23935 preserve alignment of accesses.
23937 2017-04-06 Richard Biener <rguenther@suse.de>
23939 PR tree-optimization/80262
23940 * tree-sra.c (build_ref_for_offset): Preserve address-space
23942 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23943 Drop useless address-space information on MEM_REF offsets.
23945 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23947 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23949 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23951 PR rtl-optimization/70703
23952 * ira-color.c (update_conflict_hard_regno_costs): Use
23953 HOST_WIDE_INT instead of long.
23955 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23958 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23959 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23961 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23962 for x86_64 target. Handle -m3dnowa option.
23964 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23966 PR rtl-optimization/70703
23967 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23968 (update_conflict_hard_regno_costs): Use long instead of unsigned
23969 arithmetic for cost calculation.
23971 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23972 Bernd Edlinger <bernd.edlinger@hotmail.de>
23975 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23978 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23981 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23982 ptr_mode with Pmode throughout.
23983 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23984 into probe_stack_range and use DImode.
23986 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23989 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23990 call_eh_return is true.
23992 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23994 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23995 Initialize last_match_fntype_index.
23997 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24000 * tree-nvr.c: Include internal-fn.h.
24001 (pass_return_slot::execute): Ignore internal calls without
24004 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24005 Richard Biener <rguenther@suse.de>
24008 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24009 captures used multiple times, except for the last use.
24010 * generic-match-head.c: Include gimplify.h.
24012 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24014 PR tree-optimization/79390
24015 * target.h (struct noce_if_info): Declare.
24016 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24017 * target.def (noce_conversion_profitable_p): New target hook.
24018 * ifcvt.h (struct noce_if_info): New type, moved from ...
24019 * ifcvt.c (struct noce_if_info): ... here.
24020 (noce_conversion_profitable_p): Renamed to ...
24021 (default_noce_conversion_profitable_p): ... this. No longer
24023 (noce_try_store_flag_constants, noce_try_addcc,
24024 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24025 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24026 instead of noce_conversion_profitable_p.
24027 * config/i386/i386.c: Include ifcvt.h.
24028 (ix86_option_override_internal): Don't override
24029 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24030 (ix86_noce_conversion_profitable_p): New function.
24031 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24032 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24033 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24034 * doc/tm.texi: Regenerated.
24036 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24038 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24041 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24044 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24045 instructions for small multiply cores.
24047 2017-04-04 Jeff Law <law@redhat.com>
24049 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24051 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24052 that are not set by the loop over the elements.
24054 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24057 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24058 int mode, convert_modes it to mode as unsigned, otherwise use
24059 lowpart_subreg to mode rather than SImode.
24060 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24061 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24062 Use DImode instead of SImode for the shift count operand.
24063 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24066 2017-04-04 Richard Biener <rguenther@suse.de>
24068 PR middle-end/80281
24069 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24070 arithmetic done for the negate or the plus. Simplify.
24071 (A - (-B) -> A + B): Likewise.
24072 * fold-const.c (split_tree): Make sure to not negate pointers.
24074 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24076 PR rtl-optimization/60818
24077 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24078 a compare of comparisons with the thing compared if this results
24079 in a different machine mode.
24081 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24083 * alias.c (base_alias_check): Fix typo in comment.
24084 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24085 * cgraphunit.c (symbol_table::compile): Likewise.
24086 * collect2.c (maybe_run_lto_and_relink): Likewise.
24087 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24088 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24089 * config/avr/avr.c (avr_map_op_t): Likewise.
24090 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24091 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24092 * config/epiphany/epiphany.md (movcc): Likewise.
24093 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24094 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24096 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24097 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24098 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24099 * config/sh/sh.c (sh_rtx_costs): Likewise.
24100 * fold-const.c (fold_truth_andor): Likewise.
24101 * genautomata.c (collapse_flag): Likewise.
24102 * gengtype.h (struct type::u::s): Likewise.
24103 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24104 * input.c (FORMAT_AMOUNT): Likewise.
24105 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24106 (known_aggs_to_agg_replacement_list): Likewise.
24107 * ipa-inline-analysis.c: Likewise.
24108 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24109 * ipa-polymorphic-call.c
24110 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24111 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24112 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24114 * modulo-sched.c (apply_reg_moves): Likewise.
24115 * omp-expand.c (build_omp_regions_1): Likewise.
24116 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24117 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24118 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24119 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24120 * value-prof.c: Likewise.
24121 * var-tracking.c (val_reset): Likewise.
24123 2017-04-03 Richard Biener <rguenther@suse.de>
24125 PR tree-optimization/80275
24126 * fold-const.c (split_address_to_core_and_offset): Handle
24129 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24131 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24132 descriptors is at least equal to that of functions.
24134 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24136 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24138 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24141 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24142 (mov<IMOD4:mode>): New expander.
24143 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24145 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24147 PR rtl-optimization/79405
24148 * fwprop.c (propagations_left): New variable.
24149 (forward_propagate_into): Decrement it.
24150 (fwprop_init): Initialize it.
24151 (fw_prop): If the variable has reached zero, stop propagating.
24152 (fwprop_addr): Ditto.
24154 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24157 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24158 a FUNCTION_DECL, pass it as decl instead of origin to
24161 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24163 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24166 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24169 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24170 TARGET_VSX_SMALL_INTEGER.
24172 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24174 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24175 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24177 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24179 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24180 extraction from odd-numbered MSA register.
24182 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24184 PR middle-end/80173
24185 * expmed.c (store_bit_field_1): Don't attempt to create
24186 a word subreg out of hard registers wider than word if they
24187 have HARD_REGNO_NREGS of 1 for their mode.
24189 PR middle-end/80163
24190 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24191 conversions to integer types wider than word and pointer.
24194 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24195 (rtx_equal_for_cselib_p): Pass 0 to it.
24196 * cselib.c (cselib_hasher::equal): Likewise.
24197 (rtx_equal_for_cselib_1): Add depth argument. If depth
24198 is 128, don't look up VALUE locs and punt. Increment
24199 depth in recursive calls when walking VALUE locs.
24201 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24203 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24204 (make_gcov_file_name): Use the canonical path name for generating
24206 (read_line): Fix handling of files with ascii null bytes.
24208 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24210 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24211 to initialise a vector register instead
24212 of using a const_int.
24214 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24216 PR translation/80189
24217 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24218 diagnostic messages.
24220 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24223 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24224 (dfp_diex_<mode>): Update mode of operand 1.
24225 * doc/extend.texi (dxex, dxexq): Document change to return type.
24226 (diex, diexq): Document change to argument type.
24228 2017-03-30 Martin Jambor <mjambor@suse.cz>
24231 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24232 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24233 it reflects the signature changes performed at the callee side.
24234 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24235 to cgraph_build_function_type_skip_args.
24236 (build_function_decl_skip_args): Adjust call to the above function.
24238 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24241 * config/i386/sse.md
24242 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24243 register as dest whenever it is a MEM not rtx_equal_p to the
24244 corresponding dup operand, and when forcing into reg move the
24245 reg into the memory afterwards.
24246 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24247 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24248 for the force_reg mode.
24249 (avx512vl_vextractf128<mode>): Use register as dest either
24250 always when a MEM, or when it is a MEM not rtx_equal_p to the
24251 corresponding dup operand, or even not when it is a CONST_VECTOR
24252 depending on the mode and lo vs. hi.
24253 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24255 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24256 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24257 Likewise. Require that operands[2] is even.
24258 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24259 Remove extraneous parens. Require that operands[2] is a multiple
24261 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24262 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24263 disallow memory then.
24265 2017-03-30 Richard Biener <rguenther@suse.de>
24267 PR tree-optimization/77498
24268 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24269 to non-constants over backedges.
24271 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24273 PR rtl-optimization/80233
24274 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24275 as last_combined_insn. Do not test for BARRIER_P separately.
24277 2017-03-29 Andreas Schwab <schwab@suse.de>
24280 * calls.c (prepare_call_address): Convert funexp to Pmode before
24281 copying to temp reg.
24283 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24285 PR tree-optimization/80158
24286 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24287 Handle possible future case of more than one alternate
24289 (replace_rhs_if_not_dup): Likewise.
24290 (replace_one_candidate): Likewise.
24292 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24294 PR rtl-optimization/80193
24295 * ira.c (ira): Do not check allocation for LRA.
24297 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24299 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24300 (nvptx_output_simt_exit): Declare.
24301 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24302 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24303 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24304 (nvptx_declare_function_name): ...here. Emit declaration of local
24305 memory space buffer for omp_simt_enter insn.
24306 (nvptx_output_unisimt_switch): New.
24307 (nvptx_output_softstack_switch): New.
24308 (nvptx_output_simt_enter): New.
24309 (nvptx_output_simt_exit): New.
24310 * config/nvptx/nvptx.h (struct machine_function): New fields
24311 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24312 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24313 (UNSPECV_SIMT_EXIT): Ditto.
24314 (omp_simt_enter_insn): New insn.
24315 (omp_simt_enter): New expansion.
24316 (omp_simt_exit): New insn.
24317 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24319 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24320 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24321 (expand_GOMP_SIMT_EXIT): New.
24322 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24323 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24324 (GOMP_SIMT_EXIT): Ditto.
24325 * target-insns.def (omp_simt_enter): New insn.
24326 (omp_simt_exit): Ditto.
24327 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24329 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24330 (lower_rec_input_clauses): Likewise.
24331 (lower_lastprivate_clauses): Handle SIMT privatization.
24333 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24334 (ompdevlow_adjust_simt_enter): New.
24335 (find_simtpriv_var_op): New.
24336 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24337 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24339 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24340 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24341 (copy_decl_for_dup_finish): Ditto.
24343 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24345 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24348 * config/i386/i386.c (ix86_option_override_internal): Always
24349 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24351 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24353 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24355 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24357 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24358 mark new edge's irreducible flag accordign to it.
24359 (vect_do_peeling): Check loop preheader edge's irreducible flag
24360 and pass it to function slpeel_add_loop_guard.
24362 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24364 PR tree-optimization/80218
24365 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24366 Update block frequencies and counts.
24368 2017-03-28 Richard Biener <rguenther@suse.de>
24370 PR tree-optimization/78644
24371 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24372 of a simplification result we may not use it at all.
24374 2017-03-28 Richard Biener <rguenther@suse.de>
24377 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24378 without arguments, generate default definition of a SSA name.
24380 2017-03-28 Richard Biener <rguenther@suse.de>
24382 PR middle-end/80222
24383 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24384 TYPE_REF_CAN_ALIAS_ALL references.
24385 * fold-const.c (fold_indirect_ref_1): Likewise.
24387 2017-03-28 Martin Liska <mliska@suse.cz>
24390 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24391 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24393 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24394 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24396 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24397 (EXTRA_SPECS): Define.
24398 (SUBTARGET_EXTRA_SPECS): Likewise.
24399 (SUBTARGET_CPP_SPEC): Likewise.
24400 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24401 SUBTARGET_EXTRA_SPECS.
24402 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24404 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24406 * config/arc/simdext.md (vst64_insn): Update pattern.
24407 (vld32wh_insn): Likewise.
24408 (vld32wl_insn): Likewise.
24409 (vld64_insn): Likewise.
24410 (vld32_insn): Likewise.
24412 2017-03-28 Marek Polacek <polacek@redhat.com>
24415 * fold-const.c (fold_comparison): Use protected_set_expr_location
24416 instead of SET_EXPR_LOCATION.
24418 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24420 * tree.c (add_expr): Avoid name lookup warning.
24422 2017-03-27 Jeff Law <law@redhat.com>
24424 PR tree-optimization/80216
24425 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24426 function name. Limit recursion depth.
24427 (record_temporary_equivalences): Corresponding changes.
24429 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24431 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24434 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24437 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24439 * cfgcleanup.c (reg_note_cfa_p): New array.
24440 (insns_have_identical_cfa_notes): New function.
24441 (old_insns_match_p): Don't cross-jump in between /f
24442 and non-/f instructions. If both i1 and i2 are frame related,
24443 verify all CFA notes, their order and content.
24445 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24448 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24449 HImode and SImode with zero extend to DImode to one insn.
24450 (bswap<mode>2_extenddi): Likewise.
24451 (bswapsi2_extenddi): Likewise.
24452 (bswaphi2_extendsi): Likewise.
24453 (bswaphi2): Combine bswap HImode and SImode into one insn.
24454 Separate memory insns from swapping register.
24455 (bswapsi2): Likewise.
24456 (bswap<mode>2): Likewise.
24457 (bswaphi2_internal): Delete, no longer used.
24458 (bswapsi2_internal): Likewise.
24459 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24460 store, and gpr<-gpr swap insns.
24461 (bswap<mode>2_store): Likewise.
24462 (bswaphi2_reg): Register only splitter, combine with the splitter.
24463 (bswaphi2 splitter): Likewise.
24464 (bswapsi2_reg): Likewise.
24465 (bswapsi2 splitter): Likewise.
24466 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24467 the insns into load, store, and register/register insns.
24468 (bswapdi2_ldbrx): Likewise.
24469 (bswapdi2_load): Likewise.
24470 (bswapdi2_store): Likewise.
24471 (bswapdi2_reg): Likewise.
24473 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24475 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24476 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24478 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24481 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24484 special handling for target option conflicts between dform
24485 options (-mpower9-dform, -mpower9-dform-vector,
24486 -mpower9-dform-scalar) and -mno-direct-move.
24488 2017-03-27 Richard Biener <rguenther@suse.de>
24490 PR tree-optimization/80181
24491 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24493 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24495 * config/arc/predicates.md (move_double_src_operand): Replace the
24496 call to move_double_src_operand with a call to address_operand.
24498 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24500 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24501 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24502 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24504 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24506 * config/arc/predicates.md (long_immediate_loadstore_operand):
24507 Consider scaled addresses cases.
24509 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24511 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24512 restored when in interrupt.
24513 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24514 doesn't have delay slot.
24516 2017-03-27 Richard Biener <rguenther@suse.de>
24519 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24520 inlined thunk clones.
24522 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24525 * asan.c (instrument_derefs): Copy over last operand from
24526 original COMPONENT_REF to the new COMPONENT_REF with
24527 DECL_BIT_FIELD_REPRESENTATIVE.
24528 * ubsan.c (instrument_object_size): Likewise.
24530 2017-03-27 Richard Biener <rguenther@suse.de>
24532 PR tree-optimization/80170
24533 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24534 sure DR/SCEV didnt fold in constants we do not see when looking
24535 at the reference base alignment.
24537 2017-03-27 Richard Biener <rguenther@suse.de>
24539 PR middle-end/80171
24540 * gimple-fold.c (fold_ctor_reference): Properly guard against
24541 NULL return value from canonicalize_constructor_val.
24543 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24546 * config/i386/i386.c (ix86_expand_builtin)
24547 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24548 flags reg setting and flags reg using instructions.
24549 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24550 clobbering instructions to zero extend op2.
24552 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24554 * doc/install.texi (Configuration) <--with-aix-soname>:
24555 Update link to AIX ld.
24557 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24559 PR rtl-optimization/80160
24560 PR rtl-optimization/80159
24561 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24562 reg_alternate_class into account.
24564 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24567 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24568 to consider in curr_insn_transform.
24570 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24572 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24573 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24574 and emit_mode_inner.
24576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24578 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24579 argument to the overloaded builtin variants. Use the new flag to
24580 deprecate certain builtin variants.
24581 * config/s390/s390-builtin-types.def: Add new builtin types.
24582 * config/s390/s390-builtins.h: Support new flags field for
24583 overloaded builtins.
24584 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24585 (s390_macro_to_expand): Enable vector float data type.
24586 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24587 builtins by incrementing the __VEC__ version number.
24588 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24590 (s390_resolve_overloaded_builtin): Emit error messages depending
24591 on the builtin flags.
24592 * config/s390/s390.c (s390_expand_builtin): Support additional
24593 flags argument. Change error message to match the messages
24594 emitted in s390-c.c.
24595 * config/s390/s390.md: New UNSPEC_* constants.
24596 (op_type): Add new instruction types.
24597 * config/s390/vecintrin.h: Add new builtins and test data class
24599 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24600 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24601 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24602 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24603 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24604 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24606 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24607 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24608 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24609 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24611 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24612 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24613 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24614 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24616 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24617 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24618 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24619 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24620 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24621 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24622 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24624 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24625 ("vec_scatter_element<V_HW_4:mode>_DI")
24626 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24627 ("vec_fpint<mode>", "vflls")
24628 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24629 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24630 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24631 ("*vec_cmphe<mode>_cc"): ... these.
24633 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24634 mode constant instead of magic value.
24636 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24638 * config/s390/s390.c (s390_expand_vec_compare): Support other
24639 vector floating point modes than just V2DF.
24640 (s390_expand_vcond): Likewise.
24641 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24642 (s390_cannot_change_mode_class): Prevent mode changes between TF
24643 and V1TF in vector registers.
24644 * config/s390/s390.md (DF, SF): New mode attributes.
24645 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24646 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24647 SFmode support for VRs.
24648 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24650 (VFT, VF_HW): New mode iterators.
24651 (vw, sdx): New mode attributes.
24652 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24653 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24654 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24655 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24656 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24657 also the new vector floating point modes. Renaming to ...
24659 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24660 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24661 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24662 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24663 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24664 ("vec_unordered<mode>"): ... these.
24666 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24667 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24668 ("*vec_extendv2df"): New insn definitions.
24670 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24672 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24673 ("mulditi3_2", "*muldi3_sign"): New patterns.
24674 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24675 rename the pattern definition.
24677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24679 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24681 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24683 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24685 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24686 instruction if possible.
24687 * config/s390/vector.md (vec_halfnumelts): New mode
24689 ("*vec_vllezlf<mode>"): New pattern.
24691 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24693 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24694 ("popcountv4si2", "popcountv2di2"): Rename to ...
24695 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24696 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24698 ("popcount<mode>2_vxe"): New pattern.
24700 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24702 * common/config/s390/s390-common.c (processor_flags_table): Add
24704 * config.gcc: Add arch12.
24705 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24706 Default to arch12 for unknown CPU model numbers.
24707 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24708 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24709 PROCESSOR_max sanity check.
24710 * config/s390/s390-opts.h (enum processor_type): Add
24712 * config/s390/s390.c (processor_table): Add arch12.
24713 (s390_expand_builtin): Add check for B_VXE flag.
24714 (s390_issue_rate): Add PROCESSOR_ARCH12.
24715 (s390_get_sched_attrmask): Likewise.
24716 (s390_get_unit_mask): Likewise.
24717 (s390_sched_score): Enable z13 scheduling for arch12.
24718 (s390_sched_reorder): Likewise.
24719 (s390_sched_variable_issue): Likewise.
24720 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24722 (s390_tune_attr): Use z13 scheduling also for arch12.
24723 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24724 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24725 (TARGET_VXE_P): New macros.
24726 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24727 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24728 * config/s390/s390.opt: Add arch12 as processor_type.
24730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24732 * config/s390/s390.md
24733 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24734 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24735 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24737 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24738 Rename expanders to ...
24740 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24741 ("fixuns_truncdddi2_emu"): ... these.
24743 ("fixuns_trunc<mode>si2_emu"): New expander.
24745 ("*fixuns_truncdfdi2_z13"): Rename to ...
24746 ("*fixuns_truncdfdi2_vx"): ... this.
24748 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24750 * config/s390/2964.md: Remove the single element vector compare
24751 instructions which are no longer used.
24752 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24754 (s390_canonicalize_comparison): Remove handling of DFmode
24756 (s390_expand_vec_compare_scalar): Remove function.
24757 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24758 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24760 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24764 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24766 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24767 will anyway by matched by mov<mode>_64dfp.
24769 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24771 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24772 vlef/vstef. Add missing operand to vleif.
24774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24776 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24777 pair for all vector types with 64 bit elements.
24778 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24779 * config/s390/vector.md (V_HW_64): ... here.
24780 (V_128_NOSINGLE): New mode iterator.
24781 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24782 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24783 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24784 ("*vec_load_pairv2di"): Change to ...
24785 ("*vec_load_pair<mode>"): ... this one.
24787 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24789 * config/s390/constraints.md: Add comments.
24790 (jKK): Reject element sizes > 8 bytes.
24791 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24793 * config/s390/s390.md: Add the s_operand checks formerly in
24794 s390_split_ok_p to various splitters where they are still
24796 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24797 for 128 bit vectors. Plus two splitters.
24799 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24801 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24807 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24808 error if the boundary argument is not constant.
24810 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24812 PR rtl-optimization/80112
24813 * loop-doloop.c (doloop_condition_get): Don't check condition
24814 if cmp isn't SET with IF_THEN_ELSE src.
24816 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24818 PR tree-optimization/80158
24819 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24820 replacing a candidate statement, also replace it for the
24821 candidate's alternate interpretation.
24822 (replace_rhs_if_not_dup): Likewise.
24823 (replace_one_candidate): Likewise.
24825 2017-03-24 Richard Biener <rguenther@suse.de>
24827 PR tree-optimization/80167
24828 * graphite-isl-ast-to-gimple.c
24829 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24831 (translate_isl_ast_to_gimple::get_rename): Likewise.
24833 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24836 handling of certain combinations of target options, including the
24837 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24838 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24840 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24843 * config/arm/arm.md (*load_multiple): Add reload_completed to
24844 matching condition.
24846 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24847 Richard Biener <rguenth@suse.de>
24849 PR tree-optimization/79908
24850 PR tree-optimization/80136
24851 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24852 been cast away, gimplify_and_add suffices.
24854 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24856 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24858 2017-03-23 Richard Biener <rguenther@suse.de>
24860 PR tree-optimization/80032
24861 * gimplify.c (gimple_push_cleanup): Forced unconditional
24862 cleanups still have to go to the conditional_cleanups
24865 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24867 PR tree-optimization/80072
24868 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24870 (next_operand_entry_id): Change type to unsigned int.
24871 (sort_by_operand_rank): Make sure to return the right return value
24872 even if unsigned fields are bigger than INT_MAX.
24873 (struct oecount): Change cnt and id type to unsigned int.
24874 (oecount_hasher::equal): Formatting fix.
24875 (oecount_cmp): Make sure to return the right return value
24876 even if unsigned fields are bigger than INT_MAX.
24877 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24880 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24881 TREE_READONLY on result if writing it more than once.
24884 * doc/invoke.texi (-fsanitize=thread): Document that with
24885 -fnon-call-exceptions atomics are not able to throw
24889 * tsan.c: Include tree-eh.h.
24890 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24891 maybe_clean_or_replace_eh_stmt where needed.
24892 (instrument_memory_accesses): Add cfg_changed argument.
24893 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24894 if it returned true.
24895 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24897 PR rtl-optimization/63191
24898 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24899 wrapper function, moved the whole old content into ...
24900 (ix86_delegitimize_address_1): ... this. New inline function.
24901 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24902 true as last argument instead of ix86_delegitimize_address.
24904 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24906 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24907 cortexa57_branch_cost.
24909 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24911 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24913 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24916 * doc/md.texi (Constraints): Document wA constraint.
24917 * config/rs6000/constraints.md (wA): New.
24918 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24919 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24920 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24921 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24923 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24926 * gimplify.c (is_oacc_declared): New function.
24927 (oacc_default_clause): Use it to set default flags for acc declared
24928 variables inside parallel regions.
24929 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24930 declared variables.
24931 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24932 declare attribute to any decl as necessary.
24934 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24937 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24938 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24939 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24940 (arm_arch_lpae): This.
24941 * config/arm/arm.c (arm_arch7ve): Rename into ...
24942 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24943 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24946 2017-03-22 Martin Liska <mliska@suse.cz>
24949 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24950 error message instead of an ICE.
24952 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24954 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24956 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24958 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24960 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24963 2017-03-21 Martin Sebor <msebor@redhat.com>
24965 * doc/extend.texi: Use "cannot" instead of "can't."
24966 * doc/hostconfig.texi: Same.
24967 * doc/install.texi: Same.
24968 * doc/invoke.texi: Same.
24969 * doc/loop.texi: Same.
24970 * doc/md.texi: Same.
24971 * doc/objc.texi: Same.
24972 * doc/rtl.texi: Same.
24973 * doc/tm.texi: Same.
24974 * doc/tm.texi.in: Same.
24975 * doc/trouble.texi: Same.
24977 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24980 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24981 (collect_checksum_attributes): Set it.
24982 (die_checksum_ordered): Use it.
24984 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24986 PR tree-optimization/79908
24987 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24988 change: For a VA_ARG whose LHS has been cast away, use
24989 force_gimple_operand to construct the side effects.
24991 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24993 PR translation/80001
24994 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24995 more amenable to translation.
24996 (oacc_loop_auto_partitions): Likewise.
24998 2017-03-21 Marek Polacek <polacek@redhat.com>
24999 Martin Sebor <msebor@redhat.com>
25001 PR tree-optimization/80109
25002 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25003 on INTEGRAL_TYPE_P.
25005 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25006 Segher Boessenkool <segher@kernel.crashing.org>
25009 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25010 check reg_used_between_p between insn and one of succ or succ2
25011 depending on if succ is artificial insn not inserted into insn
25014 2017-03-21 Martin Liska <mliska@suse.cz>
25016 PR gcov-profile/80081
25017 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25018 * doc/gcc.texi: Include gcov-dump stuff.
25019 * doc/gcov-dump.texi: New file.
25021 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25023 PR rtl-optimization/79150
25024 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25025 conditional jump, if the jump is the last insn of the loop.
25027 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25028 Richard Biener <rguenth@suse.de>
25030 PR tree-optimization/79908
25031 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25032 been cast away, use force_gimple_operand to construct the side
25035 2017-03-21 Martin Liska <mliska@suse.cz>
25037 PR libfortran/79956
25038 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25041 2017-03-21 Brad Spengler <spender@grsecurity.net>
25044 * plugin.c (htab_hash_plugin): New function.
25045 (add_new_plugin): Use it and adjust.
25046 (parse_plugin_arg_opt): Adjust.
25047 (init_one_plugin): Likewise.
25049 2017-03-21 Richard Biener <rguenther@suse.de>
25051 PR tree-optimization/80032
25052 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25053 if set force the cleanup to happen unconditionally.
25054 (gimplify_target_expr): Push inserted clobbers with force_uncond
25055 to avoid them being removed by control-dependent DCE.
25057 2017-03-21 Richard Biener <rguenther@suse.de>
25059 PR tree-optimization/80122
25060 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25061 va_arg_pack_len when the inlined call stmt requires pack
25063 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25065 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25068 * tsan.c (instrument_builtin_call): If the memory model argument
25069 is not a constant, assume it is valid.
25072 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25075 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25077 PR rtl-optimization/79910
25078 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25079 if its dest is used by an insn before I2 (other than the combined
25080 insns themselves, which are properly handled already).
25082 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25085 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25087 * combine.c (record_used_regs): New static function.
25088 (try_combine): Handle situations where there is an additional
25089 instruction between I2 and I3 which needs to have a LOG_LINK
25093 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25095 * combine.c (try_combine): Delete redundant i1 test. Call
25096 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25098 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25101 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25102 alternatives 13/14.
25104 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25106 PR tree-optimization/80054
25107 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25108 the optimization if a PHI or any of its arguments is not dominated
25109 by the candidate's basis. Use gphi* rather than gimple* as
25111 (replace_profitable_candidates): Clean up a gimple* variable that
25112 should be a gphi* variable.
25114 2017-03-20 Martin Sebor <msebor@redhat.com>
25117 * doc/extend.texi (attribute constructor): Document present limitation.
25119 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25122 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25123 __POWER9_VECTOR__ #ifdef control, change template definition to
25124 use Power9-specific built-in function.
25125 (vec_any_eq): Likewise.
25126 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25127 to control outcomes from this test.
25128 (vector_ae_<mode>p): For VEC_F modes, likewise.
25130 2017-03-20 Ian Lance Taylor <iant@google.com>
25132 * config/i386/i386.c (ix86_function_regparm): Save an extra
25133 register for -fsplit-stack with DECL_STATIC_CHAIN.
25135 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25138 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25139 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25141 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25143 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25145 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25148 2017-03-20 Marek Polacek <polacek@redhat.com>
25151 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25153 2017-03-20 Richard Biener <rguenther@suse.de>
25155 PR tree-optimization/80113
25156 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25157 allocate extra SSA name for PHI def.
25158 (add_close_phis_to_outer_loops): Likewise.
25159 (add_close_phis_to_merge_points): Likewise.
25160 (copy_loop_close_phi_args): Likewise.
25161 (copy_cond_phi_nodes): Likewise.
25163 2017-03-20 Martin Liska <mliska@suse.cz>
25165 PR middle-end/79753
25166 * tree-chkp.c (chkp_build_returned_bound): Do not build
25167 returned bounds for a LHS that's not a BOUNDED_P type.
25169 2017-03-20 Martin Liska <mliska@suse.cz>
25173 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25174 COMPLEX_CST and VECTOR_CST.
25176 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25179 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25180 target operand. A new splitter adds the clobber statement in case
25181 the target operand is dead anyway.
25183 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25185 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25186 to age-old versions of binutils and glibc.
25188 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25190 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25192 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25194 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25196 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25198 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25199 requirement for binutils 2.13.
25201 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25203 * combine.c (try_combine): Delete redundant i1 test. Call
25204 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25206 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25208 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25209 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25211 <riscv64-*-elf>: Re-arrange section
25212 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25213 <riscv32-*-linux>: Likewise.
25214 <riscv64-*-elf>: Likewise
25215 <riscv64-*-linux>: Likewise.
25217 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25220 * aarch64.opt(verbose-cost-dump): Fix typo.
25222 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25225 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25226 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25228 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25230 * reload.c (find_reloads): When reloading a nonoffsettable address,
25231 use RELOAD_OTHER for it and its address reloads.
25233 PR rtl-optimization/79910
25234 * combine.c (record_used_regs): New static function.
25235 (try_combine): Handle situations where there is an additional
25236 instruction between I2 and I3 which needs to have a LOG_LINK
25239 2017-03-17 Jeff Law <law@redhat.com>
25241 PR tree-optimization/71437
25242 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25243 conditional in the hash table first.
25244 (vrp_dom_walker::before_dom_children): Extract condition from
25245 ASSERT_EXPR. Record condition, its inverion and any implied
25246 conditions as well.
25248 2017-03-17 Marek Polacek <polacek@redhat.com>
25249 Markus Trippelsdorf <markus@trippelsdorf.de>
25251 PR tree-optimization/80079
25252 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25255 2017-03-17 Richard Biener <rguenther@suse.de>
25257 PR middle-end/80075
25258 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25259 Properly verify the LHS before the RHS possibly claims to be
25261 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25264 2017-03-17 Martin Jambor <mjambor@suse.cz>
25266 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25267 (List of -O2 options): Likewise.
25268 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25271 2017-03-17 Tom de Vries <tom@codesourcery.com>
25273 * gcov-dump.c (print_usage): Print bug_report_url.
25275 2017-03-17 Richard Biener <rguenther@suse.de>
25277 PR middle-end/80050
25278 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25279 (parser::peek): Likewise.
25281 2017-03-17 Richard Biener <rguenther@suse.de>
25283 PR tree-optimization/80048
25284 * sese.c (free_sese_info): Properly release rename_map and
25285 copied_bb_map elements.
25287 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25289 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25290 Add linked-list forward and backlinks. Insert on
25291 construction, remove on destruction.
25292 (class pass_store_merging): Add m_stores_head field.
25293 (pass_store_merging::terminate_and_process_all_chains):
25294 Iterate over m_stores_head list.
25295 (pass_store_merging::terminate_all_aliasing_chains):
25297 (pass_store_merging::execute): Check for debug stmts first.
25298 Push new chains onto the m_stores_head stack.
25300 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25303 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25304 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25305 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25307 2017-03-16 Jeff Law <law@redhat.com>
25309 PR tree-optimization/71437
25310 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25311 member function. Implementation moved into after_dom_children
25312 member function and into the threader's thread_outgoing_edges
25314 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25315 some code into new thread_outgoing_edges.
25316 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25317 definition. Simplify marker handling (do it here). Assume we always
25318 have the available expression and the const/copies tables.
25319 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25321 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25322 * tree-vrp.c (equiv_stack): No longer file scoped.
25323 (vrp_dom_walker): New class.
25324 (vrp_dom_walker::before_dom_children): New member function.
25325 (vrp_dom_walker::after_dom_children): Likewise.
25326 (identify_jump_threads): Setup domwalker. Use it rather than
25327 walking edges in a random order by hand. Simplify setup/finalization.
25328 (finalize_jump_threads): Remove.
25329 (vrp_finalize): Do not call identify_jump_threads here.
25330 (execute_vrp): Do it here instead and call thread_through_all_blocks
25333 PR tree-optimization/71437
25334 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25336 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25338 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25339 (dom_opt_dom_walker::thread_across_edge): Remove
25340 handle_dominating_asserts argument. All callers changed.
25341 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25342 changes. Remove calls to lhs_of_dominating_assert. Other
25343 uses of handle_dominating_asserts turn into unconditional code
25344 (simplify_control_stmt_condition_1): Likewise.
25345 (simplify_control_stmt_condition): Likewise.
25346 (thread_through_normal_block, thread_across_edge): Likewise.
25347 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25348 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25349 object if it is not an SSA_NAME.
25350 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25351 before calling into the VRP specific simplifiers.
25352 (identify_jump_threads): Remove handle_dominating_asserts
25355 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25358 * tree-diagnostic.c (default_tree_printer): No longer static.
25359 * tree-diagnostic.h (default_tree_printer): New prototype.
25361 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25363 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25364 Change ins into fmov.
25366 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25368 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25369 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25370 Use h_con constraint for operand 1.
25371 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25372 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25374 2017-03-15 Jeff Law <law@redhat.com>
25376 PR tree-optimization/71437
25377 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25378 (record_temporary_equivalences): Use it.
25380 PR tree-optimization/71437
25381 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25382 tree-ssa-scopedtables.
25383 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25384 (record_conditions, record_cond, vuse_eq): Likewise.
25385 (record_edge_info): Adjust to API tweak of record_conditions.
25386 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25387 (record_temporary_equivalences, optimize_stmt): Likewise.
25388 (eliminate_redundant_computations): Likewise.
25389 (record_equivalences_from_stmt): Likewise.
25390 * tree-ssa-scopedtables.c: Include options.h and params.h.
25391 (vuse_eq): New function, moved from tree-ssa-dom.c
25392 (build_and_record_new_cond): Likewise.
25393 (record_conditions): Likewise. Accept vector of conditions rather
25394 than edge_equivalence structure for first argument.
25395 for the first argument.
25396 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25397 from tree-ssa-dom.c.
25398 (avail_exprs_stack::record_cond): Likewise.
25399 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25400 from tree-ssa-dom.c.
25401 (avail_exprs_stack): Add new member functions lookup_avail_expr
25403 (record_conditions): Declare.
25405 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25408 * lra-constraints.c (process_alt_operands): Increase reject for
25409 reloading an input/output operand.
25411 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25414 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25415 insns to convert from signed/unsigned char/short to IEEE 128-bit
25417 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25419 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25422 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25423 subreg of inner mode for values already in registers.
25425 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25427 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25428 iteration reg is used after the loop.
25430 2017-03-14 Martin Sebor <msebor@redhat.com>
25432 PR tree-optimization/79800
25433 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25434 precision in negative-positive range.
25435 (format_floating): Call non-const overload with adjusted precision.
25437 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25440 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25443 2017-03-14 Martin Sebor <msebor@redhat.com>
25445 PR middle-end/80020
25446 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25447 * builtins.def (aligned_alloc): Use it.
25450 * Makefile.in (GTFILES): Add calls.c.
25451 * calls.c: Include "gt-calls.h".
25453 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25455 PR rtl-optimization/79728
25456 * regs.h (struct target_regs): New field
25457 x_contains_allocatable_regs_of_mode.
25458 (contains_allocatable_regs_of_mode): New macro.
25459 * reginfo.c (init_reg_sets_1): Initialize it, and change
25460 contains_reg_of_mode so it includes global regs as well.
25461 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25462 rather than contains_regs_of_mode.
25464 2017-03-14 Martin Liska <mliska@suse.cz>
25466 * doc/invoke.texi: Document options that can't be combined with
25467 -fcheck-pointer-bounds.
25469 2017-03-14 Martin Liska <mliska@suse.cz>
25471 PR middle-end/79831
25472 * doc/invoke.texi (-Wchkp): Document the option.
25474 2017-03-14 Martin Liska <mliska@suse.cz>
25476 * Makefile.in: Install gcov-dump.
25478 2017-03-14 Martin Liska <mliska@suse.cz>
25480 * multiple_target.c (expand_target_clones): Bail out for
25481 an invalid attribute.
25483 2017-03-14 Richard Biener <rguenther@suse.de>
25485 * alias.c (struct alias_set_entry): Pack properly.
25486 * cfgloop.h (struct loop): Likewise.
25487 * cse.c (struct set): Likewise.
25488 * ipa-utils.c (struct searchc_env): Likewise.
25489 * loop-invariant.c (struct invariant): Likewise.
25490 * lra-remat.c (struct cand): Likewise.
25491 * recog.c (struct change_t): Likewise.
25492 * rtl.h (struct address_info): Likewise.
25493 * symbol-summary.h (function_summary): Likewise.
25494 * tree-loop-distribution.c (struct partition): Likewise.
25495 * tree-object-size.c (struct object_size_info): Likewise.
25496 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25497 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25498 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25499 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25500 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25501 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25502 (struct _stmt_vec_info): Likewise.
25504 2017-03-14 Martin Liska <mliska@suse.cz>
25507 * multiple_target.c (create_dispatcher_calls): Check that
25508 a target can create a function dispatcher.
25510 2017-03-14 Martin Liska <mliska@suse.cz>
25513 * multiple_target.c (expand_target_clones): Drop local.local
25514 flag for default implementation.
25516 2017-03-14 Richard Biener <rguenther@suse.de>
25518 PR tree-optimization/80030
25519 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25521 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25523 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25524 gcc_fallthrough() instead of __attribute__((fallthrough));
25526 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25528 * doc/gcc.texi: Remove "up" link to (DIR).
25529 * doc/gccint.texi: Ditto.
25531 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25533 * doc/install.texi (Specific) <avr>: Remove reference to
25536 2017-03-13 Jeff Law <law@redhat.com>
25538 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25539 attribute rather than comments.
25541 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25542 match_scratch operand is highest.
25544 2017-03-13 Martin Liska <mliska@suse.cz>
25546 PR middle-end/78339
25547 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25548 is a CHKP clone, use original declaration.
25550 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25552 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25553 (arc_conditional_register_usage): Use a different allocation order
25554 when optimizing for size.
25555 * common/config/arc/arc-common.c (arc_option_optimization_table):
25556 Section anchors default on when optimizing for size.
25558 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25560 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25562 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25564 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25565 * config/arc/arc.md (cpu_facility): Add cd variant.
25566 (*movqi_insn): Add code density variant.
25567 (*movhi_insn): Likewise.
25568 (*movqi_insn): Likewise.
25569 (*addsi3_mixed): Likewise.
25570 (subsi3_insn): Likewise.
25572 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25574 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25576 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25578 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25579 expressions with MINUS and UNARY ops.
25581 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25584 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25586 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25587 between vec_select and vector argument.
25588 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25589 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25590 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25591 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25592 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25593 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25595 2017-03-13 Richard Biener <rguenther@suse.de>
25598 * params.def (vect-max-peeling-for-alignment): Fix typo.
25600 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25602 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25603 issue that only occurred with binutils below 2.18.
25605 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25607 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25608 refer to binutils 2.11/2.12 minimum.
25610 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25612 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25613 ftp.kernel.org and simplify binutils requirement.
25615 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25617 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25619 (Optimize Options): Ditto. Also remove redundancy.
25621 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25623 PR translation/79848
25624 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25626 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25627 to G_ to avoid double translation.
25629 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25631 PR translation/79923
25632 * auto-profile.c (get_combined_location): Convert leading
25633 character of diagnostics to lower case and remove trailing period.
25634 (read_profile): Likewise for various diagnostics.
25635 * config/arm/arm.c (arm_option_override): Remove trailing period
25636 from various diagnostics.
25637 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25638 (msp430_expand_delay_cycles): Likewise.
25640 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25643 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25644 full command-line argument, rather than just "str".
25645 (aarch64_validate_march): Likewise.
25646 (aarch64_validate_mtune): Likewise.
25648 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25650 PR rtl-optimization/78911
25651 * lra-assigns.c (must_not_spill_p): New function.
25652 (spill_for): Use it.
25654 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25656 PR tree-optimization/79981
25657 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25658 ATOMIC_COMPARE_EXCHANGE ifn result.
25659 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25660 IFN_ATOMIC_COMPARE_EXCHANGE.
25662 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25665 * opts.c (parse_sanitizer_options): Add missing question mark to
25666 "did you mean" message.
25668 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25670 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25672 (VMULEUH_UNS): Likewise.
25673 (VMULOUB_UNS): Likewise.
25674 (VMULOUH_UNS): Likewise.
25675 * config/rs6000/rs6000.c (builtin_function_type): Remove
25676 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25678 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25681 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25682 x with result of extra_parsing_for_operand_code_0.
25683 (function_reader::extra_parsing_for_operand_code_0): Convert
25684 return type from void to rtx, returning x. When reading
25685 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25686 larger size containing struct block_symbol.
25688 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25691 -mfloat128-hardware without -m64.
25693 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25696 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25697 entries to the case statement that marks unsigned arguments to
25698 overloaded functions.
25700 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25703 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25705 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25708 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25709 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25711 2017-03-10 Martin Liska <mliska@suse.cz>
25715 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25716 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25719 2017-03-10 Olivier Hainque <hainque@adacore.com>
25721 * tree-switch-conversion (array_value_type): Start by resetting
25722 candidate type to it's main variant.
25724 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25726 PR rtl-optimization/79909
25727 * combine.c (try_combine): Use simplify_replace_rtx on individual
25728 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25729 of the whole CALL_INSN_FUNCTION_USAGE.
25731 PR tree-optimization/79972
25732 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25733 get_range_info on SSA_NAMEs. Formatting fixes.
25735 2017-03-10 Richard Biener <rguenther@suse.de>
25736 Jakub Jelinek <jakub@redhat.com>
25738 PR tree-optimization/77975
25739 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25740 edge to be constant.
25741 (get_val_for): For constant x return it. Formatting fix.
25742 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25743 would use the same bases as the current one.
25745 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25747 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25748 instead of vec_select for V1TImode.
25749 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25751 (VSX_LE_128): Add V1TI to this mode iterator.
25752 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25753 (*vsx_le_perm_store_<mode>): Likewise.
25754 (pre-reload splitter for VSX stores): Likewise.
25755 (post-reload splitter for VSX stores): Likewise.
25756 (*vsx_xxpermdi2_le_<mode>): Likewise.
25757 (*vsx_lxvd2x2_le_<mode>): Likewise.
25758 (*vsx_stxvd2x2_le_<mode>): Likewise.
25760 2017-03-09 Michael Eager <eager@eagercon.com>
25762 Correct failures with --enable-checking=yes,rtl.
25764 * config/microblaze/microblaze.c (microblaze_expand_shift):
25765 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25766 test for const0_rtx.
25767 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25768 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25770 2017-03-09 Richard Biener <rguenther@suse.de>
25772 PR tree-optimization/79977
25773 * graphite-scop-detection.c (scop_detection::merge_sese):
25774 Handle the case of extra exits to blocks dominating the entry.
25776 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25778 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25781 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25783 PR rtl-optimization/79949
25784 * lra-constraints.c (process_alt_operands): Check memory when
25785 trying to predict a cycle. Print about the overall increase.
25787 2017-03-09 Richard Biener <rguenther@suse.de>
25789 PR middle-end/79971
25790 * gimple-expr.c (useless_type_conversion_p): Preserve
25791 TYPE_SATURATING for fixed-point types.
25793 2017-03-09 Richard Biener <rguenther@suse.de>
25796 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25797 alignment of BLKmode params.
25799 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25802 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25803 (VALL_NO_V2Q): Likewise.
25805 * config/aarch64/aarch64-simd.md
25806 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25808 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25809 VALL_NO_V2Q mode iterator.
25810 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25812 2017-03-09 Martin Liska <mliska@suse.cz>
25814 PR tree-optimization/79631
25815 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25816 tree_int_cst_sign_bit just for INTEGER constants.
25818 2017-03-09 Martin Liska <mliska@suse.cz>
25822 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25825 2017-03-09 Marek Polacek <polacek@redhat.com>
25828 * tree.c (inchash::add_expr): Handle TREE_VEC.
25830 2017-03-09 Martin Liska <mliska@suse.cz>
25833 (chkp_narrow_size_and_offset): New function.
25834 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25835 (void chkp_parse_bit_field_ref): New function.
25836 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25837 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25839 2017-03-09 Martin Liska <mliska@suse.cz>
25842 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25843 (chkp_find_bounds_1): Remove gcc_unreachable.
25845 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25848 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25849 BUILT_IN_SYNC*, determine the access type from the size suffix and
25850 always build a MEM_REF with that type. Handle forgotten
25851 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25854 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25855 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25856 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25857 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25858 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25859 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25860 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25861 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25862 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25863 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25864 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25865 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25866 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25867 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25868 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25869 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25870 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25871 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25872 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25873 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25874 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25875 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25876 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25877 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25878 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25879 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25880 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25881 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25882 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25883 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25884 definitions outside of __OPTIMIZE__ guarded section.
25887 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25888 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25889 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25890 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25893 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25895 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25896 ("vfenez<mode>"): Add missing constraints.
25898 2017-03-08 Martin Sebor <msebor@redhat.com>
25901 * config/nds32/nds32.c (nds32_option_override):
25902 Fix misspelled diagnostic.
25904 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25907 * gimplify.c (gimplify_omp_for): Replace index var in outer
25908 taskloop statement with an artificial variable and add
25909 OMP_CLAUSE_PRIVATE clause for it.
25911 2017-03-08 Richard Biener <rguenther@suse.de>
25913 PR tree-optimization/79955
25914 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25915 for accesses that are completely outside of the variable.
25917 2017-03-08 Andrew Haley <aph@redhat.com>
25919 PR tree-optimization/79943
25920 * tree-ssa-loop-split.c (compute_new_first_bound): When
25921 calculating the new upper bound, (END-BEG) should be added, not
25924 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25926 * config/avr/avr.md (setmemhi): Make sure match_dup
25927 operand number comes before match_scratch.
25929 2017-03-08 Richard Biener <rguenther@suse.de>
25931 PR tree-optimization/79920
25932 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25933 with ncopies == 1 to ...
25934 (vect_transform_slp_perm_load): ... here. Properly compute
25935 all element loads by iterating VF times over the group. Do
25936 not handle ncopies (computed in a broken way) in
25937 vect_create_mask_and_perm.
25939 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25942 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25943 is a uniform vector, use uniform_vector_p return value instead of
25944 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25946 2017-03-07 Marek Polacek <polacek@redhat.com>
25948 PR middle-end/79809
25949 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25950 (alloca_call_type): Likewise.
25952 2017-03-07 Martin Liska <mliska@suse.cz>
25954 * gcov.c (process_args): Put comment to correct location.
25956 2017-03-07 Martin Liska <mliska@suse.cz>
25958 PR middle-end/68270
25959 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25960 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25961 (chkp_narrow_bounds_for_field): Likewise.
25962 (chkp_parse_array_and_component_ref): Pass one more argument to
25965 2017-03-07 Richard Biener <rguenther@suse.de>
25967 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25970 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25972 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25973 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25975 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25978 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25979 to end of description.
25980 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25982 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25984 PR rtl-optimization/79901
25985 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25987 (*avx512f_<code><mode>3<mask_name>): ... this.
25988 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25989 iterator instead of VI8_AVX2_AVX512BW.
25991 PR rtl-optimization/79901
25992 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25993 min/max expander, expand it using expand_vec_cond_expr.
25996 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25999 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26002 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26003 to void * for PCH reasons.
26004 * dwarf2out.c (output_loc_operands, output_die): Cast
26005 v.val_vec.array to unsigned char *.
26007 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26010 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26013 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26015 PR rtl-optimization/79571
26016 * lra-constraints.c (process_alt_operands): Calculate static
26017 reject and subtract it from overall when only addresses will be
26020 2017-03-06 Julia Koval <julia.koval@intel.com>
26023 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26024 incoming stack boundary to 128 for 64-bit targets.
26026 2017-03-06 Richard Biener <rguenther@suse.de>
26028 PR tree-optimization/79894
26029 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26030 to NULL after folding it.
26032 2017-03-06 Richard Biener <rguenther@suse.de>
26034 PR tree-optimization/79824
26035 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26036 check disabling peeling for gaps.
26038 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26040 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26041 attributes): Document gettimeofday.
26043 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26045 * config/s390/s390.c (s390_option_override_internal): Set
26046 PARAM_MIN_VECT_LOOP_BOUND
26048 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26050 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26051 * config/s390/s390.md: Likewise.
26053 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26056 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26057 (<avx2_avx512>_perm<mode>): Rename to ...
26058 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26060 (<avx512>_perm<mode>_mask): Rename to ...
26061 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26063 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26064 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26065 instead of VI8F_256_512.
26066 (avx512f_perm<mode>): New define_expand.
26067 (avx512f_perm<mode>_mask): Likewise.
26068 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26069 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26071 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26073 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26074 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26076 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26078 2017-03-06 Martin Liska <mliska@suse.cz>
26081 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26082 when having a SSA NAME w/o VAR_DECL assigned to it.
26084 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26086 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26087 msa_dpsub_<su>_d): Fix MODE for vec_select.
26089 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26091 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26093 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26095 2017-03-06 Richard Biener <rguenther@suse.de>
26097 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26098 * plugin.c (register_plugin_info): Likewise.
26099 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26101 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26103 * config/i386/sse.md (sse_storehps, sse_storelps,
26104 avx_<castmode><avxsizesuffix>_<castmode>,
26105 avx512f_<castmode><avxsizesuffix>_<castmode>,
26106 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26107 in condition that at least one operand is not a MEM.
26109 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26111 PR middle-end/79805
26112 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26113 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26115 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26116 gimple_call_nothrow_p flag based on whether original builtin can throw.
26117 If it can, emit following stmts on the fallthrough edge.
26118 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26119 don't create new bb if inserting just debug stmts on the edge, try to
26120 insert them on the fallthru bb or just reset debug stmts.
26122 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26125 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26126 restore recog_data (including the operand rtxes inside it) around
26127 the call to get_insn_template.
26129 2017-03-03 Martin Sebor <msebor@redhat.com>
26131 PR tree-optimization/79699
26132 * context.c (context::~context): Free MPFR caches to avoid
26133 a memory leak on program exit.
26135 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26137 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26138 Use wide_int::ulow () instead of .elt (0).
26140 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26142 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26143 (*pushxf): Limit oF constraint to 32bit targets and add oC
26144 constraint for 64bit targets.
26145 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26146 (*pushdf): Change rmF constraint to rmC.
26148 2017-03-03 Martin Liska <mliska@suse.cz>
26150 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26151 Remove unused variable.
26153 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26156 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26157 is a memory operand, increase num_memory.
26158 (ix86_expand_args_builtin): Likewise.
26160 2017-03-03 Jan Hubicka <jh@suse.cz>
26163 * ipa-devirt.c (maybe_record_node): Properly handle
26164 __cxa_pure_virtual visibility.
26166 2017-03-03 Martin Liska <mliska@suse.cz>
26168 PR tree-optimization/79803
26169 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26171 (pass_loop_prefetch::execute): Disabled optimization if an
26172 assumption about L1 cache size is not met.
26174 2017-03-03 Martin Liska <mliska@suse.cz>
26176 PR rtl-optimization/79574
26177 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26178 (hash_scan_set): Likewise.
26179 (dump_hash_table): Likewise.
26180 (hoist_code): Likewise.
26182 2017-03-03 Richard Biener <rguenther@suse.de>
26184 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26185 in place of uhigh ().
26186 (fixed_convert_from_real): Likewise.
26188 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26191 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26193 2017-03-03 Richard Biener <rguenther@suse.de>
26195 PR middle-end/79818
26196 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26197 TYPE_OVERFLOW_UNDEFINED check.
26199 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26201 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26203 (vector_ae_<mode>_p): Likewise.
26204 (vector_nez_<mode>_p): Likewise.
26205 (vector_ne_v2di_p): Likewise.
26206 (vector_ae_v2di_p): Likewise.
26207 (vector_ne_<mode>_p): Likewise.
26208 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26210 (vsx_tsqrt<mode>2_fe): Likewise.
26212 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26215 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26217 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26219 PR rtl-optimization/79780
26220 * cprop.c (one_cprop_pass): When second and further conditional trap
26221 in a single basic block is turned into an unconditional trap, turn it
26222 into a deleted note to avoid RTL verification failures.
26224 2017-03-02 Richard Biener <rguenther@suse.de>
26226 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26228 2017-03-02 Richard Biener <rguenther@suse.de>
26230 PR tree-optimization/79345
26232 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26233 param and abort the walk, returning -1 if it is hit.
26234 (walk_aliased_vdefs): Take a limit param and pass it on.
26235 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26236 defaulting to 0 and return a signed int.
26237 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26238 (check_defs): New helper.
26239 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26240 about uninitialized memory.
26241 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26242 bogus uninitialized warning.
26243 (fixed_convert_from_real): Likewise.
26245 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26247 PR tree-optimization/66768
26248 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26249 iv_use if base object can't be determined.
26251 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26253 PR tree-optimization/79345
26254 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26255 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26256 (get_pattern_stats): Initialize it.
26257 * genemit.c (gen_expand): Verify match_scratch numbers come after
26258 match_operand/match_dup numbers.
26259 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26260 match_scratch numbers.
26261 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26263 * config/s390/s390.md (trunctdsd2): Likewise.
26265 2017-03-02 Richard Biener <rguenther@suse.de>
26267 * wide-int.h (wide_int_storage::operator=): Implement in terms
26270 2017-03-02 Richard Biener <rguenther@suse.de>
26272 PR tree-optimization/79777
26273 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26274 the to insert expression to sth existing.
26276 2017-03-01 Martin Sebor <msebor@redhat.com>
26278 PR middle-end/79692
26279 * gimple-ssa-sprintf.c
26280 (directive::known_width_and_precision): New function.
26281 (format_integer): Use it.
26282 (get_mpfr_format_length): Consider the full range of precision
26283 when computing %g output with the # flag. Set the likely byte
26284 count to 3 rather than 1 when precision is indeterminate.
26285 (format_floating): Correct the lower bound of precision.
26287 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26289 * doc/invoke.texi: Document default code model for 64-bit Linux.
26291 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26294 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26295 udiv rather than div since input pattern is unsigned.
26297 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26299 * config/i386/i386.c (print_reg): Warn for values of
26300 unsupported size in integer register.
26302 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26305 * config/rs6000/predicates.md (current_file_function_operand): Do
26306 not allow self calls to be local if the function is replaceable.
26308 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26311 * config/rs6000/altivec.h (vec_ctz and others): Change the
26312 preprocessor macro that controls conditional compilation from
26313 _ARCH_PWR9 to __POWER9_VECTOR__.
26314 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26315 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26316 control (instead of _ARCH_PWR9 control) so that template
26317 definition uses power9-specific function.
26318 (vec_any_eq): Likewise.
26319 (vec_all_ne): Change macro definition to use a power9-specific
26320 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26321 _ARCH_PWR9 control).
26322 (vec_any_eq) Likewise.
26323 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26324 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26325 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26326 support for xvcmpnedp instruction.
26327 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26328 macro expansion so that Power9 implementation of vec_all_ne does
26329 not use the AltiVec predicate framework.
26330 (VCMPNEH_P): Likewise.
26331 (VCMPNEW_P): Likewise.
26332 (VCMPNED_P): Likewise.
26333 (VCMPNEFP_P): Likewise.
26334 (VCMPNEDP_P): Likewise.
26335 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26336 implementation of vec_any_eq to not use AltiVec predicate
26338 (VCMPAEH_P): Likewise.
26339 (VCMPAEW_P): Likewise.
26340 (VCMPAED_P): Likewise.
26341 (VCMPAEFP_P): Likewise.
26342 (VCMPAEDP_P): Likewise.
26343 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26344 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26345 not use the AltiVec predicate framework.
26346 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26347 of vec_any_eq to not use AltiVec predicate framework.
26348 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26349 support for predefined __POWER9_VECTOR__ macro to indicate that
26350 Power9 instruction selection is enabled.
26351 (altivec_overloaded_builtins): Remove extraneous
26352 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26353 function argument types RS6000_BTI_bool_V16QI and
26354 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26355 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26356 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26357 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26358 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26359 Power9 for implementations of vec_cmpne. Change the signature for
26360 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26361 (representing vec_all_ne) to remove the previously described first
26362 argument of type RS6000_BTI_INTSI, as this was an artifact of
26363 reliance on the AltiVec predicate framework, which is no longer
26364 used in the implementation of these functions. Add
26365 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26366 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26367 since, unlike the AltiVec predicate framework implementation, we
26368 do not share function descriptors between vec_alle and vec_anyeq.
26369 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26370 set of modes that receive special treatment even when
26371 TARGET_P9_VECTOR is true. The special treatment emits code that
26372 does not depend on Power9 instructions.
26373 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26374 define_expand to not rely on AltiVec predicate framework.
26375 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26377 (vector_ne_v2di_p): Change this define_expand to not rely on
26378 AltiVec predicate framework.
26379 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26381 (vector_ne_<mode>_p): Change this define_expand to not rely on
26382 AltiVec predicate framework.
26383 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26385 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26386 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26387 define_insn pattern.
26388 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26389 define_insn pattern because the xvcmpne<VSs>. instruction is not
26391 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26392 instruction is not supported.
26394 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26396 * config/nvptx/nvptx.c: Include intl.h.
26398 2017-03-01 Martin Jambor <mjambor@suse.cz>
26401 * ipa-prop.h (ipa_bits): Removed field known.
26402 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26403 to pointers. Adjusted their comments to warn about their sharing.
26404 (ipcp_transformation_summary): Change bits to a vector of pointers.
26405 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26406 (ipa_get_ipa_bits_for_value): Declare.
26407 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26408 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26409 (ipa_bits_hash_table): Likewise.
26410 (ipa_vr_ggc_hash_traits): Likewise.
26411 (ipa_vr_hash_table): Likewise.
26412 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26413 being pointers and vr_known being removed.
26414 (ipa_set_jf_unknown): Likewise.
26415 (ipa_get_ipa_bits_for_value): New function.
26416 (ipa_set_jfunc_bits): Likewise.
26417 (ipa_get_value_range): New overloaded functions.
26418 (ipa_set_jfunc_vr): Likewise.
26419 (ipa_compute_jump_functions_for_edge): Use the above functions to
26420 construct bits and vr parts of jump functions.
26421 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26422 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26424 (ipcp_grow_transformations_if_necessary): Also allocate
26425 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26427 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26428 them. Fix too long lines.
26429 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26430 vr_known being removed.
26431 (ipa_read_jump_function): Use new setter functions to construct bits
26432 and vr parts of jump functions or set them to NULL.
26433 (write_ipcp_transformation_info): Adjust for bits being pointers.
26434 (read_ipcp_transformation_info): Likewise.
26435 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26437 Include gt-ipa-prop.h.
26438 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26440 (ipcp_store_bits_results): Likewise.
26441 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26442 Do not write to existing jump functions but use a temporary instead.
26444 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26447 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26448 attempt to use its first operand as BIT_FIELD_REF base.
26450 2017-03-01 Richard Biener <rguenther@suse.de>
26452 PR middle-end/79721
26453 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26454 interpolating formula in wrapping arithmetic.
26455 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26457 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26459 PR tree-optimization/79734
26460 * tree-vect-generic.c (expand_vector_condition): Optimize
26461 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26462 Handle VEC_COND_EXPR where comparison has different inner width from
26463 type's inner width.
26465 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26467 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26468 markup, and similar issues. Remove @opindex entries for things
26469 that aren't options. Add missing -mmpy-option entries.
26471 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26473 PR tree-optimization/79737
26474 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26475 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26476 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26477 instead of byte_size. Formatting fix.
26478 (shift_bytes_in_array_right): Formatting fix.
26480 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26483 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26484 condition on optimize for the leaf function test.
26486 2017-02-28 Martin Liska <mliska@suse.cz>
26489 * read-rtl-function.c (function_reader::handle_unknown_directive):
26490 Bail out when one uses -flto.
26492 2017-02-28 Martin Liska <mliska@suse.cz>
26494 * common.opt: Replace space with tabular for options of <number>
26496 * config/i386/i386.opt: Show <number> value for
26497 -mlarge-data-threshold.
26498 * opts.c (print_filtered_help): Do not display number in hexadecimal
26501 2017-02-28 Martin Liska <mliska@suse.cz>
26503 * common.opt: Fix --help=option -Q for options which are of
26506 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26508 * config/i386/i386.c (print_reg): Error out for values
26509 of 8-bit size in invalid integer register.
26511 2017-02-28 Martin Sebor <msebor@redhat.com>
26513 PR tree-optimization/79691
26514 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26516 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26519 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26520 gcc_unreachable with output_operand_lossage.
26522 2017-02-28 Richard Biener <rguenther@suse.de>
26524 PR tree-optimization/79740
26525 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26527 (visit_nary_op): Insert the nary into the hashtable if we
26528 pattern-matched sth.
26529 * tree-ssa-pre.c (eliminate_insert): Robustify.
26531 2017-02-28 Richard Biener <rguenther@suse.de>
26533 PR middle-end/79731
26534 * fold-const.c (decode_field_reference): Reject out-of-bound
26537 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26539 * config/i386/i386.c: Include intl.h.
26540 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26541 instead of just cond ? "..." : "...".
26542 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26543 * coverage.c (read_counts_file): Likewise.
26544 * omp-offload.c: Include intl.h.
26545 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26546 of just cond ? "..." : "...".
26547 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26548 of just cond ? "..." : "...".
26550 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26553 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26555 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26556 'tune for' CPU name.
26557 * config/arm/arm-cpu-data.h: Regenerated.
26559 2017-02-28 Richard Biener <rguenther@suse.de>
26561 PR tree-optimization/79732
26562 * tree-inline.c (expand_call_inline): Do not shadow var.
26564 2017-02-28 Richard Biener <rguenther@suse.de>
26566 PR tree-optimization/79723
26567 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26568 address-space properly.
26570 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26572 * doc/optinfo.texi (Optimization groups): Fix option used for
26574 * doc/invoke.texi (-fopt-info): Document "omp".
26575 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26576 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26577 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26578 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26579 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26581 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26583 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26584 OPTGROUP_OMP with "omp".
26586 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26589 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26590 for arithmetic shift of unsigned V2DI.
26592 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26594 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26595 arc/linux.h headers.
26596 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26597 (LINK_SPEC): Likewise.
26598 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26599 (EXTRA_SPECS): Likewise.
26600 (STARTFILE_SPEC): Likewise.
26601 (ENDFILE_SPEC): Likewise.
26602 (LIB_SPEC): Likewise.
26603 (TARGET_SDATA_DEFAULT): Likewise.
26604 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26605 (MULTILIB_DEFAULTS): Likewise.
26606 (DWARF2_UNWIND_INFO): Likewise.
26607 * config/arc/big.h: New file.
26608 * config/arc/elf.h: Likewise.
26609 * config/arc/linux.h: Likewise.
26610 * config/arc/t-uClibc: Remove.
26612 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26614 PR tree-optimization/77536
26615 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26616 (tree_transform_and_unroll_loop): Use above function to compute the
26617 estimated niter of unrolled loop and use it when scaling profile.
26618 Also use count info rather than frequency if it's non-zero.
26619 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26620 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26621 (vect_transform_loop): Call above function.
26623 2017-02-27 Richard Biener <rguenther@suse.de>
26625 PR tree-optimization/45397
26626 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26627 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26628 (visit_nary_op): Add pattern matching for CSEing sign-changed
26629 or truncated operations with wider ones.
26631 2017-02-27 Richard Biener <rguenther@suse.de>
26633 PR tree-optimization/79690
26634 * tree-vect-stmts.c (vectorizable_store): Use vector type
26635 built from the DR with address-space.
26637 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26639 * doc/invoke.texi (Optimize Options): Refine the description
26640 of asan-use-after-return.
26642 2017-02-25 Alan Modra <amodra@gmail.com>
26644 PR rtl-optimization/79584
26645 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26646 base, not ad->base_term, the reg within base. Remove assertion
26647 that ad->base == ad->base_term. Replace gen_int_mode using
26648 bogus mode with const0_rtx.
26650 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26652 PR middle-end/79396
26653 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26654 FMA_EXPR like tcc_binary or tcc_unary.
26656 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26659 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26661 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26662 (output_loc_operands): Handle DW_OP_call_ref and
26663 DW_OP_GNU_variable_value.
26664 (struct variable_value_struct): New type.
26665 (struct variable_value_hasher): Likewise.
26666 (variable_value_hash): New variable.
26667 (string_types): Remove.
26668 (copy_loc_descr): New function.
26669 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26670 (prepend_loc_descr_to_each): New function.
26671 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26672 instead of add_loc_descr_to_each if the first argument is single
26673 location list and the second has multiple.
26674 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26675 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26676 when looking for variable value which doesn't have other location info.
26677 (loc_list_from_tree): Formatting fix.
26678 (gen_array_type_die): Simplify DW_AT_string_length handling.
26679 (adjust_string_types): Remove.
26680 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26681 string_types. Call resolve_variable_values.
26682 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26683 (resolve_addr_in_expr): Likewise. Add A argument.
26684 (copy_deref_exprloc): Remove deref argument. Adjust for the
26685 original expression being DW_OP_GNU_variable_value with optionally
26686 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26687 optionally after it.
26688 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26689 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26690 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26691 (variable_value_hasher::hash, variable_value_hasher::equal): New
26693 (resolve_variable_value_in_expr, resolve_variable_value,
26694 resolve_variable_values, note_variable_value_in_expr,
26695 note_variable_value): New functions.
26696 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26699 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26702 * opts.h (handle_generated_option): Add GENERATED_P argument.
26703 * opts-common.c (handle_option): Adjust function comment.
26704 (handle_generated_option): Add GENERATED_P argument, pass it to
26706 (control_warning_option): Pass false to handle_generated_option
26708 * opts.c (maybe_default_option): Pass true to handle_generated_option
26710 * optc-gen.awk: Likewise.
26712 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26714 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26715 a REG, look at the REG it is a SUBREG of.
26716 (splitter for cmpeqsi_t): Ditto.
26718 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26720 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26721 the special USEs with the pattern of the insn, not the insn itself.
26723 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26726 * doc/invoke.texi: Document -mload-store-pairs.
26728 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26729 Sandra Loosemore <sandra@codesourcery.com>
26731 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26732 argument isn't a CONST_INT.
26733 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26734 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26735 (nios2_validate_compare): Bypass alternate compare logic if *op2
26736 is not a CONST_INT.
26737 (ldstwm_operation_p): Return false if first_base is not a REG or
26738 if first_offset is not a CONST_INT.
26740 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26742 * config/cris/cris.md: Use correct operand in a define_peephole2.
26744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26746 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26748 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26750 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26751 this_insn if it is an INSN or JUMP_INSN.
26752 (force_offsettable): Look at base, not at addr.
26753 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26754 on things that aren't necessarily CONST_INTs.
26756 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26758 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26759 -mfpmath=sse is the default also for x86-32 targets with SSE2
26760 instruction set when @option{-ffast-math} is enabled
26762 2017-02-24 Jeff Law <law@redhat.com>
26764 PR rtl-optimizatoin/79286
26765 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26768 2017-02-24 Richard Biener <rguenther@suse.de>
26770 PR tree-optimization/79389
26771 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26774 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26776 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26777 function comment to reflect reality.
26778 (loop_exits_before_overflow): Fix typo in function description.
26780 2017-02-24 Richard Biener <rguenther@suse.de>
26782 PR tree-optimization/79389
26783 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26784 properly that a threading opportunity exists. Detect conditional
26785 copy/constant propagation opportunities.
26787 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26789 * config/visium/visium.md (type): Add trap.
26790 (b): New mode attribute.
26791 (*btst): Rename into...
26792 (*btst<mode>): ...this and adjust.
26793 (*cbranchsi4_btst_insn): Rename into...
26794 (*cbranch<mode>4_btst_insn): ...this and adjust.
26795 (trap): New define_insn.
26797 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26799 PR tree-optimization/79389
26800 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26801 (noce_reversed_cond_code): New function.
26802 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26803 reversed_comparison_code. Formatting fix.
26804 (noce_try_store_flag): Test rev_cond != NULL in addition to
26805 reversed_comparison_code.
26806 (noce_try_store_flag_constants): Likewise.
26807 (noce_try_store_flag_mask): Likewise.
26808 (noce_try_addcc): Use rev_cond if non-NULL instead of
26809 reversed_comparison_code.
26810 (noce_try_cmove_arith): Likewise. Formatting fixes.
26811 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26812 (noce_find_if_block): Initialize rev_cond.
26813 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26814 instead of false as last argument never attempt to reverse it
26817 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26819 PR tree-optimization/79663
26820 * tree-predcom.c (combine_chains): Process refs in reverse order
26821 only for ZERO length chains, and add explaining comment.
26823 2017-02-23 Jeff Law <law@redhat.com>
26825 PR tree-optimization/79578
26826 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26827 in call to operand_equal_p.
26829 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26832 * config/i386/cpuid.h: Fix another undefined behavior.
26834 2017-02-23 Richard Biener <rguenther@suse.de>
26836 PR tree-optimization/79683
26837 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26838 vector types for data-refs.
26840 2017-02-23 Martin Liska <mliska@suse.cz>
26842 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26844 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26846 PR middle-end/79665
26847 * internal-fn.c (get_range_pos_neg): Moved to ...
26848 * tree.c (get_range_pos_neg): ... here. No longer static.
26849 * tree.h (get_range_pos_neg): New prototype.
26850 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26851 are known to be in between 0 and signed maximum inclusive, try to
26852 expand both unsigned and signed divmod and use the cheaper one from
26855 2017-02-22 Jeff Law <law@redhat.com>
26857 PR tree-optimization/79578
26858 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26859 to compare base operands.
26861 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26864 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26865 gpc_reg_operand instead of fpr_reg_operand.
26867 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26869 * config/mips/mips.c (mips_return_in_memory): Force FP
26870 vector types to be returned in memory for o32 ABI.
26872 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26874 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26875 instead of DW_TAG_member for static data member declarations and don't
26876 set no_linkage_name for static inline data members.
26877 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26880 2017-02-22 Martin Liska <mliska@suse.cz>
26882 * doc/invoke.texi: Replace inequality signs with square brackets
26885 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26887 PR tree-optimization/68644
26888 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26890 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26893 * lra-constraints.c (simplify_operand_subreg): Handle
26894 WORD_REGISTER_OPERATIONS targets.
26896 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26899 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26900 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26901 elimination by swapping fld*.
26903 2017-02-22 Richard Biener <rguenther@suse.de>
26905 PR tree-optimization/79673
26906 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26907 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26908 irrelevant address-space qualifiers and avoiding a
26909 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26911 2017-02-22 Richard Biener <rguenther@suse.de>
26913 PR tree-optimization/79666
26914 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26915 to not symbolically negate if that may introduce undefined
26918 2017-02-22 Martin Liska <mliska@suse.cz>
26921 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26922 * data-streamer-out.c (streamer_write_gcov_count_stream):
26924 * value-prof.c (stream_out_histogram_value): Make assert more
26925 precise based on type of counter.
26927 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26930 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26931 Use nonimmediate_operand instead of memory_operand for operand 1.
26932 (float-extend standard_x87sse_constant_load splitter): Ditto.
26934 2017-02-21 Jeff Law <law@redhat.com>
26936 PR tree-optimization/79621
26937 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26938 blocks with edges to themselves.
26940 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26943 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26944 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26945 Use gimple_call_builtin_p.
26948 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26949 on temporarily removed DEBUG_INSNs.
26951 PR tree-optimization/79649
26952 * tree-loop-distribution.c (classify_partition): Give up on
26953 non-generic address space loads/stores.
26955 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26957 * doc/loop.texi (Loop manipulation): Remove nonexistent
26958 tree_ssa_loop_version from the documentation.
26959 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26961 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26964 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26965 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26966 * config/rs6000/rs6000.c: Include except.h.
26967 (rs6000_expand_split_stack_prologue): Call
26968 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26970 2017-02-21 Martin Jambor <mjambor@suse.cz>
26973 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26974 have been analyzed.
26976 2017-02-21 Martin Jambor <mjambor@suse.cz>
26978 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26979 for backward compatibility only.
26980 * doc/invoke.texi (Option Summary): Remove all references to
26981 -fipa-cp-alignment.
26983 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26987 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26989 * lra-constraints.c (curr_insn_transform): Handle
26990 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26992 2017-02-21 Martin Liska <mliska@suse.cz>
26994 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26996 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26998 PR translation/79638
26999 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27001 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27004 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27005 (arm_function_ok_for_sibcall): Return false for an indirect call by
27006 descriptor if all the argument registers are used.
27007 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27008 alignment of the function.
27010 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27012 PR tree-optimization/61441
27013 * simplify-rtx.c (simplify_const_unary_operation): For
27014 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27015 the sNaN unmodified.
27017 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27019 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27020 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27021 instead of SYSTEM_HEADER_DIR.
27023 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27024 Martin Liška <mliska@suse.cz>
27026 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27027 Fix typos and grammar, use active voice, and clarify.
27029 2017-02-20 Marek Polacek <polacek@redhat.com>
27031 PR middle-end/79537
27032 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27035 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27037 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27040 * config/i386/i386.c (ix86_expand_builtin): Handle
27041 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27042 ix86_builtins_isa[fcode].isa as a requirement of those
27043 flags and any other flag in the bitmask.
27044 (ix86_init_mmx_sse_builtins): Use 0 instead of
27045 ~OPTION_MASK_ISA_64BIT as mask.
27046 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27047 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27048 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27049 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27051 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27054 * lra-constraints.c (split_reg): Check requested split mode
27055 is supported by the register.
27057 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27059 * lra-constraints.c (simplify_operand_subreg): Remove early
27062 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27065 * lra-constraints.c (curr_insn_transform): Tighten condition
27066 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27068 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27071 * lra-constraints.c (curr_insn_transform): Handle
27072 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27074 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27077 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27079 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27081 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27084 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27087 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27089 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27092 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27094 * final.c (last_columnnum, override_columnnum): New variables.
27095 (final_start_function): Set last_columnnum, pass it to begin_prologue
27096 hook and pass 0 to dwarf2out_begin_prologue.
27097 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27098 to source_line debug hook.
27099 (notice_source_line): Compute last_columnnum and for debug_column_info
27100 return true on column changes.
27101 * debug.h (struct gcc_debug_hooks): Add column argument to
27102 source_line and begin_prologue hooks.
27103 (debug_nothing_int_charstar_int_bool): Remove prototype.
27104 (debug_nothing_int_int_charstar,
27105 debug_nothing_int_int_charstar_int_bool): New prototypes.
27106 (dwarf2out_begin_prologue): Add column argument.
27107 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27108 begin_prologue hooks.
27109 (debug_nothing_int_charstar_int_bool): Remove.
27110 (debug_nothing_int_int_charstar,
27111 debug_nothing_int_int_charstar_int_bool): New functions.
27112 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27113 through to dwarf2out_source_line.
27114 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27115 (dwarf2out_source_line): Add column argument, emit it if requested.
27116 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27118 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27119 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27120 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27121 through to dwarf2out_begin_prologue.
27122 (vmsdbgout_source_line): Add column argument, pass it through to
27123 dwarf2out_source_line.
27124 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27125 dbxout_source_line caller.
27126 (dbxout_source_line): Add column argument.
27128 * common.opt (gno-column-info, gcolumn-info): New options.
27129 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27130 (check_die): Also test for multiple DW_AT_decl_column attributes.
27131 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27132 DW_AT_decl_column if requested.
27133 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27135 (gen_variable_die): Likewise.
27136 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27137 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27140 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27141 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27142 (ix86_handle_option): Handle OPT_m3dnowa.
27143 * doc/invoke.texi (-m3dnowa): Document.
27144 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27145 -m3dnowa instead of -m3dnow -march=athlon.
27148 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27149 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27151 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27154 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27155 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27156 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27157 generator for vsx_xxpermdi_<mode>_be.
27158 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27159 force big-endian semantics.
27160 (vsx_xxpermdi_<mode>_be): New define_expand with same
27161 implementation as previous version of vsx_xxpermdi_<mode>.
27163 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27165 PR tree-optimization/79327
27166 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27167 variable, its initialization and use.
27169 2017-02-17 Julia Koval <julia.koval@intel.com>
27171 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27172 (OPTION_MASK_ISA_PKU_UNSET): New.
27173 (ix86_handle_option): Handle -mrdpid.
27174 * config/i386/cpuid.h (bit_RDPID): New.
27175 * config/i386/driver-i386.c (host_detect_local_cpu):
27176 Detect RDPID feature.
27177 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27178 * config/i386/i386-c.c (ix86_target_macros_internal):
27180 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27181 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27182 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27183 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27184 * config/i386/i386.md (define_insn "rdpid"): New.
27185 * config/i386/i386.opt Add -mrdpid.
27186 * config/i386/immintrin.h (_rdpid_u32): New.
27188 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27190 PR rtl-optimization/79541
27191 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27192 instead of transforming it into USE.
27194 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27196 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27197 If HONOR_SNANS (SFmode) force the input to a register.
27198 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27199 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27200 an frsp or similar insn.
27202 2017-02-17 Martin Liska <mliska@suse.cz>
27204 PR rtl-optimization/79577
27205 * params.def (selsched-max-sched-times): Increase minimum to 1.
27207 2017-02-17 Martin Liska <mliska@suse.cz>
27209 PR rtl-optimization/79574
27210 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27212 2017-02-17 Martin Liska <mliska@suse.cz>
27214 PR tree-optimization/79529
27215 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27216 ssa_defined_default_def_p to handle cases which are implicitly
27218 * tree-ssa.c (ssa_defined_default_def_p): New function.
27219 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27220 which are implicitly defined.
27221 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27223 2017-02-17 Richard Biener <rguenther@suse.de>
27225 PR middle-end/79576
27226 * params.def (max-ssa-name-query-depth): Limit to 10.
27228 2017-02-17 Richard Biener <rguenther@suse.de>
27230 PR tree-optimization/79552
27231 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27234 2017-02-17 Richard Biener <rguenther@suse.de>
27237 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27239 2017-02-17 Marek Polacek <polacek@redhat.com>
27241 PR middle-end/79536
27242 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27243 (fold_negate_expr): New wrapper.
27245 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27247 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27248 Correct terminology and de-emphasize pre-standard behavior.
27250 2017-02-16 Alan Modra <amodra@gmail.com>
27252 PR rtl-optimization/79286
27253 * ira.c (def_dominates_uses): New function.
27254 (update_equiv_regs): Don't create an equivalence for insns that
27255 may trap where the register def does not dominate the use.
27257 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27259 PR rtl-optimization/78127
27260 * lra.c (lra): Call lra_eliminate before finish the loop after
27263 2017-02-16 Richard Biener <rguenther@suse.de>
27265 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27267 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27268 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27269 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27270 (isl_val_int_from_wi): New function.
27271 (extract_affine_gmp): Rename to ...
27272 (extract_affine_wi): ... this, take a widest_int.
27273 (extract_affine_int): Just wrap extract_affine_wi.
27274 (add_param_constraints): Use isl_val_int_from_wi.
27275 (add_loop_constraints): Likewise, and extract_affine_wi.
27277 2017-02-15 Jeff Law <law@redhat.com>
27279 PR middle-end/79521
27280 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27281 ira_init_register_move_cost_if_necessary.
27283 2017-02-15 Martin Sebor <msebor@redhat.com>
27285 PR middle-end/32003
27286 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27287 removed in a prior commit.
27289 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27291 PR tree-optimization/79347
27292 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27293 counters during peeling.
27295 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27297 * Makefile.in (site.exp): Remove "set ISLVER".
27299 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27302 * real.c (real_from_integer): Call real_convert even for decimal.
27304 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27307 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27309 2017-02-14 Andrew Pinski <apinski@cavium.com>
27311 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27312 cores and change the partno/implementer to be correct.
27313 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27314 the 'B" as the implementer.
27315 * config/aarch64/aarch64-tune.md: Regenerate.
27317 2017-02-14 Carl Love <cel@us.ibm.com>
27319 * config/rs6000/rs6000.c: Add case statement entry to make the
27320 xvcvuxdsp built-in argument unsigned.
27321 * config/rs6000/vsx.md: Fix the source and return operand types so they
27322 match the instruction definitions from the ISA document. Fix typo
27323 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27326 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27329 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27330 member early_clobber_alts.
27331 * lra-lives.c (reg_early_clobber_p): New.
27332 (process_bb_lives): Use it.
27333 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27334 (debug_operand_data): Initialize early_clobber_alts.
27335 (setup_operand_alternative): Set up early_clobber_alts.
27336 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27337 alternatives to new_insn_reg.
27338 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27340 (lra_update_insn_regno_info): Pass the new arg.
27342 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27344 PR middle-end/79505
27345 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27346 (new_oacc_loop_raw): Don't clear already cleared fields.
27349 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27350 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27351 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27353 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27356 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27358 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27361 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27362 the extra instruction to the right place to store 128-bit constant
27365 2017-02-14 Martin Sebor <msebor@redhat.com>
27367 PR middle-end/79448
27368 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27369 warning for strings of unknown length.
27371 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27373 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27375 2017-02-14 Jeff Law <law@redhat.com>
27378 * ira-costs.c (scan_one_insn): Initialize register move costs
27379 for pseudos seen in USE/CLOBBER insns.
27381 PR tree-optimization/79095
27382 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27383 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27384 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27385 if the operands are known to be not equal, then the resulting range
27387 (intersect_ranges): If the new range is ~[0,0] and the old range is
27388 wide, then prefer ~[0,0].
27389 * tree-vrp.c (overflow_comparison_p_1): New function.
27390 (overflow_comparison_p): New function.
27391 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27392 if NAME is used in an overflow test.
27393 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27394 overflow check that can be expressed as an equality test, then adjust
27395 ops to be that equality test.
27397 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27399 * config/s390/s390-builtin-types.def: Remove flags argument.
27400 * config/s390/s390.c (s390_init_builtins): Likewise.
27402 2017-02-14 Martin Liska <mliska@suse.cz>
27404 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27405 vector. Fix trailing white spaces.
27407 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27409 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27412 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27414 PR rtl-optimization/68664
27415 * config/arm/arm.c (arm_sched_can_speculate_insn):
27416 New function. Declare prototype.
27417 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27419 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27421 PR rtl-optimization/68664
27422 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27424 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27426 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27428 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27429 max skip bytes for function, loop and jump.
27431 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27433 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27434 ABS_EXPR for gimple dump.
27436 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27439 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27441 PR tree-optimization/79408
27442 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27443 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27444 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27445 also if rhs1 is INTEGER_CST.
27447 2017-02-14 Richard Biener <rguenther@suse.de>
27449 PR middle-end/79432
27450 * tree-into-ssa.c (insert_phi_nodes): When the function can
27451 have abnormal edges rewrite SSA names with broken use-def
27452 dominance out of SSA and register them for PHI insertion.
27454 2017-02-13 Martin Sebor <msebor@redhat.com>
27456 PR middle-end/79496
27457 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27458 clearing info.nowrite flag when snprintf size argument is a range.
27460 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27462 * cprop.c (cprop_jump): Add missing space in string literal.
27463 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27464 (get_constraint_for_component_ref): Likewise.
27465 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27466 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27467 * lra-constraints.c (process_alt_operands): Likewise.
27468 * ipa-inline.c (inline_small_functions): Likewise.
27469 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27470 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27471 * trans-mem.c (diagnose_tm_1_op): Likewise.
27472 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27473 (grid_parallel_clauses_gridifiable): Likewise.
27475 * config/nvptx/mkoffload.c (process): Add space in between
27478 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27479 "MOD4_SSE_REGS" and "ALL_REGS".
27481 * spellcheck.c (test_data): Add , in between "foo" and "food".
27483 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27486 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27487 boundary crossing check and subsequent code generation agree.
27489 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27491 * config/aarch64/aarch64.c (has_memory_op): Delete.
27492 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27495 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27497 PR rtl-optimization/79388
27498 PR rtl-optimization/79450
27499 * combine.c (distribute_notes): When removing TEM_INSN for which
27500 corresponding dest has last value recorded, invalidate that last
27503 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27505 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27506 of explicit '@'. Add missing assembly comment marker on branch costs
27509 2017-02-13 Nathan Sidwell <nathan@acm.org>
27511 * gengtype-lex.l (<in_struct>): Add '/'.
27513 2017-02-13 Martin Liska <mliska@suse.cz>
27516 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27518 2017-02-13 Richard Biener <rguenther@suse.de>
27520 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27522 * configure: Re-generate.
27523 * config.in: Likewise.
27524 * graphite-dependences.c: Simplify as if
27525 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27526 * graphite-isl-ast-to-gimple.c: Likewise.
27527 * graphite-optimize-isl.c: Likewise.
27528 * graphite-poly.c: Likewise.
27529 * graphite-sese-to-poly.c: Likewise.
27530 * graphite.h: Likewise.
27531 * toplev.c: Include isl/version.h and use isl_version () for
27532 printing the ISL version.
27533 * doc/install.texi: Update ISL requirement.
27535 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27537 * doc/standards.texi (Standards): Update reference to
27540 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27542 * doc/extend.texi (Named Address Spaces): sourceware.org now
27544 * doc/install.texi (Binaries): Ditto.
27547 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27549 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27550 standard terminology "stringize"/"stringizing" throughout.
27551 * doc/cppinternals.texi: Likewise.
27553 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27555 * doc/extend.texi: Fix some spelling mistakes and typos.
27556 * doc/invoke.texi: Likewise.
27558 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27561 * params.def (inline-min-speedup) Change from 10 to 8.
27563 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27565 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27568 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27571 * ipa-inline-analysis.c (get_minimal_bb): New function.
27572 (record_modified): Use it.
27573 (remap_edge_change_prob): Handle also ancestor functions.
27575 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27577 * doc/contrib.texi (Contributors): Remove broken link into
27578 the Mauve CVS repository.
27580 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27582 PR middle-end/79454
27583 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27584 result computation whenever lhs doesn't have vector mode, not
27585 just when it has BLKmode.
27587 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27589 * doc/makefile.texi (profiledbootstrap): Refer to the
27590 installation instructions only in textual form.
27592 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27595 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27597 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27599 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27600 (Specific): Update mingw-w64 reference.
27602 (Specific): Remove broken link to Renesas RX processor.
27604 2017-02-10 Richard Biener <rguenther@suse.de>
27606 * toplev.c (process_options): Do not mention obsolete graphite
27607 options when printing sorry message about missing graphite support.
27608 Mention -floop-nest-optimize.
27610 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27612 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27613 (vtst_p16): Likewise.
27614 (vtstq_p8): Likewise.
27615 (vtstq_p16): Likewise.
27617 (vtstq_p64): Likewise.
27618 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27619 (vset_lane_p64): New.
27620 (vsetq_lane_p64): New.
27622 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27624 PR tree-optimization/79411
27625 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27626 stmt operands are SSA_NAMEs used in abnormal phis.
27627 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27630 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27633 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27636 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27638 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27640 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27642 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27646 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27647 not arbitrary TREE_CONSTANT.
27650 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27651 "omp declare target link" attribute unless is_global_var.
27652 * omp-offload.c (find_link_var_op): Likewise.
27654 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27655 Chung-Lin Tang <cltang@codesourcery.com>
27657 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27659 (gimplify_adjust_omp_clauses): Don't delete TILE.
27660 (gimplify_omp_for): Deal with TILE.
27661 * internal-fn.c (expand_GOACC_TILE): New function.
27662 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27664 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27665 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27667 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27668 avoid DIV for outermost collapse var.
27669 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27670 Remove out of date comments, fix whitespace.
27671 * omp-general.c (omp_extract_for_data): Deal with tiling.
27672 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27673 adjust OLF_DIM_BASE value.
27674 (struct omp_for_data): Add tiling field.
27675 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27676 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27677 for auto loops. Remove default auto determining, moved to
27678 oacc_loop_fixed_partitions.
27679 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27680 stmts, add e_mask field.
27681 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27682 (oacc_thread_numbers): Use oacc_dim_call.
27683 (oacc_xform_tile): New.
27684 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27685 (finish_oacc_loop): Adjust for ifns vector.
27686 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27687 add case for GOACC_TILE fns.
27688 (oacc_loop_xform_loop): Delete.
27689 (oacc_loop_process): Iterate over call list directly, and add
27690 handling for GOACC_TILE fns.
27691 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27693 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27694 vector partitioning to outer loops. Assign 2 partitions to loops
27695 when available. Add TILE handling.
27696 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27697 (execite_oacc_device_lower): Process GOACC_TILE fns,
27698 ignore unknown specs.
27699 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27700 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27701 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27703 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27705 * configure.ac (ACX_BUGURL): Update.
27706 * configure: Regenerate.
27708 2017-02-09 Richard Biener <rguenther@suse.de>
27710 PR tree-optimization/69823
27711 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27712 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27714 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27716 * config/arc/arc-c.def: Add __NPS400__ definition.
27717 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27718 (TARGET_NPS400): Define.
27720 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27722 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27724 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27725 pointer, arch_info.
27726 (arc_cpu_types): Fill the arch_info field with a pointer into the
27727 arc_arch_types table.
27728 (arc_selected_cpu): Declare.
27729 * config/arc/arc.c (arc_selected_cpu): Make global.
27730 (arc_selected_arch): Delete.
27731 (arc_base_cpu): Delete.
27732 (arc_override_options): Remove references to deleted variables,
27733 update access to arch information.
27734 (ARC_OPT): Update access to arch information.
27735 (ARC_OPTX): Likewise.
27736 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27737 (TARGET_ARC600): Update access to arch information.
27738 (TARGET_ARC601): Likewise.
27739 (TARGET_ARC700): Likewise.
27740 (TARGET_EM): Likewise.
27741 (TARGET_HS): Likewise.
27742 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27745 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27748 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27749 condition/operands for integer GE/LE/GEU/LEU operations.
27751 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27753 PR translation/79397
27754 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27757 2017-02-08 Martin Jambor <mjambor@suse.cz>
27760 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27761 whether allocation happened.
27762 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27763 nothing was allocated.
27765 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27767 PR tree-optimization/79408
27768 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27769 constant, but SSA_NAME with a known integer range, use the minimum
27770 of that range instead of op1 to determine if modulo can be replaced
27771 with its first operand.
27773 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27775 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27777 2017-02-08 Richard Biener <rguenther@suse.de>
27779 PR tree-optimization/71824
27780 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27781 Check all loops contained in the merged region.
27783 2017-02-07 Andrew Pinski <apinski@cavium.com>
27785 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27787 2017-02-07 Andrew Pinski <apinski@cavium.com>
27789 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27790 (thunderxt88): Likewise.
27791 (thunderxt81): Disable LSE and change v8.1 to v8.
27792 (thunderxt83): Likewise.
27794 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27795 Richard Biener <rguenther@suse.de>
27797 PR middle-end/79399
27798 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27799 type from int to size_t.
27800 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27802 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27804 PR rtl-optimization/79386
27805 * cprop.c (bypass_conditional_jumps): Initialize
27806 bypass_last_basic_block already before splitting bbs after
27807 unconditional traps...
27808 (bypass_conditional_jumps): ... rather than here.
27811 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27812 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27813 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27814 fix -masm=intel patterns.
27816 2017-02-07 Richard Biener <rguenther@suse.de>
27818 PR tree-optimization/79256
27819 PR middle-end/79278
27820 * builtins.c (get_object_alignment_2): Use min_align_of_type
27821 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27822 and ADJUST_FIELD_ALIGN.
27824 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27826 * doc/tm.texi: Regenerate.
27827 * stor-layout.c (layout_decl): Adjust.
27828 (update_alignment_for_field): Likewise.
27829 (place_field): Likewise.
27830 (min_align_of_type): Likewise.
27831 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27832 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27833 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27834 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27835 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27836 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27837 * config/i386/i386.c (x86_field_alignment): Likewise.
27838 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27839 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27840 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27841 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27842 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27843 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27847 2017-01-30 Richard Biener <rguenther@suse.de>
27849 PR tree-optimization/79256
27850 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27851 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27854 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27856 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27857 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27858 builtins to SImode and emit a zero-extend, if necessary.
27860 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27862 * docs/invoke.texi (RISC-V Options): Alphabetize.
27864 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27866 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27869 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27871 * config/riscv/riscv.c: New file.
27872 * common/config/riscv/riscv-common.c: Likewise.
27873 * config.gcc: Likewise.
27874 * config/riscv/constraints.md: Likewise.
27875 * config/riscv/elf.h: Likewise.
27876 * config/riscv/generic.md: Likewise.
27877 * config/riscv/linux.h: Likewise.
27878 * config/riscv/multilib-generator: Likewise.
27879 * config/riscv/peephole.md: Likewise.
27880 * config/riscv/pic.md: Likewise.
27881 * config/riscv/predicates.md: Likewise.
27882 * config/riscv/riscv-builtins.c: Likewise.
27883 * config/riscv/riscv-c.c: Likewise.
27884 * config/riscv/riscv-ftypes.def: Likewise.
27885 * config/riscv/riscv-modes.def: Likewise.
27886 * config/riscv/riscv-opts.h: Likewise.
27887 * config/riscv/riscv-protos.h: Likewise.
27888 * config/riscv/riscv.h: Likewise.
27889 * config/riscv/riscv.md: Likewise.
27890 * config/riscv/riscv.opt: Likewise.
27891 * config/riscv/sync.md: Likewise.
27892 * config/riscv/t-elf-multilib: Likewise.
27893 * config/riscv/t-linux: Likewise.
27894 * config/riscv/t-linux-multilib: Likewise.
27895 * config/riscv/t-riscv: Likewise.
27896 * configure.ac: Likewise.
27897 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27898 Waterman as RISC-V maintainers.
27899 * doc/install.texi: Add RISC-V entries.
27900 * doc/invoke.texi: Add RISC-V options section.
27901 * doc/md.texi: Add RISC-V constraints section.
27902 * configure: Regenerated.
27904 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27907 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27908 false values to be constant vectors with all 0 or all 1 bits set.
27909 (vcondu<mode><mode>): Likewise.
27910 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27912 (fpmask_comparison_operator): Update comment.
27913 (vecint_comparison_operator): New predicate.
27914 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27915 vector conditionals when the true and false values are constant
27916 vectors with all 0 bits or all 1 bits set.
27918 2017-02-06 Martin Sebor <msebor@redhat.com>
27920 PR tree-optimization/79376
27921 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27923 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27925 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27926 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27927 to simplify split condition.
27929 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27931 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27932 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27935 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27937 PR rtl-optimization/68664
27938 * target.def (can_speculate_insn): New hook.
27939 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27940 * doc/tm.texi: Regenerate.
27941 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27942 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27943 (rs6000_sched_can_speculate_insn): New function.
27945 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27947 PR tree-optimization/79284
27948 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27949 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27950 vectorizable_mask_load_store, vectorizable_operation,
27951 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27952 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27953 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27954 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27955 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27956 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27957 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27958 is_gimple_assign (stmt). Replace another such test with
27959 is_gimple_assign (stmt).
27961 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27964 * config/avr/avr.c (rtl-iter.h): Include it.
27965 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27966 (avr_legitimate_combined_insn): ...and implementation.
27968 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27970 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27971 * config/s390/s390.c (s390_const_operand_ok)
27972 (s390_canonicalize_comparison, s390_extract_part)
27973 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27974 (s390_contiguous_bitmask_p, s390_rtx_costs)
27975 (legitimize_pic_address): Likewise.
27976 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27977 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27978 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27979 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27980 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27982 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27984 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27985 REGNO($0) == REGNO($1).
27987 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27989 * config/s390/linux.h(SIZE_TYPE): Add comment.
27991 2017-02-06 Julian Brown <julian@codesourcery.com>
27992 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27993 Virendra Pathak <virendra.pathak@broadcom.com>
27995 * config/aarch64/aarch64-cores.def: Change the scheduler
27997 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27998 * config/aarch64/thunderx2t99.md: New file.
28000 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28002 * doc/standards.texi (Go Language): Update link to language
28005 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28007 * tree-eh.c (lower_resx): Sanitize profile.
28008 (cleanup_empty_eh_move_lp): Likewise.
28010 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28013 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28015 * cfgloopmanip.h (loop_version): Update prototype.
28016 * modulo-sched.c (sms_schedule): Update call of loop_version.
28017 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28018 * tree-parloops.c (gen_parallel_loop): Likewise.
28019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28020 * tree-ssa-loop-split.c (split_loop): Likewise.
28021 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28022 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28024 2017-02-05 Martin Liska <mliska@suse.cz>
28027 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28029 (print_operand_address): Initialize a struct to zero.
28031 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28033 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28034 garbage collector only in textual form.
28036 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28038 * doc/extend.texi (x86 specific memory model extensions for
28039 transactional memory): Simplify a phrase.
28041 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28044 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28045 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28046 (atomic_storedi_1): Likewise.
28048 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28050 PR tree-optimization/79338
28051 * tree-parloops.c (gather_scalar_reductions): Don't call
28052 vect_analyze_loop_form for loop->inner before destroying loop's
28055 2017-02-03 Martin Sebor <msebor@redhat.com>
28057 PR tree-optimization/79327
28058 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28059 when precision has resulted in leading zeros.
28060 (format_integer): Adjust the likely counter to assume an unknown
28061 argument that may be zero is non-zero.
28063 2017-02-03 Jason Merrill <jason@redhat.com>
28066 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28067 avoid copying non-taken branch.
28069 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28071 PR tree-optimization/79340
28072 * tree-vect-loop.c (vectorizable_reduction): Release
28073 vec_defs elements after safe_splicing them into other vectors.
28076 PR tree-optimization/79327
28077 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28078 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28080 (format_integer): Use wide_int_to_tree instead of build_int_cst
28081 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28082 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28083 of shortest and longest sequence.
28085 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28087 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28088 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28090 2017-02-03 Walter Lee <walt@tilera.com>
28093 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28094 after initial stackframe link reg save.
28095 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28097 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28100 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28101 wu for stxssp alternative.
28103 2017-02-03 Martin Sebor <msebor@redhat.com>
28105 PR tree-optimization/79352
28106 * gimple-fold.c (get_range_strlen): Add argument.
28107 (get_range_strlen): Change return type to bool.
28108 (get_maxval_strlen): Pass in a dummy argument.
28109 * gimple-fold.h (get_range_strlen): Change return type to bool.
28110 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28111 * tree.h (array_at_struct_end_p): Add argument.
28112 * tree.c (array_at_struct_end_p): Handle it.
28114 2017-02-03 Martin Liska <mliska@suse.cz>
28117 * multiple_target.c (create_dispatcher_calls): Redirect edge
28118 from a caller of a dispatcher.
28119 (expand_target_clones): Make the clones local.
28120 (ipa_target_clone): Do both target clones and resolvers.
28121 (ipa_dispatcher_calls): Remove the pass.
28122 (pass_dispatcher_calls::gate): Likewise.
28123 (make_pass_dispatcher_calls): Likewise.
28124 * passes.def (pass_target_clone): Put as very first IPA early
28127 2017-02-03 Martin Liska <mliska@suse.cz>
28129 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28130 in case of a function with ifunc attribute.
28132 2017-02-03 Martin Liska <mliska@suse.cz>
28134 * cgraph.c (cgraph_node::dump): Dump function version info.
28135 * symtab.c (symtab_node::dump_base): Add missing new line.
28137 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28139 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28140 (ifcombine_ifandif): Use it.
28142 2017-02-03 Martin Liska <mliska@suse.cz>
28144 * doc/invoke.texi: Document default value for
28145 use-after-scope-direct-emission-threshold.
28147 2017-02-03 Martin Liska <mliska@suse.cz>
28149 PR tree-optimization/79339
28150 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28151 (format_floating): Likewise.
28153 2017-02-03 Martin Liska <mliska@suse.cz>
28156 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28158 (ipa_node_params_t::remove): Likewise.
28159 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28160 initialization from removed ipa_node_params_t::insert.
28161 (ipa_node_params::~ipa_node_params): Move from removed
28162 ipa_node_params_t::release.
28163 * symbol-summary.h (symbol_summary::m_released): New member.
28164 Do not release a summary twice. Do not allow to call finalizer
28165 for types of a summary that live in GGC memory.
28167 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28169 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28172 2017-02-02 Martin Sebor <msebor@redhat.com>
28174 PR middle-end/79275
28175 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28176 (format_string): Tighten up the range of output for non-constant
28177 strings and correct the expected range for wide non-constant strings.
28179 2017-02-02 Martin Sebor <msebor@redhat.com>
28181 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28183 PR middle-end/32003
28184 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28186 (-fdump-tree-@var): Add to index and document how to come up
28187 with pass-specific option and dump file names.
28188 (-fdump-passes): Clarify where to look for output.
28190 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28192 PR middle-end/77445
28193 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28194 statistics of the analyzed path; allow threading for speed when
28195 any of BBs along the path are optimized for speed.
28197 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28199 PR middle-end/78468
28200 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28201 settings of the virtual registers.
28204 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28206 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28207 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28210 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28212 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28213 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28215 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28217 * config/s390/s390.md: Add missing comments with the expanded
28219 * config/s390/vector.md: Likewise.
28220 * config/s390/vx-builtins.md: Likewise.
28222 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28225 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28226 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28227 conditions on a single line.
28229 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28231 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28232 __S390_VX__ to __VX__.
28234 2017-02-01 Andrew Pinski <apinski@cavium.com>
28236 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28237 stmt_info to record_stmt_cost.
28238 (vect_get_known_peeling_cost): Pass stmt_info if known to
28240 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28241 cpu_vector_cost field into
28242 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28243 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28244 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28245 splitting of scalar_stmt_cost and vec_stmt_cost.
28246 (thunderx_vector_cost): Likewise.
28247 (cortexa57_vector_cost): LIkewise.
28248 (exynosm1_vector_cost): Likewise.
28249 (xgene1_vector_cost): Likewise.
28250 (thunderx2t99_vector_cost): Improve after the splitting of the two
28252 (aarch64_builtin_vectorization_cost): Update for the splitting of
28253 scalar_stmt_cost and vec_stmt_cost.
28255 2017-02-01 Torvald Riegel <triegel@redhat.com>
28256 Richard Henderson <rth@redhat.com>
28258 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28259 conditional on existance of a fast atomic load.
28260 * optabs-query.c (can_atomic_load_p): New function.
28261 * optabs-query.h (can_atomic_load_p): Declare it.
28262 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28263 no fast atomic load is available for the particular size of access.
28264 (expand_atomic_compare_and_swap): Likewise.
28265 (expand_atomic_load): Likewise.
28266 (expand_atomic_store): Likewise.
28267 (expand_atomic_fetch_op): Likewise.
28268 * testsuite/lib/target-supports.exp
28269 (check_effective_target_sync_int_128): Remove x86 because it provides
28270 no fast atomic load.
28271 (check_effective_target_sync_int_128_runtime): Likewise.
28273 2017-02-01 Richard Biener <rguenther@suse.de>
28275 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28276 (graphite_transform_loops): Provide opt-info for optimized nests.
28277 * tree-parloop.c (parallelize_loops): Provide opt-info for
28278 parallelized loops.
28280 2017-02-01 Richard Biener <rguenther@suse.de>
28282 PR middle-end/79315
28283 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28284 was not set before.
28286 2017-02-01 Richard Biener <rguenther@suse.de>
28288 PR tree-optimization/71824
28289 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28290 Verify the loops are valid in the merged SESE region.
28291 (scop_detection::can_represent_loop_1): Check analyzing the
28292 evolution of the number of iterations in the region succeeds.
28294 2017-01-31 Ian Lance Taylor <iant@golang.org>
28296 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28297 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28299 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28301 PR preprocessor/79210
28302 * input.c (get_substring_ranges_for_loc): Replace line_width
28303 assertion with error-handling.
28305 2017-01-31 Richard Biener <rguenther@suse.de>
28307 PR tree-optimization/77318
28308 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28309 (create_pw_aff_from_tree): Take loop parameter.
28310 (add_condition_to_pbb): Pass loop of the condition to
28311 create_pw_aff_from_tree.
28313 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28315 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28316 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28318 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28322 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28324 (convert_int_to_float128): Likewise.
28325 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28326 (convert_int_to_float128): Likewise.
28327 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28328 (UNSPEC_IEEE128_CONVERT): Likewise.
28329 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28330 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28331 Use local variables for IBM extended format.
28332 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28333 (fix_trunc<mode>si2_fprs): Likewise.
28334 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28335 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28336 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28337 to know that we can now have integers of all sizes in vector
28339 (fix<uns>_<mode>di2_hw): Likewise.
28340 (float<uns>_<mode>si2_hw): Likewise.
28341 (fix_<mode>si2_hw): Likewise.
28342 (fixuns_<mode>si2_hw): Likewise.
28343 (float<uns>_<mode>di2_hw): Likewise.
28344 (float_<mode>di2_hw): Likewise.
28345 (float_<mode>si2_hw): Likewise.
28346 (floatuns_<mode>di2_hw): Likewise.
28347 (floatuns_<mode>si2_hw): Likewise.
28348 (xscvqp<su>wz_<mode>): Delete, no longer used.
28349 (xscvqp<su>dz_<mode>): Likewise.
28350 (xscv<su>dqp_<mode>): Likewise.
28351 (ieee128_mfvsrd_64bit): Likewise.
28352 (ieee128_mfvsrd_32bit): Likewise.
28353 (ieee128_mfvsrwz): Likewise.
28354 (ieee128_mtvsrw): Likewise.
28355 (ieee128_mtvsrd_64bit): Likewise.
28356 (ieee128_mtvsrd_32bit): Likewise.
28358 2017-01-31 Martin Liska <mliska@suse.cz>
28361 * ipa-prop.c (ipa_free_all_node_params): Call release method
28362 instead of ~sumbol_summary to not to trigger double times
28365 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28367 PR tree-optimization/71691
28368 * bitmap.h (class auto_bitmap): New.
28369 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28370 is_maybe_undefined instead of ssa_undefined_value_p.
28372 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28374 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28375 __S390_ARCH_LEVEL__ to __ARCH__.
28377 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28379 PR tree-optimization/79267
28380 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28381 if should_remove_lhs_p is true.
28383 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28386 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28387 (add_alignment_attribute): New.
28388 (base_type_die): Add alignment attribute.
28389 (subrange_type_die): Likewise.
28390 (modified_type_die): Likewise.
28391 (gen_array_type_die): Likewise.
28392 (gen_descr_array_type_die: Likewise.
28393 (gen_enumeration_type_die): Likewise.
28394 (gen_subprogram_die): Likewise.
28395 (gen_variable_die): Likewise.
28396 (gen_field_die): Likewise.
28397 (gen_ptr_to_mbr_type_die): Likewise.
28398 (gen_struct_or_union_type_die): Likewise.
28399 (gen_subroutine_type_die): Likewise.
28400 (gen_typedef_die): Likewise.
28401 (base_type_cmp): Compare alignment attribute.
28403 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28406 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28407 (setb_unsigned) New pattern for setb with CCUNS.
28408 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28409 subfc./subfe sequence to avoid overflow problems. Generate a
28410 shorter sequence with cmpld/setb for power9.
28411 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28412 for generating subfc. instruction.
28413 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28414 now uses this instruction.
28416 2017-01-30 Ian Lance Taylor <iant@google.com>
28419 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28420 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28422 2017-01-30 Martin Sebor <msebor@redhat.com>
28424 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28425 Move constant to the right of a relational operator.
28426 (get_mpfr_format_length, format_character, format_string): Ditto.
28427 (should_warn_p, maybe_warn): Same.
28429 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28431 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28434 * asan.c (get_translation_unit_decl): Remove function.
28435 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28437 2017-01-30 Martin Liska <mliska@suse.cz>
28439 PR gcov-profile/79259
28440 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28441 -fprofile-generate.
28443 2017-01-30 Martin Liska <mliska@suse.cz>
28446 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28447 Initialize variables with NULL value.
28449 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28452 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28454 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28456 2017-01-30 Richard Biener <rguenther@suse.de>
28458 PR tree-optimization/79276
28459 * tree-vrp.c (process_assert_insertions): Properly adjust common
28460 when removing a duplicate.
28462 * gcc.dg/torture/pr79276.c: New testcase.
28464 2017-01-30 Richard Biener <rguenther@suse.de>
28466 PR tree-optimization/79256
28467 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28468 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28470 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28472 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28475 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28476 ("*r<noxa>sbg_<mode>_sll_bitmask")
28477 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28478 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28479 Use contiguous_bitmask_nowrap_operand.
28481 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28484 * config/rs6000/altivec.h (vec_xl): Revise #define.
28485 (vec_xst): Likewise.
28487 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28489 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28491 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28493 PR rtl-optimization/79194
28494 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28495 traps before call to bypass_conditional_jumps.
28497 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28499 PR tree-optimization/71374
28500 * lra-constraints.c (check_conflict_input_operands): New.
28501 (match_reload): Use it.
28503 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28506 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28507 account to calculate conflict_set.
28509 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28511 PR rtl-optimization/78559
28512 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28513 other_insn in combine.
28515 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28517 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28518 uint16_type_node for BT_UINT16.
28520 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28522 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28523 "RTL Tests" to menu.
28524 (GIMPLE Tests): New node.
28525 (RTL Tests): New node.
28527 2017-01-27 Richard Biener <rguenther@suse.de>
28529 PR tree-optimization/79245
28530 * tree-loop-distribution.c (distribute_loop): Apply cost
28531 modeling also to detected patterns.
28533 2017-01-27 Richard Biener <rguenther@suse.de>
28535 PR tree-optimization/71433
28536 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28537 (compare_assert_loc): New function.
28538 (process_assert_insertions): Sort and optimize assert locations
28539 to remove duplicates and push down identical assertions on
28540 edges to their destination block.
28542 2017-01-27 Richard Biener <rguenther@suse.de>
28544 PR tree-optimization/79244
28545 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28546 out SSA names even if abnormal.
28548 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28550 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28551 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28552 instead of MPFR_RNDN.
28554 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28557 * arm.c (arm_option_override): Don't call build_target_option_node
28558 until after doing all option overrides.
28559 (arm_valid_target_attribute_tree): Likewise.
28561 2017-01-27 Martin Liska <mliska@suse.cz>
28563 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28564 for {cd}tors and C++ {cd}tors.
28566 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28568 * config/s390/s390.md ("*setmem_long_and")
28569 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28571 2017-01-26 Martin Sebor <msebor@redhat.com>
28573 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28576 2017-01-26 Martin Sebor <msebor@redhat.com>
28578 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28579 HAVE_DFmode before using XFmode or DFmode.
28580 (parse_directive): Avoid using the z length modifier to avoid
28581 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28583 PR middle-end/78703
28584 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28585 to accept adjustment as an array.
28586 (get_int_range): New function.
28587 (struct directive): Make width and prec arrays.
28588 (directive::set_width, directive::set_precision): Call get_int_range.
28589 (format_integer, format_floating): Handle width and precision ranges.
28590 (format_string, parse_directive): Same.
28592 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28595 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28596 comdat_type_p set, just clone them, but keep the children in the
28600 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28601 which have direct callers with -fvar-tracking-assignments enabled
28603 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28604 inside of type units.
28606 2017-01-26 Martin Sebor <msebor@redhat.com>
28608 PR middle-end/78703
28609 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28611 (struct format_result): Replace number_chars, number_chars_min,
28612 and number_chars_max with a single member of struct result_range.
28614 (format_result::operator+=): Adjust.
28615 (struct fmtresult): Remove bounded. Handle likely and unlikely
28617 (fmtresult::adjust_for_width_or_precision): New function.
28618 (fmtresult:type_max_digits): New function.
28619 (bytes_remaining): Handle likely and unlikely counters.
28620 (min_bytes_remaining): Remove.
28621 (format_percent): Simplify.
28622 (format_integer, format_floating): Set likely and unlikely counters.
28623 (get_string_length, format_character, format_string): Same.
28624 (format_plain, should_warn_p): New function.
28625 (maybe_warn): Call should_warn_p. Update diagnostic messages
28626 and handle those for all directives, including plain strings.
28627 (format_directive): Handle likely and unlikely counters.
28628 Remove unnecessary quoting from diagnostics. Add an informational
28630 (add_bytes): Remove.
28631 (pass_sprintf_length::compute_format_length): Simplify.
28632 (try_substitute_return_value): Handle likely and unlikely counters.
28634 2017-01-26 Carl Love <cel@us.ibm.com>
28636 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28637 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28639 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28642 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28643 endianess for subregs into account.
28644 * lra-constraints.c (lra_constraints): Do risky transformations
28645 always on the first iteration.
28646 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28647 last_call_used_reg_set.
28648 (process_bb_lives): Define and use last_call_used_reg_set.
28649 * lra.c (lra): Always continue after lra_constraints on the first
28652 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28654 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28656 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28658 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28660 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28661 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28662 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28663 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28664 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28665 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28666 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28667 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28668 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28670 2017-01-26 Marek Polacek <polacek@redhat.com>
28673 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28674 for the third operand.
28676 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28678 PR middle-end/79236
28679 * omp-low.c (struct omp_context): Add simt_stmt field.
28680 (scan_omp_for): Return omp_context *.
28681 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28682 context to the _simt_ SIMD stmt.
28683 (lower_omp_for): For combined SIMD with sibling _simt_
28684 SIMD, make sure to use the same decls in _looptemp_
28685 clauses as in the sibling.
28687 2017-01-26 David Sherwood <david.sherwood@arm.com>
28689 PR middle-end/79212
28690 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28693 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28696 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28697 emit fld b; fld a; if possible.
28699 * brig-builtins.def: Update copyright years.
28700 * config/arm/arm_acle_builtins.def: Update copyright years.
28702 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28705 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28706 constraint instead of o for the stxsd instruction.
28708 2017-01-25 Carl Love <cel@us.ibm.com>
28710 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28711 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28713 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28715 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28717 2017-01-25 Richard Biener <rguenther@suse.de>
28719 PR tree-optimization/69264
28720 * target.def (vector_alignment_reachable): Improve documentation.
28721 * doc/tm.texi: Regenerate.
28722 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28724 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28725 earlier changes with respect to TYPE_USER_ALIGN.
28726 (vector_alignment_reachable_p): Likewise. Improve dumping.
28728 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28731 * config/arm/arm.md (xordi3): Force constant operand into a register
28734 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28736 * doc/invoke.texi (-fstore-merging): Correct default optimization
28737 levels at which it is enabled.
28738 (-O): Move -fstore-merging from list to...
28741 2017-01-25 Richard Biener <rguenther@suse.de>
28744 * omp-expand.c: Include debug.h.
28745 (expand_omp_taskreg): Make sure to generate early debug before
28746 outlining anything from a function.
28747 (expand_omp_target): Likewise.
28748 (grid_expand_target_grid_body): Likewise.
28750 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28753 * asan.c (get_translation_unit_decl): New function.
28754 (asan_add_global): Extract modules file name from globals
28755 TRANSLATION_UNIT_DECL name.
28757 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28760 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28761 for long calls with APCS frame and VFP.
28763 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28765 * cfg.c (original_copy_tables_initialized_p): New function.
28766 * cfg.h (original_copy_tables_initialized_p): New decl.
28767 * cfgrtl.c (relink_block_chain): Guard the call to
28768 free_original_copy_tables with a call to
28769 original_copy_tables_initialized_p.
28770 * cgraph.h (symtab_node::native_rtl_p): New decl.
28771 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28772 (symtab_node::needed_p): Don't assert for early assembly output
28773 for __RTL functions.
28774 (cgraph_node::finalize_function): Set "force_output" for __RTL
28776 (cgraph_node::analyze): Bail out early for __RTL functions.
28777 (analyze_functions): Update assertion to support __RTL functions.
28778 (cgraph_node::expand): Bail out early for __RTL functions.
28779 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28781 * function.h (struct function): Update comment for field
28783 * gimple-expr.c: Include "tree-pass.h".
28784 (gimple_has_body_p): Return false for __RTL functions.
28785 * Makefile.in (OBJS): Add run-rtl-passes.o.
28786 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28788 (gcc::pass_manager::get_clean_slate): New accessor.
28789 * passes.c: Include "insn-addr.h".
28790 (should_skip_pass_p): Add logging. Update logic for running
28791 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28792 property-provider override so it is only done for gimple passes.
28794 (skip_pass): New function.
28795 (execute_one_pass): Call skip_pass when skipping passes.
28796 * read-md.c (md_reader::read_char): Support filtering
28797 the input to a subset of line numbers.
28798 (md_reader::md_reader): Initialize fields
28799 m_first_line and m_last_line.
28800 (md_reader::read_file_fragment): New function.
28801 * read-md.h (md_reader::read_file_fragment): New decl.
28802 (md_reader::m_first_line): New field.
28803 (md_reader::m_last_line): New field.
28804 * read-rtl-function.c (function_reader::create_function): Only
28805 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28806 curr_properties. Set DECL_INITIAL to a dummy block.
28807 (read_rtl_function_body_from_file_range): New function.
28808 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28810 * run-rtl-passes.c: New file.
28811 * run-rtl-passes.h: New file.
28813 2017-01-24 Jeff Law <law@redhat.com>
28815 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28818 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28820 PR tree-optimization/79159
28821 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28822 (record_nonwrapping_iv): Improve boundary using above function if no
28823 value range information.
28825 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28826 Martin Jambor <mjambor@suse.cz>
28828 * brig-builtins.def: New file.
28829 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28830 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28831 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28832 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28833 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28834 * builtin-types.def (BT_INT8): New.
28835 (BT_INT16): Likewise.
28836 (BT_UINT8): Likewise.
28837 (BT_UINT16): Likewise.
28838 (BT_FN_ULONG): Likewise.
28839 (BT_FN_UINT_INT): Likewise.
28840 (BT_FN_UINT_ULONG): Likewise.
28841 (BT_FN_UINT_LONG): Likewise.
28842 (BT_FN_UINT_PTR): Likewise.
28843 (BT_FN_ULONG_PTR): Likewise.
28844 (BT_FN_INT8_FLOAT): Likewise.
28845 (BT_FN_INT16_FLOAT): Likewise.
28846 (BT_FN_UINT32_FLOAT): Likewise.
28847 (BT_FN_UINT16_FLOAT): Likewise.
28848 (BT_FN_UINT8_FLOAT): Likewise.
28849 (BT_FN_UINT64_FLOAT): Likewise.
28850 (BT_FN_UINT16_UINT32): Likewise.
28851 (BT_FN_UINT32_UINT16): Likewise.
28852 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28853 (BT_FN_INT_PTR_INT): Likewise.
28854 (BT_FN_UINT_PTR_UINT): Likewise.
28855 (BT_FN_LONG_PTR_LONG): Likewise.
28856 (BT_FN_ULONG_PTR_ULONG): Likewise.
28857 (BT_FN_VOID_UINT64_UINT64): Likewise.
28858 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28859 (BT_FN_INT8_INT8_INT8): Likewise.
28860 (BT_FN_INT16_INT16_INT16): Likewise.
28861 (BT_FN_INT_INT_INT): Likewise.
28862 (BT_FN_UINT_FLOAT_UINT): Likewise.
28863 (BT_FN_FLOAT_UINT_UINT): Likewise.
28864 (BT_FN_ULONG_UINT_UINT): Likewise.
28865 (BT_FN_ULONG_UINT_PTR): Likewise.
28866 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28867 (BT_FN_UINT_UINT_UINT): Likewise.
28868 (BT_FN_VOID_UINT_PTR): Likewise.
28869 (BT_FN_UINT_UINT_PTR: Likewise.
28870 (BT_FN_UINT32_UINT64_PTR): Likewise.
28871 (BT_FN_INT_INT_UINT_UINT): Likewise.
28872 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28873 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28874 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28875 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28876 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28877 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28878 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28879 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28880 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28881 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28882 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28883 * doc/frontends.texi: List BRIG FE.
28884 * doc/install.texi (Testing): Add BRIG tesring requirements.
28885 * doc/invoke.texi (Overall Options): Mention BRIG.
28886 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28888 2017-01-24 Richard Biener <rguenther@suse.de>
28890 PR translation/79208
28891 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28893 2017-01-24 Martin Jambor <mjambor@suse.cz>
28896 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28897 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28898 and known_contexts.
28900 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28902 PR middle-end/79123
28903 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28904 casts from signed to unsigned really don't have a range.
28906 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28908 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28909 GMP_RNDx for compatiblity.
28911 2017-01-24 Martin Liska <mliska@suse.cz>
28914 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28915 that would prevent us to call alloca with -1 as argument.
28917 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28919 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28920 Avoid -Wformat-security warning.
28922 2017-01-23 Andrew Pinski <apinski@cavium.com>
28924 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28927 2017-01-23 Martin Sebor <msebor@redhat.com>
28929 PR middle-end/78703
28930 * gimple-ssa-sprintf.c (warn_level): New global.
28931 (format_integer): Use it here and throughout the rest of the file.
28932 Use the same switch to compute sign as base.
28933 (maybe_warn): New function.
28934 (format_directive): Factor out warnings into maybe_warn.
28935 Add debugging output. Use warn_level.
28936 (add_bytes): Use warn_level.
28937 (pass_sprintf_length::compute_format_length): Add debugging output.
28938 (try_substitute_return_value): Same.
28939 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28941 PR middle-end/78703
28942 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28943 (struct fmtresult, format_integer, format_floating): Adjust.
28944 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28945 (get_string_length, format_string,format_directive): Same.
28946 (pass_sprintf_length::compute_format_length): Same.
28947 (try_substitute_return_value): Simplify slightly.
28949 PR middle-end/78703
28950 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28951 (fmtresult::operator+=): Outlined.
28952 (struct fmtresult): Add ctors.
28953 (struct conversion_spec): Rename...
28954 (struct directive): ...to this. Add and remove data members.
28955 (directive::set_width, directive::set_precision): New functions.
28956 (format_percent): Use fmtresult ctor.
28957 (get_width_and_precision): Remove.
28958 (format_integer): Make naming changes. Avoid computing width and
28960 (format_floating): Same. Adjust indentation.
28961 (format_character, format_none): New functions.
28962 (format_string): Moved character handling to format_character.
28963 (format_directive): Remove arguments, change return type.
28964 (parse_directive): New function.
28965 (pass_sprintf_length::compute_format_length): Move directive
28966 parsing to parse_directive.
28968 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28970 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28971 (assign_assembler_name_if_needed): ... this.
28972 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28973 (assign_assembler_name_if_needed): ... this.
28974 (free_lang_data_in_cgraph): Adjust callers.
28975 * cgraphunit.c (cgraph_node::analyze): Likewise.
28976 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28979 2017-01-23 Richard Biener <rguenther@suse.de>
28981 PR tree-optimization/79088
28982 PR tree-optimization/79188
28983 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28984 resetting loop bounds after last path deletion. Reset loop
28985 bounds of the target loop, make code match the comments.
28986 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28987 Make sure loops need no fixups.
28989 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28991 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28992 exponent support with double type for first argument.
28993 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28994 type returned by __builtin_vec_extract_sig,
28995 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28996 functions from "vector int" to "vector unsigned int" or from
28997 "vector long long int" to "vector unsigned long long int".
28998 Changed type returned by __builtin_vec_extract_exp,
28999 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29000 functions from "vector int" to "vector unsigned int" or from
29001 "vector long long int" to "vector unsigned long long int".
29002 Changed return type of __builtin_vec_test_data_class,
29003 __builtin_vec_test_data_class_sp, and
29004 __builtin_vec_test_data_class_dp from "vector int" to
29005 "vector bool int" or from "vector long long int" to "vector bool
29006 long long int" and changed second argument type from "unsigned
29007 int" to "int". Added new overloaded function forms "vector float
29008 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29009 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29010 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29011 double, vector unsigned long long int)" and "vector double
29012 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29013 long int)". Changed return type of
29014 __builtin_scalar_test_data_class and
29015 __builtin_scalar_test_data_class_sp and
29016 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29017 int" and changed second argument from "unsigned int" to "int".
29018 Changed type returned by __builtin_scalar_test_neg,
29019 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29020 from "int" to "bool int". Added new overloaded function form
29021 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29022 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29023 exponent double-precision with floating point first argument.
29024 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29025 documentation of scalar_test_data_class, scalar_test_neg,
29026 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29027 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29028 vec_test_data_class built-in functions to reflect refinements in
29029 their type signatures.
29031 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29033 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29035 (aarch64_elf_asm_destructor): Likewise.
29037 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29039 PR rtl-optimization/78634
29040 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29041 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29042 * ifcvt.c (noce_try_cmove): Add missing cost check.
29044 PR rtl-optimization/71724
29045 * combine.c (if_then_else_cond): Look for situations where it is
29046 beneficial to undo the work of one of the recursive calls.
29048 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29050 PR tree-optimization/70754
29051 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29052 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29053 combined stmt before it if not NULL.
29054 (combine_chains): Process refs reversely and compute dominance point
29057 2017-01-23 Martin Liska <mliska@suse.cz>
29059 PR tree-optimization/79196
29060 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29061 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29063 (strlen_optimize_stmt): Call the renamed function.
29065 2017-01-23 Michael Matz <matz@suse.de>
29067 PR tree-optimization/78384
29068 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29070 2017-01-23 Richard Biener <rguenther@suse.de>
29072 PR tree-optimization/79186
29073 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29074 both incoming edges before moving an assert.
29076 2017-01-23 Martin Jambor <mjambor@suse.cz>
29078 * ipa-prop.c (load_from_param_1): Removed.
29079 (load_from_unmodified_param): Bits from load_from_param_1 put back
29081 (load_from_param): Removed.
29082 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29083 with stmt. Reverted back to use of load_from_unmodified_param.
29085 2017-01-23 Martin Jambor <mjambor@suse.cz>
29088 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29089 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29090 field a pointer to garbage collected vector, mark lattices and
29091 ipcp_orig_node with GTY((skip)).
29092 (ipa_get_param_count): Adjust to descriptors being a pointer.
29093 (ipa_get_param): Likewise.
29094 (ipa_get_type): Likewise.
29095 (ipa_get_param_move_cost): Likewise.
29096 (ipa_set_param_used): Likewise.
29097 (ipa_get_controlled_uses): Likewise.
29098 (ipa_set_controlled_uses): Likewise.
29099 (ipa_is_param_used): Likewise.
29100 (ipa_node_params_t): Move into garbage collector. New methods insert
29102 (ipa_node_params_sum): Annotate wth GTY(()).
29103 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29105 (ipa_load_from_parm_agg): Adjust declaration.
29106 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29107 * ipa-profile.c (ipa_profile): Likewise.
29108 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29109 (ipa_populate_param_decls): Make descriptors parameter garbage
29111 (ipa_dump_param): Adjust to descriptors being a pointer.
29112 (ipa_alloc_node_params): Likewise.
29113 (ipa_initialize_node_params): Likewise.
29114 (load_from_param_1): Make descriptors parameter garbage collected.
29115 (load_from_unmodified_param): Likewise.
29116 (load_from_param): Likewise.
29117 (ipa_load_from_parm_agg): Likewise.
29118 (ipa_node_params::~ipa_node_params): Removed.
29119 (ipa_free_all_node_params): Remove call to delete operator.
29120 (ipa_node_params_t::insert): New.
29121 (ipa_node_params_t::remove): Likewise.
29122 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29123 copy known_csts and known_contexts vectors.
29124 (ipa_read_node_info): Adjust to descriptors being a pointer.
29125 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29127 (ipcp_transform_function): Make descriptors variable garbage
29130 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29132 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29133 * config/i386/avx512dqintrin.h: Ditto.
29134 * config/i386/avx512fintrin.h: Ditto.
29135 * config/i386/i386.c: Handle new builtins.
29136 * config/i386/i386-builtin.def: Add new builtins.
29137 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29138 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29140 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29141 Martin Liska <mliska@suse.cz>
29143 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29144 * asan.c (asan_expand_poison_ifn): Support stores and use
29145 appropriate ASAN report function.
29146 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29147 * internal-fn.def (ASAN_POISON_USE): Declare.
29148 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29149 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29150 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29151 ASAN_POISON calls w/o LHS.
29152 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29153 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29154 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29155 * gimplify.c (asan_poison_variables): Add attribute
29156 use_after_scope_memory to variables that really needs to live
29158 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29159 having the attribute.
29161 2017-01-23 Martin Liska <mliska@suse.cz>
29163 * asan.c (create_asan_shadow_var): New function.
29164 (asan_expand_poison_ifn): Likewise.
29165 * asan.h (asan_expand_poison_ifn): New declaration.
29166 * internal-fn.c (expand_ASAN_POISON): Likewise.
29167 * internal-fn.def (ASAN_POISON): New builtin.
29168 * sanopt.c (pass_sanopt::execute): Expand
29169 asan_expand_poison_ifn.
29170 * tree-inline.c (copy_decl_for_dup_finish): Make function
29172 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29173 * tree-ssa.c (is_asan_mark_p): New function.
29174 (execute_update_addresses_taken): Rewrite local variables
29175 (identified just by use-after-scope as addressable) into SSA.
29177 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29179 * doc/install.texi (Specific): opensource.apple.com uses https
29180 now. Remove trailing slash.
29182 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29184 * README.Portability: Remove note on an Irix compatibility issue.
29186 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29188 * gcov.c (INCLUDE_ALGORITHM): Define.
29189 (INCLUDE_VECTOR): Define.
29190 No longer include <vector> and <algorithm> directly.
29192 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29194 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29196 * doc/invoke.texi (Code Gen Options): Ditto.
29198 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29201 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29203 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29205 rtl-optimization/79125
29206 * cprop.c (local_cprop_pass): Handle cases where we make an
29207 unconditional trap.
29209 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29213 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29214 read from, for big endian.
29216 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29218 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29219 register pauth builtins for LP64 only.
29221 2017-01-20 Marek Polacek <polacek@redhat.com>
29224 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29227 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29229 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29231 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29232 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29233 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29235 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29238 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29239 in big-endian mode when they are not a single duplicated value.
29241 2017-01-20 Richard Biener <rguenther@suse.de>
29243 * BASE-VER: Bump to 7.0.1.
29245 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29247 * omp-low.c (omplow_simd_context): New struct. Use it...
29248 (lower_rec_simd_input_clauses): ...here and...
29249 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29250 references to idx, lane, max_vf, is_simt.
29252 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29254 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29257 2017-01-20 Martin Jambor <mjambor@suse.cz>
29259 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29260 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29262 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29263 (GTFILES): Rename hsa.c to hsa-common.c.
29264 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29265 * hsa-dump.c: Likewise.
29266 * hsa-gen.c: Likewise.
29267 * hsa-regalloc.c: Likewise.
29268 * ipa-hsa.c: Likewise.
29269 * omp-expand.c: Likewise.
29270 * omp-low.c: Likewise.
29271 * toplev.c: Likewise.
29273 2017-01-20 Marek Polacek <polacek@redhat.com>
29276 * doc/invoke.texi: Document -Wduplicated-branches.
29277 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29278 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29279 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29280 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29281 return 0 only when not OEP_LEXICOGRAPHIC.
29282 (fold_build_cleanup_point_expr): Use the expression
29283 location when building CLEANUP_POINT_EXPR.
29284 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29285 * tree.c (add_expr): Handle error_mark_node.
29287 2017-01-20 Martin Liska <mliska@suse.cz>
29290 * tree-profile.c (init_ic_make_global_vars): Do not call
29292 (gimple_init_gcov_profiler): Likewise.
29294 2017-01-20 Martin Liska <mliska@suse.cz>
29297 * cgraph.h (maybe_create_reference): Remove argument and
29299 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29301 * ipa-cp.c (create_specialized_node): Likewise.
29302 * symtab.c (symtab_node::maybe_create_reference): Handle
29303 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29305 2017-01-20 Martin Liska <mliska@suse.cz>
29307 * read-rtl-function.c (function_reader::create_function): Use
29308 build_decl instread of build_decl_stat.
29310 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29312 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29313 * config/i386/avx512dqintrin.h: Ditto.
29314 * config/i386/avx512fintrin.h: Ditto.
29315 * config/i386/i386-builtin-types.def: Add new types.
29316 * config/i386/i386.c: Handle new types.
29317 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29318 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29319 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29320 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29321 (__builtin_ia32_kshiftridi): New.
29322 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29324 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29328 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29329 define to rs6000_init_stack_protect_guard.
29330 (rs6000_init_stack_protect_guard): New function.
29332 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29333 Yunqiang Su <yunqiang.su@imgtec.com>
29335 * config.gcc (supported_defaults): Add madd4.
29336 (with_madd4): Add validation.
29337 (all_defaults): Add madd4.
29338 * config/mips/mips.opt (mmadd4): New option.
29339 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29341 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29343 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29344 (ISA_HAS_FUSED_MADD4): Likewise.
29345 * doc/invoke.texi (-mmadd4): Document the new option.
29346 * doc/install.texi (--with-madd4): Document the new option.
29348 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29350 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29351 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29352 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29353 (aarch64_init_pauth_hint_builtins): New.
29354 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29355 (aarch64_expand_builtin): Expand new builtins.
29357 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29359 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29360 * combine-stack-adj.c (no_unhandled_cfa): Handle
29361 REG_CFA_TOGGLE_RA_MANGLE.
29362 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29363 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29364 info for return address signing.
29365 (aarch64_expand_epilogue): Likewise.
29367 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29369 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29370 * config/aarch64/aarch64-protos.h
29371 (aarch64_return_address_signing_enabled): New declaration.
29372 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29374 (aarch64_expand_prologue): Sign return address before it's pushed onto
29376 (aarch64_expand_epilogue): Authenticate return address fetched from
29378 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29379 (aarch64_attributes): New function attributes for "sign-return-address".
29380 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29381 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29382 ("*do_return"): Generate combined instructions according to key index.
29383 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29384 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29386 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29387 * config/aarch64/aarch64.opt (msign-return-address=): New.
29388 * doc/extend.texi (AArch64 Function Attributes): Documents
29389 "sign-return-address=".
29390 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29392 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29394 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29395 overall option summary.
29397 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29399 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29400 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29401 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29402 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29404 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29406 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29407 -mpower9-minmax by default for -mcpu=power9.
29408 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29409 128-bit floating point.
29411 2017-01-20 Alan Modra <amodra@gmail.com>
29413 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29414 optimizing for size.
29416 2017-01-20 Alan Modra <amodra@gmail.com>
29419 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29420 for strcmp and strncmp from corresponding builtin decl.
29422 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29424 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29425 instead of i386/rtems-64.h.
29426 * config/i386/rtems-64.h: Remove.
29428 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29432 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29434 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29436 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29438 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29439 Change int to HOST_WIDE_INT.
29440 * config/aarch64/aarch64-protos.h
29441 (aarch64_simd_gen_const_vector_dup): Likewise.
29442 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29444 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29446 * langhooks-def.h (lhd_type_for_size): New decl.
29447 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29448 * langhooks.c (lhd_type_for_size): New function, taken from
29451 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29453 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29454 define_bypass for CR latency.
29455 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29456 (power9-alu2): Add define_bypass for CR latency.
29458 (power9-mul): Update insn latency.
29459 (power9-mul-compare): Update insn latency, bypass latency and remove
29462 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29464 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29466 * config/aarch64/aarch64.md
29467 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29468 aarch64_nopcrelative_literal_loads.
29469 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29471 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29473 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29474 TARGET_LOONGSON_3A.
29475 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29477 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29480 * config.gcc (supported_defaults): Add lxc1-sxc1.
29481 (with_lxc1_sxc1): Add validation.
29482 (all_defaults): Add lxc1-sxc1.
29483 * config/mips/mips.opt (mlxc1-sxc1): New option.
29484 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29486 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29487 __mips_no_lxc1_sxc1.
29488 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29489 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29490 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29492 2017-01-19 Richard Biener <rguenther@suse.de>
29494 PR tree-optimization/72488
29495 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29496 sure to restore SSA info.
29497 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29499 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29501 PR rtl-optimization/79121
29502 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29503 of the inner type when shifting an extended value.
29505 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29508 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29509 interposable aliases.
29511 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29514 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29515 Use the evmergelohi instruction.
29516 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29517 (mov_sitf_e500_subreg8_2_be): Likewise.
29518 (mov_sitf_e500_subreg12_2_le): Likewise.
29519 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29520 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29521 (mov_sitf_e500_subreg8_2_le): Likewise.
29522 (mov_sitf_e500_subreg12_2_be): Likewise.
29524 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29526 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29527 attribute from vecsimple to vecperm.
29528 (altivec_vbpermq2): Likewise.
29530 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29533 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29535 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29536 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29537 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29538 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29539 case where N arg is SIZE_MAX.
29540 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29541 (cmpstrsi): Add pattern.
29543 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29545 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29546 __builtin_vec_revb builtins.
29547 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29548 built-in functions to support generation of the ISA 3.0 XXBR<x>
29549 vector byte reverse instructions.
29550 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29551 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29552 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29553 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29554 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29555 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29556 (P9V_BUILTIN_VEC_REVB): Likewise.
29557 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29558 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29559 (p9_xxbrq_v16qi): Likewise.
29560 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29561 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29562 (p9_xxbrh_v8hi): Likewise.
29563 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29564 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29565 vec_revb built-in functions.
29567 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29569 PR rtl-optimization/78952
29570 * config/i386/i386.md (any_extract): New code iterator.
29571 (*insvqi_2): Use any_extract for source operand.
29572 (*insvqi_3): Use any_shiftrt for source operand.
29574 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29576 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29578 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29580 2017-01-18 Matthias Klose <doko@ubuntu.com>
29582 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29584 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29586 * config/rs6000/altivec.h (vec_bperm): Change #define.
29587 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29588 (altivec_vbpermq2): New define_insn.
29589 (altivec_vbpermd): Likewise.
29590 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29591 function interface.
29592 (VBPERMD): Likewise.
29593 (VBPERM): New polymorphic function interface.
29594 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29595 Add entries for P9V_BUILTIN_VEC_VBPERM.
29596 * doc/extend.texi: Add interfaces for vec_bperm.
29598 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29600 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29601 first letter of error messages.
29602 (s390_resolve_overloaded_builtin): Likewise.
29603 * config/s390/s390.c (s390_expand_builtin): Likewise.
29604 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29605 (s390_valid_target_attribute_inner_p): Likewise.
29606 * config/s390/s390.md ("tabort"): Likewise.
29608 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29610 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29611 (ISA_AVOID_DIV_HILO): New macro.
29612 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29613 (ISA_HAS_DDIV): Likewise.
29615 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29617 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29619 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29621 * doc/invoke.texi (fabi-version): Spelling fix.
29623 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29626 * doc/invoke.texi (fabi-version): Mention mangling fix for
29629 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29632 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29634 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29637 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29638 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29640 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29642 (stack_protect_test): Ditto.
29643 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29644 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29646 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29647 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29648 -mstack-protector-guard-offset=.
29649 (RS/6000 and PowerPC Options): Ditto.
29651 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29653 * config/i386/i386.h (MASK_CLASS_P): New define.
29654 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29655 there are no registers from different register sets also when
29656 mask registers are used. Update function comment.
29657 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29658 to (*k/*r) and (*k/*km) alternatives.
29660 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29662 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29663 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29664 (EH_RETURN_HANDLER_RTX): New define.
29665 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29666 Force frame pointer in EH return functions.
29667 (aarch64_expand_epilogue): Add barrier for eh_return.
29668 (aarch64_final_eh_return_addr): Remove.
29669 (aarch64_eh_return_handler_rtx): New function.
29670 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29672 (aarch64_eh_return_handler_rtx): New prototype.
29674 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29676 * config/rs6000/altivec.h (vec_rlmi): New #define.
29677 (vec_vrlnm): Likewise.
29678 (vec_rlnm): Likewise.
29679 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29680 (UNSPEC_VRLNM): Likewise.
29681 (VIlong): New mode iterator.
29682 (altivec_vrl<VI_char>mi): New define_insn.
29683 (altivec_vrl<VI_char>nm): Likewise.
29684 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29686 (VRLDNM): Likewise.
29687 (RLNM): New polymorphic function entry.
29688 (VRLWMI): New monomorphic function entry.
29689 (VRLDMI): Likewise.
29690 (RLMI): New polymorphic function entry.
29691 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29692 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29693 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29696 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29699 * dwarf2out.c (field_byte_offset): Restore the
29700 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29701 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29702 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29705 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29708 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29710 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29713 * dwarf2out.c (add_data_member_location_attribute): For constant
29714 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29715 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29716 DW_AT_byte_size attributes.
29718 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29720 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29721 after forcing to constant memory when the code model is medium.
29723 2017-01-17 Julia Koval <julia.koval@intel.com>
29726 * config/i386/avx512fintrin.h
29727 (_mm512_i32gather_ps): Change __addr type to void const*.
29728 (_mm512_mask_i32gather_ps): Ditto.
29729 (_mm512_i32gather_pd): Ditto.
29730 (_mm512_mask_i32gather_pd): Ditto.
29731 (_mm512_i64gather_ps): Ditto.
29732 (_mm512_mask_i64gather_ps): Ditto.
29733 (_mm512_i64gather_pd): Ditto.
29734 (_mm512_mask_i64gather_pd): Ditto.
29735 (_mm512_i32gather_epi32): Ditto.
29736 (_mm512_mask_i32gather_epi32): Ditto.
29737 (_mm512_i32gather_epi64): Ditto.
29738 (_mm512_mask_i32gather_epi64): Ditto.
29739 (_mm512_i64gather_epi32): Ditto.
29740 (_mm512_mask_i64gather_epi32): Ditto.
29741 (_mm512_i64gather_epi64): Ditto.
29742 (_mm512_mask_i64gather_epi64): Ditto.
29743 (_mm512_i32scatter_ps): Change __addr type to void*.
29744 (_mm512_mask_i32scatter_ps): Ditto.
29745 (_mm512_i32scatter_pd): Ditto.
29746 (_mm512_mask_i32scatter_pd): Ditto.
29747 (_mm512_i64scatter_ps): Ditto.
29748 (_mm512_mask_i64scatter_ps): Ditto.
29749 (_mm512_i64scatter_pd): Ditto.
29750 (_mm512_mask_i64scatter_pd): Ditto.
29751 (_mm512_i32scatter_epi32): Ditto.
29752 (_mm512_mask_i32scatter_epi32): Ditto.
29753 (_mm512_i32scatter_epi64): Ditto.
29754 (_mm512_mask_i32scatter_epi64): Ditto.
29755 (_mm512_i64scatter_epi32): Ditto.
29756 (_mm512_mask_i64scatter_epi32): Ditto.
29757 (_mm512_i64scatter_epi64): Ditto.
29758 (_mm512_mask_i64scatter_epi64): Ditto.
29759 * config/i386/avx512pfintrin.h
29760 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29761 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29762 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29763 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29764 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29765 (_mm512_prefetch_i32scatter_ps): Ditto.
29766 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29767 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29768 (_mm512_prefetch_i64scatter_pd): Ditto.
29769 (_mm512_prefetch_i64scatter_ps): Ditto.
29770 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29771 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29772 * config/i386/avx512vlintrin.h
29773 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29774 (_mm_mmask_i32gather_ps): Ditto.
29775 (_mm256_mmask_i32gather_pd): Ditto.
29776 (_mm_mmask_i32gather_pd): Ditto.
29777 (_mm256_mmask_i64gather_ps): Ditto.
29778 (_mm_mmask_i64gather_ps): Ditto.
29779 (_mm256_mmask_i64gather_pd): Ditto.
29780 (_mm_mmask_i64gather_pd): Ditto.
29781 (_mm256_mmask_i32gather_epi32): Ditto.
29782 (_mm_mmask_i32gather_epi32): Ditto.
29783 (_mm256_mmask_i32gather_epi64): Ditto.
29784 (_mm_mmask_i32gather_epi64): Ditto.
29785 (_mm256_mmask_i64gather_epi32): Ditto.
29786 (_mm_mmask_i64gather_epi32): Ditto.
29787 (_mm256_mmask_i64gather_epi64): Ditto.
29788 (_mm_mmask_i64gather_epi64): Ditto.
29789 (_mm256_i32scatter_ps): Change __addr type to void*.
29790 (_mm256_mask_i32scatter_ps): Ditto.
29791 (_mm_i32scatter_ps): Ditto.
29792 (_mm_mask_i32scatter_ps): Ditto.
29793 (_mm256_i32scatter_pd): Ditto.
29794 (_mm256_mask_i32scatter_pd): Ditto.
29795 (_mm_i32scatter_pd): Ditto.
29796 (_mm_mask_i32scatter_pd): Ditto.
29797 (_mm256_i64scatter_ps): Ditto.
29798 (_mm256_mask_i64scatter_ps): Ditto.
29799 (_mm_i64scatter_ps): Ditto.
29800 (_mm_mask_i64scatter_ps): Ditto.
29801 (_mm256_i64scatter_pd): Ditto.
29802 (_mm256_mask_i64scatter_pd): Ditto.
29803 (_mm_i64scatter_pd): Ditto.
29804 (_mm_mask_i64scatter_pd): Ditto.
29805 (_mm256_i32scatter_epi32): Ditto.
29806 (_mm256_mask_i32scatter_epi32): Ditto.
29807 (_mm_i32scatter_epi32): Ditto.
29808 (_mm_mask_i32scatter_epi32): Ditto.
29809 (_mm256_i32scatter_epi64): Ditto.
29810 (_mm256_mask_i32scatter_epi64): Ditto.
29811 (_mm_i32scatter_epi64): Ditto.
29812 (_mm_mask_i32scatter_epi64): Ditto.
29813 (_mm256_i64scatter_epi32): Ditto.
29814 (_mm256_mask_i64scatter_epi32): Ditto.
29815 (_mm_i64scatter_epi32): Ditto.
29816 (_mm_mask_i64scatter_epi32): Ditto.
29817 (_mm256_i64scatter_epi64): Ditto.
29818 (_mm256_mask_i64scatter_epi64): Ditto.
29819 (_mm_i64scatter_epi64): Ditto.
29820 (_mm_mask_i64scatter_epi64): Ditto.
29821 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29822 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29823 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29824 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29825 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29826 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29827 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29828 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29829 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29830 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29831 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29832 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29833 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29834 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29835 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29836 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29837 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29838 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29839 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29840 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29841 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29842 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29843 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29844 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29845 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29846 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29847 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29848 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29849 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29850 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29851 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29852 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29853 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29854 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29855 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29856 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29857 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29858 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29859 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29860 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29861 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29862 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29863 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29864 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29865 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29866 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29867 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29868 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29869 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29870 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29871 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29872 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29873 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29874 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29875 definitions accordingly.
29877 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29878 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29881 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29884 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29887 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29888 hard regs for inner regno.
29890 2017-01-17 Martin Liska <mliska@suse.cz>
29893 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29894 assumption and add comment.
29896 2017-01-17 Nathan Sidwell <nathan@acm.org>
29898 * ipa-visibility.c (localize_node): New function, broken out of ...
29899 (function_and_variable_visibility): ... here. Call it.
29901 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29903 PR middle-end/77445
29904 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29905 correctly set frequency of oudgoing edge.
29906 (duplicate_thread_path): Fix profile updating.
29908 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29911 * configure.ac: Add GCC_BASE_VER.
29912 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29913 version from BASE-VER file.
29914 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29915 (gcc.o): Depend on $(BASEVER).
29916 * common.opt (dumpfullversion): New option.
29917 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29918 * doc/invoke.texi: Document -dumpfullversion.
29919 * doc/install.texi: Document --with-gcc-major-version-only.
29920 * configure: Regenerated.
29922 2017-01-17 Richard Biener <rguenther@suse.de>
29924 PR tree-optimization/71433
29925 * tree-vrp.c (register_new_assert_for): Merge same asserts
29926 on all incoming edges.
29927 (process_assert_insertions_for): Handle insertions at the
29930 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29932 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29933 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29935 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29938 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29941 2017-01-17 Alan Modra <amodra@gmail.com>
29944 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29945 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29946 symbolic stack limit when pic.
29948 2017-01-16 Martin Sebor <msebor@redhat.com>
29950 PR tree-optimization/78608
29951 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29953 2017-01-16 Jeff Law <law@redhat.com>
29956 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29957 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29958 for several include directories that may be relative to sysroot.
29959 * config/i386/x-mingw32 (gplus_includedir): Define.
29960 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29961 (native_system_includedir): Likewise.
29962 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29963 override if TARGET_SYSTEM_ROOT is defined.
29964 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29966 PR tree-optimization/79090
29967 PR tree-optimization/33562
29968 PR tree-optimization/61912
29969 PR tree-optimization/77485
29970 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29971 and computed trims into the dump file.
29973 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29975 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29977 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29980 * gimplify.c (gimplify_init_constructor): If want_value and
29981 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29985 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29986 sequence. Formatting fixes.
29987 (doloop_optimize): Formatting fixes.
29990 * gcc.c (debug_level_greater_than_spec_func): New function.
29991 (static_spec_functions): Add debug-level-gt spec function.
29992 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29994 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29995 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29996 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29997 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29998 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29999 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30001 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30003 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30004 QImode fixups to general and mask registers only.
30006 2017-01-16 Carl Love <cel@us.ibm.com>
30008 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30009 for built-in functions
30010 vector signed char vec_nabs (vector signed char)
30011 vector signed short vec_nabs (vector signed short)
30012 vector signed int vec_nabs (vector signed int)
30013 vector signed long long vec_nabs (vector signed long long)
30014 vector float vec_nabs (vector float)
30015 vector double vec_nabs (vector double)
30016 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30018 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30019 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30020 * doc/extend.texi: Update the documentation file for the new built-in
30023 2017-01-16 Martin Sebor <msebor@redhat.com>
30025 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30028 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30030 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30031 UNSPEC_VSX__XXSPLTD to require special splat handling.
30033 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30036 * system.h: Poison strndup.
30038 2017-01-16 Alan Modra <amodra@gmail.com>
30041 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30044 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30046 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30048 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30050 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30051 call recog here. Assert that INSN_CODE (insn) is non-negative.
30053 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30056 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30058 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30059 in the currently scheduled RTL fragment.
30061 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30063 PR rtl-optimization/78751
30064 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30067 2017-01-14 Jeff Law <law@redhat.com>
30069 PR tree-optimization/79090
30070 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30071 variable length stores.
30072 (compute_trims): Delete dead assignment to *trim_tail.
30073 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30076 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30078 PR rtl-optimization/78626
30079 PR rtl-optimization/78727
30080 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30081 of a block, and split such blocks after everything else is finished.
30083 2017-01-14 Alan Modra <amodra@gmail.com>
30086 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30087 target legitimate_combined_insn.
30088 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30089 (rs6000_legitimate_combined_insn): New function.
30090 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30092 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30093 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30094 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30096 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30098 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30100 2017-01-13 Jeff Law <law@redhat.com>
30102 PR tree-optimization/33562
30103 PR tree-optimization/61912
30104 PR tree-optimization/77485
30105 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30107 (delete_dead_assignment): Likewise.
30108 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30109 statement to delete_dead_call and delete_dead_assignment.
30111 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30114 * substring-locations.c (format_warning_va): Strengthen case 1 so
30115 that both endpoints of the substring must be within the format
30116 range for just the substring to be printed.
30118 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30120 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30121 * config/i386/i386.c (ix86_target_string): Add missing options
30122 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30123 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30124 flags_other and ix86_target_other to flags2_other. Display unknown
30126 (ix86_valid_target_attribute_inner_p): Add missing options and
30127 reorder options by implied ISAs, as in ix86_target_string.
30129 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30131 * hash-table.h (hash_table::too_empty_p): New function.
30132 (hash_table::expand): Use it.
30133 (hash_table::traverse): Likewise.
30134 (hash_table::empty_slot): Use sizeof (value_type) instead of
30135 sizeof (PTR) to convert bytes to elements. Shrink the table
30136 if the current size is excessive for the current number of
30139 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30141 * ira-costs.c (record_reg_classes): Break from the inner loop
30142 early once alt_fail is known to be true. Update outer loop
30143 handling accordingly.
30145 2017-01-13 Jeff Law <law@redhat.com>
30147 * tree-ssa-dse.c (decrement_count): New function.
30148 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30149 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30150 when we know the partially dead statement is a mem* function.
30152 PR tree-optimization/61912
30153 PR tree-optimization/77485
30154 * tree-ssa-dse.c: Include expr.h.
30155 (maybe_trim_constructor_store): New function.
30156 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30158 PR tree-optimization/33562
30159 PR tree-optimization/61912
30160 PR tree-optimization/77485
30161 * doc/invoke.texi: Document new dse-max-object-size param.
30162 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30163 * tree-ssa-dse.c: Include params.h.
30164 (dse_store_status): New enum.
30165 (initialize_ao_ref_for_dse): New, partially extracted from
30167 (valid_ao_ref_for_dse, normalize_ref): New.
30168 (setup_live_bytes_from_ref, compute_trims): Likewise.
30169 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30170 (maybe_trim_partially_dead_store): Likewise.
30171 (maybe_trim_complex_store): Likewise.
30172 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30173 Track what bytes live from the original store. Return tri-state
30174 for dead, partially dead or live.
30175 (dse_dom_walker): Add constructor, destructor and new private members.
30176 (delete_dead_call, delete_dead_assignment): New extracted from
30178 (dse_optimize_stmt): Make a member of dse_dom_walker.
30179 Use initialize_ao_ref_for_dse.
30181 PR tree-optimization/33562
30182 PR tree-optimization/61912
30183 PR tree-optimization/77485
30184 * sbitmap.h (bitmap_count_bits): Prototype.
30185 (bitmap_clear_range, bitmap_set_range): Likewise.
30186 * sbitmap.c (bitmap_clear_range): New function.
30187 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30189 2017-01-13 Martin Liska <mliska@suse.cz>
30192 * function.c (set_cfun): Add new argument force.
30193 * function.h (set_cfun): Likewise.
30194 * ipa-inline-transform.c (inline_call): Use the function when
30195 strict alising from is dropped for function we inline to.
30197 2017-01-13 Richard Biener <rguenther@suse.de>
30199 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30200 for dumping GIMPLE INTEGER_CSTs.
30202 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30204 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30205 to 201112L since C++17.
30207 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30210 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30211 if -fsanitize=kernel-address is present.
30213 2017-01-13 Richard Biener <rguenther@suse.de>
30215 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30216 as _Literal ( type ) number in case usual suffixes do not
30217 preserve all information.
30219 2017-01-13 Richard Biener <rguenther@suse.de>
30221 PR tree-optimization/77283
30222 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30223 and ssa-iterators.h.
30224 (is_feasible_trace): Implement a cost model based on joiner
30227 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30230 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30231 char or short to __float128/_Float128 directly.
30233 2017-01-12 Martin Sebor <msebor@redhat.com>
30235 to -Wformat-overflow.
30236 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30237 (min_bytes_remaining): Same.
30238 (get_string_length): Same.
30239 (format_string): Same.
30240 (format_directive): Same.
30242 (pass_sprintf_length::handle_gimple_call): Same.
30244 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30246 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30247 info.nowrite calls with no lhs that can't throw. Return bool
30248 whether gsi_remove has been called or not.
30249 (pass_sprintf_length::handle_gimple_call): Return bool whether
30250 try_substitute_return_value called gsi_remove. Formatting fix.
30251 (pass_sprintf_length::execute): Don't use gsi_remove if
30252 handle_gimple_call returned true.
30255 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30256 be removed due to side-effects, don't remove following barrier nor
30257 turn the successor edge into fallthru edge.
30259 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30262 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30263 element-reversing loads and stores as not swappable.
30265 2017-01-12 Nathan Sidwell <nathan@acm.org>
30266 Nicolai Stange <nicstange@gmail.com>
30268 * combine.c (try_combine): Don't ignore result of overlap checking
30269 loop. Combine overlap & asm check into single loop.
30271 2017-01-12 Richard Biener <rguenther@suse.de>
30273 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30274 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30276 2017-01-12 Richard Biener <rguenther@suse.de>
30278 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30279 and TS_TARGET_OPTION directly derive from TS_BASE.
30280 * tree-core.h (tree_optimization_option): Derive from tree_base.
30281 (tree_target_option): Likewise.
30283 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30285 * config/i386/i386.c (memory_address_length): Increase len
30286 only when rip_relative_addr_p returns false.
30288 2017-01-11 Julia Koval <julia.koval@intel.com>
30290 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30291 (OPTION_MASK_ISA_SGX_SET): New.
30292 (ix86_handle_option): Handle OPT_msgx.
30293 * config.gcc: Added sgxintrin.h.
30294 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30295 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30296 * config/i386/i386.c (ix86_target_string): Add -msgx.
30298 (ix86_option_override_internal): Handle new options.
30299 (ix86_valid_target_attribute_inner_p): Add sgx.
30300 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30301 * config/i386/i386.opt: Add msgx.
30302 * config/i386/sgxintrin.h: New file.
30303 * config/i386/x86intrin.h: Add sgxintrin.h.
30305 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30308 * fold-const.c (maybe_nonzero_address): Return 1 for function
30310 (tree_single_nonzero_warnv_p): Don't handle function local objects
30314 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30317 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30320 * opts.c: Include "spellcheck.h"
30321 (struct string_fragment): New struct.
30322 (struct edit_distance_traits<const string_fragment &>): New
30324 (get_closest_sanitizer_option): New function.
30325 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30327 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30329 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30331 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30332 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30333 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30334 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30335 for initial die_offset if dwarf_split_debug_info.
30336 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30337 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30339 (output_skeleton_debug_sections): Formatting fix. Use
30340 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30341 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30343 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30345 * config/arm/cortex-a53.md: Add bypasses for
30346 cortex_a53_r2f_cvt.
30347 (cortex_a53_r2f): Only use for transfers.
30348 (cortex_a53_f2r): Likewise.
30349 (cortex_a53_r2f_cvt): Add reservation for conversions.
30350 (cortex_a53_f2r_cvt): Likewise.
30352 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30354 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30355 to all inlined functions, change static to extern.
30357 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30360 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30362 (arm_assemble_integer): Likewise.
30364 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30366 * config.gcc: Use new awk script to check CPU, FPU and architecture
30367 parameters for --with-... options.
30368 * config/arm/parsecpu.awk: New file
30369 * config/arm/arm-cpus.in: New file.
30370 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30372 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30374 * config/arm/t-arm: Update dependency rules.
30375 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30376 of processing .def files.
30377 * config/arm/genopt.sh: Deleted.
30378 * config/arm/gentune.sh: Deleted.
30379 * config/arm/arm-cores.def: Deleted.
30380 * config/arm/arm-arches.def: Deleted.
30381 * config/arm/arm-fpus.def: Deleted.
30382 * config/arm/arm-tune.md: Regenerated.
30383 * config/arm/arm-tables.opt: Regenerated.
30384 * config/arm/arm-cpu.h: New generated file.
30385 * config/arm/arm-cpu-data.h: New generated file.
30386 * config/arm/arm-cpu-cdata.h: New generated file.
30388 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30391 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30393 (input_varpool_node): Unpack dynamically_initialized bit.
30395 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30397 PR rtl-optimization/79032
30398 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30399 the alignment of the adjusted memory reference against that of MODE,
30400 instead of the alignment of the original memory reference.
30402 2017-01-11 Martin Jambor <mjambor@suse.cz>
30404 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30406 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30407 decorated functions.
30409 2017-01-11 Richard Biener <rguenther@suse.de>
30411 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30412 set range/nonnull info for PHI results. Do not set it on
30413 stmts marked for removal.
30415 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30417 * expr.c (store_field): In the bitfield case, fetch the return value
30418 from the registers before applying a single big-endian adjustment.
30419 Always do a final load for a BLKmode value not larger than a word.
30421 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30424 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30425 that we correctly handle column numbers greater than
30426 LINE_MAP_MAX_COLUMN_NUMBER.
30428 2017-01-10 Martin Sebor <msebor@redhat.com>
30430 PR middle-end/78245
30431 * gimple-ssa-sprintf.c (get_destination_size): Call
30432 {init,fini}object_sizes.
30433 * tree-object-size.c (addr_object_size): Adjust.
30434 (pass_through_call): Adjust.
30435 (pass_object_sizes::execute): Adjust.
30436 * tree-object-size.h (fini_object_sizes): Declare.
30438 2017-01-10 Martin Sebor <msebor@redhat.com>
30440 PR tree-optimization/78775
30441 * builtins.c (get_size_range): Move...
30442 * calls.c: ...to here.
30443 (alloc_max_size): Accept zero argument.
30444 (operand_signed_p): Remove.
30445 (maybe_warn_alloc_args_overflow): Call get_size_range.
30446 * calls.h (get_size_range): Declare.
30448 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30450 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30451 from TI's devices.csv file as of September 2016.
30452 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30454 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30456 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30457 * doc/invoke.texi: Likewise.
30458 * doc/md.texi: Likewise.
30459 * doc/objc.texi: Likewise.
30461 2017-01-10 Joshua Conner <joshconner@google.com>
30463 * config/arm/fuchsia-elf.h: New file.
30464 * config/fuchsia.h: New file.
30465 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30466 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30468 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30470 2016-01-10 Richard Biener <rguenther@suse.de>
30472 PR tree-optimization/79034
30473 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30474 Propagate out degenerate PHIs in the joiner.
30476 2017-01-10 Martin Liska <mliska@suse.cz>
30478 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30479 (sort_congruence_classes_by_decl_uid): Likewise.
30480 (sort_congruence_class_groups_by_decl_uid): Likewise.
30481 (sem_item_optimizer::merge_classes): Sort class, groups in these
30482 classes and members in the groups by DECL_UID of declarations.
30483 This would make merge operations stable.
30485 2017-01-10 Martin Liska <mliska@suse.cz>
30487 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30488 usage of m_classes_vec.
30489 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30490 (sem_item_optimizer::get_group_by_hash): Likewise.
30491 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30492 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30493 (sem_item_optimizer::verify_classes): Likewise.
30494 (sem_item_optimizer::process_cong_reduction): Likewise.
30495 (sem_item_optimizer::dump_cong_classes): Likewise.
30496 (sem_item_optimizer::merge_classes): Likewise.
30497 * ipa-icf.h (congruence_class_hash): Rename from
30498 congruence_class_group_hash. Remove declaration of m_classes_vec.
30500 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30502 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30503 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30504 * config.gcc: Add avx512vpopcntdqintrin.h.
30505 * config/i386/avx512vpopcntdqintrin.h: New.
30506 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30507 * config/i386/i386-builtin-types.def: Add new types.
30508 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30509 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30510 __builtin_ia32_vpopcountq_v8di_mask): New.
30511 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30512 __AVX512VPOPCNTDQ__.
30513 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30514 (PTA_AVX512VPOPCNTDQ): Define.
30515 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30516 TARGET_AVX512VPOPCNTDQ_P): Define.
30517 * config/i386/i386.opt: Add mavx512vpopcntdq.
30518 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30519 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30521 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30523 PR middle-end/77484
30524 * predict.def (PRED_CALL): Set to 67.
30526 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30528 * expr.c (store_field): In the bitfield case, if the value comes from
30529 a function call and is of an aggregate type returned in registers, do
30530 not modify the field mode; extract the value in all cases if the mode
30531 is BLKmode and the size is not larger than a word.
30533 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30536 * config/i386/cpuid.h: Fix undefined behavior.
30538 2017-01-04 Jeff Law <law@redhat.com>
30540 PR tree-optimization/79007
30541 PR tree-optimization/67955
30542 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30543 conservative for pt.null when flag_non_call_exceptions is on.
30545 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30547 PR translation/79019
30548 PR translation/79020
30549 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30550 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30551 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30553 * config/avr/avr.opt (maccumulate-args): Likewise.
30554 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30555 * common.opt (freport-bug): Likewise.
30556 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30557 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30558 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30559 translatable string.
30560 * config/i386/i386.c (function_value_32): Likewise.
30561 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30562 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30564 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30565 * common/config/msp430/msp430-common.c (msp430_handle_option):
30567 * symtab.c (symtab_node::verify_base): Likewise.
30568 * opts.c (set_debug_level): Likewise.
30569 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30570 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30571 missing whitespace to translatable strings.
30572 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30573 * config/sh/superh.h: Likewise.
30574 * config/i386/xopintrin.h: Likewise.
30575 * config/i386/znver1.md: Likewise.
30576 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30577 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30578 * double-int.h (struct double_int): Likewise.
30579 * double-int.c (div_and_round_double): Likewise.
30580 * wide-int.cc: Likewise.
30581 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30582 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30583 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30584 (crossjumps_occurred): ... this.
30585 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30588 PR tree-optimization/78899
30589 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30590 returning bool return struct loop *, NULL for failure and the new
30592 (versionable_outer_loop_p): Don't version outer loop if it has
30593 dont_vectorized bit set.
30594 (tree_if_conversion): When versioning outer loop, ensure
30595 tree_if_conversion is performed also on the inner loop of the
30596 non-vectorizable outer loop copy.
30597 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30598 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30599 prevent vectorization of it.
30600 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30601 the outer loop vectorization of the non-scalar version is attempted
30602 before vectorization of the inner loop in scalar version. If
30603 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30604 vectorization of its inner loop.
30605 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30606 has 2 inner loops, rename also on edges from bb whose single pred
30607 is outer_loop->header. Fix typo in function comment.
30609 2017-01-09 Martin Sebor <msebor@redhat.com>
30612 * asan.c (asan_emit_stack_protection): Increase local buffer size
30613 to avoid snprintf truncation warning.
30615 2017-01-09 Andrew Pinski <apinski@cavium.com>
30617 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30618 to reference thunderx2t99 for the tuning structure
30619 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30621 (thunderx2t99_extra_costs): This.
30622 * config/aarch64/aarch64-tune.md: Regenerate.
30623 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30624 (vulcan_addrcost_table): This.
30625 (vulcan_regmove_cost): Rename to ...
30626 (thunderx2t99_regmove_cost): This.
30627 (vulcan_vector_cost): Rename to ...
30628 (thunderx2t99_vector_cost): this.
30629 (vulcan_branch_cost): Rename to ...
30630 (thunderx2t99_branch_cost): This.
30631 (vulcan_tunings): Rename to ...
30632 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30633 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30635 2017-01-09 Martin Jambor <mjambor@suse.cz>
30639 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30640 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30641 (propagate_vr_accross_jump_function): Use the above function for all
30642 value range computations for pass-through jump functions and type
30643 converasion from explicit value range values.
30644 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30645 parameters from TYPE_ARG_TYPES.
30646 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30647 (ipa_write_node_info): Stream type of the actual argument.
30648 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30650 2017-01-09 Martin Liska <mliska@suse.cz>
30653 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30654 (lookup_compiler): Do not show error message with have_E.
30656 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30658 PR tree-optimization/78938
30659 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30660 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30661 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30662 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30665 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30667 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30670 2017-01-09 Richard Biener <rguenther@suse.de>
30672 PR tree-optimization/78997
30673 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30674 name condition properly.
30676 2017-01-09 Richard Biener <rguenther@suse.de>
30679 * dwarf2out.c (is_cxx): New overload with context.
30680 (is_naming_typedef_decl): Use it.
30682 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30684 * invoke.texi (Option Summary): Correct spacing in option lists
30685 and add line breaks to fix over-long lines.
30687 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30689 PR middle-end/17660
30691 * extend.texi (Common Variable Attributes): Add xref to GCC
30692 Internals manual to explain mode attribute keywords.
30694 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30697 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30698 and Preprocessor Options.
30699 (Options for Linking): Document -pthread here....
30700 (RS/6000 and PowerPC Options): ...not here.
30701 (Solaris 2 Options): ...or here.
30702 * doc/cppopts.texi: Document -pthread.
30704 2017-01-08 Martin Sebor <msebor@redhat.com>
30706 PR middle-end/77708
30707 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30708 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30709 New member functions.
30710 (format_directive): Used them.
30712 (pass_sprintf_length::handle_gimple_call): Same.
30713 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30714 to avoid truncation for any argument.
30715 (extract_affine_mul): Same.
30716 * tree.c (get_file_function_name): Same.
30718 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30720 PR middle-end/77484
30721 * predict.def (PRED_INDIR_CALL): Set to 86.
30723 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30725 PR preprocessor/54124
30726 * doc/cppopts.texi: Reformat -d subtable to list the full name
30727 of the options. Add cross-reference to the docs for the general
30728 compiler -d options.
30729 * doc/invoke.texi (Developer Options): Add cross-reference to the
30730 preprocessor-specific -d option documentation.
30732 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30734 PR preprocessor/13498
30735 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30736 redudant material, and reflect new command-line options.
30737 (System Headers): Likewise.
30739 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30741 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30742 -isystem, and -idirafter. Copy-edit.
30743 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30744 default for -ftrack-macro-expansion. Delete obsolete and
30745 badly-formatted implementation details about -fdebug-cpp output.
30746 * doc/cppwarnopts.texi: Copy-edit.
30748 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30751 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30752 that the transition from a max line width >= 1<<10 to narrower
30753 lines works correctly.
30755 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30757 * doc/options.texi (PerFunction): New.
30758 * opt-functions.awk (switch_flags): Map both Optimization and
30759 PerFunction to CL_OPTIMIZATION.
30760 * opth-gen.awk: Test for PerFunction flag along with
30762 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30763 it only when the latter is present. Skip those that don't in
30764 the hash function generator.
30765 * common.opt (fvar-tracking): Mark as PerFunction instead of
30767 (fvar-tracking-assignments): Likewise.
30768 (fvar-tracking-assignments-toggle): Likewise.
30769 (fvar-tracking-uninit): Likewise.
30771 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30773 PR translation/79018
30774 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30777 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30780 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30781 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30782 TARGET_LONG_JUMP_TABLE_OFFSETS.
30783 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30784 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30785 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30786 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30787 * config/m68k/m68k.md (tablejump expander): Likewise.
30788 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30789 TARGET_LONG_JUMP_TABLE_OFFSETS.
30790 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30791 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30793 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30794 David Holsgrove <david.holsgrove@xilinx.com>
30796 * common/config/microblaze/microblaze-common.c
30797 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30798 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30800 * config/microblaze/microblaze.c (microblaze_must_save_register)
30801 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30803 (microblaze_eh_return): New function.
30804 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30805 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30806 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30807 * config/microblaze/microblaze.md (eh_return): New pattern.
30809 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30811 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30812 GCC_DIAGNOSTIC_STRINGIFY): Define.
30814 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30816 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30818 * config/arm/arm.md (<mcrr>): New.
30820 * config/arm/arm.c (arm_arch5te): New.
30821 (arm_option_override): Set arm_arch5te.
30822 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30824 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30825 (arm_mcrr_qualifiers): ... this. New.
30826 (MRRC_QUALIFIERS): Define to...
30827 (arm_mrrc_qualifiers): ... this. New.
30828 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30830 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30831 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30832 (MRRCI, mrrc, MRRC): New.
30833 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30834 VUNSPEC_MRRC2): New.
30836 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30838 * config/arm/arm.md (<mcr>): New.
30840 * config/arm/arm.c (arm_coproc_builtin_available): Add
30841 support for mcr, mrc, mcr2 and mrc2.
30842 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30843 (arm_mcr_qualifiers): ... this. New.
30844 (MRC_QUALIFIERS): Define to ...
30845 (arm_mrc_qualifiers): ... this. New.
30846 (MCR_QUALIFIERS): Define to ...
30847 (arm_mcr_qualifiers): ... this. New.
30848 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30850 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30851 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30852 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30853 VUNSPEC_MRC2): New.
30855 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30857 * config/arm/arm.md (*ldc): New.
30861 * config/arm/arm.c (arm_coproc_builtin_available): Add
30862 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30863 (arm_coproc_ldc_stc_legitimate_address): New.
30864 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30865 'qualifier_const_pointer'.
30866 (LDC_QUALIFIERS): Define to...
30867 (arm_ldc_qualifiers): ... this. New.
30868 (STC_QUALIFIERS): Define to...
30869 (arm_stc_qualifiers): ... this. New.
30870 * config/arm/arm-protos.h
30871 (arm_coproc_ldc_stc_legitimate_address): New.
30872 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30873 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30874 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30875 stc2, stcl, stc2l): New.
30876 * config/arm/constraints.md (Uz): New.
30877 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30878 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30879 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30880 VUNSPEC_STC2L): New.
30882 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30884 * config/arm/arm.md (<cdp>): New.
30885 * config/arm/arm.c (neon_const_bounds): Rename this ...
30886 (arm_const_bounds): ... this.
30887 (arm_coproc_builtin_available): New.
30888 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30889 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30890 (CDP_QUALIFIERS): Define to...
30891 (arm_cdp_qualifiers): ... this. New.
30893 (arm_expand_builtin_args): Add case for 6 arguments.
30894 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30895 (arm_const_bounds): ... this.
30896 (arm_coproc_builtin_available): New.
30897 * config/arm/arm_acle.h (__arm_cdp): New.
30899 * config/arm/arm_acle_builtins.def (cdp): New.
30901 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30902 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30903 'arm_const_bounds'.
30904 * config/arm/types.md (coproc): New.
30905 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30906 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30907 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30908 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30910 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30912 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30913 (UBINOP_QUALIFIERS): New.
30915 (acle_builtin_data): New. Change comment.
30916 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30917 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30918 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30919 arm_acle_builtins.def.
30920 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30921 (arm_init_acle_builtins): New.
30922 (CRC32_BUILTIN): Remove.
30923 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30924 crc32cb, crc32ch and crc32cw.
30925 (arm_init_crc32_builtins): Remove.
30926 (arm_init_builtins): Use arm_init_acle_builtins rather
30927 than arm_init_crc32_builtins.
30928 (arm_expand_acle_builtin): New.
30929 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30930 * config/arm/arm_acle_builtins.def: New.
30932 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30934 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30935 (arm_builtin_datum): ... this.
30936 (arm_init_neon_builtin): Rename to ...
30937 (arm_init_builtin): ... this. Add a new parameters PREFIX
30938 and USE_SIG_IN_NAME.
30939 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30940 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30941 'arm_builtin_datum'.
30942 (arm_init_vfp_builtins): Likewise.
30943 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30944 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30945 (arm_expand_neon_args): Rename to ...
30946 (arm_expand_builtin_args): ... this. Rename builtin_arg
30947 enum values and differentiate between ARG_BUILTIN_MEMORY
30948 and ARG_BUILTIN_NEON_MEMORY.
30949 (arm_expand_neon_builtin_1): Rename to ...
30950 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30951 values, arm_expand_builtin_args and add bool parameter NEON.
30952 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30953 (arm_expand_vfp_builtin): Likewise.
30954 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30956 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30958 PR middle-end/77484
30959 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30960 * predict.c (tree_estimate_probability_bb): Reverse direction of
30961 polymorphic call predictor.
30963 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30965 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30966 (determine_pass_name_match): ...this new function and...
30967 (should_skip_pass_p): ...this new function.
30969 2017-01-06 Nathan Sidwell <nathan@acm.org>
30971 * ipa-visibility.c (function_and_variable_visibility): Reformat
30972 comments and long lines. Remove extrneous if.
30973 * symtab.c (symtab_node::make_decl_local): Fix code format.
30974 (symtab_node::set_section_for_node): Fix comment typo.
30976 2017-01-06 Martin Liska <mliska@suse.cz>
30979 * lra-constraints.c: Rename invariant to lra_invariant.
30980 * predict.c (set_even_probabilities): Initialize e to NULL.
30982 2017-01-05 Martin Sebor <msebor@redhat.com>
30984 PR tree-optimization/78910
30985 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30986 (format_integer): Correct off-by-one error in the handling
30987 of precision with negative numbers in signed conversions..
30989 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30991 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30993 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30995 PR tree-optimization/71016
30996 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30997 factor_out_conditional_conversion. Formatting fix.
30998 (factor_out_conditional_conversion): Add cond_stmt argument.
30999 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31000 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31003 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31005 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31006 read-rtl-function.o, and selftest-rtl.o.
31007 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31008 (selftest::aarch64_test_loading_full_dump): New function.
31009 (selftest::aarch64_run_selftests): New function.
31010 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31011 selftest::aarch64_run_selftests.
31012 * config/i386/i386.c
31013 (selftest::ix86_test_loading_dump_fragment_1): New function.
31014 (selftest::ix86_test_loading_call_insn): New function.
31015 (selftest::ix86_test_loading_full_dump): New function.
31016 (selftest::ix86_test_loading_unspec): New function.
31017 (selftest::ix86_run_selftests): Call the new functions.
31018 * emit-rtl.c (maybe_set_max_label_num): New function.
31019 * emit-rtl.h (maybe_set_max_label_num): New decl.
31020 * function.c (instantiate_decls): Guard call to
31021 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31022 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31024 * gensupport.c (gen_reader::gen_reader): Pass "false"
31025 for new "compact" param of rtx_reader.
31026 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31027 rather than an empty string for NULL strings.
31028 * read-md.c: Potentially include config.h rather than bconfig.h.
31029 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31030 (have_error): New global, copied from errors.c.
31031 (md_reader::read_name): Rename to...
31032 (md_reader::read_name_1): ...this, adding "out_loc" param,
31033 and converting "missing name or number" to returning false, rather
31035 (md_reader::read_name): Reimplement in terms of read_name_1.
31036 (md_reader::read_name_or_nil): New function.
31037 (md_reader::read_string): Handle "(nil)" by returning NULL.
31038 (md_reader::md_reader): Add new param "compact".
31039 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31040 (md_reader::read_file): New method.
31041 * read-md.h (md_reader::md_reader): Add new param "compact".
31042 (md_reader::read_file): New method.
31043 (md_reader::is_compact): New accessor.
31044 (md_reader::read_name): Convert return type from void to file_location.
31045 (md_reader::read_name_or_nil): New decl.
31046 (md_reader::read_name_1): New decl.
31047 (md_reader::m_compact): New field.
31048 (noop_reader::noop_reader): Pass "false" for new "compact" param
31050 (rtx_reader::rtx_reader): Add new "compact" param.
31051 (rtx_reader::read_rtx_operand): Make virtual and convert return
31052 type from void to rtx.
31053 (rtx_reader::read_until): New decl.
31054 (rtx_reader::handle_any_trailing_information): New virtual function.
31055 (rtx_reader::postprocess): New virtual function.
31056 (rtx_reader::finalize_string): New virtual function.
31057 (rtx_reader::m_in_call_function_usage): New field.
31058 (rtx_reader::m_reuse_rtx_by_id): New field.
31059 * read-rtl-function.c: New file.
31060 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31061 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31062 (selftest::verify_three_block_rtl_cfg): New decl.
31063 * read-rtl-function.h: New file.
31064 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31065 For host, include function.h, memmodel.h, and emit-rtl.h.
31066 (one_time_initialization): New function.
31067 (struct compact_insn_name): New struct.
31068 (compact_insn_names): New array.
31069 (find_code): Handle insn codes in compact dumps.
31070 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31071 (bind_subst_iter_and_attr): Likewise.
31072 (add_condition_to_string): Likewise.
31073 (add_condition_to_rtx): Likewise.
31074 (apply_attribute_uses): Likewise.
31075 (add_current_iterators): Likewise.
31076 (apply_iterators): Likewise.
31077 (initialize_iterators): Guard usage of apply_subst_iterator with
31078 #ifdef GENERATOR_FILE.
31079 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31080 (md_reader::read_mapping): Likewise.
31081 (add_define_attr_for_define_subst): Likewise.
31082 (add_define_subst_attr): Likewise.
31083 (read_subst_mapping): Likewise.
31084 (check_code_iterator): Likewise.
31085 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31087 (one_time_initialization): New function.
31088 (rtx_reader::read_until): New method.
31089 (read_flags): New function.
31090 (parse_reg_note_name): New function.
31091 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31092 Handle reuse_rtx ids.
31093 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31094 Add parsing support for RTL dumps, mirroring the special-cases in
31095 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31096 values, and calling handle_any_trailing_information.
31097 (rtx_reader::read_rtx_operand): Convert return type from void
31098 to rtx, returning return_rtx. Handle case 'e'. Call
31099 finalize_string on XSTR and XTMPL fields.
31100 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31101 "(nil)" values were omitted. Call the postprocess vfunc on the
31103 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31104 class ctor. Initialize m_in_call_function_usage. Call
31105 one_time_initialization.
31106 * rtl-tests.c (selftest::test_uncond_jump): Call
31107 set_new_first_and_last_insn.
31108 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31109 * selftest-rtl.c: New file.
31110 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31111 (selftest::get_insn_by_uid): New decl.
31112 * selftest-run-tests.c (selftest::run_tests): Call
31113 read_rtl_function_c_tests.
31114 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31115 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31118 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31120 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31121 operands in a special way. Assert that pos+len <= mode precision.
31123 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31125 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31126 3 argument Alias with unlimited for the negative form.
31127 (fno-vect-cost-model): Removed.
31129 2017-01-05 Martin Liska <mliska@suse.cz>
31131 * hsa-gen.c (gen_hsa_divmod): New function.
31132 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31134 2017-01-05 Martin Liska <mliska@suse.cz>
31137 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31140 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31142 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31143 small constant length operands.
31145 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31147 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31148 between loop iterations.
31150 2017-01-05 Martin Liska <mliska@suse.cz>
31153 * gimplify.c (gimplify_decl_expr): Compare to
31154 asan_poisoned_variables instread of checking flags.
31155 (gimplify_target_expr): Likewise.
31156 (gimplify_expr): Likewise.
31157 (gimplify_function_tree): Conditionally initialize
31158 asan_poisoned_variables.
31160 2017-01-04 Jeff Law <law@redhat.com>
31162 PR tree-optimizatin/78812
31163 * rtl.h (contains_mem_rtx_p): Prototype.
31164 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31165 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31166 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31167 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31168 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31170 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31172 * input.c (assert_char_at_range): Default-initialize actual_range.
31174 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31176 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31177 to match the caller.
31179 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31181 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31182 insns after final jump in test to emit dummy move.
31184 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31186 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31187 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31189 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31191 * multiple_target.c (create_dispatcher_calls): Init e_next.
31192 * tree-ssa-loop-split.c (split_loop): Init border.
31193 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31196 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31201 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31202 (altivec_register_operand): Do not return true if the operand
31203 contains a SUBREG mixing SImode and SFmode.
31204 (vsx_register_operand): Likewise.
31205 (vsx_reg_sfsubreg_ok): New predicate.
31206 (vfloat_operand): Do not return true if the operand contains a
31207 SUBREG mixing SImode and SFmode.
31208 (vint_operand): Likewise.
31209 (vlogical_operand): Likewise.
31210 (gpc_reg_operand): Likewise.
31211 (int_reg_operand): Likewise.
31212 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31213 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31214 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31216 (rs6000_emit_move_si_sf_subreg): New helper function.
31217 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31218 fixup SUBREGs involving SImode and SFmode.
31219 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31220 numbers for the new peephole2 optimization.
31221 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31222 the GLIBC math library that do AND/IOR/XOR operations on single
31223 precision floating point.
31224 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31225 target macros to say whether we need to avoid SUBREGs mixing
31227 (TARGET_ALLOW_SF_SUBREG): Likewise.
31228 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31229 (UNSPEC_SI_FROM_SF): Likewise.
31230 (iorxor): Change spacing.
31231 (and_ior_xor): New iterator for AND, IOR, and XOR.
31232 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31233 (movdi_from_sf_zero_ext): Likewise.
31234 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31235 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31236 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31237 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31238 (fms<mode>4): Likewise.
31239 (fnma<mode>4): Likewise.
31240 (fnms<mode>4): Likewise.
31241 (nfma<mode>4): Likewise.
31242 (nfms<mode>4): Likewise.
31244 2017-01-04 Marek Polacek <polacek@redhat.com>
31247 * doc/invoke.texi: Document -Wpointer-compare.
31249 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31251 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31254 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31255 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31258 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31261 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31262 documentation of the powerpc_popcntb_ok attribute.
31263 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31264 code to issue warning messages if a requested CPU configuration is
31265 not supported by the binary (assembler and loader) toolchain.
31266 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31267 made to define a built-in function that has been disabled.
31268 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31269 made to define a built-in function that has been disabled.
31270 (altivec_init_builtins): Add comment explaining why definition
31271 of the DST built-in functions is not preceded by an assertion
31272 check. Add assertions to prevent ICE if attempts are made to
31273 define an altivec predicate or an abs* built-in function that has
31275 (htm_init_builtins): Add comment explaining why definition of the
31276 htm built-in functions is not preceded by an assertion check.
31278 2017-01-04 Jeff Law <law@redhat.com>
31280 PR tree-optimizatin/67955
31281 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31282 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31283 the points-to solution does not include pt_null. Use DECL_PT_UID
31286 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31288 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31289 Use gen_int_mode instead of gen_lopwart for const_int operands.
31291 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31293 PR tree-optimization/71563
31294 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31295 out of range value - has low bits known to be zero.
31297 2017-01-04 Alan Modra <amodra@gmail.com>
31299 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31300 * configure: Regenerate.
31301 * config.in: Regenerate.
31303 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31306 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31307 a substring of the message, but strcmp with the whole message. Ifdef
31308 ENABLE_NLS, translate the message first using dgettext.
31310 2017-01-03 Jeff Law <law@redhat.com>
31312 PR tree-optimizatin/78856
31313 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31314 (mark_threaded_blocks): Remove code to truncate thread paths that
31315 cross multiple loop headers. Instead invalidate the cached loop
31316 iteration information and handle case of a thread path walking
31317 into an irreducible region.
31319 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31322 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31323 assertions. Add support for doing the signbit if the IEEE 128-bit
31324 floating point value is in a GPR.
31325 * config/rs6000/rs6000.md (Fsignbit): Delete.
31326 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31327 Update the length attribute if the value is in a GPR.
31328 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31329 the sign or zero extension instruction, since the value is always 0/1.
31330 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31333 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31334 extracting SImode to a GPR register so that we can generate a
31335 store, limit the vector to be in a traditional Altivec register
31336 for the vextuwrx instruction.
31338 2017-01-03 Ian Lance Taylor <iant@google.com>
31340 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31342 2017-01-03 Martin Sebor <msebor@redhat.com>
31344 PR tree-optimization/78696
31345 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31346 precision. Use MPFR for %f for greater fidelity. Correct handling
31348 (pass_sprintf_length::compute_format_length): Set width and precision
31349 specified by asrerisk to void_node for vararg functions.
31350 (try_substitute_return_value): Adjust dump output.
31352 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31354 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31356 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31358 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31359 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31360 -mlra/-mno-lra was passed to the compiler.
31362 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31364 PR rtl-optimization/65618
31365 * emit-rtl.c (try_split): Move initialization of "before" and
31366 "after" to just before the call to emit_insn_after_setloc.
31368 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31370 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31372 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31374 * dwarf2out.c (gen_enumeration_type_die): When
31375 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31377 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31379 PR tree-optimization/78965
31380 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31381 Change first argument from const call_info & to call_info &. For %n
31382 set info.nowrite to false.
31384 PR middle-end/78901
31385 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31386 possibly throwing calls.
31388 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31389 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31390 and fns handling, rather than in a separate case SSA_NAME.
31392 2017-01-02 Jeff Law <law@redhat.com>
31394 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31395 fixes for first_period and second_period variables.
31397 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31400 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31401 (*insvqi_1): New insn pattern.
31402 (*insvqi_1_mem_rex64): Ditto.
31403 (*insvqi_2): Ditto.
31404 (*insvqi_3): Rename from *insvqi.
31406 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31408 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31410 * doc/cfg.texi (Edges): Remove reference to Java.
31411 (Maintaining the CFG): Ditto.
31413 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31415 PR middle-end/77674
31416 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31417 transparent aliases.
31419 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31421 PR middle-end/77484
31422 * predict.def (PRED_CALL): Update hitrate.
31423 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31424 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31425 into direct/indirect/polymorphic variants.
31427 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31429 Update copyright years.
31431 * gcc.c (process_command): Update copyright notice dates.
31432 * gcov-dump.c (print_version): Ditto.
31433 * gcov.c (print_version): Ditto.
31434 * gcov-tool.c (print_version): Ditto.
31435 * gengtype.c (create_file): Ditto.
31436 * doc/cpp.texi: Bump @copying's copyright year.
31437 * doc/cppinternals.texi: Ditto.
31438 * doc/gcc.texi: Ditto.
31439 * doc/gccint.texi: Ditto.
31440 * doc/gcov.texi: Ditto.
31441 * doc/install.texi: Ditto.
31442 * doc/invoke.texi: Ditto.
31444 Copyright (C) 2017 Free Software Foundation, Inc.
31446 Copying and distribution of this file, with or without modification,
31447 are permitted in any medium without royalty provided the copyright
31448 notice and this notice are preserved.