1 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4 Improve unaligned TImode/TFmode base/offset split.
6 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7 Alan Hayward <alan.hayward@arm.com>
8 David Sherwood <david.sherwood@arm.com>
10 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
11 * combine.c (find_single_use_1): Likewise.
12 (expand_field_assignment): Likewise.
13 (move_deaths): Likewise.
14 * lra-constraints.c (simplify_operand_subreg): Likewise.
15 (curr_insn_transform): Likewise.
16 * lra.c (collect_non_operand_hard_regs): Likewise.
17 (add_regs_to_insn_regno_info): Likewise.
18 * rtlanal.c (reg_referenced_p): Likewise.
19 (covers_regno_no_parallel_p): Likewise.
21 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
23 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
24 Don't print any bits outside the precision of the value.
25 * wide-int.cc (test_printing): Add some new tests.
27 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
30 supports -xbrace_comment option.
31 * configure: Regenerate.
32 * config.in: Regenerate.
33 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
34 (ASM_CPU_SPEC): Use it.
36 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
38 * target.def (static_rtx_alignment): New hook.
39 * targhooks.h (default_static_rtx_alignment): Declare.
40 * targhooks.c (default_static_rtx_alignment): New function.
41 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
42 * doc/tm.texi: Regenerate.
43 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
44 instead of targetm.constant_alignment. Remove call to
46 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
47 (cris_preferred_mininum_alignment): New function, split out from...
48 (cris_constant_alignment): ...here.
49 (cris_static_rtx_alignment): New function.
50 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
52 (ix86_constant_alignment): ...here.
53 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
54 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
55 (mmix_static_rtx_alignment): New function.
56 * config/spu/spu.c (spu_static_rtx_alignment): New function.
57 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
59 2017-10-26 Tamar Christina <tamar.christina@arm.com>
62 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): Add flag_trapping_math
63 and flag_fp_int_builtin_inexact.
65 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
67 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
68 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
69 mark as a sign-extending load.
70 (local_pic_load_u): Define.
72 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
75 * fold-const.c (operand_equal_for_comparison_p): Also return true
76 if ARG0 is a simple variant of ARG1 with narrower precision.
77 (fold_ternary_loc): Always pass unstripped operands to the predicate.
79 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
81 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
83 * i386.h (processor_costs): Add gather_static, gather_per_elt,
84 scatter_static, scatter_per_elt.
85 * x86-tune-costs.h: Add new cost entries.
87 2017-10-25 Richard Biener <rguenther@suse.de>
89 * tree-ssa-sccvn.h (vn_eliminate): Declare.
90 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
91 class pass_fre): Move to ...
92 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
93 class pass_fre): ... here and adjust for statistics.
95 2017-10-25 Jakub Jelinek <jakub@redhat.com>
98 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
99 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
100 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
101 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
104 2017-10-25 Richard Biener <rguenther@suse.de>
106 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
107 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
108 eliminate_push_avail, eliminate_insert): Move inside...
109 (class eliminate_dom_walker): ... this class in preparation
111 (fini_eliminate): Remove by merging with ...
112 (eliminate): ... this function. Adjust for class changes.
113 (pass_pre::execute): Remove fini_eliminate call.
114 (pass_fre::execute): Likewise.
116 2017-10-24 Jakub Jelinek <jakub@redhat.com>
119 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
120 (VPERMI2, VPERMI2I): New mode iterators.
121 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
122 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
124 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
125 mode iterator. Remove 3 old define_insn patterns.
126 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
127 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
128 VPERMI2 mode iterator, remove the other two expanders.
129 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
130 to use VPERMI2 mode iterator, add another alternative for vpermi2*
131 instructions, remove the other two patterns.
132 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
133 mode iterator, remove the other two patterns.
134 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
135 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
136 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
137 and adjust argument order accordingly.
138 (ix86_expand_vec_perm): Adjust caller.
139 (expand_vec_perm_1): Likewise.
140 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
141 (expand_vec_perm_vpermt2_vpshub2): ... this.
142 (ix86_expand_vec_perm_const_1): Adjust caller.
143 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
146 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
147 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
148 (vec_shl_<mode>): Remove unused expander.
149 (avx512bw_<shift_insn><mode>3): New define_insn.
150 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
151 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
153 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
156 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
159 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
161 PR rtl-optimization/82396
162 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
163 (autopref_multipass_init): Simplify initialization.
164 (autopref_rank_data): Simplify sort order.
165 * gcc/sched-int.h (autopref_multipass_data_): Remove
166 multi_mem_insn_p, min_offset and max_offset.
168 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
171 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
172 Check special value of flag_omit_frame_pointer.
173 (aarch64_can_eliminate): Likewise.
174 (aarch64_override_options_after_change_1): Simplify handling of
175 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
177 2017-10-24 Richard Biener <rguenther@suse.de>
179 PR tree-optimization/82697
180 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
181 zero for conditional load and unconditional store.
183 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
185 * doc/install.texi: Document bootstrap-cet.
187 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
190 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
191 ENDBR instruction at function entrance if function is only
194 2017-10-24 Jakub Jelinek <jakub@redhat.com>
197 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
198 patterns to better describe from which operation the CF is computed.
199 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
200 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
201 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
202 is 0, use _0 suffixed expanders instead of emitting a comparison
205 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
207 * config/i386/i386.md(*movsf_internal, *movdf_internal):
208 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
210 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
213 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
214 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
215 * loop-iv.c (iv_get_reaching_def): Likewise.
216 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
217 variable is promoted and the partition contains undefined values.
219 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
221 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
223 (nios2_address_cost): Define.
224 (nios2_legitimize_address): Recognize (exp + constant) directly.
225 (TARGET_ADDRESS_COST): Define.
227 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
229 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
230 (nios2_symbolic_memory_operand_p): Declare.
231 (nios2_split_large_constant): Declare.
232 (nios2_split_symbolic_memory_operand): Declare.
233 * config/nios2/nios2.c: Adjust includes.
234 (nios2_symbolic_constant_allowed): New.
235 (nios2_symbolic_constant_p): New.
236 (nios2_plus_symbolic_constant_p): New.
237 (nios2_valid_addr_expr_p): Recognize addresses involving
239 (nios2_legitimate_address_p): Likewise, also LO_SUM.
240 (nios2_symbolic_memory_operand_p): New.
241 (nios2_large_constant_p): New.
242 (nios2_split_large_constant): New.
243 (nios2_split_plus_large_constant): New.
244 (nios2_split_symbolic_memory_operand): New.
245 (nios2_legitimize_address): Code refactoring. Handle addresses
246 involving symbolic constants.
247 (nios2_emit_move_sequence): Likewise.
248 (nios2_print_operand): Improve error output.
249 (nios2_print_operand_address): Handle LO_SUM.
250 (nios2_cdx_narrow_form_p): Likewise.
251 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
252 operands involving symbolic constants.
253 (movhi_internal, movsi_internal): Likewise.
254 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
255 (extendhisi2, extendqi<mode>2): Likewise.
257 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
259 * tree-pass.h (PROP_rtl_split_insns): Define.
260 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
262 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
264 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
266 2017-10-23 Jakub Jelinek <jakub@redhat.com>
269 * target.def (const_not_ok_for_debug_p): Default to
270 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
271 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
272 * targhooks.c (default_const_not_ok_for_debug_p): New function.
273 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
274 which targetm.const_not_ok_for_debug_p returned true.
275 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
277 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
279 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
280 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
281 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
283 (ix86_const_not_ok_for_debug_p): New function.
284 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
285 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
287 2017-10-23 David Malcolm <dmalcolm@redhat.com>
290 * system.h: Conditionally include "unique-ptr.h" if
291 INCLUDE_UNIQUE_PTR is defined.
292 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
293 of defining INCLUDE_UNIQUE_PTR before including "system.h".
295 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
297 * config/rl78/rl78.md: New define_expand "subdi3".
299 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
302 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
303 DF_REF_INSN if DF_REF_INSN_INFO is false.
305 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
307 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
308 xmm_move instead of sse_move.
309 (sse_store_index): New function.
310 (ix86_register_move_cost): Be more sensible about mismatch stall;
311 model AVX moves correctly; make difference between sse->integer and
313 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
314 moves; make difference between SSE and AVX.
315 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
316 and zmm_move. Increase size of sse load and store tables;
317 add unaligned load and store tables; add ssemmx_to_integer.
318 * x86-tune-costs.h: Update all entries according to real
319 move latencies from Agner Fog's manual and chip documentation.
321 2017-10-23 Jakub Jelinek <jakub@redhat.com>
324 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
325 * config/i386/constraints.md (Wf): New constraint.
326 * config/i386/i386.md (UNSPEC_SBB): New unspec.
327 (cmp<dwi>_doubleword): Removed.
328 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
329 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
330 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
331 expand with cmp<dwi>_doubleword. For LTU and GEU use
332 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
334 * common.opt (gcolumn-info): Enable by default.
335 * doc/invoke.texi (gcolumn-info): Document new default.
337 2017-10-23 Richard Biener <rguenther@suse.de>
339 PR tree-optimization/82672
340 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
341 Fold the stmt if we propagated into it.
343 2017-10-23 Richard Biener <rguenther@suse.de>
345 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
346 (bitmap_remove_expr_from_set): ... this. All callers call this
347 for non-constant values.
348 (bitmap_set_subtract): Rename to...
349 (bitmap_set_subtract_expressions): ... this. Adjust and
351 (bitmap_set_contains_value): Remove superfluous check.
352 (bitmap_set_replace_value): Inline into single caller ...
353 (bitmap_value_replace_in_set): ... here and simplify.
354 (dependent_clean): Merge into ...
355 (clean): ... this using an overload. Adjust.
356 (prune_clobbered_mems): Adjust.
357 (compute_antic_aux): Likewise.
358 (compute_partial_antic_aux): Likewise.
360 2017-10-23 Richard Biener <rguenther@suse.de>
362 PR tree-optimization/82129
364 2017-08-01 Richard Biener <rguenther@suse.de>
366 PR tree-optimization/81181
367 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
368 (compute_antic): ... end of iteration here.
370 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
372 * target.def (starting_frame_offset): New hook.
373 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
374 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
375 * doc/tm.texi.in: Regenerate.
376 * hooks.h (hook_hwi_void_0): Declare.
377 * hooks.c (hook_hwi_void_0): New function.
378 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
379 STARTING_FRAME_OFFSET.
380 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
381 * reload1.c (reload): Likewise.
382 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
383 instead of STARTING_FRAME_OFFSET.
384 * function.c (try_fit_stack_local): Likewise.
385 (assign_stack_local_1): Likewise
386 (instantiate_virtual_regs): Likewise.
387 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
388 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
389 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
390 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
391 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
392 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
393 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
394 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
395 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
396 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
397 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
398 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
399 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
400 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
401 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
402 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
403 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
404 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
405 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
406 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
407 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
408 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
409 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
410 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
411 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
412 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
413 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
414 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
415 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
416 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
417 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
418 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
419 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
420 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
421 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
422 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
423 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
424 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
425 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
426 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
427 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
428 * config/avr/avr.c (avr_starting_frame_offset): Make static and
429 return a HOST_WIDE_INT.
430 (avr_builtin_setjmp_frame_value): Use it instead of
431 STARTING_FRAME_OFFSET.
432 (TARGET_STARTING_FRAME_OFFSET): Redefine.
433 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
434 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
436 (TARGET_STARTING_FRAME_OFFSET): Redefine.
437 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
438 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
439 (TARGET_CONSTANT_ALIGNMENT): Redefine.
440 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
441 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
442 (TARGET_STARTING_FRAME_OFFSET): Redefine.
443 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
444 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
445 (TARGET_STARTING_FRAME_OFFSET): Redefine.
446 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
447 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
449 (TARGET_STARTING_FRAME_OFFSET): Redefine.
450 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
451 * config/mips/mips.c (mips_compute_frame_info): Refer to
452 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
453 (mips_starting_frame_offset): New function.
454 (TARGET_STARTING_FRAME_OFFSET): Redefine.
455 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
456 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
457 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
458 and return a HOST_WIDE_INT.
459 (TARGET_STARTING_FRAME_OFFSET): Redefine.
460 (mmix_initial_elimination_offset): Refer to
461 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
462 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
463 * config/pa/pa.c (pa_starting_frame_offset): New function.
464 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
465 (pa_expand_prologue): Likewise.
466 (TARGET_STARTING_FRAME_OFFSET): Redefine.
467 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
468 !FRAME_GROWS_DOWNWARD handling to...
469 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
470 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
471 !FRAME_GROWS_DOWNWARD handling to...
472 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
473 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
474 !FRAME_GROWS_DOWNWARD handling to...
475 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
476 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
478 (rs6000_starting_frame_offset): New function.
479 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
480 !FRAME_GROWS_DOWNWARD handling to...
481 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
482 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
483 !FRAME_GROWS_DOWNWARD handling to...
484 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
485 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
486 !FRAME_GROWS_DOWNWARD handling to...
487 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
488 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
489 (rs6000_starting_frame_offset): New function.
490 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
491 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
492 * config/vax/vax.c (vax_starting_frame_offset): New function.
493 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
494 (TARGET_STARTING_FRAME_OFFSET): Redefine.
495 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
496 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
497 (TARGET_STARTING_FRAME_OFFSET): Redefine.
498 * system.h (STARTING_FRAME_OFFSET): Poison.
500 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
502 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
503 SCALAR_TYPE_MODE instead of TYPE_MODE.
505 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
511 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
517 2017-10-23 Richard Biener <rguenther@suse.de>
519 PR tree-optimization/82129
520 * tree-ssa-pre.c (bitmap_set_and): Remove.
521 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
522 canonicalizing expressions in the set to those with lowest
523 ID rather than taking that from the first edge.
525 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
527 * combine.c (rtx_equal_for_field_assignment_p): Use
530 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
531 Alan Hayward <alan.hayward@arm.com>
532 David Sherwood <david.sherwood@arm.com>
534 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
535 to a SUBREG_PROMOTED_VAR.
537 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
538 Alan Hayward <alan.hayward@arm.com>
539 David Sherwood <david.sherwood@arm.com>
541 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
542 (expand_debug_source_expr): Likewise.
543 * combine.c (combine_simplify_rtx): Likewise.
544 * cse.c (fold_rtx): Likewise.
545 * optabs.c (expand_float): Likewise.
546 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
547 (simplify_binary_operation_1): Likewise.
549 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
554 (record_promoted_value): Likewise.
555 * expr.c (expand_expr_real_2): Likewise.
556 * ree.c (update_reg_equal_equiv_notes): Likewise.
557 (combine_set_extension): Likewise.
558 * rtlanal.c (low_bitmask_len): Likewise.
559 * simplify-rtx.c (neg_const_int): Likewise.
560 (simplify_binary_operation_1): Likewise.
562 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
567 * regcprop.c (maybe_mode_change): Likewise.
568 * reload1.c (alter_reg): Likewise.
570 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
572 * inchash.h (inchash::hash::add_wide_int): New function.
573 * lto-streamer-out.c (hash_tree): Use it.
575 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
577 * inchash.h (inchash::hash::add_wide_int): Rename to...
578 (inchash::hash::add_hwi): ...this.
579 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
580 (polymorphic_call_target_hasher::hash): Likewise.
581 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
582 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
583 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
584 * lto-streamer-out.c (hash_tree): Likewise.
585 * optc-save-gen.awk: Likewise.
586 * tree.c (add_expr): Likewise.
588 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
591 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
592 for ordered inequality comparisons even with TARGET_IEEE_FP.
594 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
597 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
598 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
599 Expand with cmp<dwi>_doubleword.
601 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
603 * extend.texi: Add x86 specific to 'nocf_check' attribute.
605 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
606 specific to -fcf-protection option.
608 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
610 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
611 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
612 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
613 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
614 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
615 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
616 (extra_objs): Add cet.o for Linux/x86 targets.
617 (tmake_file): Add i386/t-cet for Linux/x86 targets.
618 * config/i386/cet.c: New file.
619 * config/i386/cetintrin.h: Likewise.
620 * config/i386/t-cet: Likewise.
621 * config/i386/cpuid.h (bit_SHSTK): New.
623 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
624 pass IBT and SHSTK bits.
625 * config/i386/i386-builtin-types.def
626 (VOID_FTYPE_UNSIGNED_PVOID): New.
627 (VOID_FTYPE_UINT64_PVOID): Likewise.
628 * config/i386/i386-builtin.def: Add CET intrinsics.
629 * config/i386/i386-c.c (ix86_target_macros_internal): Add
630 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
631 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
632 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
634 * config/i386/i386.c (rest_of_insert_endbranch): New.
635 (pass_data_insert_endbranch): Likewise.
636 (pass_insert_endbranch): Likewise.
637 (make_pass_insert_endbranch): Likewise.
638 (ix86_notrack_prefixed_insn_p): Likewise.
639 (ix86_target_string): Add -mibt, -mshstk flags.
640 (ix86_option_override_internal): Add flag_cf_protection
642 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
643 (ix86_print_operand): Add 'notrack' prefix output.
644 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
645 (ix86_expand_builtin): Expand CET intrinsics.
646 (x86_output_mi_thunk): Add 'endbranch' instruction.
647 * config/i386/i386.h (TARGET_IBT): New.
648 (TARGET_IBT_P): Likewise.
649 (TARGET_SHSTK): Likewise.
650 (TARGET_SHSTK_P): Likewise.
651 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
652 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
653 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
654 (builtin_setjmp_setup): New pattern.
655 (builtin_longjmp): Likewise.
656 (rdssp<mode>): Likewise.
657 (incssp<mode>): Likewise.
658 (saveprevssp): Likewise.
659 (rstorssp): Likewise.
660 (wrss<mode>): Likewise.
661 (wruss<mode>): Likewise.
662 (setssbsy): Likewise.
663 (clrssbsy): Likewise.
664 (nop_endbr): Likewise.
665 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
667 * config/i386/immintrin.h: Include <cetintrin.h>.
668 * config/i386/linux-common.h
669 (file_end_indicate_exec_stack_and_cet): New prototype.
670 (TARGET_ASM_FILE_END): New.
672 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
674 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
675 latencies instead of having separate table; make difference between
676 integer and float costs.
677 * i386.h (processor_costs): Remove scalar_stmt_cost,
678 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
679 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
681 * x86-tune-costs.h: Remove entries which has been removed in
682 procesor_costs from all tables; make cond_taken_branch_cost
683 and cond_not_taken_branch_cost COST_N_INSNS based.
685 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
687 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
689 2017-10-20 Jakub Jelinek <jakub@redhat.com>
691 * config/i386/i386.md (isa): Remove fma_avx512f.
692 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
693 <avx512>_fmadd_<mode>_mask3<round_name>,
694 <avx512>_fmsub_<mode>_mask<round_name>,
695 <avx512>_fmsub_<mode>_mask3<round_name>,
696 <avx512>_fnmadd_<mode>_mask<round_name>,
697 <avx512>_fnmadd_<mode>_mask3<round_name>,
698 <avx512>_fnmsub_<mode>_mask<round_name>,
699 <avx512>_fnmsub_<mode>_mask3<round_name>,
700 <avx512>_fmaddsub_<mode>_mask<round_name>,
701 <avx512>_fmaddsub_<mode>_mask3<round_name>,
702 <avx512>_fmsubadd_<mode>_mask<round_name>,
703 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
704 (*vec_widen_umult_even_v16si<mask_name>,
705 *vec_widen_smult_even_v16si<mask_name>): Likewise.
706 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
708 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
710 * extend.texi: Add 'nocf_check' documentation.
711 * gimple.texi: Add second parameter to
712 gimple_build_call_from_tree.
713 * invoke.texi: Add -fcf-protection documentation.
714 * rtl.texi: Add REG_CALL_NOTRACK documenation.
716 2017-10-20 Richard Biener <rguenther@suse.de>
718 PR tree-optimization/82473
719 * tree-vect-loop.c (vectorizable_reduction): Properly get at
720 the largest input type.
722 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
724 * c-attribs.c (handle_nocf_check_attribute): New function.
725 (c_common_attribute_table): Add 'nocf_check' handling.
726 * gimple-parser.c: Add second argument NULL to
727 gimple_build_call_from_tree.
728 * attrib.c (comp_type_attributes): Check nocf_check attribute.
729 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
731 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
732 * common.opt: Add fcf-protection flag.
733 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
734 * flag-types.h: Add enum cf_protection_level.
735 * gimple.c (gimple_build_call_from_tree): Add second parameter.
736 Add 'nocf_check' attribute propagation to gimple call.
737 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
738 (gimple_build_call_from_tree): Update prototype.
739 (gimple_call_nocf_check_p): New function.
740 (gimple_call_set_nocf_check): Likewise.
741 * gimplify.c: Add second argument to gimple_build_call_from_tree.
742 * ipa-icf.c: Add nocf_check attribute in statement hash.
743 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
744 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
745 * toplev.c (process_options): Add flag_cf_protection handling.
747 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
749 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
751 2017-10-20 Richard Biener <rguenther@suse.de>
753 PR tree-optimization/82603
754 * tree-if-conv.c (predicate_mem_writes): Make sure to only
755 remove false predicated stores.
757 2017-10-20 Richard Biener <rguenther@suse.de>
759 * graphite-isl-ast-to-gimple.c
760 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
761 Remove return value and simplify, dump copied stmt after lhs
763 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
764 Reduce dump verbosity.
765 (gsi_insert_earliest): Likewise.
766 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
767 * graphite.c (print_global_statistics): Adjust dumping.
768 (print_graphite_scop_statistics): Likewise.
769 (print_graphite_statistics): Do not dump loops here.
770 (graphite_transform_loops): But here.
772 2017-10-20 Nicolas Roche <roche@adacore.com>
774 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
775 * configure: Regenerate.
777 2017-10-20 Jakub Jelinek <jakub@redhat.com>
780 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
781 functions when optimizing replace GIMPLE_RETURN stmts with
782 calls to __builtin_unreachable ().
785 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
786 for -fsanitize=thread link of executables.
787 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
791 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
793 (<shift_insn><mode>3<mask_name>): Change the last of the 3
794 define_insns for logical vector shifts to use VI248_AVX512BW
795 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
796 condition, useless isa and prefix attributes. Change the first
797 2 of these define_insns to ...
798 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
799 define_insn for avx512vl.
800 (<shift_insn><mode>3): ... and this, new define_insn without
801 masking for non-avx512vl.
804 * config/i386/sse.md (*andnot<mode>3,
805 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
806 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
807 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
808 not applied use empty suffix even for TARGET_AVX512VL.
809 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
810 is applied, supply evex,evex or evex,evex,evex instead of just
813 2017-10-20 Julia Koval <julia.koval@intel.com>
815 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
816 (OPTION_MASK_ISA_GFNI_UNSET): New.
817 (ix86_handle_option): Handle OPT_mgfni.
818 * config/i386/cpuid.h (bit_GFNI): New.
819 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
820 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
821 * config/i386/i386.c (ix86_target_string): Add -mgfni.
822 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
823 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
824 * config/i386/i386.opt: Add mgfni.
826 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
828 * config/msp430/msp430.c (msp430_option_override): Disable
829 -fdelete-null-pointer-checks.
830 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
832 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
834 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
835 of x87 and SSE instructions.
837 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
839 * asan.c (create_cond_insert_point): Do not update edge count.
840 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
841 (afdo_propagate_circuit): Likewise.
842 (afdo_calculate_branch_prob): Likewise.
843 (afdo_annotate_cfg): Likewise.
844 * basic-block.h (struct edge_def): Remove count.
845 (edge_def::count): New accessor.
846 * bb-reorder.c (rotate_loop): Update.
847 (find_traces_1_round): Update.
848 (connect_traces): Update.
849 (sanitize_hot_paths): Update.
850 * cfg.c (unchecked_make_edge): Update.
851 (make_single_succ_edge): Update.
852 (check_bb_profile): Update.
853 (dump_edge_info): Update.
854 (update_bb_profile_for_threading): Update.
855 (scale_bbs_frequencies_int): Update.
856 (scale_bbs_frequencies_gcov_type): Update.
857 (scale_bbs_frequencies_profile_count): Update.
858 (scale_bbs_frequencies): Update.
859 * cfganal.c (connect_infinite_loops_to_exit): Update.
860 * cfgbuild.c (compute_outgoing_frequencies): Update.
861 (find_many_sub_basic_blocks): Update.
862 * cfgcleanup.c (try_forward_edges): Update.
863 (try_crossjump_to_edge): Update
864 * cfgexpand.c (expand_gimple_cond): Update
865 (expand_gimple_tailcall): Update
866 (construct_exit_block): Update
867 * cfghooks.c (verify_flow_info): Update
868 (redirect_edge_succ_nodup): Update
870 (make_forwarder_block): Update
871 (duplicate_block): Update
872 (account_profile_record): Update
873 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
874 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
875 * cfgloopmanip.c (scale_loop_profile): Update.
877 (lv_adjust_loop_entry_edge): Update.
878 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
879 (force_nonfallthru_and_redirect): Update.
880 (purge_dead_edges): Update.
881 (rtl_flow_call_edges_add): Update.
882 * cgraphunit.c (init_lowered_empty_function): Update.
883 (cgraph_node::expand_thunk): Update.
884 * gimple-pretty-print.c (dump_probability): Update.
885 (dump_edge_probability): Update.
886 * gimple-ssa-isolate-paths.c (isolate_path): Update.
887 * haifa-sched.c (sched_create_recovery_edges): Update.
888 * hsa-gen.c (convert_switch_statements): Update.
889 * ifcvt.c (dead_or_predicable): Update.
890 * ipa-inline-transform.c (inline_transform): Update.
891 * ipa-split.c (split_function): Update.
892 * ipa-utils.c (ipa_merge_profiles): Update.
893 * loop-doloop.c (add_test): Update.
894 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
895 * lto-streamer-in.c (input_cfg): Update.
896 (input_function): Update.
897 * lto-streamer-out.c (output_cfg): Update.
898 * modulo-sched.c (sms_schedule): Update.
899 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
900 * predict.c (maybe_hot_edge_p): Update.
901 (unlikely_executed_edge_p): Update.
902 (probably_never_executed_edge_p): Update.
903 (dump_prediction): Update.
904 (drop_profile): Update.
905 (propagate_unlikely_bbs_forward): Update.
906 (determine_unlikely_bbs): Update.
907 (force_edge_cold): Update.
908 * profile.c (compute_branch_probabilities): Update.
909 * reg-stack.c (better_edge): Update.
910 * shrink-wrap.c (handle_simple_exit): Update.
911 * tracer.c (better_p): Update.
912 * trans-mem.c (expand_transaction): Update.
913 (split_bb_make_tm_edge): Update.
914 * tree-call-cdce.c: Update.
915 * tree-cfg.c (gimple_find_sub_bbs): Update.
916 (gimple_split_edge): Update.
917 (gimple_duplicate_sese_region): Update.
918 (gimple_duplicate_sese_tail): Update.
919 (gimple_flow_call_edges_add): Update.
920 (insert_cond_bb): Update.
921 (execute_fixup_cfg): Update.
922 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
923 * tree-complex.c (expand_complex_div_wide): Update.
924 * tree-eh.c (lower_resx): Update.
925 (unsplit_eh): Update.
926 (cleanup_empty_eh_move_lp): Update.
927 * tree-inline.c (copy_edges_for_bb): Update.
928 (freqs_to_counts): Update.
929 (copy_cfg_body): Update.
930 * tree-ssa-dce.c (remove_dead_stmt): Update.
931 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
932 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
933 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
934 (unloop_loops): Update.
935 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
936 * tree-ssa-loop-split.c (connect_loops): Update.
937 (split_loop): Update.
938 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
939 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
940 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
941 * tree-ssa-reassoc.c (branch_fixup): Update.
942 * tree-ssa-tail-merge.c (replace_block_by): Update.
943 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
944 (compute_path_counts): Update.
945 (update_profile): Update.
946 (recompute_probabilities): Update.
947 (update_joiner_offpath_counts): Update.
948 (estimated_freqs_path): Update.
949 (freqs_to_counts_path): Update.
950 (clear_counts_path): Update.
951 (ssa_fix_duplicate_block_edges): Update.
952 (duplicate_thread_path): Update.
953 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
954 (case_bit_test_cmp): Update.
955 (collect_switch_conv_info): Update.
956 (gen_inbound_check): Update.
957 (do_jump_if_equal): Update.
958 (emit_cmp_and_jump_insns): Update.
959 * tree-tailcall.c (decrease_profile): Update.
960 (eliminate_tail_call): Update.
961 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
962 (vect_do_peeling): Update.
963 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
964 * ubsan.c (ubsan_expand_null_ifn): Update.
965 (ubsan_expand_ptr_ifn): Update.
966 * value-prof.c (gimple_divmod_fixed_value): Update.
967 (gimple_mod_pow2): Update.
968 (gimple_mod_subtract): Update.
970 (gimple_stringop_fixed_value): Update.
972 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
975 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
977 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
979 PR rtl-optimization/82395
980 * ira-color.c (allocno_priority_compare_func): Fix comparison step
981 based on non_spilled_static_chain_regno_p.
983 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
985 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
986 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
987 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
989 2017-10-19 Martin Sebor <msebor@redhat.com>
991 PR tree-optimization/82596
992 * tree.c (array_at_struct_end_p): Handle STRING_CST.
994 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
996 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
997 (get_mem_refs_of_builtin_call): Likewise.
998 * builtins.c (expand_builtin_apply): Adjust call to
999 allocate_dynamic_stack_space.
1000 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1001 the third argument to allocate_dynamic_stack_space, otherwise -1.
1002 (expand_builtin): Deal with all alloca variants.
1003 (is_inexpensive_builtin): Likewise.
1004 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1005 * calls.c (special_function_p): Deal with all alloca variants.
1006 (initialize_argument_information): Adjust call to
1007 allocate_dynamic_stack_space.
1008 (expand_call): Likewise.
1009 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1010 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1011 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1012 use it for the stack usage computation.
1013 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1014 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1015 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1016 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1017 (in_loop_p): Remove first argument and useless check.
1018 (pass_walloca::execute): Remove useless test and adjust call to above.
1019 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1020 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1021 (gimplify_call_expr): Deal with all alloca variants.
1022 * hsa-gen.c (gen_hsa_alloca): Likewise.
1023 (gen_hsa_insns_for_call): Likewise.
1024 * ipa-pure-const.c (special_builtin_state): Likewise.
1025 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1026 * tree-object-size.c (alloc_object_size): Likewise.
1027 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1028 (call_may_clobber_ref_p_1): Likewise.
1029 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1030 (ccp_fold_stmt): Likewise.
1031 (optimize_stack_restore): Likewise.
1032 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1033 (mark_all_reaching_defs_necessary_1): Likewise.
1034 (propagate_necessity): Likewise.
1035 (eliminate_unnecessary_stmts): Likewise.
1036 * tree.c (build_common_builtin_nodes): Build
1037 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1038 (build_alloca_call_expr): New function.
1039 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1040 (CASE_BUILT_IN_ALLOCA): Likewise.
1041 (build_alloca_call_expr): Declare.
1042 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1044 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1047 * dwarf2out.c (new_die_raw): New static inline function.
1048 (new_die): Use it to create the DIE.
1049 (add_AT_external_die_ref): Likewise.
1050 (clone_die): Likewise.
1051 (clone_as_declaration): Likewise.
1052 (dwarf2out_vms_debug_main_pointer): Likewise.
1053 (base_type_die): Likewise. Remove early return for corner cases.
1054 Do not call add_pubtype on the DIE here.
1055 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1056 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1057 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1058 native order exists for base types, attach the DIE manually and call
1059 add_pubtype on it. Do not equate a reverse order DIE to the type.
1061 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1063 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1064 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1066 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1068 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1070 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1071 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1072 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1073 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1074 * opts.c (sanitizer_opts): Add builtin.
1075 * ubsan.c (instrument_builtin): New function.
1076 (pass_ubsan::execute): Call it.
1077 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1078 * doc/invoke.texi: Document -fsanitize=builtin.
1080 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1081 builtins, store max (log2 (align), 0) into uchar field instead of
1082 align into uptr field.
1083 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1084 store uchar 0 field instead of uptr 0 field.
1085 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1086 instead of passing one address of struct with 2 locations pass
1087 two addresses of structs with 1 location each.
1088 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1089 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1090 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1091 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1092 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1093 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1094 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1095 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1097 2017-10-19 Martin Liska <mliska@suse.cz>
1100 * file-find.c (remove_prefix): Remove.
1101 * file-find.h (remove_prefix): Likewise.
1102 * gcc-ar.c: Remove smartness of lookup.
1104 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1106 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1107 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1108 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1110 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1113 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1114 (setcc + and to xor + setcc): New peephole2.
1116 2017-10-19 Tom de Vries <tom@codesourcery.com>
1118 * doc/sourcebuild.texi (Test Directives, Variants of
1119 dg-require-support): Add dg-require-stack-size.
1121 2017-10-19 Martin Liska <mliska@suse.cz>
1124 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1125 that have a large alignment.
1126 (gimplify_target_expr): Likewise.
1128 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1130 PR rtl-optimization/82602
1131 * ira.c (rtx_moveable_p): Return false for volatile asm.
1133 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1136 * config/i386/i386-modes.def (CCGZ): New CC mode.
1137 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1138 * config/i386/predicates.md (ix86_comparison_operator):
1140 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1141 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1142 with double-word subtraction.
1143 (put_condition_code): Handle CCGZmode.
1145 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1147 * wide-int.cc (debug (const wide_int &)): New.
1148 (debug (const wide_int *)): New.
1149 (debug (const widest_int &)): New.
1150 (debug (const widest_int *)): New.
1152 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1155 * lra-constraints.c (curr_insn_transform): Use non-input operand
1156 instead of output one for matched reload.
1158 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1160 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1161 (tree-ssa-loop-ivopts.h): New header file.
1162 (struct builtin_info): New fields.
1163 (classify_builtin_1): Compute and record base and offset parts for
1164 memset builtin partition by calling strip_offset.
1165 (offset_cmp, fuse_memset_builtins): New functions.
1166 (finalize_partitions): Fuse adjacent memset partitions by calling
1168 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1169 Expose the interface.
1170 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1172 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1174 PR tree-optimization/82574
1175 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1176 that data reference must be executed exactly once per iteration
1177 against the outermost loop in nest.
1178 (classify_partition): Update call to above function.
1180 2017-10-18 Richard Biener <rguenther@suse.de>
1182 PR tree-optimization/82591
1183 * graphite.c (graphite_transform_loops): Move code gen message
1185 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1186 Here. Handle scop_to_isl_ast failing.
1187 (scop_to_isl_ast): Limit the number of ISL operations.
1189 2017-10-18 Richard Biener <rguenther@suse.de>
1191 * graphite-isl-ast-to-gimple.c
1192 (translate_isl_ast_to_gimple::set_rename): Simplify.
1193 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1194 (graphite_copy_stmts_from_block): ... here.
1195 (copy_bb_and_scalar_dependences): Simplify.
1196 (add_parameters_to_ivs_params): Canonicalize.
1197 (generate_entry_out_of_ssa_copies): Simplify.
1198 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1199 by passing in ISL dimension.
1200 (parameter_index_in_region_1): Rename to ...
1201 (parameter_index_in_region): ... this.
1202 (extract_affine): Adjust assert, pass down parameter index.
1203 (add_param_constraints): Use range-info when available.
1204 (build_scop_context): Adjust.
1205 * sese.c (new_sese_info): Adjust.
1206 (free_sese_info): Likewise.
1207 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1208 Remove unused typedefs.
1209 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1211 2017-10-18 Martin Liska <mliska@suse.cz>
1213 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1215 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1217 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1218 (s390_sched_init): Do not reset s390_sched_state if we entered the
1219 current basic block via a fallthru edge and all others are unlikely.
1221 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1223 * config/s390/s390.c (NUM_SIDES): New variable.
1224 (LONGRUNNING_THRESHOLD): New variable.
1225 (LATENCY_FACTOR): New variable.
1226 (s390_sched_score): Decrease score for long-running instructions on
1228 (s390_sched_variable_issue): Perform bookkeeping for long-running
1231 2017-10-18 Richard Biener <rguenther@suse.de>
1233 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1234 Simplify with removal of the parameter rename map.
1235 (set_rename): Likewise.
1236 (should_copy_to_new_region): Likewise.
1237 (graphite_copy_stmts_from_block): Likewise.
1238 (copy_bb_and_scalar_dependences): Remove initialization of
1239 unused copied_bb_map.
1241 (copy_internal_parameters): Likewise.
1242 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1243 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1244 Use INTEGRAL_TYPE_P.
1245 (parameter_index_in_region_1): Rename to ...
1246 (assign_parameter_index_in_region): ... this. Assert we have
1247 a parameter we handle.
1248 (scan_tree_for_params): Adjust.
1249 * sese.h (parameter_rename_map_t): Remove.
1250 (struct sese_info_t): Remove unused parameter_rename_map and
1251 copied_bb_map members.
1252 * sese.c (new_sese_info): Adjust.
1253 (free_sese_info): Likewise.
1255 2017-10-18 Martin Liska <mliska@suse.cz>
1258 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1259 on an abnormal edge.
1261 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1263 * doc/invoke.texi (ffunction-sections and fdata-sections):
1266 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1268 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1269 the use statement can throw internally.
1271 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1273 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1274 any RTX present on the RHS of a SET.
1275 * compare-elim.c (try_eliminate_compare): Restore comment.
1277 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1279 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1280 may be also called on tcc_type nodes.
1281 * langhooks.c (lhd_tree_size): Likewise.
1283 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1285 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1286 format_warning_at_substring.
1287 (maybe_warn): Convert source_range * param to a location_t. Pass
1288 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1289 (format_directive): Remove code to extract source_ranges and
1290 source_range * in favor of just a location_t.
1291 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1293 * substring-locations.c (format_warning_va): Convert
1294 source_range * param to a location_t.
1295 (format_warning_at_substring): Likewise.
1296 * substring-locations.h (format_warning_va): Likewise.
1297 (format_warning_at_substring): Likewise.
1299 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1301 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1303 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1304 and scatter/gather ops.
1306 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1307 vec_gather_load and vec_scatter_store.
1308 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1309 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1310 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1311 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1312 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1313 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1315 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1317 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1318 Detect FTST instruction and handle its register pops. Only pop
1319 second operand if can_pop_second_op is true.
1320 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1321 set can_pop_second_op to false in the compare_for_stack_reg call.
1323 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1324 output_fp_compare for stack register operands.
1325 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1326 instructions here. Do not emit stack register pops here. Assert
1327 that FCOMPP pops next to top stack register. Rewrite function.
1329 2017-10-17 Nathan Sidwell <nathan@acm.org>
1332 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1333 use DECL_ASSEMBLER_NAME_RAW.
1336 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1339 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1340 Wilco Dijkstra <wilco.dijkstra@arm.com>
1342 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1343 converstion to Pmode from the buf_addr.
1345 2017-10-17 Richard Biener <rguenther@suse.de>
1347 * graphite-dependences.c (scop_get_reads_and_writes): Change
1348 output parameters to references.
1350 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1352 PR 71026/tree-optimization
1353 * fold-const.c (distribute_real_division): Removed.
1354 (fold_binary_loc): Remove calls to distribute_real_divison.
1356 2017-10-17 Richard Biener <rguenther@suse.de>
1358 * graphite-scop-detection.c
1359 (scop_detection::stmt_has_simple_data_refs_p): Always use
1360 the full nest as region.
1361 (try_generate_gimple_bb): Likewise.
1362 * sese.c (scalar_evolution_in_region): Simplify now that
1363 SCEV can handle instantiation in regions.
1364 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1365 in the non-loop part of a function if requested.
1367 2017-10-17 Richard Biener <rguenther@suse.de>
1369 PR tree-optimization/82563
1370 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1372 (graphite_regenerate_ast_isl): Call it.
1373 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1375 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1377 PR tree-optimization/82549
1378 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1379 Formatting fixes. Instead of calling make_bit_field_ref with negative
1382 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1384 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1385 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1386 _mm_maskz_reduce_ss): New.
1387 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1388 __builtin_ia32_reducess_mask): Ditto..
1389 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1390 * config/i386/sse.md (reduces<mode>): Renamed to ...
1391 (reduces<mode><mask_scalar_name>): ... this.
1392 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1394 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1395 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1397 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1399 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1400 * selftest-run-tests.c (selftest::run_tests): Call
1401 selftest::unique_ptr_tests_cc_tests.
1402 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1403 * unique-ptr-tests.cc: New file.
1405 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1408 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1410 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1411 (make_hard_regno_born, make_hard_regno_dead): Update
1412 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1414 2017-10-16 Jeff Law <law@redhat.com>
1416 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1418 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1420 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1422 2017-10-16 Olivier Hainque <hainque@adacore.com>
1424 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1425 with_cpu if we were configured for an e500v2 target cpu name.
1427 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1429 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1430 * doc/invoke.texi: Document +nodsp as a valid extension for
1433 2017-10-16 Martin Liska <mliska@suse.cz>
1435 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1436 (test_set_range): Likewise.
1437 (test_range_functions): Rename to ...
1438 (test_bit_in_range): ... this.
1439 (sbitmap_c_tests): Add new test.
1441 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1443 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1444 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1445 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1447 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1449 * config/aarch64/aarch64-builtins.c
1450 (aarch64_types_quadopu_lane_qualifiers): New.
1451 (TYPES_QUADOPU_LANE): New.
1452 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1453 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1454 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1455 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1456 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1457 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1458 (Vdottype, DOTPROD): New.
1459 (sur): Add SDOT and UDOT.
1461 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1463 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1464 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1465 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1466 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1467 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1468 (cortex-a75.cortex-a55): Likewise.
1469 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1471 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1473 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1474 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1475 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1476 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1477 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1478 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1479 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1480 * config/arm/types.md (neon_dot, neon_dot_q): New.
1481 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1483 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1485 * config/arm/arm.h (TARGET_DOTPROD): New.
1486 * config/arm/arm.c (arm_arch_dotprod): New.
1487 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1488 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1489 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1490 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1491 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1492 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1493 * doc/invoke.texi (armv8.2-a): Document dotprod
1495 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1497 * i386.c (ix86_vec_cost): New function.
1498 (ix86_rtx_costs): Handle vector operations better.
1499 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1500 * x86-tune-costs.h: Add new costs to all tables.
1502 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1504 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1506 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1507 divsd, sqrtss and sqrtsd
1508 * x86-tune-costs.h: Add new entries to all costs.
1509 (znver1_cost): Fix to match real instruction latencies.
1511 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1512 Michael Collison <michael.collison@arm.com>
1514 * compare-elim.c: Include emit-rtl.h.
1515 (can_merge_compare_into_arith): New function.
1516 (try_validate_parallel): Likewise.
1517 (try_merge_compare): Likewise.
1518 (try_eliminate_compare): Call the above when no previous clobber
1520 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1523 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1527 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1528 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1532 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1533 to be any operand_equal_p operands. For & (B - 1) require
1534 B to be power of 2. Recognize
1535 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1537 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1540 * optabs.c (expand_memory_blockage): Fix call of
1541 targetm.have_memory_blockage.
1543 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1546 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1547 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1548 objects to extra_objs instead of overwriting it.
1550 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1552 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1553 Use any_fp_register_operand as operand[3] predicate. Simplify
1554 equality test for operands[2] and operands[4] memory location.
1555 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1556 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1557 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1558 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1559 any_fp_register_operand as operand[1] predicate. Simplify
1560 equality test for operands[0] and operands[3] memory location.
1561 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1562 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1563 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1565 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1567 * target-insns.def: Add memory_blockage.
1568 * optabs.c (expand_memory_blockage): New function.
1569 (expand_asm_memory_barrier): Rename ...
1570 (expand_asm_memory_blockage): ... to this.
1571 (expand_mem_thread_fence): Call expand_memory_blockage
1572 instead of expand_asm_memory_barrier.
1573 (expand_mem_singnal_fence): Ditto.
1574 (expand_atomic_load): Ditto.
1575 (expand_atomic_store): Ditto.
1576 * doc/md.texi (Standard Pattern Names For Generation):
1577 Document memory_blockage instruction pattern.
1579 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1581 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1582 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1583 * config/rl78/rl78.md: New define_expand "adddi3".
1585 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1587 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1590 2017-10-13 Jeff Law <law@redhat.com>
1592 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1594 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1597 * internal-fn.c (expand_mul_overflow): If both operands have
1598 the same highpart of -1 or 0 and the topmost bit of lowpart
1599 is different, overflow is if res <= 0 rather than res < 0.
1601 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1603 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1604 TARGET_P9_VECTOR code for unaligned_load case.
1606 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1608 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1610 2017-10-13 Nathan Sidwell <nathan@acm.org>
1612 * tree-core.h (tree_contains_struct): Make bool.
1613 * tree.c (tree_contains_struct): Likewise.
1614 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1615 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1616 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1617 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1618 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1620 2017-10-13 Richard Biener <rguenther@suse.de>
1622 * graphite-isl-ast-to-gimple.c
1623 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1624 parameters and dominance check.
1625 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1626 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1627 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1628 Do not update SSA form here or do intermediate IL verification.
1629 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1630 (graphite_initialize): Remove check on the number of loops in
1631 the function and inline into graphite_transform_loops.
1632 (graphite_finalize): Inline into graphite_transform_loops.
1633 (graphite_transform_loops): Perform SSA update and IL verification
1635 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1637 2017-10-13 Richard Biener <rguenther@suse.de>
1639 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1640 graphite_expression_type_precision): Avoid global constructor
1642 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1643 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1644 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1645 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1646 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1647 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1649 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1652 * config/i386/i386.h (ix86_red_zone_size): New.
1653 * config/i386/i386.md (push peephole2s): Replace
1654 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1656 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1657 Alan Hayward <alan.hayward@arm.com>
1658 David Sherwood <david.sherwood@arm.com>
1660 * combine.c (can_change_dest_mode): Reject changes in
1661 REGMODE_NATURAL_SIZE.
1663 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1664 Alan Hayward <alan.hayward@arm.com>
1665 David Sherwood <david.sherwood@arm.com>
1667 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1668 (expand_debug_source_expr): Likewise.
1669 * combine.c (combine_simplify_rtx): Likewise.
1670 * cse.c (fold_rtx): Likewise.
1671 * fwprop.c (canonicalize_address): Likewise.
1672 * targhooks.c (default_shift_truncation_mask): Likewise.
1674 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1675 Alan Hayward <alan.hayward@arm.com>
1676 David Sherwood <david.sherwood@arm.com>
1678 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1679 (widened_mode): Likewise.
1680 (expand_unop): Likewise.
1681 * ree.c (transform_ifelse): Likewise.
1682 (merge_def_and_ext): Likewise.
1683 (combine_reaching_defs): Likewise.
1684 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1686 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1687 Alan Hayward <alan.hayward@arm.com>
1688 David Sherwood <david.sherwood@arm.com>
1690 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1691 * combine.c (gen_lowpart_for_combine): Likewise.
1692 * dwarf2out.c (rtl_for_decl_location): Likewise.
1693 * final.c (alter_subreg): Likewise.
1694 * rtlhooks.c (gen_lowpart_general): Likewise.
1695 (gen_lowpart_if_possible): Likewise.
1697 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1698 Alan Hayward <alan.hayward@arm.com>
1699 David Sherwood <david.sherwood@arm.com>
1701 * calls.c (expand_call): Use subreg_lowpart_offset.
1702 * cse.c (cse_insn): Likewise.
1703 * regcprop.c (copy_value): Likewise.
1704 (copyprop_hardreg_forward_1): Likewise.
1706 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1709 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1710 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1711 =Q constraints to +Q and into insn condition add check
1712 that operands[0] and operands[1] are equal.
1713 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1714 =Q constraints to +Q and into insn condition add check
1715 that operands[0] is equal to either operands[1] or operands[2].
1718 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1719 instead of handling MINUS_EXPR twice (once for each argument),
1720 canonicalize operand order and handle just once, use rtype where
1721 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1724 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1725 any values of __C while still being pattern recognizable as a simple
1728 2017-10-13 Richard Biener <rguenther@suse.de>
1730 PR tree-optimization/82451
1732 2017-10-02 Richard Biener <rguenther@suse.de>
1734 PR tree-optimization/82355
1735 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1736 a mapping for the enclosing loop but avoid generating one for
1738 (copy_bb_and_scalar_dependences): Remove premature codegen
1739 error on PHIs in blocks duplicated into multiple places.
1740 * graphite-scop-detection.c
1741 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1742 in the region use it as loop and nest to analyze the DR in.
1743 (try_generate_gimple_bb): Likewise.
1744 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1745 (add_loop_constraints): For blocks in a loop not in the region
1746 create a dimension with a single iteration.
1747 * sese.h (gbb_loop_at_index): Remove assert.
1749 * cfgloop.c (loop_preheader_edge): For the loop tree root
1750 return the single successor of the entry block.
1751 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1752 Reset the SCEV hashtable and niters.
1753 * graphite-scop-detection.c
1754 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1755 assert that we only have POLYNOMIAL_CHREC that vary in loops
1756 contained in the region.
1757 (scop_detection::graphite_can_represent_expr): Adjust.
1758 (scop_detection::stmt_has_simple_data_refs_p): For loops
1759 not in the region set loop to NULL. The nest is now the
1760 entry edge to the region.
1761 (try_generate_gimple_bb): Likewise.
1762 * sese.c (scalar_evolution_in_region): Adjust for
1763 instantiate_scev change.
1764 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1765 Make nest parameter the edge into the region.
1766 (create_data_ref): Likewise.
1767 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1768 entry edge into a region and adjust instantiate_scev calls.
1769 (create_data_ref): Likewise.
1770 (graphite_find_data_references_in_stmt): Likewise.
1771 (find_data_references_in_stmt): Pass the loop preheader edge
1772 from the nest argument.
1773 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1774 parameter the edge into the region.
1775 (instantiate_parameters): Use the loop preheader edge as entry.
1776 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1778 (get_instantiated_value_entry): Make instantiate_below parameter
1779 the edge into the region.
1780 (instantiate_scev_name): Likewise. Adjust dominance checks,
1781 when we cannot use loop-based instantiation instantiate by
1782 walking use-def chains.
1783 (instantiate_scev_poly): Adjust.
1784 (instantiate_scev_binary): Likewise.
1785 (instantiate_scev_convert): Likewise.
1786 (instantiate_scev_not): Likewise.
1787 (instantiate_array_ref): Remove.
1788 (instantiate_scev_3): Likewise.
1789 (instantiate_scev_2): Likewise.
1790 (instantiate_scev_1): Likewise.
1791 (instantiate_scev_r): Do not blindly handle N-operand trees.
1792 Do not instantiate array-refs. Handle all constants and invariants.
1793 (instantiate_scev): Make instantiate_below parameter
1794 the edge into the region.
1795 (resolve_mixers): Use the loop preheader edge for the region
1796 parameter to instantiate_scev_r.
1797 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1799 2017-10-13 Richard Biener <rguenther@suse.de>
1801 PR tree-optimization/82525
1802 * graphite-isl-ast-to-gimple.c
1803 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1805 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1806 Fail code generation when we cannot represent the isl integer.
1807 (binary_op_to_tree): Elide modulo operations that are no-ops
1808 in the type we code generate. Remove now superfluous code
1811 2017-10-13 Richard Biener <rguenther@suse.de>
1813 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1814 (scop_detection::harmful_loop_in_region): Remove premature
1815 IV type restriction.
1816 (scop_detection::graphite_can_represent_scev): We can handle
1817 pointer IVs just fine.
1819 2017-10-13 Alan Modra <amodra@gmail.com>
1821 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1822 "Clobbers and Scratch Registers". Add paragraph on
1823 alternative to clobbers for scratch registers and OpenBLAS
1826 2017-10-13 Alan Modra <amodra@gmail.com>
1828 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1829 example of a memory input for a string of known length. Move
1830 commentary out of table. Add a number of new examples
1831 covering array memory inputs.
1833 2017-10-12 Martin Liska <mliska@suse.cz>
1835 PR tree-optimization/82493
1836 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1837 (test_range_functions): New function.
1838 (sbitmap_c_tests): Likewise.
1839 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1840 * selftest.h (sbitmap_c_tests): New function.
1842 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1844 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1846 * config/rs6000/amo.h: Fix spacing issue.
1848 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1851 * config/i386/i386.md (*ashl<mode>3_mask_1,
1852 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1853 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1856 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1858 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1859 (profile_probability): Set max_probability
1860 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1862 * profile-count.c (profile_probability::differs_from_p): Do not
1863 rely on max_probaiblity == 10000
1865 2017-10-12 Jeff Law <law@redhat.com>
1867 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1870 2017-10-12 Martin Sebor <msebor@redhat.com>
1874 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1875 (handle_alias_pairs): Call it.
1876 * common.opt (-Wattribute-alias): New option.
1877 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1878 * doc/invoke.texi (-Wattribute-alias): Document.
1880 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1883 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1885 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1887 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1888 (make_hard_regno_born, make_hard_regno_dead): Update
1889 bb_killed_pseudos and bb_gen_pseudos.
1891 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1893 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1895 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1897 * config/alpha/alpha.c (alpha_split_conditional_move):
1898 Use std::swap instead of manually swapping.
1899 (alpha_stdarg_optimize_hook): Ditto.
1900 (alpha_canonicalize_comparison): Ditto.
1902 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1904 * tree-loop-distribution.c (struct builtin_info): New struct.
1905 (struct partition): Refactor fields into struct builtin_info.
1906 (partition_free): Free struct builtin_info.
1907 (build_size_arg_loc, build_addr_arg_loc): Delete.
1908 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1909 information from struct builtin_info.
1910 (find_single_drs): New function refactored from classify_partition.
1911 Also moved builtin validity checks to this function.
1912 (compute_access_range, alloc_builtin): New functions.
1913 (classify_builtin_st, classify_builtin_ldst): New functions.
1914 (classify_partition): Refactor code into functions find_single_drs,
1915 classify_builtin_st and classify_builtin_ldst.
1916 (distribute_loop): Don't do runtime alias check when distributing
1918 (find_seed_stmts_for_distribution): New function.
1919 (pass_loop_distribution::execute): Refactor code finding seed
1920 stmts into above function. Support distribution for the innermost
1921 two-level loop nest. Adjust dump information.
1923 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1925 * tree-loop-distribution.c: Adjust the general comment.
1926 (NUM_PARTITION_THRESHOLD): New macro.
1927 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1928 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1929 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1930 in call to build_partition_graph.
1931 (finalize_partitions): New parameter. Make loop distribution more
1932 conservative by fusing more partitions.
1933 (distribute_loop): Don't do runtime alias check in case of loop nest
1935 (find_seed_stmts_for_distribution): New function.
1936 (prepare_perfect_loop_nest): New function.
1937 (pass_loop_distribution::execute): Refactor code finding seed stmts
1938 and loop nest into above functions. Support loop nest distribution.
1939 Adjust dump information accordingly.
1941 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1943 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1944 and set PTYPE_SEQUENTIAL for merged partition.
1946 2017-10-12 Richard Biener <rguenther@suse.de>
1948 PR tree-optimization/69728
1950 2017-09-19 Richard Biener <rguenther@suse.de>
1952 PR tree-optimization/69728
1953 * graphite-sese-to-poly.c (schedule_error): New global.
1954 (add_loop_schedule): Handle empty domain by failing the
1956 (build_original_schedule): Handle schedule_error.
1958 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1959 domain by returning an unchanged schedule.
1961 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1963 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1964 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1966 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1968 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1971 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1974 * expr.c (store_field): Don't optimize away bitsize == 0 store
1975 from CALL_EXPR with addressable return type.
1977 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1979 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1980 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1981 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1982 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1983 TARGET_ISEL instead of TARGET_ISEL<sel>.
1985 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1987 * config/rs6000/rs6000.c
1988 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1990 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1992 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1994 (reg_or_cint_operand): Fix comment.
1995 (reg_or_zero_operand): New predicate.
1996 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1997 * config/rs6000/rs6000.c (output_isel): Delete.
1998 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1999 instead of reg_or_cint_operand. Output instruction directly (not via
2001 (isel_unsigned_<mode>): Ditto.
2002 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2003 gpc_reg_operand. Add an instruction alternative for this. Output
2004 instruction directly.
2005 (*isel_reversed_unsigned_<mode>): Ditto.
2007 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2009 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2010 (TARGET_CANONICALIZE_COMPARISON): Define.
2012 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2015 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2016 Check whether the dest is REG before adding REG_EQUIV note.
2018 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2021 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2023 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2024 (make_hard_regno_born, make_hard_regno_dead): Update
2025 bb_killed_pseudos and bb_gen_pseudos.
2027 2017-10-11 Nathan Sidwell <nathan@acm.org>
2029 * incpath.h (enum incpath_kind): Name enum, prefix values.
2030 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2031 * incpath.c (heads, tails): Use INC_MAX.
2032 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2033 (merge_include_chains, split_quote_chain,
2034 register_include_chains): Update incpath_kind names.
2035 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2036 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2038 (add_framework_path, darwin_register_objc_includes): Likewise.
2039 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2041 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2043 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2044 Do not use float_operator operator predicate.
2045 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2046 * config/i386/predicates.md (float_operator): Remove predicate.
2048 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2050 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2051 (*jccxf_i387): Ditto.
2052 (*jcc<mode>_i387): Ditto.
2053 (*jccu<mode>_i387): Ditto.
2054 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2055 (*jcc_*_i387 splitters): Remove.
2056 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2057 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2058 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2061 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2063 * profile-count.h (slow_safe_scale_64bit): New function.
2064 (safe_scale_64bit): New inline.
2065 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2066 * profile-count.c: Include wide-int.h
2067 (slow_safe_scale_64bit): New.
2069 2017-10-11 Nathan Sidwell <nathan@acm.org>
2071 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2072 HAS_DECL_ASSEMBLER_NAME_P.
2073 * gimple-expr.c (gimple_decl_printable_name: Check
2074 HAS_DECL_ASSEMBLER_NAME_P too.
2075 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2076 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2077 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2078 * passes.c (rest_of_decl_compilation): Check
2079 HAS_DECL_ASSEMBLER_NAME_P too.
2080 * recog.c (verify_changes): Likewise.
2081 * tree-pretty-print.c (dump_decl_name): Likewise.
2082 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2084 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2085 (SET_DECL_ASSEMBLER_NAME): Use it.
2086 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2087 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2088 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2090 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2092 * config.gcc (i386, x86_64): Add extra objects.
2093 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2094 (ix86_min_insn_size): Declare.
2095 (ix86_issue_rate): Declare.
2096 (ix86_adjust_cost): Declare.
2097 (ia32_multipass_dfa_lookahead): Declare.
2098 (ix86_macro_fusion_p): Declare.
2099 (ix86_macro_fusion_pair_p): Declare.
2100 (ix86_bd_has_dispatch): Declare.
2101 (ix86_bd_do_dispatch): Declare.
2102 (ix86_core2i7_init_hooks): Declare.
2103 (ix86_atom_sched_reorder): Declare.
2104 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2105 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2106 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2107 (rip_relative_addr_p): Rename to ...
2108 (ix86_rip_relative_addr_p): ... this one; export.
2109 (memory_address_length): Update.
2110 (ix86_issue_rate): Move to x86-tune-sched.c.
2111 (ix86_flags_dependent): Move to x86-tune-sched.c.
2112 (ix86_agi_dependent): Move to x86-tune-sched.c.
2113 (exact_dependency_1): Move to x86-tune-sched.c.
2114 (exact_store_load_dependency): Move to x86-tune-sched.c.
2115 (ix86_adjust_cost): Move to x86-tune-sched.c.
2116 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2117 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2118 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2119 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2120 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2121 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2122 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2123 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2124 (min_insn_size): Rename to ...
2125 (ix86_min_insn_size): ... this one; export.
2126 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2127 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2128 (core2i7_first_cycle_multipass_backtrack): Move to
2129 x86-tune-sched-core.c.
2130 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2131 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2132 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2133 (ix86_avoid_jump_mispredicts): Update.
2134 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2135 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2136 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2137 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2138 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2139 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2140 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2141 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2142 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2143 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2144 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2145 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2146 (BIG): Move to ix86-tune-sched-bd.c.
2147 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2148 (enum insn_path): Move to ix86-tune-sched-bd.c.
2149 (get_mem_group): Move to ix86-tune-sched-bd.c.
2150 (is_cmp): Move to ix86-tune-sched-bd.c.
2151 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2152 (is_branch): Move to ix86-tune-sched-bd.c.
2153 (is_prefetch): Move to ix86-tune-sched-bd.c.
2154 (init_window): Move to ix86-tune-sched-bd.c.
2155 (allocate_window): Move to ix86-tune-sched-bd.c.
2156 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2157 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2158 (process_end_window): Move to ix86-tune-sched-bd.c.
2159 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2160 (find_constant): Move to ix86-tune-sched-bd.c.
2161 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2162 (has_immediate): Move to ix86-tune-sched-bd.c.
2163 (get_insn_path): Move to ix86-tune-sched-bd.c.
2164 (get_insn_group): Move to ix86-tune-sched-bd.c.
2165 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2166 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2167 (add_insn_window): Move to ix86-tune-sched-bd.c.
2168 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2169 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2170 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2171 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2172 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2173 (do_dispatch): Move to ix86-tune-sched-bd.c.
2174 (has_dispatch): Move to ix86-tune-sched-bd.c.
2175 * config/i386/t-i386: Add new object files.
2176 * config/i386/x86-tune-costs.h: New file.
2177 * config/i386/x86-tune-sched-atom.c: New file.
2178 * config/i386/x86-tune-sched-bd.c: New file.
2179 * config/i386/x86-tune-sched-core.c: New file.
2180 * config/i386/x86-tune-sched.c: New file.
2182 2017-10-11 Liu Hao <lh_mouse@126.com>
2184 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2185 the generic version below instead.
2186 (should_colorize): Recognize Windows consoles as terminals
2188 * pretty-print.c [__MINGW32__] (write_all): New function.
2189 [__MINGW32__] (find_esc_head): Likewise.
2190 [__MINGW32__] (find_esc_terminator): Likewise.
2191 [__MINGW32__] (eat_esc_sequence): Likewise.
2192 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2194 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2197 2017-10-11 Richard Biener <rguenther@suse.de>
2199 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2200 Properly call analyze_scalar_evolution with the loop of the stmt.
2202 2017-10-11 Richard Biener <rguenther@suse.de>
2204 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2205 * tree-core.h (tree_base): Add chrec_var union member.
2206 * tree.h (CHREC_VAR): Remove.
2207 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2208 * tree-chrec.h (build_polynomial_chrec): Adjust.
2209 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2210 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2212 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2214 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2215 * match.pd: ... here.
2216 ((T) X == (T) Y): Relax condition.
2218 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2220 PR tree-optimization/82472
2221 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2223 (break_alias_scc_partitions): Update postorder number.
2225 2017-10-11 Martin Liska <mliska@suse.cz>
2228 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2230 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2232 2017-10-11 Martin Liska <mliska@suse.cz>
2234 * print-rtl.c (print_insn): Move declaration of idbuf
2235 to same scope as name.
2237 2017-10-11 Martin Liska <mliska@suse.cz>
2242 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2245 2017-10-11 Martin Liska <mliska@suse.cz>
2248 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2251 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2253 * config.gcc (default_gnu_indirect_function): Default to yes for
2254 arm*-*-linux* with glibc.
2256 2017-10-11 Richard Biener <rguenther@suse.de>
2258 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2259 default-defs and types we do not want to analyze.
2260 (interpret_loop_phi): Replace unreachable code with an assert.
2261 (compute_scalar_evolution_in_loop): Remove and inline ...
2262 (analyze_scalar_evolution_1): ... here, replacing condition with
2263 what makes the intent clearer. Remove handling of cases
2264 get_scalar_evolution now handles.
2266 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2268 PR rtl-optimization/81434
2269 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2270 comment for main loop. In sched_group_found if, also add checks for
2271 pass and min_cost_group.
2273 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2275 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2276 (rs6000_insn_cost): New function.
2277 * config/rs6000/rs6000.md (cost): New attribute.
2279 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2280 H.J. Lu <hongjiu.lu@intel.com>
2284 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2285 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2286 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2287 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2288 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2289 OPTION_MASK_ISA_MMX and some other bit set require both
2290 mmx and the ISAs without the mmx bit.
2291 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2292 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2293 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2294 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2295 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2296 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2297 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2298 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2299 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2300 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2301 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2302 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2303 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2304 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2305 Add OPTION_MASK_ISA_MMX.
2307 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2309 * config.gcc (armv7*-*-freebsd*): New target.
2310 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2312 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2314 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2315 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2318 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2320 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2321 a template parameter.
2322 (WIDE_INT_REF_FOR): Update accordingly.
2323 * tree.h (wi::int_traits <const_tree>): Delete.
2324 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2325 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2326 (wi::tree_to_wide_ref): New typedef.
2327 (wi::to_wide): New function.
2328 * calls.c (get_size_range): Use wi::to_wide when operating on
2330 * cgraph.c (cgraph_node::create_thunk): Likewise.
2331 * config/i386/i386.c (ix86_data_alignment): Likewise.
2332 (ix86_local_alignment): Likewise.
2333 * dbxout.c (stabstr_O): Likewise.
2334 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2335 * expr.c (const_vector_from_tree): Likewise.
2336 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2337 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2338 (fold_negate_expr_1, int_const_binop_1, const_binop)
2339 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2340 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2341 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2342 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2343 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2344 (fold_not_const, round_up_loc): Likewise.
2345 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2346 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2347 (alloca_call_type): Likewise.
2348 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2349 * godump.c (go_output_typedef): Likewise.
2350 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2351 * internal-fn.c (get_min_precision): Likewise.
2352 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2353 * ipa-polymorphic-call.c
2354 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2355 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2356 (ipa_modify_call_arguments): Likewise.
2357 * match.pd: Likewise.
2358 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2359 * print-tree.c (print_node_brief, print_node): Likewise.
2360 * stmt.c (expand_case): Likewise.
2361 * stor-layout.c (layout_type): Likewise.
2362 * tree-affine.c (tree_to_aff_combination): Likewise.
2363 * tree-cfg.c (group_case_labels_stmt): Likewise.
2364 * tree-data-ref.c (dr_analyze_indices): Likewise.
2365 (prune_runtime_alias_test_list): Likewise.
2366 * tree-dump.c (dequeue_and_dump): Likewise.
2367 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2368 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2369 * tree-pretty-print.c (dump_generic_node): Likewise.
2370 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2371 (simple_iv_with_niters): Likewise.
2372 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2373 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2374 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2375 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2376 (refine_value_range_using_guard, number_of_iterations_ne_max)
2377 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2378 (get_cst_init_from_scev, record_nonwrapping_iv)
2379 (scev_var_range_cant_overflow): Likewise.
2380 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2381 * tree-ssa-pre.c (compute_avail): Likewise.
2382 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2383 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2384 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2385 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2386 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2387 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2388 (dump_case_nodes, try_switch_expansion): Likewise.
2389 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2390 (vect_do_peeling): Likewise.
2391 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2392 * tree-vect-stmts.c (vectorizable_load): Likewise.
2393 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2394 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2395 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2396 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2397 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2398 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2399 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2400 (evrp_dom_walker::before_dom_children): Likewise.
2401 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2402 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2403 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2404 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2405 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2406 (get_range_pos_neg): Likewise.
2407 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2408 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2409 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2410 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2411 * config/avr/avr.c (avr_fold_builtin): Likewise.
2412 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2413 * config/msp430/msp430.c (msp430_attr): Likewise.
2414 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2415 * config/powerpcspe/powerpcspe-c.c
2416 (altivec_resolve_overloaded_builtin): Likewise.
2417 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2418 (rs6000_expand_ternop_builtin): Likewise.
2419 * config/rs6000/rs6000-c.c
2420 (altivec_resolve_overloaded_builtin): Likewise.
2421 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2422 (rs6000_expand_ternop_builtin): Likewise.
2423 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2425 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2427 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2428 when copying loop nest with only one inner loop.
2430 2017-10-10 Richard Biener <rguenther@suse.de>
2432 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2433 blocks if SCEV is active.
2434 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2436 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2437 (scev_initialize): Assert we are not yet initialized.
2439 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2441 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2442 inner loop's exit stmt by making it always exit the loop, otherwise
2443 we would generate an infinite empty loop.
2445 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2447 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2448 renaming variables in new preheader if it's deleted.
2450 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2452 * tree-loop-distribution.c (struct partition): Remove unused field
2453 loops of the structure.
2454 (partition_alloc, partition_free): Ditto.
2455 (build_rdg_partition_for_vertex): Ditto.
2457 2017-10-09 Jeff Law <law@redhat.com>
2459 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2460 return type to match prototype and documentation.
2462 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2464 * config/rs6000/rs6000.c (processor_costs): Move to ...
2465 * config/rs6000/rs6000.h: ... here.
2466 (rs6000_cost): Declare.
2468 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2470 * except.c (setjmp_fn): New global variable.
2471 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2472 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2473 if DONT_USE_BUILTIN_SETJMP is defined.
2475 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2477 * target.def (insn_cost): New hook.
2478 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2479 * doc/tm.texi: Regenerate.
2480 * rtlanal.c (insn_cost): Use the new hook.
2482 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2484 * combine.c (combine_validate_cost): Compute the new insn_cost,
2485 not just pattern_cost.
2486 (try_combine): Adjust comment.
2488 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2490 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2492 * combine.c (uid_insn_cost): Adjust comment.
2493 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2495 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2496 * dse.c (find_shift_sequence): Ditto.
2497 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2498 (bb_valid_for_noce_process_p): Use pattern_cost.
2499 * rtl.h (insn_rtx_cost): Delete.
2500 (pattern_cost): New prototype.
2501 (insn_cost): New prototype.
2502 * rtlanal.c (insn_rtx_cost): Rename to...
2503 (pattern_cost): ... this.
2506 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2508 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2509 (*jcc<mode>_0_r_i387): Ditto.
2510 (*jccxf_r_i387): Ditto.
2511 (*jcc<mode>_r_i387): Ditto.
2512 (*jccu<mode>_r_i387): Ditto.
2513 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2514 (*jcc): Rename from *jcc_1.
2516 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2518 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2519 deferred rescans after the lvx/stvx recombination pre-pass.
2521 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2523 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2524 memory operation instruction support.
2525 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2526 (rs6000-ibm-aix[789]*): Likewise.
2527 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2528 Document new functions.
2530 2017-10-09 Richard Biener <rguenther@suse.de>
2532 PR tree-optimization/82397
2533 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2534 equality only for semantically equal trees.
2536 2017-10-09 Richard Biener <rguenther@suse.de>
2538 PR tree-optimization/82449
2539 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2540 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2541 allow constant addresses.
2542 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2545 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2547 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2550 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2553 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2556 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2559 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2561 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2564 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2565 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2567 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2569 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2570 (WI_BINARY_PREDICATE_RESULT): Likewise.
2571 (wi::binary_traits::operator_result): New type.
2572 (wi::binary_traits::predicate_result): Likewise.
2573 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2574 (generic_wide_int::operator==, generic_wide_int::operator!=)
2575 (generic_wide_int::operator&, generic_wide_int::and_not)
2576 (generic_wide_int::operator|, generic_wide_int::or_not)
2577 (generic_wide_int::operator^, generic_wide_int::operator+
2578 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2580 (operator~, unary operator-, operator==, operator!=, operator&)
2581 (operator|, operator^, operator+, binary operator-, operator*): New
2583 * expr.c (get_inner_reference): Use wi::bit_and_not.
2584 * fold-const.c (fold_binary_loc): Likewise.
2585 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2586 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2587 (bit_value_binop): Likewise.
2588 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2589 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2590 (extract_range_from_binary_expr_1): Likewise.
2591 (masked_increment): Likewise.
2592 (simplify_bit_ops_using_ranges): Likewise.
2594 2017-10-09 Martin Jambor <mjambor@suse.cz>
2597 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2598 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2599 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2601 (hsa_fixup_mov_insn_type): New function.
2602 (hsa_op_with_type::get_in_type): Use it.
2603 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2604 immediates in an assert.
2605 (hsa_op_with_type::extend_int_to_32bit): New method.
2606 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2607 types. Convert to dest type if necessary.
2608 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2609 (reg_for_gimple_ssa): Pass false as min32int to
2610 hsa_type_for_scalar_tree_type.
2611 (gen_hsa_addr): Fixup type when creating addresable temporary.
2612 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2613 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2614 necessary. Call hsa_fixup_mov_insn_type.
2615 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2616 extend operands and convert to dest type if necessary.
2617 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2618 to dest type if necessary.
2619 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2620 if conversion nt necessary and size matches.
2621 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2622 to dest type if necessary.
2623 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2624 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2626 (gen_hsa_clrsb): Likewise.
2627 (gen_hsa_ffs): Likewise.
2628 (gen_hsa_divmod): Extend operands and convert to dest type if
2630 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2632 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2634 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2635 Remove empty default arguments. Use a brace block as output
2637 (conditional return): Ditto.
2639 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2640 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2642 (group_ending_nop): Ditto.
2643 (doloop_end): Ditto.
2644 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2645 (splitters for those): Ditto.
2647 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2649 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2650 a conditional jump (and the compare for it) so that pc_rtx is the
2652 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2653 for the deleted and renamed ctr<mode>_internal[234] patterns.
2654 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2655 Delete second conditional return pattern.
2656 (ctr<mode>_internal2): Delete this second bdnz pattern.
2657 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2658 (ctr<mode>_internal4): Delete this second bdz pattern.
2660 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2662 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2663 (always_initialized_rtx_for_ssa_name_p): New predicate.
2664 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2665 (finish_out_of_ssa): Free new field of SA.
2666 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2667 * tree-ssa-coalesce.c: Include tree-ssa.h.
2668 (get_parm_default_def_partitions): Remove extern keyword.
2669 (get_undefined_value_partitions): New function.
2670 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2671 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2673 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2675 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2677 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2679 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2681 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2682 for targets that preffer 128bit.
2684 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2686 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2688 2017-10-08 Olivier Hainque <hainque@adacore.com>
2690 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2691 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2693 (thumb_set_return_address): Likewise.
2695 2017-10-08 Olivier Hainque <hainque@adacore.com>
2697 * common/config/arm/arm-common.c (arm_except_unwind_info):
2698 Handle DWARF2_UNWIND_INFO.
2700 2017-10-07 Michael Collison <michael.collison@arm.com>
2702 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2705 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2707 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2708 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2709 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2710 defined, force the creation of a new block for a dispatch label.
2712 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2714 * invoke.texi (Wsuggest-attribute=cold): Document.
2715 * common.opt (Wsuggest-attribute=cold): New
2716 * ipa-pure-const.c (warn_function_cold): New function.
2717 * predict.c (compute_function_frequency): Use it.
2718 * predict.h (warn_function_cold): Declare.
2720 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2722 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2725 2017-10-06 Martin Liska <mliska@suse.cz>
2727 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2728 keyword for member functions.
2729 (struct sanopt_tree_couple): New struct.
2730 (struct sanopt_tree_couple_hash): New function.
2731 (struct sanopt_ctx): Add new hash_map.
2732 (has_dominating_ubsan_ptr_check): New function.
2733 (record_ubsan_ptr_check_stmt): Likewise.
2734 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2735 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2736 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2738 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2741 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2742 aarch64_simd_valid_immediate on CONST_VECTORs.
2743 (aarch64_reg_or_bic_imm): Likewise.
2745 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2747 PR rtl-optimization/82396
2748 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2750 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2752 * graphite-dependences.c (scop_get_reads): Move code to...
2753 (scop_get_must_writes): Move code to...
2754 (scop_get_may_writes): Move code to...
2755 (scop_get_reads_and_writes): ... here.
2756 (scop_get_dependences): Call scop_get_reads_and_writes.
2758 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2760 PR tree-optimization/82434
2761 * fold-const.h (can_native_encode_type_p,
2762 can_native_encode_string_p): Remove.
2763 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2764 don't encode anything, just return what would be otherwise returned.
2765 (native_encode_fixed, native_encode_complex, native_encode_vector):
2767 (native_encode_string): Likewise. Inline by hand
2768 can_native_encode_string_p.
2769 (can_native_encode_type_p): Remove.
2770 (can_native_encode_string_p): Remove.
2771 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2772 STRING_CSTs using can_native_encode_string_p, test all
2773 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2774 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2775 argument from native_encode_expr.
2776 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2777 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2780 2017-10-06 Richard Biener <rguenther@suse.de>
2782 PR tree-optimization/82397
2783 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2784 operand_equal_p but rely on data_ref_compare_tree for detecting
2786 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2787 to match up with dr_group_sort_cmp.
2789 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2792 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2794 * config/s390/s390-builtin-types.def: Regenerate.
2796 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2799 * config/s390/s390-builtin-types.def: Regenerate.
2800 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2801 Change flag from B_VXE to B_VX.
2802 (s390_vec_min_dbl): Remove B_VXE flag.
2804 2017-10-06 Richard Biener <rguenther@suse.de>
2806 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2807 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2808 translate_isl_ast_to_gimple::is_valid_rename,
2809 translate_isl_ast_to_gimple::get_rename,
2810 translate_isl_ast_to_gimple::get_def_bb_for_const,
2811 translate_isl_ast_to_gimple::get_new_name,
2812 translate_isl_ast_to_gimple::collect_all_ssa_names,
2813 translate_isl_ast_to_gimple::copy_loop_phi_args,
2814 translate_isl_ast_to_gimple::collect_all_ssa_names,
2815 translate_isl_ast_to_gimple::copy_loop_phi_args,
2816 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2817 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2818 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2819 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2820 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2821 translate_isl_ast_to_gimple::copy_cond_phi_args,
2822 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2823 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2824 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2825 translate_isl_ast_to_gimple::rename_uses,
2826 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2827 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2828 (set_rename_for_each_def): Likewise.
2829 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2830 here. Handle rewriting SCEV analyzable uses here.
2831 (copy_bb_and_scalar_dependences): Generate code for PHI
2833 (graphite_regenerate_ast_isl): Adjust.
2834 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2835 (add_write, add_read): New functions.
2836 (build_cross_bb_scalars_def): Use it and simplify.
2837 (build_cross_bb_scalars_use): Likewise.
2838 (graphite_find_cross_bb_scalar_vars): Inline into...
2839 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2840 simulating out-of-SSA. Compute liveout and add dependencies.
2841 (build_scops): Force an empty entry block.
2842 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2844 (sese_build_liveouts): Declare.
2845 (sese_trivially_empty_bb_p): Likewise.
2846 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2847 compute liveout and debug_liveout.
2848 (sese_bad_liveouts_use): Remove.
2849 (sese_reset_debug_liveouts_bb): Likewise.
2850 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2851 (sese_build_liveouts): Build liveout and debug_liveout and store
2853 (new_sese_info): Adjust.
2854 (free_sese_info): Likewise.
2855 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2856 do not build liveout here.
2857 (move_sese_in_condition): Adjust region entry.
2858 (scev_analyzable_p): Match up with chrec_apply requirements.
2859 (sese_trivially_empty_bb_p): New.
2860 * tree-into-ssa.c (get_reaching_def): Properly support generating
2861 default-defs for incremental rewrite of anonymous names.
2863 2017-10-06 Richard Biener <rguenther@suse.de>
2865 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2866 precision do not perform modulo reduction.
2868 2017-10-06 Richard Biener <rguenther@suse.de>
2870 PR tree-optimization/82436
2871 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2872 conservatively choose the vectorization factor when checking
2873 whether we can perform the required load permutation.
2874 (vect_transform_slp_perm_load): Assert when we may not fail.
2876 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2878 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2879 message for incompatible -msdata=* and -mcall-* options.
2881 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2883 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2884 rate for post-reload scheduling.
2886 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2888 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2890 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2892 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2893 to improve monte carlo in scimark.
2895 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2897 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2898 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2899 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2900 pentium4_cost, nocona_cost): Set reassociation width to 1.
2901 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2902 width to 2 for fp operations and 1 otherwise.
2903 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2905 (atom_cost): Set reassociation width to 2.
2906 (slm_cost, generic_cost): Set fp reassociation width
2907 to 2 and 1 otherwise.
2908 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2909 (core_cost): Set fp reassociation width to 4 and vector to 2.
2910 (ix86_reassociation_width): Rewrite using cost table; special case
2911 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2912 and TARGET_AVX128_OPTIMAL.
2913 * config/i386/i386.h (processor_costs): Add
2914 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2915 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2916 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2917 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2918 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2919 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2921 2017-10-05 Nathan Sidwell <nathan@acm.org>
2923 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2925 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2927 * config/arm/arm.c (arm_test_fpu_data): New.
2928 (arm_run_selftests): Call arm_test_fpu_data.
2930 2017-10-04 Nathan Sidwell <nathan@acm.org>
2932 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2934 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2936 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2938 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2939 check type for aarch64_simd_valid_immediate.
2940 (aarch64_output_simd_mov_immediate): Update prototype.
2941 (aarch64_simd_valid_immediate): Update prototype.
2942 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2943 support for ORR-immediate.
2944 (and<mode>3): modified pattern to add support for BIC-immediate.
2945 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2946 now checks for valid immediate for BIC and ORR based on new enum
2948 (aarch64_output_simd_mov_immediate): Function now used to output
2949 BIC/ORR imm as well based on new enum argument.
2950 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2952 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2953 (aarch64_reg_or_bic_imm): Likewise.
2955 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2957 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2958 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2960 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2964 PR rtl-optimization/82396
2965 * haifa-sched.c (autopref_multipass_init): Simplify
2967 (autopref_rank_data): Simplify sort order.
2968 * sched-int.h (autopref_multipass_data_): Remove
2969 multi_mem_insn_p, min_offset and max_offset.
2971 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2973 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2975 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2977 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2980 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2982 * doc/sourcebuild.texi: Document vect_long_mult.
2984 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2986 PR tree-optimization/82413
2987 * fold-const.c (build_range_check): Use widest_int when comparing
2988 the maximum ETYPE value with HIGH.
2990 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2992 PR rtl-optimization/82396
2993 * haifa-sched.c (autopref_multipass_init): Simplify
2995 (autopref_rank_data): Simplify sort order.
2996 * sched-int.h (autopref_multipass_data_): Remove
2997 multi_mem_insn_p, min_offset and max_offset.
2999 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3001 PR tree-optimization/82381
3002 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3003 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3006 PR tree-optimization/82374
3007 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3008 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3009 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3010 current_function_decl to the new decl.
3012 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3014 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3015 helper macro for IEEE float128 hardware built-in functions.
3016 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3018 (TRUNCF128_ODD): Likewise.
3019 (ADDF128_ODD): Likewise.
3020 (SUBF128_ODD): Likewise.
3021 (MULF128_ODD): Likewise.
3022 (DIVF128_ODD): Likewise.
3023 (FMAF128_ODD): Likewise.
3024 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3025 UNSPEC_TRUNC_ROUND_TO_ODD.
3026 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3027 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3028 floating point round to odd instructions.
3029 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3030 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3031 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3032 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3033 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3034 (trunc<mode>sf2_hw): Change the truncate with round to odd
3035 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3036 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3037 to odd hardware instructions.
3038 (sub<mode>3_odd): Likewise.
3039 (mul<mode>3_odd): Likewise.
3040 (div<mode>3_odd): Likewise.
3041 (sqrt<mode>2_odd): Likewise.
3042 (fma<mode>4_odd): Likewise.
3043 (fms<mode>4_odd): Likewise.
3044 (nfma<mode>4_odd): Likewise.
3045 (nfms<mode>4_odd): Likewise.
3046 (trunc<mode>df2_odd): Change the truncate with round to odd
3047 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3049 * doc/extend.texi (PowerPC built-in functions): Update documentation
3050 for existing IEEE float128-bit built-in functions. Add built-in
3051 functions that generate the IEEE 128-bit floating point round to
3054 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3056 PR rtl-optimization/77729
3057 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3058 to (X&(C1&~C2))|C2 transformations.
3060 2017-10-03 Martin Jambor <mjambor@suse.cz>
3062 PR tree-optimization/82363
3063 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3064 mismatch, mark lacc written regardless of racc.
3066 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3068 PR tree-optimization/82381
3069 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3070 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3071 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3072 fallthrough into reassoc_stmt_dominates_stmt_p.
3075 * combine.c (combine_instructions): Don't combine in unreachable
3078 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3082 function to not use the have_cpu variable. Do not set cpu_index,
3083 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3085 (rs6000_valid_attribute_p): Remove duplicate initializations of
3086 old_optimize and func_optimize.
3087 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3088 (rs6000_activate_target_options): Make global.
3089 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3092 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3094 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3095 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3096 Set *poffset to 0 and *psize and *pmax_size to -1 if
3097 *poffset + *psize overflows in HOST_WIDE_INT.
3099 PR tree-optimization/82387
3100 PR tree-optimization/82388
3101 PR tree-optimization/82389
3102 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3103 instead of live_bytes non-NULL.
3105 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3108 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3111 2017-10-02 Richard Biener <rguenther@suse.de>
3113 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3114 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3115 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3117 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3119 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3120 requested precision matches the type's.
3121 * calls.c (alloc_max_size): Calculate the new candidate size as
3122 a widest_int and use wi::to_widest when comparing it with the
3123 current candidate size.
3124 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3125 zero rather than integer_zero_node.
3126 * match.pd: Check for a no-op conversion before using wi::add
3127 rather than after. Use tree_to_uhwi when summing small shift
3128 counts into an unsigned int.
3130 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3131 Alan Hayward <alan.hayward@arm.com>
3132 David Sherwood <david.sherwood@arm.com>
3135 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3136 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3137 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3138 POINTER_AND_FP_REGS.
3140 2017-10-02 Richard Biener <rguenther@suse.de>
3142 PR tree-optimization/82355
3143 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3144 a mapping for the enclosing loop but avoid generating one for
3146 (copy_bb_and_scalar_dependences): Remove premature codegen
3147 error on PHIs in blocks duplicated into multiple places.
3148 * graphite-scop-detection.c
3149 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3150 in the region use it as loop and nest to analyze the DR in.
3151 (try_generate_gimple_bb): Likewise.
3152 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3153 (add_loop_constraints): For blocks in a loop not in the region
3154 create a dimension with a single iteration.
3155 * sese.h (gbb_loop_at_index): Remove assert.
3157 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3159 * omp-expand.c (adjust_context_scope): New function.
3160 (expand_parallel_call): Call adjust_context_scope.
3162 2017-10-01 Jeff Law <law@redhat.com>
3164 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3165 dom_opt_dom_walker class with direct access to private members.
3166 Add comments. Call test_for_singularity.
3167 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3168 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3169 m_dummy_cond anymore.
3170 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3172 (pass_dominator:execute): Build the dummy_cond here and pass it
3173 to the dom_opt_dom_walker ctor.
3174 (test_for_singularity): New function.
3176 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3177 Maya Rashish <coypu@sdf.org>
3179 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3180 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3181 (alpha*-*-netbsd*) Use nbsd_tm_file.
3182 (arm*-*-netbsdelf*) Likewise.
3183 (i[34567]86-*-netbsdelf*) Likewise.
3184 (x86_64-*-netbsd*) Likewise.
3185 (mips*-*-netbsd*) Likewise.
3186 (powerpc-*-netbsd*) Likewise.
3187 (sh*-*-netbsd*) Likewise.
3188 (sparc-*-netbsdelf*) Likewise.
3189 (sparc64-*-netbsd*) Likewise.
3190 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3192 (vax-*-netbsdelf*) Likewise.
3193 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3194 (UINT_FAST8_TYPE) Likewise.
3195 (INT_FAST16_TYPE) Check CHAR_FAST16.
3196 (UINT_FAST16_TYPE) Likewise.
3198 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3201 * config/i386/i386.md
3202 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3203 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3204 *divmodsi4_zext_2): New define_insn_and_split.
3205 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3206 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3207 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3208 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3209 New define_insn_and_split.
3210 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3211 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3212 operands[1] having DImode when mode is SImode.
3214 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3215 always SImode for DIV and MOD in REG_EQUAL notes.
3217 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3220 * match.pd: Fix handling of NaNs in pattern.
3222 2017-09-29 Jeff Law <law@redhat.com>
3224 * sbitmap.c (bitmap_bit_in_range_p): New function.
3225 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3226 * tree-ssa-dse.c (live_bytes_read): New function.
3227 (dse_classify_store): Ignore reads of dead bytes.
3229 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3230 typos and whitespace errors.
3231 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3232 * config/s390/s390.c (s390_emit_prologue): Likewise.
3234 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3237 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3238 with a symbol for LRA.
3240 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3242 PR rtl-optimization/82338
3243 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3245 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3247 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3248 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3249 (qsort_chk): Declare.
3250 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3251 (qsort_chk): New function.
3253 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3255 PR tree-optimization/82337
3256 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3257 phi definition if the PHI result appears in an abnormal PHI.
3258 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3259 basis appears in an abnormal PHI.
3261 2017-09-29 Richard Biener <rguenther@suse.de>
3263 * graphite-isl-ast-to-gimple.c
3264 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3265 (binary_op_to_tree): Use it.
3266 (get_rename_from_scev): Likewise.
3267 (copy_loop_phi_nodes): Likewise.
3268 (copy_bb_and_scalar_dependences): Likewise.
3269 (translate_pending_phi_nodes): Likewise.
3271 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3274 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3275 for movabsq $(i32 << shift), r64.
3277 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3279 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3280 index when encoding %esp as %rsp to avoid 0x67 prefix.
3282 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3284 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3285 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3287 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3289 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3290 Extensions with more than 16 double VFP registers.
3291 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3292 to_clear_mask and all code related to it. Replace the remaining
3293 entry by a sbitmap and adapt code accordingly.
3295 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3297 * brig-builtins.def: Change pure attributes to const.
3299 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3301 * config.gcc (default_gnu_indirect_function): Default to yes for
3302 sparc*-*-linux* with glibc.
3304 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3306 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3307 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3308 when creating .init_array and .fini_array sections with priority
3311 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3314 * config/aarch64/aarch64.c
3315 (aarch64_builtin_support_vector_misalignment): Always return false
3316 when misalignment is unknown.
3318 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3320 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3321 this function to return false if the definition used by the swap
3322 instruction is artificial, or if the memory address from which the
3323 constant value is loaded is not represented by a base address held
3324 in a register or if the base address register is a frame or stack
3325 pointer. Additionally, return false if the base address of the
3326 loaded constant is a SYMBOL_REF but is not considered to be a
3328 (replace_swapped_load_constant): New function.
3329 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3330 loaded constant vector with a load of a swapped constant vector.
3332 2017-09-27 Carl Love <cel@us.ibm.com>
3334 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3335 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3336 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3339 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3341 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3342 first, always call autopref_rank_data otherwise.
3344 2017-09-27 Richard Biener <rguenther@suse.de>
3346 * graphite-scop-detection.c (find_scop_parameters): Move
3347 loop bound handling ...
3348 (gather_bbs::before_dom_children): ... here, avoiding the need
3349 to build scop_info->loop_nest.
3350 (record_loop_in_sese): Remove.
3351 * sese.h (sese_info_t::loop_nest): Remove.
3352 * sese.c (new_sese_info): Do not allocate loop_nest.
3353 (free_sese_info): Do not free loop_nest.
3355 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3358 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3359 lhs from calls if the lhs has addressable type.
3361 2017-09-27 Richard Biener <rguenther@suse.de>
3363 * graphite.h (scop::max_alias_set): New member.
3364 * graphite-scop-detection.c: Remove references to non-existing
3365 --param in comments.
3366 (build_alias_sets): Record the maximum alias set used for drs.
3367 (build_scops): Support zero as unlimited for
3368 --param graphite-max-arrays-per-scop.
3369 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3371 (build_poly_sr_1): ... here. Compute alias set based on the
3372 maximum alias set used for drs rather than
3373 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3375 2017-09-27 Richard Biener <rguenther@suse.de>
3377 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3378 --param loop-block-tile-size=0 to disable tiling.
3380 2017-09-27 Richard Biener <rguenther@suse.de>
3382 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3383 (graphite-max-nb-scop-params): Document special value zero.
3384 * domwalk.h (dom_walker::STOP): New symbolical constant.
3385 (dom_walker::dom_walker): Add optional parameter for bb to
3387 (dom_walker::~dom_walker): Declare.
3388 (dom_walker::before_dom_children): Document STOP return value.
3389 (dom_walker::m_user_bb_to_rpo): New member.
3390 (dom_walker::m_bb_to_rpo): Likewise.
3391 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3392 mapping here if not provided by the user.
3393 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3394 provided by the user.
3395 (dom_walker::STOP): Define.
3396 (dom_walker::walk): Do not compute bb to RPO mapping here.
3397 Support STOP return value from before_dom_children to stop
3399 * graphite-optimize-isl.c (optimize_isl): If the schedule
3400 is the same still generate code if -fgraphite-identity
3401 or -floop-parallelize-all are given.
3402 * graphite-scop-detection.c: Include cfganal.h.
3403 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3405 (gather_bbs::before_dom_children): Return STOP for BBs
3407 (build_scops): Compute bb to RPO mapping and pass it to
3408 the domwalk. Treat --param graphite-max-nb-scop-params=0
3409 as not limiting the number of params.
3410 * graphite.c (graphite_initialize): Remove limit on the
3411 number of basic-blocks in a function.
3412 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3413 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3414 default value of 10.
3416 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3418 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3419 Adjust code to eliminate needing to do the shift right 32-bits
3420 operation after XSCVDPSPN.
3422 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3424 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3425 ((X / Y) != 0 -> X >= Y): Likewise.
3427 2017-09-26 Carl Love <cel@us.ibm.com>
3429 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3430 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3431 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3432 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3433 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3434 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3435 definitions and overloading.
3436 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3437 statement for P9V_BUILTIN_XST_LEN_R.
3438 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3439 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3440 define_expand and define_insn for the instructions and builtins.
3441 * doc/extend.texi: Update the built-in documentation file for the new
3443 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3444 define_insn for the instructions
3446 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3449 * gcc/config/netbsd-protos.h: New file.
3450 * gcc/config/netbsd.c: New file.
3451 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3452 * gcc/config/t-netbsd: New file.
3453 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3454 (tmake_file) Add t-netbsd.
3455 (extra_objs) Add netbsd.o.
3457 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3461 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3463 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3465 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3466 sign extension from a vector register to a GPR by doing a 32-bit
3467 direct move and then an EXTSW.
3468 (extendsi<mode>2 splitter): Likewise.
3469 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3470 right or vector extract after doing XSCVDPSPN. Use
3471 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3473 (movdi_from_sf_zero_ext): Likewise.
3474 (reload_gpr_from_vsxsf): Likewise.
3475 (p8_mfvsrd_4_disf): Delete, no longer used.
3476 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3477 then needing to move the SFmode to a GPR to use the XSCVDPSP
3478 instruction instead of FRSP and XSCVDPSPN.
3479 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3480 it is adjacent to the other XSCVSPDP insns.
3481 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3482 SFmode to be in traditional Altivec registers.
3483 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3484 (vsx_xscvspdpn): Likewise.
3485 (vsx_xscvspdpn_scalar): Likewise.
3487 2017-09-26 Martin Jambor <mjambor@suse.cz>
3489 * tree-sra.c (compare_access_positions): Put integral types first,
3490 stabilize sorting of integral types, remove conditions putting
3491 non-full-precision integers last.
3492 (sort_and_splice_var_accesses): Disable scalarization if a
3493 non-integert would be represented by a non-full-precision integer.
3495 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3497 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3498 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3499 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3500 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3501 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3502 conditionals inside the function instead of around it. Call
3503 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3504 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3506 2017-09-26 Richard Biener <rguenther@suse.de>
3508 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3510 (scop_detection::build_scop_breadth): ... this. Removed.
3511 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3512 (scop_detection::harmful_stmt_in_bb): Likewise.
3513 (scop_detection::graphite_can_represent_stmt): Likewise.
3514 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3515 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3516 (scop_detection::can_represent_loop_1): ... this. Removed.
3517 (scop_detection::harmful_loop_in_region): Simplify after inlining
3518 the above and remove more quadraticness.
3519 (build_scops): Adjust.
3520 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3523 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3526 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3527 REGNO (base) == SP_REG if base is a REG.
3530 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3531 if it is different SSA_NAME.
3532 (optimize_range_tests_cmp_bitwise): New function.
3533 (optimize_range_tests): Call it.
3535 2017-09-26 Richard Biener <rguenther@suse.de>
3537 PR tree-optimization/82321
3538 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3539 for the def being inside the loop.
3541 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3543 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3545 * config/s390/s390-builtins.def: Fix constraint on op4.
3547 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3549 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3550 independent expanders.
3551 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3552 ("vec_ordered", "vec_unordered"): New expanders.
3554 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3556 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3559 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3561 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3562 vec_unpacks_lo_v16qi.
3563 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3565 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3567 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3568 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3569 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3571 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3573 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3575 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3577 ("*vec_slb<mode>"): New insn pattern.
3578 ("vec_shr_<mode>"): New expander.
3579 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3580 and force the shift count operand to V16QImode.
3581 ("vec_srb<mode>"): Set shift count mode to V16QI.
3583 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3585 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3586 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3587 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3589 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3592 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3594 2017-09-26 Richard Biener <rguenther@suse.de>
3596 PR tree-optimization/82320
3597 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3600 2017-09-25 Jeff Law <law@redhat.com>
3602 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3603 prototype for new argument.
3604 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3605 mostly extracted from rs6000_emit_allocate_stack.
3606 (rs6000_emit_probe_stack_range_stack_clash): New function.
3607 (rs6000_emit_allocate_stack): Call
3608 rs6000_emit_probe_stack_range_stack_clash as needed.
3609 (rs6000_emit_probe_stack_range): Add additional argument
3610 to call to gen_probe_stack_range{si,di}.
3611 (output_probe_stack_range): New.
3612 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3613 (output_probe_stack_range_stack_clash): New.
3614 (rs6000_emit_prologue): Emit notes into dump file as requested.
3615 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3616 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3617 Add additional operand and pass it to output_probe_stack_range.
3619 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3621 PR tree-optimization/82163
3622 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3623 (checking_verify_loop_closed_ssa): New parameter.
3624 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3625 (check_loop_closed_ssa_stmt): Delete.
3626 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3627 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3628 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3631 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3633 * brig-builtins.def: Treat HSAIL barrier builtins as
3634 setjmp/longjump style functions.
3636 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3638 * target.def (constant_alignment): New hook.
3639 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3640 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3641 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3642 * doc/tm.texi: Regenerate.
3643 * targhooks.h (default_constant_alignment): Declare.
3644 (constant_alignment_word_strings): Likewise.
3645 * targhooks.c (default_constant_alignment): New function.
3646 (constant_alignment_word_strings): Likewise.
3647 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3648 instead of CONSTANT_ALIGNMENT.
3649 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3650 (force_const_mem): Likewise.
3651 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3652 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3653 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3654 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3655 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3657 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3658 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3659 constant_alignment_word_strings.
3660 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3661 (CONSTANT_ALIGNMENT): Likewise.
3662 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3663 (arm_constant_alignment): New function.
3664 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3665 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3666 constant_alignment_word_strings.
3667 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3668 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3669 constant_alignment_word_strings.
3670 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3671 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3672 (cris_constant_alignment): New function.
3673 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3674 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3675 (epiphany_constant_alignment): New function.
3676 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3677 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3678 constant_alignment_word_strings.
3679 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3680 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3681 constant_alignment_word_strings.
3682 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3683 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3684 constant_alignment_word_strings.
3685 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3686 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3687 * config/i386/i386.c (ix86_constant_alignment): Make static.
3688 Use the same interface as the target hook.
3689 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3690 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3691 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3692 constant_alignment_word_strings.
3693 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3694 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3695 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3696 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3697 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3698 constant_alignment_word_strings.
3699 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3700 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3701 constant_alignment_word_strings.
3702 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3703 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3704 constant_alignment_word_strings.
3705 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3706 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3708 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3709 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3710 * config/mips/mips.c (mips_constant_alignment): New function.
3711 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3712 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3713 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3714 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3715 (mmix_constant_alignment): Make static. Use the same interface
3717 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3718 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3719 constant_alignment_word_strings.
3720 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3721 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3722 constant_alignment_word_strings.
3723 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3724 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3725 constant_alignment_word_strings.
3726 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3727 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3728 (rs6000_constant_alignment): New function.
3729 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3730 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3731 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3732 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3733 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3734 (rs6000_constant_alignment): New function.
3735 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3736 * config/s390/s390.c (s390_constant_alignment): New function.
3737 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3738 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3739 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3740 constant_alignment_word_strings.
3741 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3742 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3743 (sparc_constant_alignment): New function.
3744 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3745 * config/spu/spu.c (spu_constant_alignment): New function.
3746 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3747 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3748 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3749 constant_alignment_word_strings.
3750 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3751 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3752 constant_alignment_word_strings.
3753 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3754 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3755 constant_alignment_word_strings.
3756 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3757 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3758 (visium_constant_alignment): New function.
3759 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3760 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3761 (xtensa_constant_alignment): New function.
3762 * system.h (CONSTANT_ALIGNMENT): Poison.
3764 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3767 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3768 (rs6000_builtin_valid_without_lhs): New helper function.
3769 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3770 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3772 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3774 * target.h (vec_perm_indices): Use unsigned short rather than
3776 (auto_vec_perm_indices): Likewise.
3777 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3778 Use unsigned int rather than unsigned char.
3779 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3781 2017-09-25 Richard Biener <rguenther@suse.de>
3783 * cfgloop.h (sort_sibling_loops): Declare.
3784 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3785 (sort_sibling_loops): New function sorting the sibling loop list
3787 * graphite.c (graphite_transform_loops): Sort sibling loops.
3789 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3791 * target.def (vec_perm_const_ok): Change sel parameter to
3793 * optabs-query.c (can_vec_perm_p): Update accordingly.
3794 * doc/tm.texi: Regenerate.
3795 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3796 auto_vec_perm_indices and remove separate nelt field.
3797 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3798 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3799 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3800 (aarch64_expand_vec_perm_const): Update accordingly.
3801 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3802 to vec_perm_indices.
3803 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3804 auto_vec_perm_indices and remove separate nelt field.
3805 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3806 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3807 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3809 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3810 to vec_perm_indices.
3811 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3812 sel to vec_perm_indices.
3813 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3814 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3815 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3817 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3820 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3823 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3824 on the FUNCTION_DECL function context if it has a DIE that is a
3827 2017-09-25 Richard Biener <rguenther@suse.de>
3829 PR tree-optimization/82285
3830 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3833 2017-09-25 Tom de Vries <tom@codesourcery.com>
3837 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3840 2017-09-25 Richard Biener <rguenther@suse.de>
3842 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3843 ISL errors other than isl_error_quota happen. Dump if the
3844 schedule is the same.
3845 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3846 errors instead of aborting inside ISL.
3848 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3851 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3852 of libgcc_eh for m64.
3853 * config/i386/darwin64.h: Likewise.
3855 2017-09-25 Richard Biener <rguenther@suse.de>
3858 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3859 attribute for incomplete types nor twice for complete ones.
3861 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3864 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3865 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3868 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3871 * config/i386/i386.opt (mprefer-avx256): Use
3872 ix86_target_flags variable.
3873 * config/i386/i386.c (ix86_target_string): Move
3874 -mprefer-avx256 to flag2_opts.
3876 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3879 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3880 and x != -1 | y != -1 into (x & y) != -1.
3882 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3884 * config.gcc: Add new case statement to set
3885 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3886 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3887 s390x-*-linux* case statements. Added aarch64 to the list of
3888 supported architectures.
3890 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3892 PR tree-optimization/82289
3893 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3894 STMT_VINFO_RELEVANT_P.
3896 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3897 Alan Hayward <alan.hayward@arm.com>
3898 David Sherwood <david.sherwood@arm.com>
3900 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3901 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3902 (extract_range_from_binary_expr_1): Don't call
3903 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3905 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3906 Alan Hayward <alan.hayward@arm.com>
3907 David Sherwood <david.sherwood@arm.com>
3909 * target.def (preferred_vector_alignment): New hook.
3910 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3912 * doc/tm.texi: Regenerate.
3913 * targhooks.h (default_preferred_vector_alignment): Declare.
3914 * targhooks.c (default_preferred_vector_alignment): New function.
3915 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3917 (DR_TARGET_ALIGNMENT): New macro.
3918 (aligned_access_p): Update commentary.
3919 (vect_known_alignment_in_bytes): New function.
3920 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3922 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3923 Calculate the misalignment based on the target alignment rather than
3925 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3926 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3927 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3928 the target alignment, rather than masking the element misalignment
3929 with the number of elements in a vector. Also use the target
3930 alignment when calculating the maximum number of peels.
3931 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3932 instead of TYPE_ALIGN_UNIT.
3933 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3934 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3935 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3936 (vect_create_data_ref_ptr): Likewise.
3937 (vect_setup_realignment): Realign by ANDing with
3938 -DR_TARGET_MISALIGNMENT.
3939 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3940 the number of peels based on DR_TARGET_ALIGNMENT.
3941 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3942 with the guaranteed alignment boundary when deciding whether
3944 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3945 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3946 (ensure_base_align): Remove stmt_info parameter. Get the
3947 target base alignment from DR_TARGET_ALIGNMENT.
3948 (vectorizable_store): Update call accordingly. Interpret
3949 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3951 (vectorizable_load): Likewise.
3953 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3954 Alan Hayward <alan.hayward@arm.com>
3955 David Sherwood <david.sherwood@arm.com>
3957 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3958 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3959 (vect_enhance_data_refs_alignment): Likewise.
3961 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3963 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3964 error. Only quit immediately if parsing is complete.
3965 (BEGIN): Initialize fatal_err and parse_done.
3966 (begin fpu, end fpu): Check number of arguments.
3967 (begin arch, end arch): Likewise.
3968 (begin cpu, end cpu): Likewise.
3969 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3970 (optalias): Likewise.
3972 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3974 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3975 * config/arm/arm-isa.h: Delete. Move definitions to ...
3976 * arm-cpus.in: ... here. Use new feature and fgroup values.
3977 * config/arm/arm.c (arm_option_override): Use lower case for feature
3979 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3980 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3981 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3983 (print_isa_bits_for): New function.
3984 (gen_isa): New function.
3985 (gen_comm_data): Use print_isa_bits_for.
3986 (define feature): New keyword.
3987 (define fgroup): New keyword.
3988 * config/arm/t-arm (TM_H): Remove.
3989 (GTM_H): Add arm-isa.h.
3990 (arm-isa.h): Add rule to generate file.
3991 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3992 case for feature bit names.
3994 2017-09-22 Richard Biener <rguenther@suse.de>
3996 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3998 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3999 print of no dependency loops ...
4000 * graphite.c (graphite_transform_loops): ... here.
4001 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4003 (same_close_phi_node, remove_duplicate_close_phi,
4004 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4005 (canonicalize_loop_closed_ssa): ... here and simplify.
4006 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4007 (optimize_isl): Use dump_printf_loc to tell when we stopped
4008 optimizing because of an ISL timeout.
4010 2017-09-22 Richard Biener <rguenther@suse.de>
4012 PR tree-optimization/82291
4013 * tree-if-conv.c (predicate_mem_writes): Make sure to
4014 remove writes in blocks predicated with false.
4016 2017-09-22 Richard Biener <rguenther@suse.de>
4018 * sese.c: Include cfganal.h.
4019 (if_region_set_false_region): Remove.
4020 (create_if_region_on_edge): Likewise.
4021 (move_sese_in_condition): Re-implement without destroying
4024 2017-09-22 Richard Biener <rguenther@suse.de>
4026 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4027 Verify both BBs contain loop PHI nodes before dispatching to
4029 (graphite_regenerate_ast_isl): Do not recompute dominators,
4030 do not verify three times. Restructure for clarity.
4031 * graphite-scop-detection.c (same_close_phi_node,
4032 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4033 defined_in_loop_p, canonicalize_loop_closed_ssa,
4034 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4035 checking and SSA rewrite, move to ...
4036 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4037 (graphite_initialize): Do not pass in ctx, do not reset the
4038 SCEV cache, compute only dominators.
4039 (graphite_transform_loops): Allocate ISL ctx after
4040 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4041 Maintain post-dominators only around build_scops.
4042 * sese.c (if_region_set_false_region): Make static. Free
4043 and recompute dominators.
4044 (move_sese_in_condition): Assert we don't get called with
4045 post-dominators computed.
4046 * sese.h (if_region_set_false_region): Remove.
4048 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4050 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4051 mode attribute for TARGET_AVX512VL.
4053 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4055 * config/i386/i386.opt (mprefer-avx256): New option.
4056 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4058 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4059 for TARGET_PREFER_AVX256.
4060 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4062 2017-09-21 Jeff Law <law@redhat.com>
4064 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4065 Fix dump output if the only stack space is for pushed registers.
4067 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4069 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4072 2017-09-21 Martin Sebor <msebor@redhat.com>
4075 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4076 code (in C++) or code that triggers warnings.
4078 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4080 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4082 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4084 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4085 * sched-rgn.c: Ditto.
4086 * sel-sched-ir.c: Ditto.
4088 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4090 * toplev.h (set_random_seed): Adjust return type.
4091 * toplev.c (init_local_tick): Move eager initialization of random_seed
4092 to get_random_seed. Adjust comment.
4093 (init_random_seed): Inline to get_random_seed, delete.
4094 (get_random_seed): Initialize random_seed lazily.
4095 (set_random_seed): Do not return previous value.
4096 (print_switch_value): Do not call get_random_seed.
4098 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4100 * cgraph.c (delete_function_version): New, broken out from...
4101 (cgraph_node::delete_function_version): ...here. Rename to
4102 cgraph_node::delete_function_version_by_decl. Update all uses.
4103 (cgraph_node::remove): Call delete_function_version.
4105 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4108 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4109 VAR_DECLs to which addressable non-volatile parameters are mapped
4110 and for id->retvar after the return value assignment. Clear
4111 id->retval and id->retbnd after inlining.
4113 2017-09-21 Richard Biener <rguenther@suse.de>
4115 PR tree-optimization/82276
4116 PR tree-optimization/82244
4117 * tree-vrp.c (build_assert_expr_for): Set
4118 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4120 (remove_range_assertions): Revert earlier change.
4122 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4125 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4127 2017-09-21 Richard Biener <rguenther@suse.de>
4129 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4130 Restore valid IL after code generation errors.
4131 * graphite.c (graphite_transform_loops): Diagnose code
4132 generation issues as MSG_MISSED_OPTIMIZATION and continue
4133 with processing SCOPs.
4135 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4136 Alan Hayward <alan.hayward@arm.com>
4137 David Sherwood <david.sherwood@arm.com>
4139 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4140 rather than choosing between plus_constant and gen_rtx_<CODE>.
4141 * expr.c (emit_push_insn): Likewise.
4142 (expand_expr_real_2): Likewise.
4144 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4145 Alan Hayward <alan.hayward@arm.com>
4146 David Sherwood <david.sherwood@arm.com>
4148 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4150 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4151 Alan Hayward <alan.hayward@arm.com>
4152 David Sherwood <david.sherwood@arm.com>
4154 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4155 calling tree_to_uhwi.
4157 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4158 Alan Hayward <alan.hayward@arm.com>
4159 David Sherwood <david.sherwood@arm.com>
4161 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4162 INTEGER_CST rather than a negative test for ADDR_EXPR.
4164 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4165 Alan Hayward <alan.hayward@arm.com>
4166 David Sherwood <david.sherwood@arm.com>
4168 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4169 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4171 2017-09-21 Richard Biener <rguenther@suse.de>
4173 PR tree-optimization/71351
4174 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4175 graphite_create_new_loop_guard): Remove, fold remaining parts
4177 (translate_isl_ast_node_for): ... here and simplify.
4179 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4182 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4183 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4184 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4185 alternative always use QI mode, for -Os imov (=R,R) alternative
4186 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4189 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4190 Jeff Law <law@redhat.com>
4192 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4193 (allocate_stack_space): New function, partially extracted from
4195 (s390_emit_prologue): Track offset to most recent stack probe.
4196 Code to allocate space moved into allocate_stack_space.
4197 Dump actions when no stack is allocated.
4198 (s390_prologue_plus_offset): New function.
4199 (s390_emit_stack_probe): Likewise.
4201 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4203 * common.opt (Wa, Wl, Wp, g, gz=): Add
4205 (gno-column-info): Remove.
4206 (gcolumn-info): Drop RejectNegative.
4208 (gno-record-gcc-switches): Remove.
4209 (grecord-gcc-switches): Drop RejectNegative.
4210 (gno-split-dwarf): Remove.
4211 (gsplit-dwarf): Drop RejectNegative.
4212 (gno-strict-dwarf): Remove.
4213 (gstrict-dwarf): Drop RejectNegative.
4214 * config/darwin.opt (gfull, gused): Add RejectNegative.
4215 * dwarf2out.c (gen_producer_string): Drop
4216 gno-record-gcc-switches handler.
4217 * optc-gen.awk: Add g to prefixes with negative forms.
4218 * opts-common.c (remapping_prefix_p): New.
4219 (find_opt): Check it.
4220 (generate_canonical_option): Test g prefix.
4221 (option_map): Add -gno- mapping.
4222 (add_misspelling_candidates): Check remapping_prefix_p.
4224 2017-09-20 Jeff Law <law@redhat.com>
4226 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4227 thinko in stack clash protection support.
4229 * explow.c (compute_stack_clash_protection_loop_data): Use
4230 CONST_INT_P instead of explicit test. Verify object is a
4231 CONST_INT_P before looking at INTVAL.
4232 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4233 instead of explicit test.
4235 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4238 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4239 address instead of to r1 and r11.
4241 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4243 * config.gcc: Support "knm".
4244 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4245 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4247 * config/i386/i386.c (m_KNM): Define.
4248 (processor_target_table): Add "knm".
4250 (ix86_option_override_internal): Add "knm".
4251 (ix86_issue_rate): Add PROCESSOR_KNM.
4252 (ix86_adjust_cost): Ditto.
4253 (ia32_multipass_dfa_lookahead): Ditto.
4254 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4255 (fold_builtin_cpu): Add M_INTEL_KNM.
4256 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4257 (processor_type): Add PROCESSOR_KNM.
4258 * config/i386/x86-tune.def: Add m_KNM.
4259 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4261 2017-09-20 Richard Biener <rguenther@suse.de>
4263 PR tree-optimization/80213
4264 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4265 are allowed in empty BBs as well.
4266 (canonicalize_loop_closed_ssa): Also look for other complex
4268 (scop_detection::get_sese): Include the loop-closed PHI block
4270 (scop_detection::merge_sese): Remove code adding extra blocks.
4271 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4272 (build_scops): Assert the final returned scop is invalid.
4274 2017-09-20 Richard Biener <rguenther@suse.de>
4276 PR tree-optimization/82264
4277 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4279 (vn_phi_lookup): Likewise.
4280 (vn_phi_insert): Likewise.
4282 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4284 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4285 that fits into uhwi or shwi, add DW_AT_const_value regardless
4286 of early_dwarf without going through RTL, using add_AT_unsigned
4289 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4290 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4291 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4292 (DEBUG_MACRO_SECTION): Likewise.
4293 (DEBUG_LTO_MACRO_SECTION): Likewise.
4294 (DEBUG_STR_DWO_SECTION): Likewise.
4295 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4296 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4297 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4298 (DEBUG_LTO_LINE_STR_SECTION): Define.
4299 (init_sections_and_labels): Initialize debug_line_str_section
4300 variable. Initialize debug_loc_section for -gdwarf-5 to
4301 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4303 2017-09-20 Richard Biener <rguenther@suse.de>
4305 * graphite-sese-to-poly.c (extract_affine): Properly handle
4306 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4308 2017-09-20 Richard Biener <rguenther@suse.de>
4310 PR tree-optimization/81373
4311 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4312 Force SESE live-out defs to be handled even if they are
4315 2017-09-19 Jeff Law <law@redhat.com>
4317 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4318 nothing for stack adjustments with REG_STACK_CHECK.
4319 * sched-deps.c (parse_add_or_inc): Reject insns with
4320 REG_STACK_CHECK from dependency breaking.
4321 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4322 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4323 * reg-notes.def (STACK_CHECK): New note.
4325 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4326 (ix86_expand_prologue): Dump stack clash info as needed.
4327 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4329 * function.c (dump_stack_clash_frame_info): New function.
4330 * function.h (dump_stack_clash_frame_info): Prototype.
4331 (enum stack_clash_probes): New enum.
4333 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4334 flag_stack_clash_protection.
4335 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4336 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4337 (arm_frame_pointer_required): Likewise.
4338 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4339 (ia64_expand_prologue): Likewise.
4340 * config/mips/mips.c (mips_expand_prologue): Likewise.
4341 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4342 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4343 (sparc_flat_expand_prologue): Likewise.
4344 * config/spu/spu.c (spu_expand_prologue): Likewise.
4346 * explow.c: Include "params.h".
4347 (anti_adjust_stack_and_probe_stack_clash): New function.
4348 (get_stack_check_protect): Likewise.
4349 (compute_stack_clash_protection_loop_data): Likewise.
4350 (emit_stack_clash_protection_loop_start): Likewise.
4351 (emit_stack_clash_protection_loop_end): Likewise.
4352 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4353 Use anti_adjust_stack_and_probe_stack_clash.
4354 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4355 (emit_stack_clash_protection_loop_start): Likewise.
4356 (emit_stack_clash_protection_loop_end): Likewise.
4357 * rtl.h (get_stack_check_protect): Prototype.
4358 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4359 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4360 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4362 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4364 * doc/tm.texi: Rebuilt.
4365 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4366 get_stack_check_protect.
4367 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4368 * config/arm/arm.c (arm_expand_prologue): Likewise.
4369 (arm_frame_pointer_required): Likewise.
4370 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4371 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4372 * config/mips/mips.c (mips_expand_prologue): Likewise.
4373 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4374 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4375 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4376 (sparc_flat_expand_prologue): Likewise.
4378 * common.opt (-fstack-clash-protection): New option.
4379 * flag-types.h (enum stack_check_type): Note difference between
4380 -fstack-check= and -fstack-clash-protection.
4381 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4382 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4383 * toplev.c (process_options): Issue warnings/errors for cases
4384 not handled with -fstack-clash-protection.
4385 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4386 (-fstack-check): Note additional problem with -fstack-check=generic.
4387 Note that -fstack-check is primarily for Ada and refer users
4388 to -fstack-clash-protection for stack-clash-protection.
4389 Document new params for stack clash protection.
4391 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4393 * config/i386/i386.c (ix86_split_long_move): Do not handle
4394 address used for LEA in a special way.
4396 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4398 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4400 2017-09-19 Martin Sebor <msebor@redhat.com>
4403 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4404 of incompatible types.
4406 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4409 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4410 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4411 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4413 2017-09-19 Richard Biener <rguenther@suse.de>
4415 PR tree-optimization/82244
4416 * tree-vrp.c (remove_range_assertions): Do not propagate
4417 a constant to abnormals but replace the assert with a copy.
4419 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4421 PR rtl-optimization/57878
4422 PR rtl-optimization/68988
4423 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4424 avoidance test involving non_reload_pseudos. Move frequency test
4425 below the general fragmentation avoidance test.
4427 2017-09-19 Richard Biener <rguenther@suse.de>
4429 PR tree-optimization/69728
4430 * graphite-sese-to-poly.c (schedule_error): New global.
4431 (add_loop_schedule): Handle empty domain by failing the
4433 (build_original_schedule): Handle schedule_error.
4435 2017-09-19 Richard Biener <rguenther@suse.de>
4437 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4438 Do not iterate to sibling loops but only to siblings of inner
4441 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4444 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4445 that the registers are different.
4447 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4449 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4450 to processor_model and "amdfam17h" to arch_names_table.
4451 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4453 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4456 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4458 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4459 Alan Hayward <alan.hayward@arm.com>
4460 David Sherwood <david.sherwood@arm.com>
4462 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4464 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4466 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4467 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4468 vect_schedule_slp_instance.
4469 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4470 Update call to vect_slp_analyze_node_operations. Simplify return
4472 (vect_slp_analyze_bb_1): Update call accordingly.
4473 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4474 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4475 (vect_schedule_slp): Update call accordingly.
4477 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4478 Alan Hayward <alan.hayward@arm.com>
4479 David Sherwood <david.sherwood@arm.com>
4481 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4482 with types that aren't in fact scalar.
4484 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4486 * tree-vect-slp.c (vect_record_max_nunits): New function,
4488 (vect_build_slp_tree_1): ...here.
4489 (vect_build_slp_tree_2): Call it for phis too.
4491 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4493 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4494 to vect_get_vec_def_for_operand when getting the mask operand.
4496 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4497 Alan Hayward <alan.hayward@arm.com>
4498 David Sherwood <david.sherwood@arm.com>
4500 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4503 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4504 Alan Hayward <alan.hayward@arm.com>
4505 David Sherwood <david.sherwood@arm.com>
4507 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4508 calculation for vector booleans.
4510 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4511 Alan Hayward <alan.hayward@arm.com>
4512 David Sherwood <david.sherwood@arm.com>
4514 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4516 (vect_transform_stmt): ...here.
4517 (vect_analyze_stmt): Use it instead of calling
4518 vectorizable_live_operation directly.
4520 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4522 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4523 non-SIMT targets in acc vector loops.
4525 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4527 * configure.ac: Add arc and check if assembler supports gdwarf2.
4528 * configure: Regenerate.
4530 2017-09-18 Richard Biener <rguenther@suse.de>
4532 PR tree-optimization/82220
4533 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4534 epilogue niters from the min_profitable_iters compute.
4536 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4539 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4540 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4541 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4543 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4546 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4547 switching to a new text section.
4549 2017-09-18 Richard Biener <rguenther@suse.de>
4551 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4553 (build_alias_set): Reject aliases with no access function.
4555 2017-09-18 Richard Biener <rguenther@suse.de>
4557 PR tree-optimization/79622
4558 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4560 (build_cross_bb_scalars_use): Likewise.
4562 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4564 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4566 2017-09-18 Alan Modra <amodra@gmail.com>
4569 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4570 stack_pointer_rtx for count 0. Update comments. Break up
4571 large rtl expression.
4573 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4575 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4576 Increase to 20 bytes.
4577 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4578 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4579 or avx version of the stub.
4581 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4584 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4585 compute the minimum stack alignment. Also update preferred stack
4586 boundary for leaf functions.
4588 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4590 PR tree-optimization/82228
4591 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4594 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4596 * common/config/nds32/nds32-common.c
4597 (nds32_option_optimization_table): Refine formatting.
4598 (nds32_option_optimization_table): Use -fsched-pressure and
4599 -fomit-frame-pointer for specific optimization level.
4601 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4603 * config/nds32/nds32.c: Refine formatting and comments.
4604 * config/nds32/nds32.h: Likewise.
4605 * config/nds32/nds32.md: Likewise.
4606 * config/nds32/nds32-cost.c: Likewise.
4607 * config/nds32/nds32-isr.c: Likewise.
4608 * config/nds32/nds32-md-auxiliary.c: Likewise.
4609 * config/nds32/nds32-multiple.md: Likewise.
4610 * config/nds32/nds32-predicates.c: Likewise.
4612 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4613 Jakub Jelinek <jakub@redhat.com>
4615 Add support for -std=c++2a.
4616 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4618 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4620 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4623 * doc/extend.texi (Common Function Attributes): Add
4624 references to ARM, AArch64, and S/390 specific attributes.
4625 (Function Specific Option Pragmas): Add AArch64 and S/390
4626 to list of back ends that support the target pragma.
4628 2017-09-15 Nathan Sidwell <nathan@acm.org>
4630 * doc/standards.texi: Fix C++17 description. Update URLs for
4633 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4635 * common.opt (Wcast-align=strict): New warning option.
4636 * doc/invoke.texi: Document -Wcast-align=strict.
4638 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4640 * cgraph.h (cgraph_thunk_info): Add comments.
4641 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4642 assert for VIRTUAL_* arguments stricter.
4644 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4646 PR tree-optimization/71026
4647 * match.pd: Move RDIV patterns from fold-const.c
4648 * fold-const.c (distribute_real_division): Removed.
4649 (fold_binary_loc): Remove calls to distribute_real_divison.
4651 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4653 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4654 c++1z and gnu++1z as deprecated. Change other references to
4655 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4656 Change -Wc++1z-compat to -Wc++17-compat.
4657 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4658 * dwarf2out.c (highest_c_language): Handle C++17.
4659 (gen_compile_unit_die): Likewise.
4661 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4663 PR rtl-optimization/82192
4664 * combine.c (make_extraction): Don't look through non-paradoxical
4665 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4668 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4669 Alan Hayward <alan.hayward@arm.com>
4670 David Sherwood <david.sherwood@arm.com>
4672 * target.def (function_arg_offset): New hook.
4673 * targhooks.h (default_function_arg_offset): Declare.
4674 * targhooks.c (default_function_arg_offset): New function.
4675 * function.c (locate_and_pad_parm): Use
4676 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4677 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4678 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4679 * doc/tm.texi: Regenerate.
4680 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4681 * config/spu/spu.c (spu_function_arg_offset): New function.
4682 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4683 * system.h (FUNCTION_ARG_OFFSET): Poison.
4685 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayard <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * target.def (truly_noop_truncation): New hook.
4690 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4691 than TRULY_NOOP_TRUNCATION.
4692 * hooks.h (hook_bool_uint_uint_true): Declare.
4693 * hooks.c (hook_bool_uint_uint_true): New function.
4694 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4695 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4696 * doc/tm.texi: Regenerate.
4697 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4698 rather than TRULY_NOOP_TRUNCATION in comments.
4699 (simplify_comparison): Likewise.
4700 (record_truncated_value): Likewise.
4701 * expmed.c (extract_bit_field_1): Likewise.
4702 (extract_split_bit_field): Likewise.
4703 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4704 instead of TRULY_NOOP_TRUNCATION.
4705 * function.c (assign_parm_setup_block): Likewise.
4706 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4707 * rtlhooks.c: Include target.h.
4708 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4709 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4710 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4711 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4712 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4713 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4714 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4715 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4716 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4717 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4718 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4719 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4720 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4721 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4722 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4723 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4724 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4725 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4726 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4727 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4728 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4729 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4730 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4731 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4732 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4733 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4734 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4735 rather than TRULY_NOOP_TRUNCATION in comments.
4736 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4737 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4738 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4739 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4740 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4741 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4742 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4743 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4744 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4745 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4746 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4747 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4748 rather than TRULY_NOOP_TRUNCATION in comments.
4749 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4750 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4751 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4752 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4753 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4754 TRULY_NOOP_TRUNCATION condition.
4755 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4756 (TRULY_NOOP_TRUNCATION): Delete.
4757 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4758 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4759 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4760 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4761 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4762 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4763 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4764 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4765 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4766 rather than TRULY_NOOP_TRUNCATION in comments.
4767 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4768 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4769 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4770 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4771 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4772 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4774 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4777 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4778 (*cmp_ior): Likewise.
4779 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4780 (*ior_scc_scc_cmp): Likewise.
4781 (*and_scc_scc): Likewise.
4782 (*and_scc_scc_cmp): Likewise.
4784 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4785 Alan Hayard <alan.hayward@arm.com>
4786 David Sherwood <david.sherwood@arm.com>
4788 * target.def (can_change_mode_class): New hook.
4789 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4790 (hard_regno_nregs): Likewise.
4791 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4792 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4793 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4794 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4795 (LOAD_EXTEND_OP): Update accordingly.
4796 * doc/tm.texi: Regenerate.
4797 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4798 CANNOT_CHANGE_MODE_CLASS.
4799 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4800 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4801 * combine.c (simplify_set): Update accordingly.
4802 * emit-rtl.c (validate_subreg): Likewise.
4803 * recog.c (general_operand): Likewise.
4804 * regcprop.c (mode_change_ok): Likewise.
4805 * reload1.c (choose_reload_regs): Likewise.
4806 (inherit_piecemeal_p): Likewise.
4807 * rtlanal.c (simplify_subreg_regno): Likewise.
4808 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4809 instead of CANNOT_CHANGE_MODE_CLASS.
4810 (reload_cse_simplify_operands): Likewise.
4811 * reload.c (push_reload): Use targetm.can_change_mode_class
4812 instead of CANNOT_CHANGE_MODE_CLASS.
4813 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4814 REG_CANNOT_CHANGE_MODE_P.
4815 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4816 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4817 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4818 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4819 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4820 (arm_can_change_mode_class): New function.
4821 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4822 than CANNOT_CHANGE_MODE_CLASS in comments.
4823 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4824 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4825 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4826 (ix86_can_change_mode_class): ...this new function, inverting the
4827 sense of the return value.
4828 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4829 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4830 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4831 (ia64_can_change_mode_class): New function.
4832 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4833 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4834 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4835 (m32c_can_change_mode_class): ...this new function, inverting the
4836 sense of the return value.
4837 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4838 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4839 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4840 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4841 (mips_can_change_mode_class): ...this new function, inverting the
4842 sense of the return value.
4843 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4844 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4845 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4846 (msp430_can_change_mode_class): New function.
4847 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4848 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4849 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4850 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4851 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4852 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4853 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4854 (pa_cannot_change_mode_class): Replace with...
4855 (pa_can_change_mode_class): ...this new function, inverting the
4856 sense of the return value.
4857 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4858 than CANNOT_CHANGE_MODE_CLASS in comments.
4859 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4860 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4861 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4862 (pdp11_cannot_change_mode_class): Replace with...
4863 (pdp11_can_change_mode_class): ...this new function, inverting the
4864 sense of the return value.
4865 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4866 * config/powerpcspe/powerpcspe-protos.h
4867 (rs6000_cannot_change_mode_class_ptr): Delete.
4868 * config/powerpcspe/powerpcspe.c
4869 (rs6000_cannot_change_mode_class_ptr): Delete.
4870 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4871 (rs6000_option_override_internal): Assign to
4872 targetm.can_change_mode_class instead of
4873 rs6000_cannot_change_mode_class_ptr.
4874 (rs6000_cannot_change_mode_class): Replace with...
4875 (rs6000_can_change_mode_class): ...this new function, inverting the
4876 sense of the return value.
4877 (rs6000_debug_cannot_change_mode_class): Replace with...
4878 (rs6000_debug_can_change_mode_class): ...this new function.
4879 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4880 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4881 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4882 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4883 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4885 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4886 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4887 (rs6000_option_override_internal): Assign to
4888 targetm.can_change_mode_class instead of
4889 rs6000_cannot_change_mode_class_ptr.
4890 (rs6000_cannot_change_mode_class): Replace with...
4891 (rs6000_can_change_mode_class): ...this new function, inverting the
4892 sense of the return value.
4893 (rs6000_debug_cannot_change_mode_class): Replace with...
4894 (rs6000_debug_can_change_mode_class): ...this new function.
4895 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4896 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4897 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4898 (s390_can_change_mode_class): ...this new function, inverting the
4899 sense of the return value.
4900 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4901 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4902 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4903 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4904 (sh_cannot_change_mode_class): Replace with...
4905 (sh_can_change_mode_class): ...this new function, inverting the
4906 sense of the return value.
4907 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4908 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4909 (sparc_can_change_mode_class): New function.
4910 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4911 * config/spu/spu.c (spu_can_change_mode_class): New function.
4912 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4913 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4914 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4915 (visium_can_change_mode_class): New function.
4916 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4918 2017-09-15 Richard Biener <rguenther@suse.de>
4920 PR tree-optimization/82217
4921 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4922 but not undefined case.
4924 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4927 * postreload.c (reload_cse_simplify_operands): Skip
4928 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4930 2017-09-15 Richard Biener <rguenther@suse.de>
4932 PR tree-optimization/68823
4933 * graphite-scop-detection.c (build_alias_set): If we have a
4934 possible dependence check whether we can handle them by just
4935 looking at the DRs DR_ACCESS_FNs.
4936 (build_scops): If build_alias_set fails, fail the SCOP.
4938 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4940 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4941 to support float128 built-in functions that require the ISA 3.0
4943 (BU_FLOAT128_3_HW): Likewise.
4944 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4946 (FMAF128): Likewise.
4948 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4949 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4950 floating point instructions.
4951 (rs6000_invalid_builtin): Likewise.
4952 (rs6000_builtin_mask_names): Likewise.
4953 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4954 (RS6000_BTM_FLOAT128_HW): Likewise.
4955 (RS6000_BTM_COMMON): Likewise.
4956 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4958 * doc/extend.texi (RS/6000 built-in functions): Document the
4959 IEEE 128-bit floating point square root and fused multiply-add
4962 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4964 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4965 reg (r2) isn't in the set of registers defined in the prologue.
4967 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4968 Alan Hayward <alan.hayward@arm.com>
4969 David Sherwood <david.sherwood@arm.com>
4971 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4972 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4973 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4974 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4975 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4977 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4978 max_vectorization_factor.
4979 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4981 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4982 Alan Hayward <alan.hayward@arm.com>
4983 David Sherwood <david.sherwood@arm.com>
4985 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4986 (vect_worthwhile_without_simd_p): Declare.
4987 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4988 (vectorizable_reduction): Use it.
4989 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4990 (vectorizable_operation): Likewise.
4992 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4993 Alan Hayward <alan.hayward@arm.com>
4994 David Sherwood <david.sherwood@arm.com>
4996 * tree-vectorizer.h (vect_get_num_copies): New function.
4997 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4998 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4999 (vectorizable_induction): Likewise.
5000 (vectorizable_live_operation): Likewise.
5001 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5002 (vectorizable_bswap): Likewise.
5003 (vectorizable_call): Likewise.
5004 (vectorizable_conversion): Likewise.
5005 (vectorizable_assignment): Likewise.
5006 (vectorizable_shift): Likewise.
5007 (vectorizable_operation): Likewise.
5008 (vectorizable_store): Likewise.
5009 (vectorizable_load): Likewise.
5010 (vectorizable_condition): Likewise.
5011 (vectorizable_comparison): Likewise.
5012 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5014 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5015 Alan Hayward <alan.hayward@arm.com>
5016 David Sherwood <david.sherwood@arm.com>
5018 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5019 of vect_init_vector.
5021 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5022 Alan Hayward <alan.hayward@arm.com>
5023 David Sherwood <david.sherwood@arm.com>
5025 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5026 an inline wrapper that provides a location.
5027 (gimple_build_vector): Likewise.
5028 * gimple-fold.c (gimple_build_vector_from_val): New function.
5029 (gimple_build_vector): Likewise.
5030 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5031 functions to build the initial value. Always return a gimple value.
5032 (get_initial_defs_for_reduction): Likewise. Only compute
5034 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5035 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5036 (vectorizable_induction): Use gimple_build_vector rather than
5039 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5040 Alan Hayward <alan.hayward@arm.com>
5041 David Sherwood <david.sherwood@arm.com>
5043 * target.h (vec_perm_indices): New typedef.
5044 (auto_vec_perm_indices): Likewise.
5045 * optabs-query.h: Include target.h
5046 (can_vec_perm_p): Take a vec_perm_indices *.
5047 * optabs-query.c (can_vec_perm_p): Likewise.
5048 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5049 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5050 * tree-vect-generic.c (lower_vec_perm): Likewise.
5051 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5052 (vect_grouped_load_supported): Likewise.
5053 (vect_shift_permute_load_chain): Likewise.
5054 (vect_permute_store_chain): Use auto_vec_perm_indices.
5055 (vect_permute_load_chain): Likewise.
5056 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5057 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5058 Update uses of can_vec_perm_p.
5059 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5060 mode with a number of elements. Take a vec_perm_indices *.
5061 (vect_create_epilog_for_reduction): Update accordingly.
5062 Use auto_vec_perm_indices.
5063 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5064 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5065 (vect_transform_slp_perm_load): Likewise.
5066 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5067 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5068 (vect_gen_perm_mask_checked): Likewise.
5069 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5070 (vect_gen_perm_mask_checked): Likewise.
5071 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5072 (vectorizable_store): Likewise.
5073 (vectorizable_load): Likewise.
5074 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5075 (vectorizable_bswap): Likewise.
5077 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5078 Alan Hayward <alan.hayward@arm.com>
5079 David Sherwood <david.sherwood@arm.com>
5081 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5082 * tree.c (build_vector): Likewise.
5083 (build_vector_from_ctor): Update accordingly.
5084 (build_vector_from_val): Likewise.
5085 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5086 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5087 * tree-vect-generic.c (add_rshift): Likewise.
5088 (expand_vector_divmod): Likewise.
5089 (optimize_vector_constructor): Likewise.
5090 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5091 (vect_transform_slp_perm_load): Likewise.
5092 (vect_schedule_slp_instance): Likewise.
5093 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5094 (vectorizable_call): Likewise.
5095 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5096 * expmed.c (make_tree): Likewise.
5097 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5098 a vector passed to build_vector.
5099 (fold_convert_const): Likewise.
5100 (exact_inverse): Likewise.
5101 (fold_ternary_loc): Likewise.
5102 (fold_relational_const): Likewise.
5103 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5104 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5105 (const_unop): Likewise. Store the reduction accumulator in a
5106 variable rather than an array.
5107 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5108 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5109 the new vector, rather than constructing it after the input arrays.
5110 (native_interpret_vector): Use auto_vec<tree> when building
5111 a vector passed to build_vector. Add elements in order.
5112 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5113 auto_vec<tree> when building a vector passed to build_vector.
5114 (vect_create_epilog_for_reduction): Likewise.
5115 (vectorizable_induction): Likewise.
5116 (get_initial_def_for_reduction): Likewise. Fix indentation of
5118 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5120 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5121 passed to build_vector.
5123 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5124 Alan Hayward <alan.hayward@arm.com>
5125 David Sherwood <david.sherwood@arm.com>
5127 * tree-core.h (tree_base::u): Add an "nelts" field.
5128 (tree_vector): Use VECTOR_CST_NELTS as the length.
5129 * tree.c (tree_size): Likewise.
5130 (make_vector): Initialize VECTOR_CST_NELTS.
5131 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5132 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5133 TYPE_VECTOR_SUBPARTS.
5134 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5135 as the number of units, setting it from VECTOR_CST_NELTS.
5136 (const_vector_from_tree): Likewise.
5137 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5138 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5139 (fold_negate_expr_1): Likewise.
5140 (fold_convert_const): Likewise.
5141 (const_binop): Likewise. Differentiate the number of output and
5143 (const_unop): Likewise.
5144 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5145 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5146 in cases that did the opposite.
5148 2017-09-14 Richard Biener <rguenther@suse.de>
5150 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5153 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5155 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5157 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5160 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5161 if and where to split a bb, except for splitting before debug insn
5162 sequences followed by non-label real insn. Delete debug insns
5163 in between basic blocks.
5165 * combine.c (make_compound_operation_int): Formatting fixes.
5167 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5168 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5169 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5170 * config/sol2.h (LINK_EH_SPEC): Likewise.
5171 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5172 * config/s390/linux.h (LINK_SPEC): Likewise.
5173 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5174 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5175 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5176 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5177 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5178 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5179 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5180 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5182 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5184 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5186 (ENDFILE_LINUX_SPEC): Likewise.
5187 (LINK_EH_SPEC): Likewise.
5188 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5189 (LINK_OS_LINUX_SPEC32): Likewise.
5190 (LINK_OS_LINUX_SPEC64): Likewise.
5191 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5192 (LINK_OS_LINUX_SPEC): Likewise.
5194 2017-09-13 Martin Liska <mliska@suse.cz>
5197 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5198 CASE_HIGH is NULL_TREE.
5200 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5201 Alan Hayward <alan.hayward@arm.com>
5202 David Sherwood <david.sherwood@arm.com>
5204 * target.def (secondary_memory_needed): New hook.
5205 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5206 instead of SECONDARY_MEMORY_NEEDED.
5207 (secondary_memory_needed_mode): Likewise.
5208 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5209 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5210 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5211 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5212 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5213 * doc/tm.texi: Regenerate.
5214 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5215 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5216 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5217 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5218 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5219 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5220 mode argument first and change the reg_class arguments to reg_class_t.
5221 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5222 Make static. Update the call to inline_secondary_memory_needed.
5223 (ix86_register_move_cost): Update the call to
5224 inline_secondary_memory_needed.
5225 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5226 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5228 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5229 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5231 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5232 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5233 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5234 and match hook interface. Add comment from mips.h.
5235 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5236 * config/mmix/mmix.md (truncdfsf2): Refer to
5237 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5239 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5240 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5241 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5242 (pa_secondary_memory_needed): New function.
5243 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5244 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5245 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5246 (pdp11_secondary_memory_needed): Make static and match hook interface.
5247 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5248 * config/powerpcspe/powerpcspe-protos.h
5249 (rs6000_secondary_memory_needed_ptr): Delete.
5250 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5252 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5253 (rs6000_option_override_internal): Assign to
5254 targetm.secondary_memory_needed rather than
5255 rs6000_secondary_memory_needed_ptr.
5256 (rs6000_secondary_memory_needed): Match hook interface.
5257 (rs6000_debug_secondary_memory_needed): Likewise.
5258 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5259 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5260 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5261 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5262 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5263 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5265 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5266 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5267 (rs6000_option_override_internal): Assign to
5268 targetm.secondary_memory_needed rather than
5269 rs6000_secondary_memory_needed_ptr.
5270 (rs6000_secondary_memory_needed): Match hook interface.
5271 (rs6000_debug_secondary_memory_needed): Likewise.
5272 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5273 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5274 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5275 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5276 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5277 (sparc_secondary_memory_needed): New function.
5278 * lra-constraints.c (check_and_process_move): Refer to
5279 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5281 (curr_insn_transform): Likewise.
5282 (process_alt_operands): Use targetm.secondary_memory_needed
5283 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5284 (check_secondary_memory_needed_p): Likewise.
5285 (choose_split_class): Likewise.
5286 * reload.c: Unconditionally include code that was previously
5287 conditional on SECONDARY_MEMORY_NEEDED.
5288 (push_secondary_reload): Use targetm.secondary_memory_needed
5289 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5290 (push_reload): Likewise.
5291 * reload1.c: Unconditionally include code that was previously
5292 conditional on SECONDARY_MEMORY_NEEDED.
5293 (choose_reload_regs): Use targetm.secondary_memory_needed
5294 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5295 (gen_reload): Likewise.
5296 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5298 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5299 Alan Hayward <alan.hayward@arm.com>
5300 David Sherwood <david.sherwood@arm.com>
5302 * target.def (secondary_memory_needed_mode): New hook:
5303 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5304 * targhooks.h (default_secondary_memory_needed_mode): New function.
5305 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5306 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5307 * doc/tm.texi: Regenerate.
5308 * lra-constraints.c (check_and_process_move): Use
5309 targetm.secondary_memory_needed_mode instead of
5310 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5311 (curr_insn_transform): Likewise.
5312 * reload.c (get_secondary_mem): Likewise.
5313 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5314 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5316 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5317 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5318 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5319 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5320 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5322 * config/powerpcspe/powerpcspe-protos.h
5323 (rs6000_secondary_memory_needed_mode): Delete.
5324 * config/powerpcspe/powerpcspe.c
5325 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5326 (rs6000_secondary_memory_needed_mode): Make static.
5327 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5328 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5330 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5332 (rs6000_secondary_memory_needed_mode): Make static.
5333 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5334 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5335 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5336 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5337 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5339 (sparc_secondary_memory_needed_mode): New function.
5340 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5342 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5344 * config/aarch64/constraints.md (Umq): New constraint.
5345 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5347 (mov<mode>): Update condition.
5349 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5351 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5352 when bitposition is the same.
5354 2017-09-13 Richard Biener <rguenther@suse.de>
5356 * dwarf2out.c (output_die_symbol): Remove.
5357 (output_die): Do not output a DIEs symbol.
5359 2017-09-13 Richard Biener <rguenther@suse.de>
5362 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5363 default-def to avoid breaking iterator update with the weird
5364 interaction with cgraph_update_edges_for_call_stmt_node.
5366 2017-09-13 Richard Biener <rguenther@suse.de>
5368 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5369 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5370 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5371 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5373 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5375 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5376 Disable pc relative literal load irrespective of
5377 TARGET_FIX_ERR_A53_84341 for default.
5379 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5381 * config/sparc/sparc.c (output_return): Output the source location of
5382 the insn in the delay slot, if any.
5383 (output_sibcall): Likewise.
5385 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5388 * common.opt (-static-pie): New alias.
5389 (shared): Negate static-pie.
5390 (-no-pie): Update help text.
5392 (static-pie): New option.
5393 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5394 -static-pie support.
5395 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5396 (LINK_EH_SPEC): Likewise.
5397 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5398 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5399 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5400 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5401 (init_gcc_specs): Likewise.
5402 (init_spec): Likewise.
5403 (display_help): Update help message for -pie.
5404 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5407 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5409 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5410 (movdi_aarch64): Likewise.
5411 (movti_aarch64): Likewise.
5413 2017-09-12 Simon Wright <simon@pushface.org>
5416 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5417 calculation of the minor version, always output as 0.
5419 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5422 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5423 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5424 on it early, rather than manual conversion late. For
5425 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5426 instead of performing manual conversion.
5428 2017-09-12 Carl Love <cel@us.ibm.com>
5430 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5431 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5433 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5434 VMULOSW): Add definitions.
5435 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5436 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5437 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5438 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5439 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5441 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5443 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5445 (movti_aarch64): Likewise.
5446 (movdf_aarch64): Likewise.
5447 (movtf_aarch64): Likewise.
5448 (load_pairdi): Likewise.
5449 (store_pairdi): Likewise.
5450 (load_pairdf): Likewise.
5451 (store_pairdf): Likewise.
5452 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5453 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5454 (ldr_got_small_<mode>): Likewise.
5455 (ldr_got_small_28k_<mode>): Likewise.
5456 (ldr_got_tiny): Likewise.
5457 * config/aarch64/iterators.md (ldst_sz): New.
5458 (ldpstp_sz): Likewise.
5459 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5461 (thunderx_load): Split load_8 to load_16.
5462 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5464 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5465 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5466 (xgene1_store_pair): Split store_8 to store_16.
5467 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5468 (falkor_st_0_st_sd): Split store_8 to store_16.
5470 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5472 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5473 and store1/2/3/4 to store_4/8/12/16.
5474 * config/aarch64/aarch64.md: Update for rename.
5475 * config/arm/arm.md: Likewise.: Likewise.
5476 * config/arm/arm.c: Likewise.
5477 * config/arm/thumb1.md: Likewise.
5478 * config/arm/thumb2.md: Likewise.
5479 * config/arm/vfp.md: Likewise.
5480 * config/arm/arm-generic.md: Likewise.
5481 * config/arm/arm1020e.md: Likewise.
5482 * config/arm/arm1026ejs.md: Likewise.
5483 * config/arm/arm1136jfs.md: Likewise.
5484 * config/arm/arm926ejs.md: Likewise.
5485 * config/arm/cortex-a15.md: Likewise.
5486 * config/arm/cortex-a17.md: Likewise.
5487 * config/arm/cortex-a5.md: Likewise.
5488 * config/arm/cortex-a53.md: Likewise.
5489 * config/arm/cortex-a57.md: Likewise.
5490 * config/arm/cortex-a7.md: Likewise.
5491 * config/arm/cortex-a8.md: Likewise.
5492 * config/arm/cortex-a9.md: Likewise.
5493 * config/arm/cortex-m4.md: Likewise.
5494 * config/arm/cortex-m7.md: Likewise.
5495 * config/arm/cortex-r4.md: Likewise.
5496 * config/arm/exynos-m1.md: Likewise.
5497 * config/arm/fa526.md: Likewise.
5498 * config/arm/fa606te.md: Likewise.
5499 * config/arm/fa626te.md: Likewise.
5500 * config/arm/fa726te.md: Likewise.
5501 * config/arm/fmp626.md: Likewise.
5502 * config/arm/iwmmxt.md: Likewise.
5503 * config/arm/ldmstm.md: Likewise.
5504 * config/arm/marvell-pj4.md: Likewise.
5505 * config/arm/xgene1.md: Likewise.
5506 * config/aarch64/thunderx.md: Likewise.
5507 * config/aarch64/thunderx2t99.md: Likewise.
5508 * config/aarch64/falkor.md: Likewise.
5510 2017-09-12 Martin Liska <mliska@suse.cz>
5512 * attribs.c (private_lookup_attribute): New function.
5513 * attribs.h (private_lookup_attribute): Declared here.
5514 (lookup_attribute): Called from this place.
5516 2017-09-12 Richard Biener <rguenther@suse.de>
5518 PR tree-optimization/82157
5519 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5520 stmts with side-effects.
5522 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5523 Alan Hayward <alan.hayward@arm.com>
5524 David Sherwood <david.sherwood@arm.com>
5526 * target.def (hard_regno_nregs): New hook.
5527 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5528 * targhooks.h (default_hard_regno_nregs): Declare.
5529 * targhooks.c (default_hard_regno_nregs): New function.
5530 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5531 (TARGET_HARD_REGNO_NREGS): ...this hook.
5532 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5533 (CLASS_MAX_NREGS): Likewise.
5534 * doc/tm.texi: Regenerate.
5535 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5536 instead of HARD_REGNO_NREGS.
5537 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5538 HARD_REGNO_NREGS in the comment.
5539 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5540 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5541 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5542 Return an unsigned int.
5543 (TARGET_HARD_REGNO_NREGS): Redefine.
5544 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5545 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5546 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5547 (arc_hard_regno_nregs): New function.
5548 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5549 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5550 (arm_hard_regno_nregs): New function.
5551 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5552 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5553 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5554 (TARGET_HARD_REGNO_NREGS): Redefine.
5555 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5556 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5557 (HARD_REGNO_NREGS): Delete.
5558 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5559 (cr16_hard_regno_nregs): New function.
5560 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5561 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5562 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5563 (cris_hard_regno_nregs): New function.
5564 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5565 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5566 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5567 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5568 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5569 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5570 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5571 (frv_hard_regno_nregs): Make static. Take and return an
5573 (frv_class_max_nregs): Remove outdated copy of documentation.
5574 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5575 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5576 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5577 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5578 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5579 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5580 (TARGET_HARD_REGNO_NREGS): Redefine.
5581 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5582 (CLASS_MAX_NREGS): Update comment.
5583 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5584 (ia64_hard_regno_nregs): New function.
5585 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5586 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5587 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5588 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5589 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5591 (m32c_hard_regno_nregs): Likewise. Make static.
5592 (TARGET_HARD_REGNO_NREGS): Redefine.
5593 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5594 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5595 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5596 (m68k_hard_regno_nregs): New function.
5597 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5598 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5599 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5600 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5601 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5602 Take and return an unsigned int.
5603 (TARGET_HARD_REGNO_NREGS): Redefine.
5604 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5605 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5606 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5607 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5608 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5609 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5610 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5611 (msp430_hard_regno_nregs): Make static. Take and return an
5613 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5614 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5615 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5616 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5617 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5618 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5619 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5620 (TARGET_HARD_REGNO_NREGS): Redefine.
5621 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5622 (PA_HARD_REGNO_NREGS): ...this.
5623 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5624 (PA_HARD_REGNO_NREGS): ...this.
5625 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5626 (pa_hard_regno_nregs): New function.
5627 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5628 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5629 (pdp11_hard_regno_nregs): New function.
5630 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5631 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5632 (rs6000_hard_regno_nregs_hook): New function.
5633 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5634 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5635 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5636 Take and return an unsigned int. Move earlier in file.
5637 (TARGET_HARD_REGNO_NREGS): Redefine.
5638 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5639 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5640 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5641 (rl78_hard_regno_nregs): Make static. Take and return an
5643 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5644 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5645 (rs6000_hard_regno_nregs_hook): New function.
5646 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5647 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5648 (TARGET_HARD_REGNO_NREGS): Redefine.
5649 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5650 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5651 instead of HARD_REGNO_NREGS.
5652 (s390_hard_regno_nregs): New function.
5653 (s390_hard_regno_mode_ok): Add comment from s390.h.
5654 (TARGET_HARD_REGNO_NREGS): Redefine.
5655 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5656 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5657 (sh_hard_regno_nregs): New function.
5658 (sh_pass_in_reg_p): Use it.
5659 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5660 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5661 (sparc_hard_regno_nregs): New function.
5662 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5663 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5664 (spu_function_arg_advance): Use it, supplying a valid register number.
5665 (TARGET_HARD_REGNO_NREGS): Redefine.
5666 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5667 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5668 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5669 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5670 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5671 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5672 (CLASS_MAX_NREGS): Remove copy of old documentation.
5673 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5674 (visium_hard_regno_nregs): New function.
5675 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5676 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5677 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5678 xtensa_hard_regno_nregs): New function.
5679 * system.h (HARD_REGNO_NREGS): Poison.
5681 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5683 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5684 hard_regno_nregs instead of HARD_REGNO_NREGS.
5685 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5686 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5687 (c6x_expand_epilogue): Likewise.
5688 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5689 (frv_read_iacc_argument): Likewise.
5690 * config/sh/sh.c: Include regs.h.
5691 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5692 (regs_used): Likewise.
5693 (output_stack_adjust): Likewise.
5694 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5695 * expmed.c: Include regs.h.
5696 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5697 * ree.c: Include regs.h.
5698 (combine_reaching_defs): Use hard_regno_nregs instead of
5700 (add_removable_extension): Likewise.
5702 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5704 * regs.h (hard_regno_nregs): Turn into a function.
5705 (end_hard_regno): Update accordingly.
5706 * caller-save.c (setup_save_areas): Likewise.
5707 (save_call_clobbered_regs): Likewise.
5708 (replace_reg_with_saved_mem): Likewise.
5709 (insert_restore): Likewise.
5710 (insert_save): Likewise.
5711 * combine.c (can_change_dest_mode): Likewise.
5712 (move_deaths): Likewise.
5713 (distribute_notes): Likewise.
5714 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5715 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5716 (rs6000_split_multireg_move): Likewise.
5717 (rs6000_register_move_cost): Likewise.
5718 (rs6000_memory_move_cost): Likewise.
5719 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5720 (rs6000_split_multireg_move): Likewise.
5721 (rs6000_register_move_cost): Likewise.
5722 (rs6000_memory_move_cost): Likewise.
5723 * cselib.c (cselib_reset_table): Likewise.
5724 (cselib_lookup_1): Likewise.
5725 * emit-rtl.c (set_mode_and_regno): Likewise.
5726 * function.c (aggregate_value_p): Likewise.
5727 * ira-color.c (setup_profitable_hard_regs): Likewise.
5728 (check_hard_reg_p): Likewise.
5729 (calculate_saved_nregs): Likewise.
5730 (assign_hard_reg): Likewise.
5731 (improve_allocation): Likewise.
5732 (calculate_spill_cost): Likewise.
5733 * ira-emit.c (modify_move_list): Likewise.
5734 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5735 (ira_hard_reg_in_set_p): Likewise.
5736 * ira.c (setup_reg_mode_hard_regset): Likewise.
5737 (clarify_prohibited_class_mode_regs): Likewise.
5738 (check_allocation): Likewise.
5739 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5740 (lra_setup_reg_renumber): Likewise.
5741 (setup_try_hard_regno_pseudos): Likewise.
5742 (spill_for): Likewise.
5743 (assign_hard_regno): Likewise.
5744 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5745 * lra-constraints.c (in_class_p): Likewise.
5746 (lra_constraint_offset): Likewise.
5747 (simplify_operand_subreg): Likewise.
5748 (lra_constraints): Likewise.
5749 (split_reg): Likewise.
5750 (split_if_necessary): Likewise.
5751 (invariant_p): Likewise.
5752 (inherit_in_ebb): Likewise.
5753 * lra-lives.c (process_bb_lives): Likewise.
5754 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5755 (get_hard_regs): Likewise.
5756 (do_remat): Likewise.
5757 * lra-spills.c (assign_spill_hard_regs): Likewise.
5758 * mode-switching.c (create_pre_exit): Likewise.
5759 * postreload.c (reload_combine_recognize_pattern): Likewise.
5760 * recog.c (peep2_find_free_register): Likewise.
5761 * regcprop.c (kill_value_regno): Likewise.
5762 (set_value_regno): Likewise.
5763 (copy_value): Likewise.
5764 (maybe_mode_change): Likewise.
5765 (find_oldest_value_reg): Likewise.
5766 (copyprop_hardreg_forward_1): Likewise.
5767 * regrename.c (check_new_reg_p): Likewise.
5768 (regrename_do_replace): Likewise.
5769 * reload.c (push_reload): Likewise.
5770 (combine_reloads): Likewise.
5771 (find_dummy_reload): Likewise.
5772 (operands_match_p): Likewise.
5773 (find_reloads): Likewise.
5774 (find_equiv_reg): Likewise.
5775 (reload_adjust_reg_for_mode): Likewise.
5776 * reload1.c (count_pseudo): Likewise.
5777 (count_spilled_pseudo): Likewise.
5778 (find_reg): Likewise.
5779 (clear_reload_reg_in_use): Likewise.
5780 (free_for_value_p): Likewise.
5781 (allocate_reload_reg): Likewise.
5782 (choose_reload_regs): Likewise.
5783 (reload_adjust_reg_for_temp): Likewise.
5784 (emit_reload_insns): Likewise.
5785 (delete_output_reload): Likewise.
5786 * rtlanal.c (subreg_get_info): Likewise.
5787 * sched-deps.c (sched_analyze_reg): Likewise.
5788 * sel-sched.c (init_regs_for_mode): Likewise.
5789 (mark_unavailable_hard_regs): Likewise.
5790 (choose_best_reg_1): Likewise.
5791 (verify_target_availability): Likewise.
5792 * valtrack.c (dead_debug_insert_temp): Likewise.
5793 * var-tracking.c (track_loc_p): Likewise.
5794 (emit_note_insn_var_location): Likewise.
5795 * varasm.c (make_decl_rtl): Likewise.
5796 * reginfo.c (choose_hard_reg_mode): Likewise.
5797 (init_reg_modes_target): Refer directly to
5798 this_target_regs->x_hard_regno_nregs.
5800 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5802 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5803 instead of hard_regno_nregs.
5805 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5807 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5808 end_hard_regno instead of hard_regno_nregs.
5809 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5810 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5811 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5812 * ira-color.c (improve_allocation): Likewise.
5813 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5814 * lra-lives.c (mark_regno_live): Likewise.
5815 (mark_regno_dead): Likewise.
5816 * lra-remat.c (operand_to_remat): Likewise.
5817 * lra.c (collect_non_operand_hard_regs): Likewise.
5818 * postreload.c (reload_combine_note_store): Likewise.
5819 (move2add_valid_value_p): Likewise.
5820 * reload.c (regno_clobbered_p): Likewise.
5822 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5824 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5826 * config/v850/v850.c (v850_reorg): Likewise.
5827 * reload.c (refers_to_regno_for_reload_p): Likewise.
5828 (find_equiv_reg): Likewise.
5829 * reload1.c (reload_reg_reaches_end_p): Likewise.
5831 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5833 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5835 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5836 * config/arm/arm.c (output_move_neon): Likewise.
5837 (arm_attr_length_move_neon): Likewise.
5838 (neon_split_vcombine): Likewise.
5839 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5840 (c6x_mark_reg_written): Likewise.
5841 (c6x_dwarf_register_span): Likewise.
5842 * config/i386/i386.c (ix86_save_reg): Likewise.
5843 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5844 (rws_access_reg): Likewise.
5845 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5846 * mode-switching.c (create_pre_exit): Likewise.
5847 * ree.c (combine_reaching_defs): Likewise.
5848 (add_removable_extension): Likewise.
5849 * regcprop.c (find_oldest_value_reg): Likewise.
5850 (copyprop_hardreg_forward_1): Likewise.
5851 * reload.c (reload_inner_reg_of_subreg): Likewise.
5852 (push_reload): Likewise.
5853 (combine_reloads): Likewise.
5854 (find_dummy_reload): Likewise.
5855 (reload_adjust_reg_for_mode): Likewise.
5856 * reload1.c (find_reload_regs): Likewise.
5857 (forget_old_reloads_1): Likewise.
5858 (reload_reg_free_for_value_p): Likewise.
5859 (reload_adjust_reg_for_temp): Likewise.
5860 (emit_reload_insns): Likewise.
5861 (delete_output_reload): Likewise.
5862 * sel-sched.c (choose_best_reg_1): Likewise.
5863 (choose_best_pseudo_reg): Likewise.
5865 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5866 Alan Hayward <alan.hayward@arm.com>
5867 David Sherwood <david.sherwood@arm.com>
5869 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5870 * target.def (slow_unaligned_access): New hook.
5871 * targhooks.h (default_slow_unaligned_access): Declare.
5872 * targhooks.c (default_slow_unaligned_access): New function.
5873 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5874 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5875 * doc/tm.texi: Regenerate.
5876 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5877 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5878 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5880 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5881 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5883 (rs6000_slow_unaligned_access): New function.
5884 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5885 (expand_block_compare): Likewise.
5886 (expand_strn_compare): Likewise.
5887 (rs6000_rtx_costs): Likewise.
5888 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5889 (riscv_slow_unaligned_access): Likewise.
5890 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5891 (riscv_slow_unaligned_access_p): ...this and make static.
5892 (riscv_option_override): Update accordingly.
5893 (riscv_slow_unaligned_access): New function.
5894 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5895 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5896 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5897 (rs6000_slow_unaligned_access): New function.
5898 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5899 (rs6000_rtx_costs): Likewise.
5900 * config/rs6000/rs6000-string.c (expand_block_compare)
5901 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5902 of SLOW_UNALIGNED_ACCESS.
5903 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5904 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5905 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5906 of SLOW_UNALIGNED_ACCESS.
5907 * expmed.c (simple_mem_bitfield_p): Likewise.
5908 * expr.c (alignment_for_piecewise_move): Likewise.
5909 (emit_group_load_1): Likewise.
5910 (emit_group_store): Likewise.
5911 (copy_blkmode_from_reg): Likewise.
5912 (emit_push_insn): Likewise.
5913 (expand_assignment): Likewise.
5914 (store_field): Likewise.
5915 (expand_expr_real_1): Likewise.
5916 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5917 * lra-constraints.c (simplify_operand_subreg): Likewise.
5918 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5919 * gimple-ssa-store-merging.c: Likewise in block comment at start
5921 * tree-ssa-strlen.c: Include target.h.
5922 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5923 of SLOW_UNALIGNED_ACCESS.
5924 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5926 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5928 PR rtl-optimization/82185
5929 * expmed.c (emit_store_flag_int): Only test tem if it has been
5932 2017-09-12 Richard Biener <rguenther@suse.de>
5935 * match.pd ((FTYPE) N CMP CST): Fix typo.
5937 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5939 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5941 (mips_near_type_p): Add 'short_call' attribute as a synonym
5943 * doc/extend.texi (short_call): Document new function attribute.
5945 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5948 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5949 assertion check that in the condition.
5950 (get_atomic_generic_size): Likewise. Before testing if parameter
5951 has pointer type, if it has array type, call for C++
5952 default_conversion to perform array-to-pointer conversion.
5954 2017-09-12 Richard Biener <rguenther@suse.de>
5956 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5957 for operations we cannot scalarize.
5959 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5961 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5962 vectors heap vectors. Clean up comments.
5963 Make visited_bbs a reference.
5964 (profitable_jump_thread_path): Make GC
5965 vectors heap vectors. Clean up comments.
5967 (convert_and_register_jump_thread_path): Make GC vectors heap
5969 (check_subpath_and_update_thread_path): Same. Clean up comments.
5970 Make visited_bbs a reference.
5971 (handle_phi): Abstract common code to to
5972 register_jump_thread_path_if_profitable.
5973 Rename VAR_BB to DEF_BB.
5975 Make GC vectors heap vectors.
5976 Make visited_bbs a reference.
5977 (handle_assignment): Same.
5978 (register_jump_thread_path_if_profitable): New.
5979 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5981 Make GC vectors heap vectors. Clean up comments.
5982 Make visited_bbs a reference.
5983 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5984 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5987 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5990 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5991 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5993 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5995 Revert r251800 and r251799.
5997 2017-09-11 Martin Jambor <mjambor@suse.cz>
6000 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6001 arguments in advance.
6002 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6003 use it to find predecessor edges.
6004 (naive_outof_ssa): Collect vector of predecessors.
6006 2017-09-08 Jason Merrill <jason@redhat.com>
6008 PR c++/70029 - ICE with ref-qualifier and -flto
6009 * langhooks.h (struct lang_hooks_for_types): Add
6010 copy_lang_qualifiers.
6011 * attribs.c (build_type_attribute_qual_variant): Use it.
6012 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6014 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6015 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6017 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6020 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6021 (*mulsi3_sp64): New instruction.
6022 (mulsi3): New expander.
6024 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6026 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6028 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6030 * sancov.c: Include memmodel.h.
6032 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6035 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6038 2017-09-07 Carl Love <cel@us.ibm.com>
6040 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6041 the sldi instruction.
6043 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6045 * sancov.c: Include tm_p.h.
6047 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6050 * output.h (switch_to_other_text_partition): New declaration.
6051 * varasm.c (switch_to_other_text_partition): New function.
6052 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6053 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6054 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6055 to the other text partition before emitting LCL label and switch back
6056 after emitting the word after it.
6058 2017-09-07 Richard Biener <rguenther@suse.de>
6060 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6061 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6062 critical edges here, after loop init.
6063 (pass_data_pre): Remove PROP_no_crit_edges flags.
6064 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6065 for valueization of call args to avoid leaking VN_TOP.
6066 (visit_use): Assert we do not visit default defs.
6067 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6068 Use error_mark_node to more easily detect leaking VN_TOP.
6069 All default-defs are varying, not VN_TOP. Mark them visited.
6070 (run_scc_vn): Make code match comment.
6072 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6074 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6075 OPTION_MASK_FLOAT128_KEYWORD.
6076 (POWERPC_MASKS): Likewise.
6077 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6078 support for the -mfloat128-type option, and make -mfloat128
6079 default on PowerPC Linux systems. Define or undefine
6080 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6081 Define __float128 to be __ieee128 if IEEE 128-bit support is
6082 enabled, or undefine it.
6083 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6084 Delete defining __FLOAT128_TYPE__.
6085 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6086 -mfloat128-type option and make -mfloat128 default on PowerPC
6088 (TARGET_FLOAT128_TYPE): Likewise.
6089 (-mfloat128-type): Likewise.
6090 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6091 Delete the -mfloat128-type option and make -mfloat128 default on
6092 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6093 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6094 128-bit floating point is enabled. Change tests from using
6095 -mfloat128-type to -mfloat128.
6096 (rs6000_mangle_type): Use the correct mangling for the __float128
6097 type even if normal long double is restricted to 64-bits.
6098 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6100 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6101 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6102 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6104 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6105 documentation for -mfloat128.
6107 2017-09-06 Olivier Hainque <hainque@adacore.com>
6109 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6111 2017-09-06 Wish Wu <wishwu007@gmail.com>
6112 Jakub Jelinek <jakub@redhat.com>
6114 * asan.c (initialize_sanitizer_builtins): Add
6115 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6116 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6117 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6118 BT_FN_VOID_UINT64_PTR variables.
6119 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6120 (BT_FN_VOID_UINT16_UINT16): Likewise.
6121 (BT_FN_VOID_UINT32_UINT32): Likewise.
6122 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6123 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6124 (BT_FN_VOID_UINT64_PTR): Likewise.
6125 * common.opt (flag_sanitize_coverage): New variable.
6126 (fsanitize-coverage=trace-pc): Remove.
6127 (fsanitize-coverage=): Add.
6128 * flag-types.h (enum sanitize_coverage_code): New enum.
6129 * fold-const.c (fold_range_test): Disable non-short-circuit
6130 optimization if flag_sanitize_coverage.
6131 (fold_truth_andor): Likewise.
6132 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6133 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6134 (coverage_sanitizer_opts): New array.
6135 (get_closest_sanitizer_option): Add OPTS argument, handle also
6136 OPT_fsanitize_coverage_.
6137 (parse_sanitizer_options): Adjusted to also handle
6138 OPT_fsanitize_coverage_.
6139 (common_handle_option): Add OPT_fsanitize_coverage_.
6140 * sancov.c (instrument_comparison, instrument_switch): New function.
6141 (sancov_pass): Add trace-cmp support.
6142 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6143 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6144 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6145 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6146 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6147 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6148 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6149 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6150 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6151 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6153 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6155 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6156 error. Only quit immediately if parsing is complete.
6157 (BEGIN): Initialize fatal_err and parse_done.
6158 (begin fpu, end fpu): Check number of arguments.
6159 (begin arch, end arch): Likewise.
6160 (begin cpu, end cpu): Likewise.
6161 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6162 (optalias): Likewise.
6164 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6166 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6167 * config/arm/arm-isa.h: Delete. Move definitions to ...
6168 * arm-cpus.in: ... here. Use new feature and fgroup values.
6169 * config/arm/arm.c (arm_option_override): Use lower case for feature
6171 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6172 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6173 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6175 (print_isa_bits_for): New function.
6176 (gen_isa): New function.
6177 (gen_comm_data): Use print_isa_bits_for.
6178 (define feature): New keyword.
6179 (define fgroup): New keyword.
6180 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6181 (arm-isa.h): Add rule to generate file.
6182 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6183 case for feature bit names.
6185 2017-09-06 Richard Biener <rguenther@suse.de>
6187 * tree-ssa-pre.c (NECESSARY): Remove.
6188 (create_expression_by_pieces): Do not touch pass-local flags.
6189 (insert_into_preds_of_block): Likewise.
6190 (do_pre_regular_insertion): Likewise.
6191 (eliminate_insert): Likewise.
6192 (eliminate_dom_walker::before_dom_children): Likewise.
6193 (fini_eliminate): Do not look at inserted_exprs.
6194 (mark_operand_necessary): Remove.
6195 (remove_dead_inserted_code): Replace with simple work-list
6196 algorithm based on inserted_exprs and SSA uses.
6197 (pass_pre::execute): Re-order fini_eliminate and
6198 remove_dead_inserted_code.
6200 2017-09-06 Olivier Hainque <hainque@adacore.com>
6202 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6203 for VxWorks 7. Adjust surrounding comments.
6205 2017-09-06 Richard Biener <rguenther@suse.de>
6207 * gimple-ssa-strength-reduction.c
6208 (find_candidates_dom_walker::before_dom_children): Also allow
6211 2017-09-06 Richard Biener <rguenther@suse.de>
6213 PR tree-optimization/82108
6214 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6215 for gap in the non-permutation SLP case.
6217 2017-09-06 Martin Jambor <mjambor@suse.cz>
6219 PR tree-optimization/82078
6220 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6221 add_access_to_work_queue...
6222 (build_accesses_from_assign): ...here.
6223 (propagate_all_subaccesses): Make sure racc is the group
6224 representative, if there is one.
6226 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6229 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6232 2017-09-06 Richard Biener <rguenther@suse.de>
6234 * gimple-ssa-strength-reduction.c
6235 (find_candidates_dom_walker::before_doom_children): Use a
6236 type and not a mode check.
6238 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6241 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6243 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6245 2017-09-05 Jeff Law <law@redhat.com>
6247 PR tree-optimization/64910
6248 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6249 cases where we have 3 or more operands.
6251 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6254 * omp-low.c (lower_omp_for): Recompute tree invariant if
6255 gimple_omp_for_initial/final is ADDR_EXPR.
6258 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6259 into gimple val before gimplification fo the COND_EXPR.
6261 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6263 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6264 REGION_COPY argument.
6265 (thread_through_all_blocks): Remove unused argument to
6266 duplicate_thread_path.
6268 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6269 Alan Hayward <alan.hayward@arm.com>
6270 David Sherwood <david.sherwood@arm.com>
6272 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6273 Take a scalar_mode rather than a machine_mode.
6274 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6275 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6276 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6277 (aarch64_gen_adjusted_ldpstp): Likewise.
6278 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6280 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6281 Alan Hayward <alan.hayward@arm.com>
6282 David Sherwood <david.sherwood@arm.com>
6284 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6285 Take a scalar_int_mode instead of a machine_mode.
6286 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6287 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6288 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6289 (aarch64_simd_attr_length_rglist): Delete.
6290 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6291 a scalar_int_mode instead of a machine_mode.
6292 (aarch64_add_offset): Likewise.
6293 (aarch64_internal_mov_immediate): Likewise
6294 (aarch64_add_constant_internal): Likewise.
6295 (aarch64_add_constant): Likewise.
6296 (aarch64_movw_imm): Likewise.
6297 (aarch64_rtx_arith_op_extract_p): Likewise.
6298 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6299 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6300 Remove assert that the mode isn't a vector.
6301 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6302 (aarch64_expand_mov_immediate): Update calls after above changes.
6303 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6304 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6305 (aarch64_move_imm): Likewise.
6306 (aarch64_can_const_movi_rtx_p): Likewise.
6307 (aarch64_strip_extend): Likewise.
6308 (aarch64_extr_rtx_p): Likewise.
6309 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6310 a CONST_INT when the mode parameter is VOIDmode.
6311 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6313 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6315 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6316 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6317 (bitwise_type_for_mode): Update accordingly.
6319 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6321 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6322 * stor-layout.c (mode_for_size_tree): Likewise.
6323 (mode_for_array): Update accordingly.
6324 (layout_decl): Likewise.
6325 (compute_record_mode): Likewise. Only set the mode once.
6327 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6329 * target.def (get_mask_mode): Change return type to opt_mode.
6331 * doc/tm.texi: Regenerate.
6332 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6333 * targhooks.c (default_get_mask_mode): Likewise.
6334 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6335 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6336 targetm.get_mask_mode.
6337 * tree.c (build_truth_vector_type): Likewise.
6339 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6341 * machmode.h (mode_for_vector): Return an opt_mode.
6342 * stor-layout.c (mode_for_vector): Likewise.
6343 (mode_for_int_vector): Update accordingly.
6344 (layout_type): Likewise.
6345 * config/i386/i386.c (emit_memmov): Likewise.
6346 (ix86_expand_set_or_movmem): Likewise.
6347 (ix86_expand_vector_init): Likewise.
6348 (ix86_get_mask_mode): Likewise.
6349 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6351 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6352 * expmed.c (extract_bit_field_1): Likewise.
6353 * expr.c (expand_expr_real_2): Likewise.
6354 * optabs-query.c (can_vec_perm_p): Likewise.
6355 (can_vec_mask_load_store_p): Likewise.
6356 * optabs.c (expand_vec_perm): Likewise.
6357 * targhooks.c (default_get_mask_mode): Likewise.
6358 * tree-vect-stmts.c (vectorizable_store): Likewise.
6359 (vectorizable_load): Likewise.
6360 (get_vectype_for_scalar_type_and_size): Likewise.
6362 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6364 * machmode.h (mode_for_int_vector): New function.
6365 * stor-layout.c (mode_for_int_vector): Likewise.
6366 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6367 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6368 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6369 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6370 (s390_expand_vcond): Likewise.
6372 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6374 * machmode.h (opt_machine_mode): New type.
6375 (opt_mode<T>): Allow construction from anything that can be
6377 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6378 (mode_for_size): Return an opt_machine_mode.
6379 * stor-layout.c (mode_for_size): Likewise.
6380 (mode_for_size_tree): Update call accordingly.
6381 (bitwise_mode_for_mode): Likewise.
6382 (make_fract_type): Likewise.
6383 (make_accum_type): Likewise.
6384 * caller-save.c (replace_reg_with_saved_mem): Update call
6386 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6387 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6388 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6389 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6390 * expmed.c (extract_bit_field_1): Likewise.
6391 * reload.c (get_secondary_mem): Likewise.
6392 * varasm.c (assemble_integer): Likewise.
6393 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6396 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6398 * machmode.h (decimal_float_mode_for_size): New function.
6399 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6400 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6401 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6402 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6403 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6404 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6406 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6408 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6409 (get_builtin_sync_mode): Likewise.
6410 (expand_ifn_atomic_compare_exchange): Likewise.
6411 (expand_builtin_atomic_clear): Likewise.
6412 (expand_builtin_atomic_test_and_set): Likewise.
6413 (fold_builtin_atomic_always_lock_free): Likewise.
6414 * calls.c (compute_argument_addresses): Likewise.
6415 (emit_library_call_value_1): Likewise.
6416 (store_one_arg): Likewise.
6417 * combine.c (combine_instructions): Likewise.
6418 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6419 * config/arm/arm.c (arm_function_value): Likewise.
6420 (aapcs_allocate_return_reg): Likewise.
6421 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6422 * config/i386/i386.c (construct_container): Likewise.
6423 (ix86_gimplify_va_arg): Likewise.
6424 (ix86_expand_sse_cmp): Likewise.
6425 (emit_memmov): Likewise.
6426 (emit_memset): Likewise.
6427 (expand_small_movmem_or_setmem): Likewise.
6428 (ix86_expand_pextr): Likewise.
6429 (ix86_expand_pinsr): Likewise.
6430 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6431 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6433 * config/mips/mips.c (mips_function_value_1) Likewise.
6434 (mips_block_move_straight): Likewise.
6435 (mips_expand_ins_as_unaligned_store): Likewise.
6436 * config/powerpcspe/powerpcspe.c
6437 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6438 (rs6000_darwin64_record_arg_flush): Likewise.
6439 * config/rs6000/rs6000.c
6440 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6441 (rs6000_darwin64_record_arg_flush): Likewise.
6442 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6443 (sparc_function_value_1): Likewise.
6444 * config/spu/spu.c (adjust_operand): Likewise.
6445 (spu_emit_branch_or_set): Likewise.
6446 (arith_immediate_p): Likewise.
6447 * emit-rtl.c (gen_lowpart_common): Likewise.
6448 * expr.c (expand_expr_real_1): Likewise.
6449 * function.c (assign_parm_setup_block): Likewise.
6450 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6451 * reload1.c (alter_reg): Likewise.
6452 * stor-layout.c (mode_for_vector): Likewise.
6453 (layout_type): Likewise.
6455 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6457 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6458 (spu_convert_move): Likewise.
6459 * lower-subreg.c (resolve_simple_move): Likewise.
6461 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6464 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6465 define_insn to a define_expand.
6466 (altivec_vsum2sws_direct): New define_insn.
6467 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6469 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6471 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6474 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6479 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6480 realign stack if stack alignment needed is less than incoming
6483 2017-09-05 Marek Polacek <polacek@redhat.com>
6486 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6489 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6491 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6493 2017-09-05 Richard Biener <rguenther@suse.de>
6495 PR tree-optimization/82084
6496 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6498 2017-09-05 Richard Biener <rguenther@suse.de>
6500 PR tree-optimization/82102
6501 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6503 2017-09-05 Martin Liska <mliska@suse.cz>
6505 PR tree-optimization/82032
6506 * tree-cfg.c (generate_range_test): New function.
6507 * tree-cfg.h (generate_range_test): Declared here.
6508 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6509 (cleanup_control_expr_graph): Use it.
6510 * tree-switch-conversion.c (try_switch_expansion): Remove
6512 (emit_case_nodes): Use generate_range_test.
6514 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6517 * config/i386/i386.md (*<btsc><mode>_mask): Add
6518 TARGET_USE_BT to insn constraint.
6519 (*btr<mode>_mask): Ditto.
6521 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6523 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6524 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6526 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6529 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6530 TARGET_NEON and TARGET_IWMMXT.
6531 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6532 TARGET_NEON and TARGET_IWMMXT.
6533 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6535 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6537 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6538 (ix86_rewrite_tls_address): Ditto.
6539 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6540 (ix86_rewrite_tls_address_1): Ditto.
6541 (ix86_rewrite_tls_address): Ditto.
6542 * config/i386/predicates.md (tls_address_pattern): New predicate.
6543 * config/i386/i386.md (TLS address splitter): New splitter.
6545 2017-09-04 Richard Biener <rguenther@suse.de>
6547 PR tree-optimization/82084
6548 * fold-const.h (can_native_encode_string_p): Declare.
6549 * fold-const.c (can_native_encode_string_p): Factor out from ...
6550 (native_encode_string): ... here.
6551 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6552 vectorizing stores from constants we later cannot handle.
6554 2017-09-04 Marek Polacek <polacek@redhat.com>
6557 * doc/invoke.texi: Update -Wtautological-compare documentation.
6559 2017-09-04 Jeff Law <law@redhat.com>
6561 PR tree-optimization/64910
6562 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6563 swap the first and last operand if the last is a constant.
6565 2017-09-04 Marek Polacek <polacek@redhat.com>
6568 * convert.c (do_narrow): When sanitizing signed integer overflows,
6569 bail out for signed types.
6570 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6572 2017-09-04 Richard Biener <rguenther@suse.de>
6574 PR tree-optimization/82060
6575 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6576 Move devirtualization after stmt folding and before EH/AB/noreturn
6577 cleanup to get the stmt refs canonicalized. Use a bool instead
6578 of gimple_modified_p since that doesn't work for NOPs. Schedule
6579 NOPs generated by folding for removal.
6581 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6582 Alan Hayward <alan.hayward@arm.com>
6583 David Sherwood <david.sherwood@arm.com>
6585 * coretypes.h (pad_direction): New enum.
6586 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6587 (FUNCTION_ARG_PADDING): Likewise.
6588 * target.def (function_arg_padding): New hook.
6589 * targhooks.h (default_function_arg_padding): Declare.
6590 * targhooks.c (default_function_arg_padding): New function.
6591 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6592 (TARGET_FUNCTION_ARG_PADDING): ...this.
6593 * doc/tm.texi: Regenerate.
6594 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6595 instead of direction.
6596 (compute_argument_addresses): Likewise.
6597 (load_register_parameters): Likewise.
6598 (emit_library_call_value_1): Likewise.
6599 (store_one_arg): Use targetm.calls.function_arg_padding instead
6600 of FUNCTION_ARG_PADDING.
6601 (must_pass_in_stack_var_size_or_pad): Likewise.
6602 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6603 (emit_group_store): Likewise.
6604 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6605 instead of FUNCTION_ARG_PADDING.
6606 (emit_push_insn): Likewise, and propagate enum change throughout
6608 * function.h (direction): Delete.
6609 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6611 * function.c (assign_parm_find_stack_rtl): Likewise.
6612 (assign_parm_setup_block_p): Likewise.
6613 (assign_parm_setup_block): Likewise.
6614 (gimplify_parameters): Likewise.
6615 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6616 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6618 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6619 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6620 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6621 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6622 (aarch64_function_arg_padding): ...this new function.
6623 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6624 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6625 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6626 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6627 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6628 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6629 (arm_pad_arg_upward): Replace with...
6630 (arm_function_arg_padding): ...this new function.
6631 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6633 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6634 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6635 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6636 (ia64_hpux_function_arg_padding): Replace with...
6637 (ia64_function_arg_padding): ...this new function. Use pad_direction
6638 instead of direction. Check for TARGET_HPUX.
6639 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6640 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6641 (iq2000_function_arg_padding): New function.
6642 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6643 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6644 (mips_function_arg_padding): ...this new function.
6645 (mips_pad_reg_upward): Update accordingly.
6646 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6647 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6648 targetm.calls.function_arg_padding.
6649 (FUNCTION_ARG_PADDING): Delete.
6650 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6651 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6652 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6653 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6654 (nios2_block_reg_padding): Return pad_direction instead of direction.
6655 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6656 instead of direction.
6657 (nios2_function_arg_padding): Likewise. Make static.
6658 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6659 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6660 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6661 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6662 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6663 (pa_function_arg_padding): Make static. Return pad_direction instead
6665 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6666 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6667 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6668 instead of direction. Use targetm.calls.function_arg_padding.
6669 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6670 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6671 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6672 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6673 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6675 (function_arg_padding): Rename to...
6676 (rs6000_function_arg_padding): ...this. Make static. Return
6677 pad_direction instead of direction.
6678 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6679 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6680 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6681 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6682 instead of direction. Use targetm.calls.function_arg_padding.
6683 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6684 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6685 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6686 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6687 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6688 (function_arg_padding): Rename to...
6689 (rs6000_function_arg_padding): ...this. Make static. Return
6690 pad_direction instead of direction.
6691 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6692 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6693 * config/s390/s390.c (s390_function_arg_padding): New function.
6694 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6695 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6696 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6697 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6698 (function_arg_padding): Rename to...
6699 (sparc_function_arg_padding): ...this. Make static. Return
6700 pad_direction instead of direction.
6701 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6702 * config/spu/spu.c (spu_function_arg_padding): New function.
6703 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6704 * system.h (FUNCTION_ARG_PADDING): Poison.
6706 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6707 Alan Hayward <alan.hayward@arm.com>
6708 David Sherwood <david.sherwood@arm.com>
6710 * target.def (modes_tieable_p): New hook.
6711 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6712 (TARGET_MODES_TIEABLE_P): ...this.
6713 * doc/tm.texi.in: Regenerate.
6714 * hooks.h (hook_bool_mode_mode_true): Declare.
6715 * hooks.c (hook_bool_mode_mode_true): New function.
6716 * combine.c (subst): Use targetm.modes_tieable_p instead of
6718 * dse.c (find_shift_sequence): Likewise.
6719 * expmed.c (extract_low_bits): Likewise.
6720 * lower-subreg.c: Include target.h.
6721 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6723 * rtlanal.c (rtx_cost): Likewise.
6724 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6725 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6726 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6727 (TARGET_MODES_TIEABLE_P): Redefine.
6728 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6729 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6730 (TARGET_MODES_TIEABLE_P): Redefine.
6731 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6732 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6733 (arc_modes_tieable_p): New function.
6734 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6735 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6736 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6737 (arm_modes_tieable_p): Make static.
6738 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6739 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6740 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6741 (TARGET_MODES_TIEABLE_P): Redefine.
6742 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6743 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6744 (TARGET_MODES_TIEABLE_P): Redefine.
6745 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6746 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6747 (cr16_modes_tieable_p): New function.
6748 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6749 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6750 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6751 (TRULY_NOOP_TRUNCATION): Update comment.
6752 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6753 (TRULY_NOOP_TRUNCATION): Update comment.
6754 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6755 (frv_modes_tieable_p): New function.
6756 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6757 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6758 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6759 (TARGET_MODES_TIEABLE_P): Redefine.
6760 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6761 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6762 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6763 (TARGET_MODES_TIEABLE_P): Redefine.
6764 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6765 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6766 (ia64_modes_tieable_p): New function.
6767 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6768 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6769 (iq2000_modes_tieable_p): New function.
6770 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6771 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6772 (lm32_modes_tieable_p): New function.
6773 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6774 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6775 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6776 (TARGET_MODES_TIEABLE_P): Redefine.
6777 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6778 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6779 (m32r_modes_tieable_p): New function.
6780 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6781 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6782 (m68k_modes_tieable_p): New function.
6783 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6784 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6785 (mcore_modes_tieable_p): New function.
6786 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6787 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6789 (TARGET_MODES_TIEABLE_P): Redefine.
6790 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6791 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6792 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6793 (TARGET_MODES_TIEABLE_P): Redefine.
6794 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6795 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6796 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6797 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6798 (mn10300_modes_tieable_p): ...this and make static.
6799 (TARGET_MODES_TIEABLE_P): Redefine.
6800 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6801 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6802 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6803 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6804 (msp430_modes_tieable_p): Make static.
6805 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6806 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6807 (TARGET_MODES_TIEABLE_P): Redefine.
6808 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6809 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6810 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6811 (TARGET_MODES_TIEABLE_P): Redefine.
6812 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6813 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6814 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6815 (TARGET_MODES_TIEABLE_P): Redefine.
6816 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6817 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6818 (pdp11_modes_tieable_p): New function.
6819 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6820 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6821 (rs6000_modes_tieable_p): New function.
6822 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6823 * config/powerpcspe/powerpcspe.md: Update comment.
6824 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6825 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6826 (TARGET_MODES_TIEABLE_P): Redefine.
6827 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6828 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6829 (rl78_modes_tieable_p): New function.
6830 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6831 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6832 (rs6000_modes_tieable_p): New function.
6833 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6834 * config/rs6000/rs6000.md: Update comment.
6835 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6836 * config/rx/rx.c (rx_modes_tieable_p): New function.
6837 (TARGET_MODES_TIEABLE_P): Redefine.
6838 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6839 * config/s390/s390.c (s390_modes_tieable_p): New function.
6840 (TARGET_MODES_TIEABLE_P): Redefine.
6841 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6842 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6843 (sh_modes_tieable_p): New function.
6844 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6845 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6846 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6847 (sparc_modes_tieable_p): Make static.
6848 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6849 * config/spu/spu.c (spu_modes_tieable_p): New function.
6850 (TARGET_MODES_TIEABLE_P): Redefine.
6851 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6852 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6853 (TARGET_MODES_TIEABLE_P): Redefine.
6854 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6855 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6856 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6857 * config/v850/v850.c (v850_modes_tieable_p): New function.
6858 (TARGET_MODES_TIEABLE_P): Redefine.
6859 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6860 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6861 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6862 (visium_modes_tieable_p): New function.
6863 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6864 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6865 (xtensa_modes_tieable_p): New function.
6866 * system.h (MODES_TIEABLE_P): Poison.
6868 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6869 Alan Hayward <alan.hayward@arm.com>
6870 David Sherwood <david.sherwood@arm.com>
6872 * target.def (hard_regno_mode_ok): New hook.
6873 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6874 (TARGET_HARD_REGNO_MODE_OK): ...this.
6875 * doc/tm.texi.in: Regenerate.
6876 * hooks.h (hook_bool_uint_mode_true): Declare.
6877 * hooks.c (hook_bool_uint_mode_true): New function.
6878 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6880 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6881 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6882 instead of HARD_REGNO_MODE_OK.
6883 * caller-save.c: Include target.h.
6884 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6886 * combine.c (can_combine_p): Likewise.
6887 (combinable_i3pat): Likewise.
6888 (can_change_dest_mode): Likewise.
6889 * expr.c (init_expr_target): Likewise.
6890 (convert_move): Likewise.
6891 (convert_modes): Likewise.
6892 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6893 (setup_prohibited_mode_move_regs): Likewise.
6894 * ira.h (target_ira): Likewise.
6895 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6896 * lra-constraints.c (process_alt_operands): Likewise.
6897 (split_reg): Likewise.
6898 * recog.c (peep2_find_free_register): Likewise.
6899 * ree.c (combine_reaching_defs): Likewise.
6900 * regcprop.c (maybe_mode_change): Likewise.
6901 * reginfo.c (init_reg_sets_1): Likewise.
6902 (choose_hard_reg_mode): Likewise.
6903 (simplifiable_subregs): Likewise.
6904 * regrename.c (check_new_reg_p): Likewise.
6905 * reload.c (find_valid_class): Likewise.
6906 (find_valid_class_1): Likewise.
6907 (reload_inner_reg_of_subreg): Likewise.
6908 (push_reload): Likewise.
6909 (combine_reloads): Likewise.
6910 (find_dummy_reload): Likewise.
6911 (find_reloads): Likewise.
6912 * reload1.c (find_reg): Likewise.
6913 (set_reload_reg): Likewise.
6914 (allocate_reload_reg): Likewise.
6915 (choose_reload_regs): Likewise.
6916 (reload_adjust_reg_for_temp): Likewise.
6917 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6918 (simplify_subreg_regno): Likewise.
6919 * sel-sched.c (init_regs_for_mode): Likewise.
6920 * varasm.c (make_decl_rtl): Likewise.
6921 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6922 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6924 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6925 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6926 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6927 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6928 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6929 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6930 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6931 (arc_mode_class): Delete.
6932 (HARD_REGNO_MODE_OK): Delete.
6933 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6934 (arc_hard_regno_mode_ok): Rename old array to...
6935 (arc_hard_regno_mode_ok_modes): ...this.
6936 (arc_conditional_register_usage): Update accordingly.
6937 (arc_mode_class): Make static.
6938 (arc_hard_regno_mode_ok): New function.
6939 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6940 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6941 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6942 (arm_hard_regno_mode_ok): Make static.
6943 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6945 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6946 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6947 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6949 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6950 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6951 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6952 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6953 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6954 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6955 * config/bfin/predicates.md (valid_reg_operand): Use
6956 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6957 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6958 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6959 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6960 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6961 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6962 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6963 (cr16_hard_regno_mode_ok): Make static and return a bool.
6964 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6965 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6966 (cris_hard_regno_mode_ok): New function.
6967 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6968 (epiphany_mode_class): Delete.
6969 (HARD_REGNO_MODE_OK): Delete.
6970 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6971 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6972 (hard_regno_mode_ok): Rename to...
6973 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6974 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6975 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6977 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6978 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6979 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6980 (frv_hard_regno_mode_ok): Make static and return a bool.
6981 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6983 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6984 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6985 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6986 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6989 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6990 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6991 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6993 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6994 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6995 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6996 (ia64_hard_regno_mode_ok): New function.
6997 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6998 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6999 (iq2000_hard_regno_mode_ok): New function.
7000 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7001 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7002 (lm32_hard_regno_mode_ok): New function.
7003 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7004 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7005 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7006 instead of HARD_REGNO_MODE_OK.
7007 (m32c_hard_regno_ok): Rename to...
7008 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7009 (m32c_cannot_change_mode_class): Update accordingly.
7010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7011 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7012 (m32r_mode_class): Delete.
7013 (HARD_REGNO_MODE_OK): Delete.
7014 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7015 (m32r_hard_regno_mode_ok): Rename to...
7016 (m32r_hard_regno_modes): ...this.
7017 (m32r_mode_class): Make static.
7018 (m32r_hard_regno_mode_ok): New function.
7019 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7020 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7021 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7022 (m68k_hard_regno_mode_ok): Make static.
7023 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7024 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7025 (mcore_hard_regno_mode_ok): New function.
7026 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7027 (HARD_REGNO_MODE_OK): Delete.
7028 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7030 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7031 (microblaze_hard_regno_mode_ok): New function.
7032 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7033 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7034 (mips_hard_regno_mode_ok): Delete.
7035 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7036 (mips_hard_regno_mode_ok_p): ...this and make static.
7037 (mips_hard_regno_mode_ok_p): Rename to...
7038 (mips_hard_regno_mode_ok_uncached): ...this.
7039 (mips_hard_regno_mode_ok): New function.
7040 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7041 of HARD_REGNO_MODE_OK.
7042 (mips_option_override): Update after above name changes.
7043 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7044 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7045 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7046 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7047 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7048 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7049 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7050 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7051 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7052 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7053 (msp430_hard_regno_mode_ok): Make static and return a bool.
7054 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7055 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7056 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7058 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7059 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7060 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7061 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7062 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7063 (PA_HARD_REGNO_MODE_OK): ...this
7064 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7065 (PA_HARD_REGNO_MODE_OK): ...this.
7066 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7067 (pa_hard_regno_mode_ok): New function.
7068 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7069 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7070 (pdp11_hard_regno_mode_ok): New function.
7071 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7072 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7074 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7076 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7077 (rs6000_hard_regno_mode_ok): Rename to...
7078 (rs6000_hard_regno_mode_ok_uncached): ...this.
7079 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7080 (rs6000_hard_regno_mode_ok): New function.
7081 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7082 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7083 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7084 (riscv_hard_regno_mode_ok): ...this and make static.
7085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7086 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7087 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7088 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7089 (rl78_hard_regno_mode_ok): Make static and return bool.
7090 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7091 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7093 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7094 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7095 (rs6000_hard_regno_mode_ok): Rename to...
7096 (rs6000_hard_regno_mode_ok_uncached): ...this.
7097 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7098 (rs6000_hard_regno_mode_ok): New function.
7099 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7100 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7102 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7103 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7104 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7106 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7107 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7108 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7109 (sh_hard_regno_mode_ok): Make static.
7110 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7111 instead of HARD_REGNO_MODE_OK.
7112 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7113 (sparc_mode_class): Delete.
7114 (HARD_REGNO_MODE_OK): Delete.
7115 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7116 (hard_regno_mode_classes): Make static.
7117 (sparc_mode_class): Likewise.
7118 (sparc_hard_regno_mode_ok): New function.
7119 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7120 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7121 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7124 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7125 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7126 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7127 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7128 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7129 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7130 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7131 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7132 (visium_hard_regno_mode_ok): New function.
7133 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7134 instead of HARD_REGNO_MODE_OK.
7135 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7136 (HARD_REGNO_MODE_OK): Delete.
7137 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7138 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7139 (xtensa_option_override): Update accordingly.
7140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7141 (xtensa_hard_regno_mode_ok): New function.
7142 * system.h (HARD_REGNO_MODE_OK): Poison.
7144 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7145 Alan Hayward <alan.hayward@arm.com>
7146 David Sherwood <david.sherwood@arm.com>
7148 * target.def (hard_regno_call_part_clobbered): New hook.
7149 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7150 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7151 * doc/tm.texi: Regenerate.
7152 * hooks.h (hook_bool_uint_mode_false): Declare.
7153 * hooks.c (hook_bool_uint_mode_false): New function.
7154 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7155 * cselib.c (cselib_process_insn): Use
7156 targetm.hard_regno_call_part_clobbered instead of
7157 HARD_REGNO_CALL_PART_CLOBBERED.
7158 * ira-conflicts.c (ira_build_conflicts): Likewise.
7159 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7160 * lra-constraints.c (need_for_call_save_p): Likewise.
7161 * lra-lives.c: Include target.h.
7162 (check_pseudos_live_through_calls): Use
7163 targetm.hard_regno_call_part_clobbered instead of
7164 HARD_REGNO_CALL_PART_CLOBBERED.
7165 * regcprop.c: Include target.h.
7166 (copyprop_hardreg_forward_1): Use
7167 targetm.hard_regno_call_part_clobbered instead of
7168 HARD_REGNO_CALL_PART_CLOBBERED.
7169 * reginfo.c (choose_hard_reg_mode): Likewise.
7170 * regrename.c (check_new_reg_p): Likewise.
7171 * reload.c (find_equiv_reg): Likewise.
7172 * reload1.c (emit_reload_insns): Likewise.
7173 * sched-deps.c (deps_analyze_insn): Likewise.
7174 * sel-sched.c (init_regs_for_mode): Likewise.
7175 (mark_unavailable_hard_regs): Likewise.
7176 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7177 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7178 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7180 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7181 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7182 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7184 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7186 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7187 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7188 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7190 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7191 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7192 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7194 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7195 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7197 * config/powerpcspe/powerpcspe.c
7198 (rs6000_hard_regno_call_part_clobbered): New function.
7199 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7200 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7201 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7203 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7204 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7205 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7207 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7208 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7209 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7211 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7212 Alan Hayward <alan.hayward@arm.com>
7213 David Sherwood <david.sherwood@arm.com>
7215 * rtl.h (subreg_memory_offset): Declare.
7216 * emit-rtl.c (subreg_memory_offset): New function.
7217 * expmed.c (store_bit_field_1): Use it.
7218 * expr.c (undefined_operand_subword_p): Likewise.
7219 * simplify-rtx.c (simplify_subreg): Likewise.
7221 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7223 PR rtl-optimization/57448
7226 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7227 using atomic_load pattern.
7228 (expand_atomic_store): Likewise.
7230 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7233 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7234 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7235 handling. Use replace_call_with_value with NULL instead of
7236 gsi_replace, unlink_stmt_vdef and release_defs.
7238 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7241 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7243 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7246 * rtl.h (emit_library_call_value_1): Declare.
7247 (emit_library_call): Replace declaration with a series of overloads.
7248 Remove the parameter count argument.
7249 (emit_library_call_value): Likewise.
7250 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7251 with an "rtx_mode_t *".
7252 (emit_library_call_value): Delete.
7253 (emit_library_call): Likewise.
7254 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7255 (asan_emit_allocas_unpoison): Likewise.
7256 * builtins.c (expand_builtin_powi): Likewise.
7257 (expand_asan_emit_allocas_unpoison): Likewise.
7258 * cfgexpand.c (expand_main_function): Likewise.
7259 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7260 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7261 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7262 * config/arm/arm.c (arm_trampoline_init): Likewise.
7263 (arm_call_tls_get_addr): Likewise.
7264 (arm_expand_divmod_libfunc): Likewise.
7265 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7266 (smulsi3_highpart): Likewise.
7267 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7268 (c6x_expand_compare): Likewise.
7269 (c6x_expand_movmem): Likewise.
7270 * config/frv/frv.c (frv_trampoline_init): Likewise.
7271 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7272 (ix86_expand_divmod_libfunc): Likewise.
7273 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7274 (ia64_expand_compare): Likewise.
7275 (ia64_profile_hook): Likewise.
7276 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7277 (nonlocal_goto): Likewise.
7278 (restore_stack_nonlocal): Likewise.
7279 * config/m32r/m32r.c (block_move_call): Likewise.
7280 (m32r_trampoline_init): Likewise.
7281 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7282 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7283 (m68k_call_m68k_read_tp): Likewise.
7284 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7285 (microblaze_expand_divide): Likewise.
7286 * config/mips/mips.h (mips_args): Likewise.
7287 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7288 (MIPS_ICACHE_SYNC): Likewise.
7289 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7290 (nios2_trampoline_init): Likewise.
7291 * config/pa/pa.c (hppa_tls_call): Likewise.
7292 (pa_trampoline_init): Likewise.
7293 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7294 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7295 (expand_strn_compare): Likewise.
7296 (rs6000_generate_compare): Likewise.
7297 (rs6000_expand_float128_convert): Likewise.
7298 (output_profile_hook): Likewise.
7299 (rs6000_trampoline_init): Likewise.
7300 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7301 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7302 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7303 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7304 (rs6000_generate_compare): Likewise.
7305 (rs6000_expand_float128_convert): Likewise.
7306 (output_profile_hook): Likewise.
7307 (rs6000_trampoline_init): Likewise.
7308 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7309 * config/sh/sh.c (sh_trampoline_init): Likewise.
7310 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7311 (sparc_emit_float_lib_cmp): Likewise.
7312 (sparc32_initialize_trampoline): Likewise.
7313 (sparc64_initialize_trampoline): Likewise.
7314 (sparc_profile_hook): Likewise.
7315 * config/spu/spu.c (ea_load_store): Likewise.
7316 * config/spu/spu.md (floatunssidf2): Likewise.
7317 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7318 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7319 * config/visium/visium.c (expand_block_move_4): Likewise.
7320 (expand_block_move_2): Likewise.
7321 (expand_block_move_1): Likewise.
7322 (expand_block_set_4): Likewise.
7323 (expand_block_set_2): Likewise.
7324 (expand_block_set_1): Likewise.
7325 (visium_trampoline_init): Likewise.
7326 (visium_profile_hook): Likewise.
7327 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7328 (xtensa_setup_frame_addresses): Likewise.
7329 (xtensa_trampoline_init): Likewise.
7330 * except.c (sjlj_emit_function_enter): Likewise.
7331 (sjlj_emit_function_exit): Likewise.
7332 * explow.c (allocate_dynamic_stack_space): Likewise.
7333 (probe_stack_range): Likewise.
7334 * expr.c (convert_mode_scalar): Likewise.
7335 * optabs.c (expand_binop): Likewise.
7336 (expand_twoval_binop_libfunc): Likewise.
7337 (expand_unop): Likewise.
7338 (prepare_cmp_insn): Likewise.
7339 (prepare_float_lib_cmp): Likewise.
7340 (expand_float): Likewise.
7341 (expand_fix): Likewise.
7342 (expand_fixed_convert): Likewise.
7343 (maybe_emit_sync_lock_test_and_set): Likewise.
7344 (expand_atomic_compare_and_swap): Likewise.
7345 (expand_mem_thread_fence): Likewise.
7346 (expand_atomic_fetch_op): Likewise.
7348 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7350 * doc/generic.texi (OpenACC): Adjust URL.
7351 * doc/invoke.texi (C Dialect Options): Ditto.
7353 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7355 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7356 predicate for operand 1. Add (m,<S>) constraint.
7357 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7358 Prevent memory operand 1 with register operand 2.
7360 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7362 PR rtl-optimization/82024
7363 * combine.c (try_combine): If the combination result is a PARALLEL,
7364 and we only need to retain the SET in there that would be placed
7365 at I2, check that we can place that at I3 instead, before doing so.
7367 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7370 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7372 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7373 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7376 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7377 Jeff Law <law@redhat.com>
7379 * varasm.c (bss_initializer_p): Do not put constants into .bss
7380 (categorize_decl_for_section): Handle bss_initializer_p returning
7381 false when DECL_INITIAL is NULL.
7383 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7386 * config/s390/s390.c (s390_can_inline_p): New function.
7388 2017-09-01 Jeff Law <law@redhat.com>
7390 PR tree-optimization/82052
7391 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7392 Always initialize the returned slot after a hash table miss
7393 when INSERT is true.
7395 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7397 * config/s390/s390.md (mem_signal_fence): Remove.
7398 * doc/md.texi (mem_signal_fence): Remove.
7399 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7401 * target-insns.def (mem_signal_fence): Remove.
7403 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7406 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7409 * asan.c (create_odr_indicator): Strip name encoding from assembler
7410 name before appending it after __odr_asan_.
7412 2017-09-01 Martin Liska <mliska@suse.cz>
7414 PR tree-optimization/82059
7415 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7416 frequency only when an edge is redirected.
7418 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7420 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7421 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7422 (arc_conditional_register_usage): Remove ARC600 lp_count
7424 (arc_file_start): Emit Tag_ARC_CPU_variation.
7425 (arc_can_use_doloop_p): New conditions to use ZOLs.
7426 (hwloop_fail): New function.
7427 (hwloop_optimize): Likewise.
7428 (hwloop_pattern_reg): Likewise.
7429 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7430 (arc_reorg_loops): New function, calls reorg_loops.
7431 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7432 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7434 (arc_loop_hazard): Remove function, functionality moved into
7436 (arc_hazard): Remove arc_loop_hazard call.
7437 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7438 into hwloop_optimize.
7439 (arc_label_align): Remove ZOL handling.
7440 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7441 * config/arc/arc.md (doloop_begin): Remove pattern.
7442 (doloop_begin_i): Likewise.
7443 (doloop_end_i): Likewise.
7444 (doloop_fallback): Likewise.
7445 (doloop_fallback_m): Likewise.
7446 (doloop_end): Reimplement expand.
7447 (arc_lp): New pattern for LP instruction.
7448 (loop_end): New pattern.
7449 (loop_fail): Likewise.
7450 (decrement_and_branch_until_zero): Likewise.
7451 * config/arc/arc.opt (mlpc-width): New option.
7452 * doc/invoke.texi (mlpc-width): Document option.
7454 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7456 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7457 (arc_ccfsm_advance): Fix checking for delay slots.
7458 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7460 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7462 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7464 (movhi_insn): Update store instruction constraint which are saving
7465 6-bit short immediates.
7466 (movsi_insn): Consider also short scaled load operations.
7467 (zero_extendhisi2_i): Use Usd constraint instead of T.
7468 (extendhisi2_i): Add q constraint.
7469 (arc_clzsi2): Add type and length attributes.
7470 (arc_ctzsi2): Likewise.
7471 * config/arc/constraints.md (Usc): Update constraint, the
7472 assembler can parse two relocations for a single instruction.
7474 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7476 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7477 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7479 2017-08-31 Olivier Hainque <hainque@adacore.com>
7481 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7482 match as powerpc-wrs-vxworks*.
7484 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7486 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7487 register constraint for by-element operand.
7488 (aarch64_mls_elt_merge<mode>): Likewise.
7490 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7492 * config/arc/arc.c (arc_can_follow_jump): Check for short
7495 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7497 * config.gcc: Use g.opt for arc.
7498 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7499 functionality moved to ...
7500 (legitimate_scaled_address_p): New function, ...here.
7501 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7502 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7503 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7505 (arc_override_options): Handle G option.
7506 (arc_output_pic_addr_const): Correct function definition.
7507 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7508 (arc_decl_anon_ns_mem_p): Delete.
7509 (arc_in_small_data_p): Overhaul this function to take into
7510 consideration the value given via G option.
7511 (arc_rewrite_small_data_1): Renamed and corrected old
7512 arc_rewrite_small_data function.
7513 (arc_rewrite_small_data): New function.
7514 (small_data_pattern): Don't use pic_offset_table_rtx.
7515 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7516 * config/arc/simdext.md (movmisalignv2hi): Use
7517 prepare_move_operands function.
7519 (movmisalign*): Likewise.
7520 * doc/invoke.texi (ARC options): Document -G option.
7522 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7524 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7526 * config/arc/arc.c (arc_print_operand): Output scalled address for
7527 sdata whenever is possible.
7528 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7529 load/stores are available.
7530 (compact_sda_memory_operand): Check for the alignment required by
7531 code density instructions.
7532 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7534 * config/arc/constraints.md (Usd): Update constraint.
7535 (Us0): New constraint.
7536 (Usc): Update constraint.
7538 2017-08-31 Richard Biener <rguenther@suse.de>
7541 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7544 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7546 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7547 Resize type_signature.
7549 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7550 Alan Hayward <alan.hayward@arm.com>
7551 David Sherwood <david.sherwood@arm.com>
7553 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7554 subregs whose inner modes can be stored in GPRs.
7555 (aarch64_classify_index): Likewise.
7557 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7562 (V_INT_EQUIV): ...this.
7563 (v_cmp_result): Rename to...
7564 (v_int_equiv): ...this.
7565 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7566 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7567 (copysign<mode>3): Likewise.
7568 (aarch64_simd_bsl<mode>_internal): Likewise.
7569 (aarch64_simd_bsl<mode>): Likewise.
7570 (vec_cmp<mode><mode>): Likewise.
7571 (vcond<mode><mode>): Likewise.
7572 (vcond<v_cmp_mixed><mode>): Likewise.
7573 (vcondu<mode><v_cmp_mixed>): Likewise.
7574 (aarch64_cm<optab><mode>): Likewise.
7575 (aarch64_cmtst<mode>): Likewise.
7576 (aarch64_fac<optab><mode>): Likewise.
7577 (vec_perm_const<mode>): Likewise.
7578 (vcond_mask_<mode><v_cmp_result>): Rename to...
7579 (vcond_mask_<mode><v_int_equiv>): ...this.
7580 (vec_cmp<mode><v_cmp_result>): Rename to...
7581 (vec_cmp<mode><v_int_equiv>): ...this.
7583 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7584 Alan Hayward <alan.hayward@arm.com>
7585 David Sherwood <david.sherwood@arm.com>
7587 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7589 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7590 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7591 (UNSPEC_LD4_DREG): New unspecs.
7592 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7593 (aarch64_ld2<mode>_dreg_be): Replace with...
7594 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7596 (aarch64_ld3<mode>_dreg_le)
7597 (aarch64_ld3<mode>_dreg_be): Replace with...
7598 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7600 (aarch64_ld4<mode>_dreg_le)
7601 (aarch64_ld4<mode>_dreg_be): Replace with...
7602 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7605 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7607 PR tree-optimization/81987
7608 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7609 insert an initializer in a location not dominated by the stride
7612 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7614 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7615 on the entire header of the finally block in the fallthru case.
7617 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7619 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7621 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7623 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7624 rs6000_emit_move_from_cr and call renamed function.
7625 (rs6000_emit_prologue): Call renamed functions.
7626 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7627 movesi_from_cr, remove volatile CRs.
7629 2017-08-30 Jon Beniston <jon@beniston.com>
7630 Richard Biener <rguenther@suse.de>
7632 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7633 of VECTOR_MODE_P check.
7634 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7635 element vector types.
7637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7639 * df.h (df_read_modify_subreg_p): Remove in favor of...
7640 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7641 const_rtx instead of an rtx.
7642 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7643 * df-problems.c (df_word_lr_mark_ref): Likewise.
7644 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7645 (mark_pseudo_reg_dead): Likewise.
7646 (mark_ref_dead): Likewise.
7647 * reginfo.c (init_subregs_of_mode): Likewise.
7648 * sched-deps.c (sched_analyze_1): Likewise.
7649 * df-scan.c (df_def_record_1): Likewise.
7650 (df_uses_record): Likewise.
7651 (df_read_modify_subreg_p): Remove in favor of...
7652 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7653 const_rtx instead of an rtx.
7655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7656 Alan Hayward <alan.hayward@arm.com>
7657 David Sherwood <david.sherwood@arm.com>
7659 * rtl.h (partial_subreg_p): New function.
7660 * caller-save.c (save_call_clobbered_regs): Use it.
7661 * calls.c (expand_call): Likewise.
7662 * combine.c (combinable_i3pat): Likewise.
7663 (simplify_set): Likewise.
7664 (make_extraction): Likewise.
7665 (make_compound_operation_int): Likewise.
7666 (gen_lowpart_or_truncate): Likewise.
7667 (force_to_mode): Likewise.
7668 (make_field_assignment): Likewise.
7669 (reg_truncated_to_mode): Likewise.
7670 (record_truncated_value): Likewise.
7671 (move_deaths): Likewise.
7672 * cse.c (record_jump_cond): Likewise.
7673 (cse_insn): Likewise.
7674 * cselib.c (cselib_lookup_1): Likewise.
7675 * expmed.c (extract_bit_field_using_extv): Likewise.
7676 * function.c (assign_parm_setup_reg): Likewise.
7677 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7678 * ira-build.c (create_insn_allocnos): Likewise.
7679 * lra-coalesce.c (merge_pseudos): Likewise.
7680 * lra-constraints.c (match_reload): Likewise.
7681 (simplify_operand_subreg): Likewise.
7682 (curr_insn_transform): Likewise.
7683 * lra-lives.c (process_bb_lives): Likewise.
7684 * lra.c (new_insn_reg): Likewise.
7685 (lra_substitute_pseudo): Likewise.
7686 * regcprop.c (mode_change_ok): Likewise.
7687 (maybe_mode_change): Likewise.
7688 (copyprop_hardreg_forward_1): Likewise.
7689 * reload.c (push_reload): Likewise.
7690 (find_reloads): Likewise.
7691 (find_reloads_subreg_address): Likewise.
7692 * reload1.c (alter_reg): Likewise.
7693 (eliminate_regs_1): Likewise.
7694 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7696 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7698 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7699 back to if statements, including unpack.
7701 2017-08-30 Martin Liska <mliska@suse.cz>
7704 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7706 (func_checker::compare_asm_inputs_outputs): ... this function.
7707 (func_checker::compare_gimple_asm): Use the function to compare
7708 also ASM constrains.
7709 * ipa-icf-gimple.h: Rename the function.
7711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7712 Alan Hayward <alan.hayward@arm.com>
7713 David Sherwood <david.sherwood@arm.com>
7715 * coretypes.h (complex_mode): New type.
7716 * gdbhooks.py (build_pretty_printer): Handle it.
7717 * machmode.h (complex_mode): New class.
7718 (complex_mode::includes_p): New function.
7719 (is_complex_int_mode): Likewise.
7720 (is_complex_float_mode): Likewise.
7721 * genmodes.c (get_mode_class): Handle complex mode classes.
7722 * function.c (expand_function_end): Use is_complex_int_mode.
7724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7725 Alan Hayward <alan.hayward@arm.com>
7726 David Sherwood <david.sherwood@arm.com>
7728 * coretypes.h (scalar_mode_pod): New typedef.
7729 * gdbhooks.py (build_pretty_printer): Handle it.
7730 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7731 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7732 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7733 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7737 Alan Hayward <alan.hayward@arm.com>
7738 David Sherwood <david.sherwood@arm.com>
7740 * machmode.h (mode_for_vector): Take a scalar_mode instead
7742 * stor-layout.c (mode_for_vector): Likewise.
7743 * explow.c (promote_mode): Use as_a <scalar_mode>.
7744 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7747 Alan Hayward <alan.hayward@arm.com>
7748 David Sherwood <david.sherwood@arm.com>
7750 * target.def (preferred_simd_mode): Take a scalar_mode
7751 instead of a machine_mode.
7752 * targhooks.h (default_preferred_simd_mode): Likewise.
7753 * targhooks.c (default_preferred_simd_mode): Likewise.
7754 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7755 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7756 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7757 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7758 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7759 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7760 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7761 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7763 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7764 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7765 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7766 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7767 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7768 * doc/tm.texi: Regenerate.
7769 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7773 Alan Hayward <alan.hayward@arm.com>
7774 David Sherwood <david.sherwood@arm.com>
7776 * target.def (scalar_mode_supported_p): Take a scalar_mode
7777 instead of a machine_mode.
7778 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7779 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7780 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7781 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7782 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7783 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7784 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7785 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7786 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7787 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7788 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7789 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7790 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7791 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7793 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7794 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7795 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7796 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7797 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7799 * doc/tm.texi: Regenerate.
7801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7802 Alan Hayward <alan.hayward@arm.com>
7803 David Sherwood <david.sherwood@arm.com>
7805 * coretypes.h (opt_scalar_mode): New typedef.
7806 * gdbhooks.py (build_pretty_printers): Handle it.
7807 * machmode.h (mode_iterator::get_2xwider): Add overload for
7809 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7811 * expr.c (convert_mode_scalar): Likewise.
7812 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7813 * optabs.c (expand_float): Likewise.
7814 (expand_fix): Likewise.
7815 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7818 Alan Hayward <alan.hayward@arm.com>
7819 David Sherwood <david.sherwood@arm.com>
7821 * optabs.c (expand_float): Explicitly check for scalars before
7822 using a branching expansion.
7823 (expand_fix): Likewise.
7825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7826 Alan Hayward <alan.hayward@arm.com>
7827 David Sherwood <david.sherwood@arm.com>
7829 * expr.c (convert_mode): Split scalar handling out into...
7830 (convert_mode_scalar): ...this new function. Treat the modes
7833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7834 Alan Hayward <alan.hayward@arm.com>
7835 David Sherwood <david.sherwood@arm.com>
7837 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7839 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7842 Alan Hayward <alan.hayward@arm.com>
7843 David Sherwood <david.sherwood@arm.com>
7845 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7846 rather than a machine_mode.
7847 (fixed_from_string): Likewise.
7848 (fixed_convert): Likewise.
7849 (fixed_convert_from_int): Likewise.
7850 (fixed_convert_from_real): Likewise.
7851 (real_convert_from_fixed): Likewise.
7852 * fixed-value.c (fixed_from_double_int): Likewise.
7853 (fixed_from_string): Likewise.
7854 (fixed_convert): Likewise.
7855 (fixed_convert_from_int): Likewise.
7856 (fixed_convert_from_real): Likewise.
7857 (real_convert_from_fixed): Likewise.
7858 * config/avr/avr.c (avr_out_round): Use as_a <scalar_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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7865 of separate mode class checks. Do not allow vector modes here.
7866 (immed_wide_int_const): Use as_a <scalar_mode>.
7867 * explow.c (trunc_int_for_mode): Likewise.
7868 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7869 (wi::shwi): Likewise.
7870 (wi::min_value): Likewise.
7871 (wi::max_value): Likewise.
7872 * dwarf2out.c (loc_descriptor): Likewise.
7873 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7877 Alan Hayward <alan.hayward@arm.com>
7878 David Sherwood <david.sherwood@arm.com>
7880 * tree.h (SCALAR_TYPE_MODE): New macro.
7881 * expr.c (expand_expr_addr_expr_1): Use it.
7882 (expand_expr_real_2): Likewise.
7883 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7884 (fold_convert_const_fixed_from_int): Likewise.
7885 (fold_convert_const_fixed_from_real): Likewise.
7886 (native_encode_fixed): Likewise
7887 (native_encode_complex): Likewise
7888 (native_encode_vector): Likewise.
7889 (native_interpret_fixed): Likewise.
7890 (native_interpret_real): Likewise.
7891 (native_interpret_complex): Likewise.
7892 (native_interpret_vector): Likewise.
7893 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7894 (simd_clone_adjust_argument_types): Likewise.
7895 (simd_clone_init_simd_arrays): Likewise.
7896 (simd_clone_adjust): Likewise.
7897 * stor-layout.c (layout_type): Likewise.
7898 * tree.c (build_minus_one_cst): Likewise.
7899 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7900 * tree-inline.c (estimate_move_cost): Likewise.
7901 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7902 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7903 (vectorizable_reduction): Likewise.
7904 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7905 (vect_recog_mixed_size_cond_pattern): Likewise.
7906 (check_bool_pattern): Likewise.
7907 (adjust_bool_pattern): Likewise.
7908 (search_type_for_mask_1): Likewise.
7909 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7910 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7911 (vectorizable_load): Likewise.
7912 (vectorizable_store): Likewise.
7913 * ubsan.c (ubsan_encode_value): Likewise.
7914 * varasm.c (output_constant): Likewise.
7916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7917 Alan Hayward <alan.hayward@arm.com>
7918 David Sherwood <david.sherwood@arm.com>
7920 * coretypes.h (scalar_mode): New class.
7921 * machmode.h (scalar_mode): Likewise.
7922 (scalar_mode::includes_p): New function.
7923 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7924 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7925 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7926 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7927 * expmed.c (store_bit_field_1): Likewise.
7928 (extract_bit_field_1): Likewise.
7929 * expr.c (write_complex_part): Likewise.
7930 (read_complex_part): Likewise.
7931 (emit_move_complex_push): Likewise.
7932 (expand_expr_real_2): Likewise.
7933 * function.c (assign_parm_setup_reg): Likewise.
7934 (assign_parms_unsplit_complex): Likewise.
7935 * optabs.c (expand_binop): Likewise.
7936 * rtlanal.c (subreg_get_info): Likewise.
7937 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7938 * varasm.c (output_constant_pool_2): Likewise.
7940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7941 Alan Hayward <alan.hayward@arm.com>
7942 David Sherwood <david.sherwood@arm.com>
7944 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7946 (expmed_mult_highpart_optab): Likewise.
7947 (expmed_mult_highpart): Likewise.
7949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7950 Alan Hayward <alan.hayward@arm.com>
7951 David Sherwood <david.sherwood@arm.com>
7953 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7954 instead of a machine_mode.
7955 (builtin_memset_read_str): Likewise.
7956 * builtins.c (c_readstr): Likewise.
7957 (builtin_memcpy_read_str): Likewise.
7958 (builtin_strncpy_read_str): Likewise.
7959 (builtin_memset_read_str): Likewise.
7960 (builtin_memset_gen_str): Likewise.
7961 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7962 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7963 instead of a machine_mode.
7964 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7966 (make_extraction): Likewise.
7967 (try_widen_shift_mode): Take and return scalar_int_modes instead
7969 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7970 a scalar_int_mode instead of a machine_mode.
7971 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7972 (avr_addr_space_pointer_mode): Likewise.
7973 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7974 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7975 (msp430_unwind_word_mode): Likewise.
7976 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7977 (spu_addr_space_pointer_mode): Likewise.
7978 (spu_addr_space_address_mode): Likewise.
7979 (spu_libgcc_cmp_return_mode): Likewise.
7980 (spu_libgcc_shift_count_mode): Likewise.
7981 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7982 (rl78_addr_space_pointer_mode): Likewise.
7983 (fl78_unwind_word_mode): Likewise.
7984 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7986 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7987 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7988 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7989 (mips_valid_pointer_mode): Likewise.
7990 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7991 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7992 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7994 (ft32_addr_space_address_mode): Likewise.
7995 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7996 scalar_int_mode instead of a machine_mode.
7997 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7999 (m32c_addr_space_address_mode): Likewise.
8000 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8001 (rs6000_eh_return_filter_mode): Likewise.
8002 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8003 (rs6000_eh_return_filter_mode): Likewise.
8004 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8005 (s390_libgcc_shift_count_mode): Likewise.
8006 (s390_unwind_word_mode): Likewise.
8007 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8009 * target.def (mode_rep_extended): Likewise.
8010 (valid_pointer_mode): Likewise.
8011 (addr_space.valid_pointer_mode): Likewise.
8012 (eh_return_filter_mode): Return a scalar_int_mode rather than
8014 (libgcc_cmp_return_mode): Likewise.
8015 (libgcc_shift_count_mode): Likewise.
8016 (unwind_word_mode): Likewise.
8017 (addr_space.pointer_mode): Likewise.
8018 (addr_space.address_mode): Likewise.
8019 * doc/tm.texi: Regenerate.
8020 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8022 (do_jump): Use scalar_int_mode for local variables.
8023 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8024 rather than a machine_mode.
8025 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8026 (scompare_loc_descriptor_wide): Likewise.
8027 (scompare_loc_descriptor_narrow): Likewise.
8028 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8030 * except.c (sjlj_emit_dispatch_table): Likewise.
8031 (expand_builtin_eh_copy_values): Likewise.
8032 * explow.c (convert_memory_address_addr_space_1): Likewise.
8033 Take a scalar_int_mode rather than a machine_mode.
8034 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8035 than a machine_mode.
8036 (memory_address_addr_space): Use scalar_int_mode for local variables.
8037 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8038 rather than a machine_mode.
8039 * expmed.c (mask_rtx): Likewise.
8040 (init_expmed_one_conv): Likewise.
8041 (expand_mult_highpart_adjust): Likewise.
8042 (extract_high_half): Likewise.
8043 (expmed_mult_highpart_optab): Likewise.
8044 (expmed_mult_highpart): Likewise.
8045 (expand_smod_pow2): Likewise.
8046 (expand_sdiv_pow2): Likewise.
8047 (emit_store_flag_int): Likewise.
8048 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8050 (extract_low_bits): Likewise.
8051 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8053 * expr.c (pieces_addr::adjust): Likewise.
8054 (can_store_by_pieces): Likewise.
8055 (store_by_pieces): Likewise.
8056 (clear_by_pieces_1): Likewise.
8057 (expand_expr_addr_expr_1): Likewise.
8058 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8059 (expand_expr_real_1): Likewise.
8060 (try_casesi): Likewise.
8061 * final.c (shorten_branches): Likewise.
8062 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8063 type of "mode" to machine_mode.
8064 * internal-fn.c (expand_arith_overflow_result_store): Take a
8065 scalar_int_mode rather than a machine_mode.
8066 (expand_mul_overflow): Use scalar_int_mode for local variables.
8067 * loop-doloop.c (doloop_modify): Likewise.
8068 (doloop_optimize): Likewise.
8069 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8070 than a machine_mode.
8071 (expand_doubleword_shift_condmove): Likewise.
8072 (expand_doubleword_shift): Likewise.
8073 (expand_doubleword_clz): Likewise.
8074 (expand_doubleword_popcount): Likewise.
8075 (expand_doubleword_parity): Likewise.
8076 (expand_absneg_bit): Use scalar_int_mode for local variables.
8077 (prepare_float_lib_cmp): Likewise.
8078 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8079 rather than a machine_mode.
8080 (convert_memory_address_addr_space): Likewise.
8081 (get_mode_bounds): Likewise.
8082 (get_address_mode): Return a scalar_int_mode rather than a
8084 * rtlanal.c (get_address_mode): Likewise.
8085 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8086 than a machine_mode.
8087 * targhooks.c (default_mode_rep_extended): Likewise.
8088 (default_valid_pointer_mode): Likewise.
8089 (default_addr_space_valid_pointer_mode): Likewise.
8090 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8091 than a machine_mode.
8092 (default_libgcc_cmp_return_mode): Likewise.
8093 (default_libgcc_shift_count_mode): Likewise.
8094 (default_unwind_word_mode): Likewise.
8095 (default_addr_space_pointer_mode): Likewise.
8096 (default_addr_space_address_mode): Likewise.
8097 * targhooks.h (default_eh_return_filter_mode): Likewise.
8098 (default_libgcc_cmp_return_mode): Likewise.
8099 (default_libgcc_shift_count_mode): Likewise.
8100 (default_unwind_word_mode): Likewise.
8101 (default_addr_space_pointer_mode): Likewise.
8102 (default_addr_space_address_mode): Likewise.
8103 (default_mode_rep_extended): Take a scalar_int_mode rather than
8105 (default_valid_pointer_mode): Likewise.
8106 (default_addr_space_valid_pointer_mode): Likewise.
8107 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8109 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8110 rather than a machine_mode.
8111 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8112 for local variables.
8113 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8114 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8115 than a machine_mode.
8117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8118 Alan Hayward <alan.hayward@arm.com>
8119 David Sherwood <david.sherwood@arm.com>
8121 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8122 the mode argument to scalar_int_mode.
8123 (do_jump_by_parts_zero_rtx): Likewise.
8124 (do_jump_by_parts_equality_rtx): Likewise.
8125 (do_jump_by_parts_greater): Take a mode argument.
8126 (do_jump_by_parts_equality): Likewise.
8127 (do_jump_1): Update calls accordingly.
8129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8130 Alan Hayward <alan.hayward@arm.com>
8131 David Sherwood <david.sherwood@arm.com>
8133 * is-a.h (safe_dyn_cast): New function.
8134 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8135 (jump_table_for_label): Likewise.
8136 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8137 instead of an rtx_insn *.
8138 (shorten_branches): Use dyn_cast instead of LABEL_P and
8139 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8140 rtx_jump_table_data::get_data_mode.
8141 (final_scan_insn): Likewise.
8143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8144 Alan Hayward <alan.hayward@arm.com>
8145 David Sherwood <david.sherwood@arm.com>
8147 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8148 trying to combine a full-register integer set with a subreg
8151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8152 Alan Hayward <alan.hayward@arm.com>
8153 David Sherwood <david.sherwood@arm.com>
8155 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8156 that is always either address_mode or pointer_mode.
8158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8159 Alan Hayward <alan.hayward@arm.com>
8160 David Sherwood <david.sherwood@arm.com>
8162 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8163 when the two are known to be equal.
8165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8166 Alan Hayward <alan.hayward@arm.com>
8167 David Sherwood <david.sherwood@arm.com>
8169 * simplify-rtx.c (simplify_const_unary_operation): Use
8170 is_a <scalar_int_mode> instead of checking for a nonzero
8171 precision. Forcibly convert op_mode to a scalar_int_mode
8172 in that case. More clearly differentiate the operand and
8173 result modes and use the former when deciding what the value
8174 of a count-bits operation should be. Use is_int_mode instead
8175 of checking for a MODE_INT. Remove redundant check for whether
8176 this mode has a zero precision.
8178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8179 Alan Hayward <alan.hayward@arm.com>
8180 David Sherwood <david.sherwood@arm.com>
8182 * optabs.c (widen_leading): Change the type of the mode argument
8183 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8184 (widen_bswap): Likewise.
8185 (expand_parity): Likewise.
8186 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8187 (expand_ffs): Likewise.
8188 (epand_unop): Check for scalar integer modes before calling the
8191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8192 Alan Hayward <alan.hayward@arm.com>
8193 David Sherwood <david.sherwood@arm.com>
8195 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8197 (expand_expr_real_1): Update call accordingly.
8199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8200 Alan Hayward <alan.hayward@arm.com>
8201 David Sherwood <david.sherwood@arm.com>
8203 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8204 value_mode arguments. Use scalar_int_mode internally.
8205 (store_bit_field_1): Rename the new integer mode from imode
8206 to op0_mode and use it instead of GET_MODE (op0). Update calls
8207 to store_split_bit_field, store_bit_field_using_insv and
8208 store_fixed_bit_field.
8209 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8210 Use scalar_int_mode internally. Use a bit count rather than a mode
8211 when calculating the largest bit size for get_best_mode.
8212 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8213 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8214 Remove assertion that OP0 has a scalar integer mode.
8215 (store_split_bit_field): Add op0_mode and value_mode arguments.
8216 Update calls to extract_fixed_bit_field.
8217 (extract_bit_field_using_extv): Add an op0_mode argument.
8218 Use scalar_int_mode internally.
8219 (extract_bit_field_1): Rename the new integer mode from imode to
8220 op0_mode and use it instead of GET_MODE (op0). Update calls to
8221 extract_split_bit_field, extract_bit_field_using_extv and
8222 extract_fixed_bit_field.
8223 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8224 to extract_split_bit_field and extract_fixed_bit_field_1.
8225 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8226 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8228 (extract_split_bit_field): Add an op0_mode argument. Update call
8229 to extract_fixed_bit_field.
8231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8232 Alan Hayward <alan.hayward@arm.com>
8233 David Sherwood <david.sherwood@arm.com>
8235 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8236 * explow.c (hard_function_value): Likewise.
8237 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8238 convert_to_mode call outside the loop.
8239 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8240 for the mode iterator. Require the mode specified by max_pieces
8242 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8244 (copy_blkmode_to_reg): Likewise.
8245 (set_storage_via_setmem): Likewise.
8246 * optabs.c (prepare_cmp_insn): Likewise.
8247 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8248 * stor-layout.c (finish_bitfield_representative): Likewise.
8250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8251 Alan Hayward <alan.hayward@arm.com>
8252 David Sherwood <david.sherwood@arm.com>
8254 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8255 * expr.c (convert_move): Use them.
8256 (convert_modes): Likewise.
8257 (store_expr_with_bounds): 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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8264 parameter for the mode of "x". Remove the "known_x", "known_mode"
8265 and "known_ret" arguments. Change the type of the mode argument
8267 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8268 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8269 (reg_num_sign_bit_copies_for_combine): Likewise.
8270 * rtlanal.c (nonzero_bits1): Likewise.
8271 (num_sign_bit_copies1): Likewise.
8272 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8273 (reg_num_sign_bit_copies_general): Likewise.
8274 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8275 (reg_nonzero_bits_general): Likewise.
8277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8278 Alan Hayward <alan.hayward@arm.com>
8279 David Sherwood <david.sherwood@arm.com>
8281 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8282 than in subroutines. Return 1 for non-integer modes.
8283 (cached_num_sign_bit_copies): Change the type of the mode parameter
8285 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8286 classes. Handle CONST_INT_P first and then check whether X also
8287 has a scalar integer mode. Check the same thing for inner registers
8288 of a SUBREG and for values that are being extended or truncated.
8290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8291 Alan Hayward <alan.hayward@arm.com>
8292 David Sherwood <david.sherwood@arm.com>
8294 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8295 in subroutines. Return the mode mask for non-integer modes.
8296 (cached_nonzero_bits): Change the type of the mode parameter
8298 (nonzero_bits1): Likewise. Remove early exit for other mode
8299 classes. Handle CONST_INT_P first and then check whether X
8300 also has a scalar integer mode.
8302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8303 Alan Hayward <alan.hayward@arm.com>
8304 David Sherwood <david.sherwood@arm.com>
8306 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8307 Return a scalar_int_mode and assert that the size is greater than
8309 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8310 (op_by_pieces_d::op_by_pieces_d): Likewise.
8311 (op_by_pieces_d::run): Likewise.
8312 (can_store_by_pieces): Likewise.
8314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8315 Alan Hayward <alan.hayward@arm.com>
8316 David Sherwood <david.sherwood@arm.com>
8318 * combine.c (extract_left_shift): Add a mode argument and update
8320 (make_compound_operation_int): Change the type of the mode parameter
8321 to scalar_int_mode and update the call to extract_left_shift.
8323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8324 Alan Hayward <alan.hayward@arm.com>
8325 David Sherwood <david.sherwood@arm.com>
8327 * combine.c (simplify_and_const_int): Change the type of the mode
8328 parameter to scalar_int_mode.
8329 (simplify_and_const_int_1): Likewise. Update recursive call.
8331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8332 Alan Hayward <alan.hayward@arm.com>
8333 David Sherwood <david.sherwood@arm.com>
8335 * combine.c (simplify_compare_const): Check that the mode is a
8336 scalar_int_mode (rather than VOIDmode) before testing its
8338 (simplify_comparison): Move COMPARISON_P handling out of the
8339 loop and restrict the latter part of the loop to scalar_int_modes.
8340 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8341 and when considering SUBREG_REGs. Use is_int_mode instead of
8342 checking GET_MODE_CLASS against MODE_INT.
8344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8345 Alan Hayward <alan.hayward@arm.com>
8346 David Sherwood <david.sherwood@arm.com>
8348 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8349 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8350 shift_unit_mode and for modes involved in scalar shifts.
8352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8353 Alan Hayward <alan.hayward@arm.com>
8354 David Sherwood <david.sherwood@arm.com>
8356 * combine.c (force_int_to_mode): New function, split out from...
8357 (force_to_mode): ...here. Keep xmode up-to-date and use it
8358 instead of GET_MODE (x).
8360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8361 Alan Hayward <alan.hayward@arm.com>
8362 David Sherwood <david.sherwood@arm.com>
8364 * optabs-query.h (extraction_insn::struct_mode): Change type to
8365 opt_scalar_int_mode and update comment.
8366 (extraction_insn::field_mode): Change type to scalar_int_mode.
8367 (extraction_insn::pos_mode): Likewise.
8368 * combine.c (make_extraction): Update accordingly.
8369 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8370 (get_optab_extraction_insn): Likewise.
8371 * recog.c (simplify_while_replacing): Likewise.
8372 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8373 parameter to opt_scalar_int_mode.
8375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8376 Alan Hayward <alan.hayward@arm.com>
8377 David Sherwood <david.sherwood@arm.com>
8379 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8380 to a scalar_int_mode instead of a machine_mode.
8381 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8382 (get_best_mode): Return a boolean and use a pointer argument to store
8383 the selected mode. Replace the limit mode parameter with a bit limit.
8384 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8385 for the values returned by bit_field_mode_iterator::next_mode.
8386 (store_bit_field): Update call to get_best_mode.
8387 (store_fixed_bit_field): Likewise.
8388 (extract_fixed_bit_field): Likewise.
8389 * expr.c (optimize_bitfield_assignment_op): Likewise.
8390 * fold-const.c (optimize_bit_field_compare): Likewise.
8391 (fold_truth_andor_1): Likewise.
8392 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8393 Update for new type of m_mode.
8394 (get_best_mode): As above.
8396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8397 Alan Hayward <alan.hayward@arm.com>
8398 David Sherwood <david.sherwood@arm.com>
8400 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8401 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8402 (store_bit_field): Check is_a <scalar_int_mode> before calling
8403 strict_volatile_bitfield_p.
8404 (extract_bit_field): Likewise.
8406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8407 Alan Hayward <alan.hayward@arm.com>
8408 David Sherwood <david.sherwood@arm.com>
8410 * target.def (cstore_mode): Return a scalar_int_mode.
8411 * doc/tm.texi: Regenerate.
8412 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8413 * targhooks.h (default_cstore_mode): Likewise.
8414 * targhooks.c (default_cstore_mode): Likewise, using a forced
8416 * expmed.c (emit_cstore): Expect the target of the cstore to be
8419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8420 Alan Hayward <alan.hayward@arm.com>
8421 David Sherwood <david.sherwood@arm.com>
8423 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8425 (niter_desc): Likewise mode.
8426 (iv_analyze): Add a mode parameter.
8428 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8429 and change its type to scalar_int_mode.
8430 * loop-iv.c: Update commentary at head of file.
8431 (iv_constant): Pass the mode paraeter before the rtx it describes
8432 and change its type to scalar_int_mode. Remove VOIDmode handling.
8433 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8434 (iv_extend): Likewise.
8435 (shorten_into_mode): Likewise.
8436 (iv_add): Use scalar_int_mode.
8437 (iv_mult): Likewise.
8438 (iv_shift): Likewise.
8439 (canonicalize_iv_subregs): Likewise.
8440 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8441 it describes and change its mode to scalar_int_mode. Also change
8442 the type of the returned inner_mode to scalar_int_mode.
8443 (get_biv_step): Likewise, turning outer_mode from a pointer
8444 into a direct parameter. Update call to get_biv_step_1.
8445 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8446 iv_constant and get_biv_step.
8447 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8448 and change its type to scalar_int_mode. Don't initialise iv->mode
8449 to VOIDmode and remove later checks for its still being VOIDmode.
8450 Update calls to iv_analyze_op and iv_analyze_expr. Check
8451 is_a <scalar_int_mode> when changing the mode under consideration.
8452 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8453 Update call to iv_analyze_expr.
8454 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8455 inner register is not also a scalar_int_mode. Update call to
8457 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8458 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8459 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8460 separate mode class checks. Update calls to iv_analyze. Remove
8461 fix-up of VOIDmodes after iv_analyze_biv.
8462 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8463 don't have a scalar_int_mode. Update call to biv_p.
8465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8466 Alan Hayward <alan.hayward@arm.com>
8467 David Sherwood <david.sherwood@arm.com>
8469 * cfgexpand.c (convert_debug_memory_address): Use
8470 as_a <scalar_int_mode>.
8471 * combine.c (expand_compound_operation): Likewise.
8472 (make_extraction): Likewise.
8473 (change_zero_ext): Likewise.
8474 (simplify_comparison): Likewise.
8475 * cse.c (cse_insn): Likewise.
8476 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8477 (mem_loc_descriptor): Likewise.
8478 (loc_descriptor): Likewise.
8479 * expmed.c (init_expmed_one_mode): Likewise.
8480 (synth_mult): Likewise.
8481 (emit_store_flag_1): Likewise.
8482 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8483 of a comparison with size.
8484 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8485 (reduce_to_bit_field_precision): Likewise.
8486 * function.c (expand_function_end): Likewise.
8487 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8488 * loop-doloop.c (doloop_modify): Likewise.
8489 * optabs.c (expand_binop): Likewise.
8490 (expand_unop): Likewise.
8491 (expand_copysign_absneg): Likewise.
8492 (prepare_cmp_insn): Likewise.
8493 (maybe_legitimize_operand): Likewise.
8494 * recog.c (const_scalar_int_operand): Likewise.
8495 * rtlanal.c (get_address_mode): Likewise.
8496 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8497 (simplify_cond_clz_ctz): Likewise.
8498 * tree-nested.c (get_nl_goto_field): Likewise.
8499 * tree.c (build_vector_type_for_mode): Likewise.
8500 * var-tracking.c (use_narrower_mode): Likewise.
8502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8503 Alan Hayward <alan.hayward@arm.com>
8504 David Sherwood <david.sherwood@arm.com>
8506 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8507 * builtins.c (expand_builtin_signbit): Use it.
8508 * cfgexpand.c (expand_debug_expr): Likewise.
8509 * dojump.c (do_jump): Likewise.
8510 (do_compare_and_jump): Likewise.
8511 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8512 * expmed.c (make_tree): Likewise.
8513 * expr.c (expand_expr_real_2): Likewise.
8514 (expand_expr_real_1): Likewise.
8515 (try_casesi): Likewise.
8516 * fold-const-call.c (fold_const_call_ss): Likewise.
8517 * fold-const.c (unextend): Likewise.
8518 (extract_muldiv_1): Likewise.
8519 (fold_single_bit_test): Likewise.
8520 (native_encode_int): Likewise.
8521 (native_encode_string): Likewise.
8522 (native_interpret_int): Likewise.
8523 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8524 * internal-fn.c (expand_addsub_overflow): Likewise.
8525 (expand_neg_overflow): Likewise.
8526 (expand_mul_overflow): Likewise.
8527 (expand_arith_overflow): Likewise.
8528 * match.pd: Likewise.
8529 * stor-layout.c (layout_type): Likewise.
8530 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8531 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8532 * tree-ssanames.c (get_range_info): Likewise.
8533 * tree-switch-conversion.c (array_value_type) Likewise.
8534 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8535 (vect_recog_divmod_pattern): Likewise.
8536 (vect_recog_mixed_size_cond_pattern): Likewise.
8537 * tree-vrp.c (extract_range_basic): Likewise.
8538 (simplify_float_conversion_using_ranges): Likewise.
8539 * tree.c (int_fits_type_p): Likewise.
8540 * ubsan.c (instrument_bool_enum_load): Likewise.
8541 * varasm.c (mergeable_string_section): Likewise.
8542 (narrowing_initializer_constant_valid_p): Likewise.
8543 (output_constant): Likewise.
8545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8546 Alan Hayward <alan.hayward@arm.com>
8547 David Sherwood <david.sherwood@arm.com>
8549 * machmode.h (NARROWEST_INT_MODE): New macro.
8550 * expr.c (alignment_for_piecewise_move): Use it instead of
8551 GET_CLASS_NARROWEST_MODE (MODE_INT).
8552 (push_block): Likewise.
8553 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8555 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8558 Alan Hayward <alan.hayward@arm.com>
8559 David Sherwood <david.sherwood@arm.com>
8561 * postreload.c (move2add_valid_value_p): Change the type of the
8562 mode parameter to scalar_int_mode.
8563 (move2add_use_add2_insn): Add a mode parameter and use it instead
8565 (move2add_use_add3_insn): Likewise.
8566 (reload_cse_move2add): Update accordingly.
8568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8574 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8575 * optabs.c (expand_unop): Likewise.
8577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8578 Alan Hayward <alan.hayward@arm.com>
8579 David Sherwood <david.sherwood@arm.com>
8581 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8582 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8583 (popcount_loc_descriptor): Likewise.
8584 (bswap_loc_descriptor): Likewise.
8585 (rotate_loc_descriptor): Likewise.
8586 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8587 calling the functions above.
8589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8590 Alan Hayward <alan.hayward@arm.com>
8591 David Sherwood <david.sherwood@arm.com>
8593 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8595 (try_combine): Likewise.
8596 (simplify_if_then_else): Likewise.
8597 * cse.c (cse_insn): Likewise.
8598 * dwarf2out.c (mem_loc_descriptor): Likewise.
8599 * emit-rtl.c (gen_lowpart_common): Likewise.
8600 * simplify-rtx.c (simplify_truncation): Likewise.
8601 (simplify_binary_operation_1): Likewise.
8602 (simplify_const_relational_operation): Likewise.
8603 (simplify_ternary_operation): Likewise.
8604 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8611 * reload.c (push_reload): Likewise.
8612 (find_reloads): Likewise.
8614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8615 Alan Hayward <alan.hayward@arm.com>
8616 David Sherwood <david.sherwood@arm.com>
8618 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8619 (make_compound_operation_int): Likewise.
8620 (change_zero_ext): Likewise.
8621 * expr.c (convert_move): Likewise.
8622 (convert_modes): Likewise.
8623 * fwprop.c (forward_propagate_subreg): Likewise.
8624 * loop-iv.c (get_biv_step_1): Likewise.
8625 * optabs.c (widen_operand): Likewise.
8626 * postreload.c (move2add_valid_value_p): Likewise.
8627 * recog.c (simplify_while_replacing): Likewise.
8628 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8629 (simplify_binary_operation_1): Likewise. Remove redundant
8630 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
8637 is_a <scalar_int_mode>.
8638 (simplify_if_then_else): Likewise.
8639 (make_field_assignment): Likewise.
8640 (simplify_comparison): Likewise.
8641 * ifcvt.c (noce_try_bitop): Likewise.
8642 * loop-invariant.c (canonicalize_address_mult): Likewise.
8643 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8646 Alan Hayward <alan.hayward@arm.com>
8647 David Sherwood <david.sherwood@arm.com>
8649 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8650 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8657 instead of != VOIDmode.
8658 * combine.c (if_then_else_cond): Likewise.
8659 (change_zero_ext): Likewise.
8660 * dwarf2out.c (mem_loc_descriptor): Likewise.
8661 (loc_descriptor): Likewise.
8662 * rtlanal.c (canonicalize_condition): Likewise.
8663 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8666 Alan Hayward <alan.hayward@arm.com>
8667 David Sherwood <david.sherwood@arm.com>
8669 * simplify-rtx.c (simplify_binary_operation_1): Use
8670 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8673 Alan Hayward <alan.hayward@arm.com>
8674 David Sherwood <david.sherwood@arm.com>
8676 * wide-int.h (int_traits<unsigned char>) New class.
8677 (int_traits<unsigned short>) Likewise.
8678 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8679 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8681 * combine.c (set_nonzero_bits_and_sign_copies): Use
8682 is_a <scalar_int_mode>.
8683 (find_split_point): Likewise.
8684 (combine_simplify_rtx): Likewise.
8685 (simplify_logical): Likewise.
8686 (expand_compound_operation): Likewise.
8687 (expand_field_assignment): Likewise.
8688 (make_compound_operation): Likewise.
8689 (extended_count): Likewise.
8690 (change_zero_ext): Likewise.
8691 (simplify_comparison): Likewise.
8692 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8693 (ucompare_loc_descriptor): Likewise.
8694 (minmax_loc_descriptor): Likewise.
8695 (mem_loc_descriptor): Likewise.
8696 (loc_descriptor): Likewise.
8697 * expmed.c (init_expmed_one_mode): Likewise.
8698 * lra-constraints.c (lra_constraint_offset): Likewise.
8699 * optabs.c (prepare_libcall_arg): Likewise.
8700 * postreload.c (move2add_note_store): Likewise.
8701 * reload.c (operands_match_p): Likewise.
8702 * rtl.h (load_extend_op): Likewise.
8703 * rtlhooks.c (gen_lowpart_general): Likewise.
8704 * simplify-rtx.c (simplify_truncation): Likewise.
8705 (simplify_unary_operation_1): Likewise.
8706 (simplify_binary_operation_1): Likewise.
8707 (simplify_const_binary_operation): Likewise.
8708 (simplify_const_relational_operation): Likewise.
8709 (simplify_subreg): Likewise.
8710 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8711 * var-tracking.c (adjust_mems): Likewise.
8712 (prepare_call_arguments): Likewise.
8714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8715 Alan Hayward <alan.hayward@arm.com>
8716 David Sherwood <david.sherwood@arm.com>
8718 * machmode.h (is_int_mode): New fuction.
8719 * combine.c (find_split_point): Use it.
8720 (combine_simplify_rtx): Likewise.
8721 (simplify_if_then_else): Likewise.
8722 (simplify_set): Likewise.
8723 (simplify_shift_const_1): Likewise.
8724 (simplify_comparison): Likewise.
8725 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8726 * cse.c (notreg_cost): Likewise.
8727 (cse_insn): Likewise.
8728 * cselib.c (cselib_lookup_1): Likewise.
8729 * dojump.c (do_jump_1): Likewise.
8730 (do_compare_rtx_and_jump): Likewise.
8731 * dse.c (get_call_args): Likewise.
8732 * dwarf2out.c (rtl_for_decl_init): Likewise.
8733 (native_encode_initializer): Likewise.
8734 * expmed.c (emit_store_flag_1): Likewise.
8735 (emit_store_flag): Likewise.
8736 * expr.c (convert_modes): Likewise.
8737 (store_field): Likewise.
8738 (expand_expr_real_1): Likewise.
8739 * fold-const.c (fold_read_from_constant_string): Likewise.
8740 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8741 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8742 * optabs.c (expand_binop): Likewise.
8743 (expand_unop): Likewise.
8744 (expand_abs_nojump): Likewise.
8745 (expand_one_cmpl_abs_nojump): Likewise.
8746 * simplify-rtx.c (mode_signbit_p): Likewise.
8747 (val_signbit_p): Likewise.
8748 (val_signbit_known_set_p): Likewise.
8749 (val_signbit_known_clear_p): Likewise.
8750 (simplify_relational_operation_1): Likewise.
8751 * tree.c (vector_type_mode): Likewise.
8753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8754 Alan Hayward <alan.hayward@arm.com>
8755 David Sherwood <david.sherwood@arm.com>
8757 * machmode.h (smallest_mode_for_size): Fix formatting.
8758 (smallest_int_mode_for_size): New function.
8759 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8760 instead of smallest_mode_for_size.
8761 * combine.c (make_extraction): Likewise.
8762 * config/arc/arc.c (arc_expand_movmem): Likewise.
8763 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8764 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8765 * config/s390/s390.c (s390_expand_insv): Likewise.
8766 * config/sparc/sparc.c (assign_int_registers): Likewise.
8767 * config/spu/spu.c (spu_function_value): Likewise.
8768 (spu_function_arg): Likewise.
8769 * coverage.c (get_gcov_type): Likewise.
8770 (get_gcov_unsigned_t): Likewise.
8771 * dse.c (find_shift_sequence): Likewise.
8772 * expmed.c (store_bit_field_1): Likewise.
8773 * expr.c (convert_move): Likewise.
8774 (store_field): Likewise.
8775 * internal-fn.c (expand_arith_overflow): Likewise.
8776 * optabs-query.c (get_best_extraction_insn): Likewise.
8777 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8778 * stor-layout.c (layout_type): Likewise.
8779 (initialize_sizetypes): Likewise.
8780 * targhooks.c (default_get_mask_mode): Likewise.
8781 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8784 Alan Hayward <alan.hayward@arm.com>
8785 David Sherwood <david.sherwood@arm.com>
8787 * machmode.h (opt_mode::else_blk): New function.
8788 (int_mode_for_mode): Declare.
8789 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8790 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8792 * cfgexpand.c (expand_debug_expr): Likewise.
8793 * combine.c (gen_lowpart_or_truncate): Likewise.
8794 (gen_lowpart_for_combine): Likewise.
8795 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8796 * config/avr/avr.c (avr_to_int_mode): Likewise.
8797 (avr_out_plus_1): Likewise.
8798 (avr_out_plus): Likewise.
8799 (avr_out_round): Likewise.
8800 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8801 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8802 (s390_expand_vcond): Likewise.
8803 * config/spu/spu.c (spu_split_immediate): Likewise.
8804 (spu_expand_mov): Likewise.
8805 * dse.c (get_stored_val): Likewise.
8806 * expmed.c (store_bit_field_1): Likewise.
8807 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8809 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8810 (extract_low_bits): Likewise.
8811 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8812 handling rather than repeating the check.
8813 (emit_group_store): Likewise.
8814 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8815 * optabs.c (expand_absneg_bit): Likewise.
8816 (expand_copysign_absneg): Likewise.
8817 (expand_copysign_bit): Likewise.
8818 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8819 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8820 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8821 * var-tracking.c (prepare_call_arguments): Likewise.
8822 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8823 int_mode_for_mode instead of mode_for_size.
8824 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8827 Alan Hayward <alan.hayward@arm.com>
8828 David Sherwood <david.sherwood@arm.com>
8830 * machmode.h (int_mode_for_size): New function.
8831 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8832 instead of mode_for_size.
8833 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8835 * combine.c (expand_field_assignment): Use int_mode_for_size
8836 instead of mode_for_size.
8837 (make_extraction): Likewise.
8838 (simplify_shift_const_1): Likewise.
8839 (simplify_comparison): Likewise.
8840 * dojump.c (do_jump): Likewise.
8841 * dwarf2out.c (mem_loc_descriptor): Likewise.
8842 * emit-rtl.c (init_derived_machine_modes): Likewise.
8843 * expmed.c (flip_storage_order): Likewise.
8844 (convert_extracted_bit_field): Likewise.
8845 * expr.c (copy_blkmode_from_reg): Likewise.
8846 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8847 * internal-fn.c (expand_mul_overflow): Likewise.
8848 * lower-subreg.c (simple_move): Likewise.
8849 * optabs-libfuncs.c (init_optabs): Likewise.
8850 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8851 * tree.c (vector_type_mode): Likewise.
8852 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8853 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8854 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8855 * tree-vect-stmts.c (vectorizable_load): Likewise.
8856 (vectorizable_store): Likewise.
8858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8859 Alan Hayward <alan.hayward@arm.com>
8860 David Sherwood <david.sherwood@arm.com>
8862 * coretypes.h (pod_mode): New type.
8863 (scalar_int_mode_pod): New typedef.
8864 * machmode.h (pod_mode): New class.
8865 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8866 * genmodes.c (emit_mode_int_n): Update accordingly.
8867 * lower-subreg.h (target_lower_subreg): Change type to
8868 scalar_int_mode_pod.
8869 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8870 scalar_int_mode_pod.
8872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8873 Alan Hayward <alan.hayward@arm.com>
8874 David Sherwood <david.sherwood@arm.com>
8876 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8877 machine_mode to scalar_int_mode.
8878 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8879 (rs6000_option_override_internal): Remove cast to int.
8880 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8881 machine_mode to scalar_int_mode.
8882 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8883 (rs6000_option_override_internal): Remove cast to int.
8884 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8885 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8887 * config/s390/s390.c (s390_expand_builtin): Likewise.
8888 * coretypes.h (scalar_int_mode): New type.
8889 (opt_scalar_int_mode): New typedef.
8890 * machmode.h (scalar_int_mode): New class.
8891 (scalar_int_mode::includes_p): New function.
8892 (byte_mode): Change type to scalar_int_mode.
8893 (word_mode): Likewise.
8894 (ptr_mode): Likewise.
8895 * emit-rtl.c (byte_mode): Likewise.
8896 (word_mode): Likewise.
8897 (ptr_mode): Likewise.
8898 (init_derived_machine_modes): Update accordingly.
8899 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8900 and MODE_PARTIAL_INT.
8901 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8902 opt_scalar_int_mode.
8904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8905 Alan Hayward <alan.hayward@arm.com>
8906 David Sherwood <david.sherwood@arm.com>
8908 * target.def (libgcc_floating_mode_supported_p): Take a
8910 * doc/tm.texi: Regenerate.
8911 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8913 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8914 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8918 Alan Hayward <alan.hayward@arm.com>
8919 David Sherwood <david.sherwood@arm.com>
8921 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8922 * doc/tm.texi: Regenerate.
8923 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8924 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8925 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8926 * targhooks.h (default_floatn_mode): Likewise.
8927 * targhooks.c (default_floatn_mode): Likewise.
8928 * tree.c (build_common_tree_nodes): Update accordingly.
8930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8931 Alan Hayward <alan.hayward@arm.com>
8932 David Sherwood <david.sherwood@arm.com>
8934 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8935 (mode_iterator::iterate_p): Likewise.
8936 (mode_iterator::get_wider): Likewise.
8937 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8940 Alan Hayward <alan.hayward@arm.com>
8941 David Sherwood <david.sherwood@arm.com>
8943 * coretypes.h (opt_scalar_float_mode): New typedef.
8944 * machmode.h (float_mode_for_size): New function.
8945 * emit-rtl.c (double_mode): Delete.
8946 (init_emit_once): Use float_mode_for_size.
8947 * stor-layout.c (layout_type): Likewise.
8948 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8951 Alan Hayward <alan.hayward@arm.com>
8952 David Sherwood <david.sherwood@arm.com>
8954 * output.h (assemble_real): Take a scalar_float_mode.
8955 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8956 * config/arm/arm.md (consttable_4): Likewise.
8957 (consttable_8): Likewise.
8958 (consttable_16): Likewise.
8959 * config/mips/mips.md (consttable_float): Likewise.
8960 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8961 * varasm.c (assemble_real): Take a scalar_float_mode.
8962 (output_constant_pool_2): Update accordingly.
8963 (output_constant): Likewise.
8965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8966 Alan Hayward <alan.hayward@arm.com>
8967 David Sherwood <david.sherwood@arm.com>
8969 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8970 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8971 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8972 (native_encode_real): Likewise.
8973 (native_interpret_real): Likewise.
8974 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8975 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8978 Alan Hayward <alan.hayward@arm.com>
8979 David Sherwood <david.sherwood@arm.com>
8981 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8982 <scalar_float_mode>. Simplify.
8983 (gen_extend_conv_libfunc): Likewise.
8985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8986 Alan Hayward <alan.hayward@arm.com>
8987 David Sherwood <david.sherwood@arm.com>
8989 * coretypes.h (scalar_float_mode): New type.
8990 * machmode.h (mode_traits::from_int): Use machine_mode if
8991 USE_ENUM_MODES is defined.
8992 (is_a): New function.
8994 (dyn_cast): Likewise.
8995 (scalar_float_mode): New class.
8996 (scalar_float_mode::includes_p): New function.
8997 (is_float_mode): Likewise.
8998 * gdbhooks.py (MachineModePrinter): New class.
8999 (build_pretty_printer): Use it for scalar_float_mode.
9000 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9001 (format_helper::format_helper): Turn into a template.
9002 * genmodes.c (get_mode_class): New function.
9003 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9004 or machine_mode if none.
9005 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9006 as_a <scalar_float_mode>.
9007 * dwarf2out.c (mem_loc_descriptor): Likewise.
9008 (insert_float): Likewise.
9009 (add_const_value_attribute): Likewise.
9010 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9011 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9012 (expand_unop): Update accordingly.
9013 (expand_abs_nojump): Likewise.
9014 (expand_copysign_absneg): Take a scalar_float_mode.
9015 (expand_copysign_bit): Likewise.
9016 (expand_copysign): Update accordingly.
9018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9019 Alan Hayward <alan.hayward@arm.com>
9020 David Sherwood <david.sherwood@arm.com>
9022 * coretypes.h (opt_mode): New class.
9023 * machmode.h (opt_mode): Likewise.
9024 (opt_mode::else_void): New function.
9025 (opt_mode::require): Likewise.
9026 (opt_mode::exists): Likewise.
9027 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9028 (GET_MODE_2XWIDER_MODE): Likewise.
9029 (mode_iterator::get_wider): Update accordingly.
9030 (mode_iterator::get_2xwider): Likewise.
9031 (mode_iterator::get_known_wider): Likewise, turning into a template.
9032 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9033 forcing a wider mode to exist.
9034 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9035 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9036 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9037 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9038 * lower-subreg.c (init_lower_subreg): Likewise.
9039 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9040 on the final iteration.
9041 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9042 a wider mode exists before asking for a move pattern.
9043 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9044 forcing a wider mode to exist.
9045 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9046 returning false if no such mode exists.
9047 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9048 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9049 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9050 Avoid checking for a MODE_INT if we already know the mode is not a
9052 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9053 forcing a wider mode to exist.
9054 (expmed_mult_highpart_optab): Likewise.
9055 (expmed_mult_highpart): Likewise.
9056 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9058 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9059 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9060 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9061 * internal-fn.c (expand_mul_overflow): Update use of
9062 GET_MODE_2XWIDER_MODE.
9063 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9064 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9065 GET_MODE_WIDER_MODE.
9066 (convert_plusminus_to_widen): Likewise.
9067 * tree-switch-conversion.c (array_value_type): Likewise.
9068 * var-tracking.c (emit_note_insn_var_location): Likewise.
9069 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9070 Return false inside rather than outside the loop if no wider mode
9072 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9073 and GET_MODE_2XWIDER_MODE
9074 (can_compare_p): Use else_void.
9075 * gdbhooks.py (OptMachineModePrinter): New class.
9076 (build_pretty_printer): Use it for opt_mode.
9078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9079 Alan Hayward <alan.hayward@arm.com>
9080 David Sherwood <david.sherwood@arm.com>
9082 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9083 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9086 Alan Hayward <alan.hayward@arm.com>
9087 David Sherwood <david.sherwood@arm.com>
9089 * machmode.h (mode_traits): New structure.
9090 (get_narrowest_mode): New function.
9091 (mode_iterator::start): Likewise.
9092 (mode_iterator::iterate_p): Likewise.
9093 (mode_iterator::get_wider): Likewise.
9094 (mode_iterator::get_known_wider): Likewise.
9095 (mode_iterator::get_2xwider): Likewise.
9096 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9097 (FOR_EACH_MODE): Likewise.
9098 (FOR_EACH_MODE_FROM): Likewise.
9099 (FOR_EACH_MODE_UNTIL): Likewise.
9100 (FOR_EACH_WIDER_MODE): Likewise.
9101 (FOR_EACH_2XWIDER_MODE): Likewise.
9102 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9103 * combine.c (simplify_comparison): Likewise
9104 * config/i386/i386.c (type_natural_mode): Likewise.
9105 * cse.c (cse_insn): Likewise.
9106 * dse.c (find_shift_sequence): Likewise.
9107 * emit-rtl.c (init_derived_machine_modes): Likewise.
9108 (init_emit_once): Likewise.
9109 * explow.c (hard_function_value): Likewise.
9110 * expmed.c (extract_fixed_bit_field_1): Likewise.
9111 (extract_bit_field_1): Likewise.
9112 (expand_divmod): Likewise.
9113 (emit_store_flag_1): Likewise.
9114 * expr.c (init_expr_target): Likewise.
9115 (convert_move): Likewise.
9116 (alignment_for_piecewise_move): Likewise.
9117 (widest_int_mode_for_size): Likewise.
9118 (emit_block_move_via_movmem): Likewise.
9119 (copy_blkmode_to_reg): Likewise.
9120 (set_storage_via_setmem): Likewise.
9121 (compress_float_constant): Likewise.
9122 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9123 * optabs-query.c (get_best_extraction_insn): Likewise.
9124 * optabs.c (expand_binop): Likewise.
9125 (expand_twoval_unop): Likewise.
9126 (expand_twoval_binop): Likewise.
9127 (widen_leading): Likewise.
9128 (widen_bswap): Likewise.
9129 (expand_parity): Likewise.
9130 (expand_unop): Likewise.
9131 (prepare_cmp_insn): Likewise.
9132 (prepare_float_lib_cmp): Likewise.
9133 (expand_float): Likewise.
9134 (expand_fix): Likewise.
9135 (expand_sfix_optab): Likewise.
9136 * postreload.c (move2add_use_add2_insn): Likewise.
9137 * reg-stack.c (reg_to_stack): Likewise.
9138 * reginfo.c (choose_hard_reg_mode): Likewise.
9139 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9140 * stor-layout.c (mode_for_size): Likewise.
9141 (smallest_mode_for_size): Likewise.
9142 (mode_for_vector): Likewise.
9143 (finish_bitfield_representative): Likewise.
9144 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9145 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9146 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9147 * var-tracking.c (prepare_call_arguments): Likewise.
9149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9150 Alan Hayward <alan.hayward@arm.com>
9151 David Sherwood <david.sherwood@arm.com>
9153 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9154 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9155 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9156 * machmode.h (mode_size): Move earlier in file.
9157 (mode_precision): Likewise.
9158 (mode_inner): Likewise.
9159 (mode_nunits): Likewise.
9160 (mode_unit_size): Likewise.
9161 (unit_unit_precision): Likewise.
9162 (mode_wider): Likewise.
9163 (mode_2xwider): Likewise.
9164 (machine_mode): New class.
9165 (mode_to_bytes): New function.
9166 (mode_to_bits): Likewise.
9167 (mode_to_precision): Likewise.
9168 (mode_to_inner): Likewise.
9169 (mode_to_unit_size): Likewise.
9170 (mode_to_unit_precision): Likewise.
9171 (mode_to_nunits): Likewise.
9172 (GET_MODE_SIZE): Use mode_to_bytes.
9173 (GET_MODE_BITSIZE): Use mode_to_bits.
9174 (GET_MODE_PRECISION): Use mode_to_precision.
9175 (GET_MODE_INNER): Use mode_to_inner.
9176 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9177 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9178 (GET_MODE_NUNITS): Use mode_to_nunits.
9179 * system.h (ALWAYS_INLINE): New macro.
9180 * config/powerpcspe/powerpcspe-c.c
9181 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9182 int for arg1_mode and arg2_mode.
9184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9185 Alan Hayward <alan.hayward@arm.com>
9186 David Sherwood <david.sherwood@arm.com>
9188 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9189 Prefix mode names with E_ in case statements.
9190 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9191 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9192 (aarch64_split_simd_move): Likewise.
9193 (aarch64_gen_storewb_pair): Likewise.
9194 (aarch64_gen_loadwb_pair): Likewise.
9195 (aarch64_gen_store_pair): Likewise.
9196 (aarch64_gen_load_pair): Likewise.
9197 (aarch64_get_condition_code_1): Likewise.
9198 (aarch64_constant_pool_reload_icode): Likewise.
9199 (get_rsqrte_type): Likewise.
9200 (get_rsqrts_type): Likewise.
9201 (get_recpe_type): Likewise.
9202 (get_recps_type): Likewise.
9203 (aarch64_gimplify_va_arg_expr): Likewise.
9204 (aarch64_simd_container_mode): Likewise.
9205 (aarch64_emit_load_exclusive): Likewise.
9206 (aarch64_emit_store_exclusive): Likewise.
9207 (aarch64_expand_compare_and_swap): Likewise.
9208 (aarch64_gen_atomic_cas): Likewise.
9209 (aarch64_emit_bic): Likewise.
9210 (aarch64_emit_atomic_swap): Likewise.
9211 (aarch64_emit_atomic_load_op): Likewise.
9212 (aarch64_evpc_trn): Likewise.
9213 (aarch64_evpc_uzp): Likewise.
9214 (aarch64_evpc_zip): Likewise.
9215 (aarch64_evpc_ext): Likewise.
9216 (aarch64_evpc_rev): Likewise.
9217 (aarch64_evpc_dup): Likewise.
9218 (aarch64_gen_ccmp_first): Likewise.
9219 (aarch64_gen_ccmp_next): Likewise.
9220 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9221 (alpha_emit_xfloating_libcall): Likewise.
9222 (emit_insxl): Likewise.
9223 (alpha_arg_type): Likewise.
9224 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9225 (arc_preferred_simd_mode): Likewise.
9226 (arc_secondary_reload): Likewise.
9227 (get_arc_condition_code): Likewise.
9228 (arc_print_operand): Likewise.
9229 (arc_legitimate_constant_p): Likewise.
9230 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9231 * config/arc/arc.md (casesi_load): Likewise.
9232 (casesi_compact_jump): Likewise.
9233 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9234 (cc_use_register): Likewise.
9235 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9236 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9237 (arm_init_iwmmxt_builtins): Likewise.
9238 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9239 (neon_expand_vector_init): Likewise.
9240 (arm_attr_length_move_neon): Likewise.
9241 (maybe_get_arm_condition_code): Likewise.
9242 (arm_emit_vector_const): Likewise.
9243 (arm_preferred_simd_mode): Likewise.
9244 (arm_output_iwmmxt_tinsr): Likewise.
9245 (thumb1_output_casesi): Likewise.
9246 (thumb2_output_casesi): Likewise.
9247 (arm_emit_load_exclusive): Likewise.
9248 (arm_emit_store_exclusive): Likewise.
9249 (arm_expand_compare_and_swap): Likewise.
9250 (arm_evpc_neon_vuzp): Likewise.
9251 (arm_evpc_neon_vzip): Likewise.
9252 (arm_evpc_neon_vrev): Likewise.
9253 (arm_evpc_neon_vtrn): Likewise.
9254 (arm_evpc_neon_vext): Likewise.
9255 (arm_validize_comparison): Likewise.
9256 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9257 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9258 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9259 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9260 (c6x_preferred_simd_mode): Likewise.
9261 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9262 (epiphany_rtx_costs): Likewise.
9263 * config/epiphany/predicates.md (proper_comparison_operator):
9265 * config/frv/frv.c (condexec_memory_operand): Likewise.
9266 (frv_emit_move): Likewise.
9267 (output_move_single): Likewise.
9268 (output_condmove_single): Likewise.
9269 (frv_hard_regno_mode_ok): Likewise.
9270 (frv_matching_accg_mode): Likewise.
9271 * config/h8300/h8300.c (split_adds_subs): Likewise.
9272 (h8300_rtx_costs): Likewise.
9273 (h8300_print_operand): Likewise.
9274 (compute_mov_length): Likewise.
9275 (output_logical_op): Likewise.
9276 (compute_logical_op_length): Likewise.
9277 (compute_logical_op_cc): Likewise.
9278 (h8300_shift_needs_scratch_p): Likewise.
9279 (output_a_shift): Likewise.
9280 (compute_a_shift_length): Likewise.
9281 (compute_a_shift_cc): Likewise.
9282 (expand_a_rotate): Likewise.
9283 (output_a_rotate): Likewise.
9284 * config/i386/i386.c (classify_argument): Likewise.
9285 (function_arg_advance_32): Likewise.
9286 (function_arg_32): Likewise.
9287 (function_arg_64): Likewise.
9288 (function_value_64): Likewise.
9289 (ix86_gimplify_va_arg): Likewise.
9290 (ix86_legitimate_constant_p): Likewise.
9291 (put_condition_code): Likewise.
9292 (split_double_mode): Likewise.
9293 (ix86_avx256_split_vector_move_misalign): Likewise.
9294 (ix86_expand_vector_logical_operator): Likewise.
9295 (ix86_split_idivmod): Likewise.
9296 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9297 (ix86_build_const_vector): Likewise.
9298 (ix86_build_signbit_mask): Likewise.
9299 (ix86_match_ccmode): Likewise.
9300 (ix86_cc_modes_compatible): Likewise.
9301 (ix86_expand_branch): Likewise.
9302 (ix86_expand_sse_cmp): Likewise.
9303 (ix86_expand_sse_movcc): Likewise.
9304 (ix86_expand_int_sse_cmp): Likewise.
9305 (ix86_expand_vec_perm_vpermi2): Likewise.
9306 (ix86_expand_vec_perm): Likewise.
9307 (ix86_expand_sse_unpack): Likewise.
9308 (ix86_expand_int_addcc): Likewise.
9309 (ix86_split_to_parts): Likewise.
9310 (ix86_vectorize_builtin_gather): Likewise.
9311 (ix86_vectorize_builtin_scatter): Likewise.
9312 (avx_vpermilp_parallel): Likewise.
9313 (inline_memory_move_cost): Likewise.
9314 (ix86_tieable_integer_mode_p): Likewise.
9315 (x86_maybe_negate_const_int): Likewise.
9316 (ix86_expand_vector_init_duplicate): Likewise.
9317 (ix86_expand_vector_init_one_nonzero): Likewise.
9318 (ix86_expand_vector_init_one_var): Likewise.
9319 (ix86_expand_vector_init_concat): Likewise.
9320 (ix86_expand_vector_init_interleave): Likewise.
9321 (ix86_expand_vector_init_general): Likewise.
9322 (ix86_expand_vector_set): Likewise.
9323 (ix86_expand_vector_extract): Likewise.
9324 (emit_reduc_half): Likewise.
9325 (ix86_emit_i387_round): Likewise.
9326 (ix86_mangle_type): Likewise.
9327 (ix86_expand_round_sse4): Likewise.
9328 (expand_vec_perm_blend): Likewise.
9329 (canonicalize_vector_int_perm): Likewise.
9330 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9331 (expand_vec_perm_1): Likewise.
9332 (expand_vec_perm_interleave3): Likewise.
9333 (expand_vec_perm_even_odd_pack): Likewise.
9334 (expand_vec_perm_even_odd_1): Likewise.
9335 (expand_vec_perm_broadcast_1): Likewise.
9336 (ix86_vectorize_vec_perm_const_ok): Likewise.
9337 (ix86_expand_vecop_qihi): Likewise.
9338 (ix86_expand_mul_widen_hilo): Likewise.
9339 (ix86_expand_sse2_abs): Likewise.
9340 (ix86_expand_pextr): Likewise.
9341 (ix86_expand_pinsr): Likewise.
9342 (ix86_preferred_simd_mode): Likewise.
9343 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9344 * config/i386/sse.md (*andnot<mode>3): Likewise.
9345 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9346 (*<code><mode>3): Likewise.
9347 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9348 (ia64_expand_atomic_op): Likewise.
9349 (ia64_arg_type): Likewise.
9350 (ia64_mode_to_int): Likewise.
9351 (ia64_scalar_mode_supported_p): Likewise.
9352 (ia64_vector_mode_supported_p): Likewise.
9353 (expand_vec_perm_broadcast): Likewise.
9354 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9355 (iq2000_function_arg_advance): Likewise.
9356 (iq2000_function_arg): Likewise.
9357 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9358 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9359 (m68k_libcall_value): Likewise.
9360 (m68k_function_value): Likewise.
9361 (sched_attr_op_type): Likewise.
9362 * config/mcore/mcore.c (mcore_output_move): Likewise.
9363 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9365 (microblaze_function_arg): Likewise.
9366 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9367 (mips_print_operand): Likewise.
9368 (mips_mode_ok_for_mov_fmt_p): Likewise.
9369 (mips_vector_mode_supported_p): Likewise.
9370 (mips_preferred_simd_mode): Likewise.
9371 (mips_expand_vpc_loongson_even_odd): Likewise.
9372 (mips_expand_vec_unpack): Likewise.
9373 (mips_expand_vi_broadcast): Likewise.
9374 (mips_expand_vector_init): Likewise.
9375 (mips_expand_vec_reduc): Likewise.
9376 (mips_expand_msa_cmp): Likewise.
9377 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9378 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9379 (cc_flags_for_mode): Likewise.
9380 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9381 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9382 (nds32_output_casesi_pc_relative): Likewise.
9383 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9384 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9385 (nvptx_gen_unpack): Likewise.
9386 (nvptx_gen_pack): Likewise.
9387 (nvptx_gen_shuffle): Likewise.
9388 (nvptx_gen_wcast): Likewise.
9389 (nvptx_preferred_simd_mode): Likewise.
9390 * config/pa/pa.c (pa_secondary_reload): Likewise.
9391 * config/pa/predicates.md (base14_operand): Likewise.
9392 * config/powerpcspe/powerpcspe-c.c
9393 (altivec_resolve_overloaded_builtin): Likewise.
9394 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9396 (rs6000_preferred_simd_mode): Likewise.
9397 (output_vec_const_move): Likewise.
9398 (rs6000_expand_vector_extract): Likewise.
9399 (rs6000_split_vec_extract_var): Likewise.
9400 (reg_offset_addressing_ok_p): Likewise.
9401 (rs6000_legitimate_offset_address_p): Likewise.
9402 (rs6000_legitimize_address): Likewise.
9403 (rs6000_emit_set_const): Likewise.
9404 (rs6000_const_vec): Likewise.
9405 (rs6000_emit_move): Likewise.
9406 (spe_build_register_parallel): Likewise.
9407 (rs6000_darwin64_record_arg_recurse): Likewise.
9408 (swap_selector_for_mode): Likewise.
9409 (spe_init_builtins): Likewise.
9410 (paired_init_builtins): Likewise.
9411 (altivec_init_builtins): Likewise.
9412 (do_load_for_compare): Likewise.
9413 (rs6000_generate_compare): Likewise.
9414 (rs6000_expand_float128_convert): Likewise.
9415 (emit_load_locked): Likewise.
9416 (emit_store_conditional): Likewise.
9417 (rs6000_output_function_epilogue): Likewise.
9418 (rs6000_handle_altivec_attribute): Likewise.
9419 (rs6000_function_value): Likewise.
9420 (emit_fusion_gpr_load): Likewise.
9421 (emit_fusion_p9_load): Likewise.
9422 (emit_fusion_p9_store): Likewise.
9423 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9424 (fusion_gpr_mem_load): Likewise.
9425 (fusion_addis_mem_combo_load): Likewise.
9426 (fusion_addis_mem_combo_store): Likewise.
9427 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9428 (fusion_gpr_mem_load): Likewise.
9429 (fusion_addis_mem_combo_load): Likewise.
9430 (fusion_addis_mem_combo_store): Likewise.
9431 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9433 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9434 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9435 (rs6000_preferred_simd_mode): Likewise.
9436 (output_vec_const_move): Likewise.
9437 (rs6000_expand_vector_extract): Likewise.
9438 (rs6000_split_vec_extract_var): Likewise.
9439 (reg_offset_addressing_ok_p): Likewise.
9440 (rs6000_legitimate_offset_address_p): Likewise.
9441 (rs6000_legitimize_address): Likewise.
9442 (rs6000_emit_set_const): Likewise.
9443 (rs6000_const_vec): Likewise.
9444 (rs6000_emit_move): Likewise.
9445 (rs6000_darwin64_record_arg_recurse): Likewise.
9446 (swap_selector_for_mode): Likewise.
9447 (paired_init_builtins): Likewise.
9448 (altivec_init_builtins): Likewise.
9449 (rs6000_expand_float128_convert): Likewise.
9450 (emit_load_locked): Likewise.
9451 (emit_store_conditional): Likewise.
9452 (rs6000_output_function_epilogue): Likewise.
9453 (rs6000_handle_altivec_attribute): Likewise.
9454 (rs6000_function_value): Likewise.
9455 (emit_fusion_gpr_load): Likewise.
9456 (emit_fusion_p9_load): Likewise.
9457 (emit_fusion_p9_store): Likewise.
9458 * config/rx/rx.c (rx_gen_move_template): Likewise.
9459 (flags_from_mode): Likewise.
9460 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9461 (s390_slb_comparison): Likewise.
9462 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9463 (s390_vector_mode_supported_p): Likewise.
9464 (s390_cc_modes_compatible): Likewise.
9465 (s390_match_ccmode_set): Likewise.
9466 (s390_canonicalize_comparison): Likewise.
9467 (s390_emit_compare_and_swap): Likewise.
9468 (s390_branch_condition_mask): Likewise.
9469 (s390_rtx_costs): Likewise.
9470 (s390_secondary_reload): Likewise.
9471 (__SECONDARY_RELOAD_CASE): Likewise.
9472 (s390_expand_cs): Likewise.
9473 (s390_preferred_simd_mode): Likewise.
9474 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9475 * config/sh/sh.c (sh_print_operand): Likewise.
9476 (dump_table): Likewise.
9477 (sh_secondary_reload): Likewise.
9478 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9479 * config/sh/sh.md (casesi_worker_1): Likewise.
9480 (casesi_worker_2): Likewise.
9481 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9482 (fcc_comparison_operator): Likewise.
9483 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9484 (emit_soft_tfmode_cvt): Likewise.
9485 (sparc_preferred_simd_mode): Likewise.
9486 (output_cbranch): Likewise.
9487 (sparc_print_operand): Likewise.
9488 (sparc_expand_vec_perm_bmask): Likewise.
9489 (vector_init_bshuffle): Likewise.
9490 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9491 (spu_vector_mode_supported_p): Likewise.
9492 (spu_expand_insv): Likewise.
9493 (spu_emit_branch_or_set): Likewise.
9494 (spu_handle_vector_attribute): Likewise.
9495 (spu_builtin_splats): Likewise.
9496 (spu_builtin_extract): Likewise.
9497 (spu_builtin_promote): Likewise.
9498 (spu_expand_sign_extend): Likewise.
9499 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9500 (tilegx_simd_int): Likewise.
9501 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9502 (tilepro_simd_int): Likewise.
9503 * config/v850/v850.c (const_double_split): Likewise.
9504 (v850_print_operand): Likewise.
9505 (ep_memory_offset): Likewise.
9506 * config/vax/vax.c (vax_rtx_costs): Likewise.
9507 (vax_output_int_move): Likewise.
9508 (vax_output_int_add): Likewise.
9509 (vax_output_int_subtract): Likewise.
9510 * config/visium/predicates.md (visium_branch_operator): Likewise.
9511 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9512 (visium_print_operand_address): Likewise.
9513 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9514 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9515 (xtensa_expand_conditional_branch): Likewise.
9516 (xtensa_copy_incoming_a7): Likewise.
9517 (xtensa_output_literal): Likewise.
9518 * dfp.c (decimal_real_maxval): Likewise.
9519 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9522 Alan Hayward <alan.hayward@arm.com>
9523 David Sherwood <david.sherwood@arm.com>
9525 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9526 (mode_nunits_inline): Likewise.
9527 (mode_inner_inline): Likewise.
9528 (mode_unit_size_inline): Likewise.
9529 (mode_unit_precision_inline): Likewise.
9530 (emit_insn_modes_h): Likewise. Also emit a #define of the
9532 (emit_mode_wider): Add an E_ prefix to mode names.
9533 (emit_mode_complex): Likewise.
9534 (emit_mode_inner): Likewise.
9535 (emit_mode_adjustments): Likewise.
9536 (emit_mode_int_n): Likewise.
9537 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9538 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9539 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9540 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9541 (CRC32_BUILTIN, ENTRY): Likewise.
9542 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9543 (aarch64_pop_regs): Likewise.
9544 (aarch64_process_components): Likewise.
9545 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9546 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9547 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9548 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9549 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9550 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9552 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9553 (pcmpistr): Likewise.
9554 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9555 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9556 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9558 * config/rl78/rl78.c (mduc_regs): Likewise.
9559 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9560 (htm_expand_builtin): Likewise.
9561 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9562 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9563 * config/xtensa/xtensa.c (print_operand): Likewise.
9564 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9565 (NUM_MODE_VECTOR_INT): Likewise.
9566 * genoutput.c (null_operand): Likewise.
9567 (output_operand_data): Likewise.
9568 * genrecog.c (print_parameter_value): Likewise.
9569 * lra.c (debug_operand_data): Likewise.
9571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9573 * dwarf2out.c (scompare_loc_descriptor_wide)
9574 (scompare_loc_descriptor_narrow): New functions, split out from...
9575 (scompare_loc_descriptor): ...here.
9576 * expmed.c (emit_store_flag_int): New function, split out from...
9577 (emit_store_flag): ...here.
9579 2017-08-30 Richard Biener <rguenther@suse.de>
9581 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9582 (dwarf2out_early_finish): Move setting of AT_pubnames from
9583 early debug output to early finish.
9585 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9587 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9588 and -mdata-region to the assembler.
9590 2017-08-30 Richard Biener <rguenther@suse.de>
9592 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9594 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9595 (dwarf2out_early_global_decl): Only generate a DIE for the
9596 abstract origin if it doesn't already exist or is a declaration DIE.
9597 (resolve_addr): Do not add the linkage name twice when
9598 generating a stub DIE for the DW_TAG_GNU_call_site target.
9600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9603 Use machine_mode rather than int for arg1_mode.
9605 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9608 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9609 that the second argument of the built-in functions to unpack
9610 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9611 switch statement instead a lot of if statements.
9612 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9613 Allow 64-bit values to be in Altivec registers as well as
9614 traditional floating point registers.
9615 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9617 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9619 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9620 MAX_REGS_PER_ADDRESS == 1.
9622 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9624 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9625 * config/i386/i386.c (ix86_option_override_internal): Simplify
9626 setting of opts->x_flag_entry.
9628 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9629 Jakub Jelinek <jakub@redhat.com>
9630 Richard Biener <rguenther@suse.de>
9632 PR tree-optimization/81503
9633 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9634 folded constant fits in the target type; reorder tests for clarity.
9636 2017-08-29 Martin Liska <mliska@suse.cz>
9638 * passes.def: Include pass_lower_switch.
9639 * stmt.c (dump_case_nodes): Remove and move to
9640 tree-switch-conversion.
9641 (case_values_threshold): Likewise.
9642 (expand_switch_as_decision_tree_p): Likewise.
9643 (emit_case_decision_tree): Likewise.
9644 (expand_case): Likewise.
9645 (balance_case_nodes): Likewise.
9646 (node_has_low_bound): Likewise.
9647 (node_has_high_bound): Likewise.
9648 (node_is_bounded): Likewise.
9649 (emit_case_nodes): Likewise.
9650 (struct simple_case_node): New struct.
9651 (add_case_node): Remove.
9652 (emit_case_dispatch_table): Use vector instead of case_list.
9653 (reset_out_edges_aux): Remove.
9654 (compute_cases_per_edge): Likewise.
9655 (expand_case): Build list of simple_case_node.
9656 (expand_sjlj_dispatch_table): Use it.
9657 * tree-switch-conversion.c (struct case_node): Moved from
9658 stmt.c and adjusted.
9659 (emit_case_nodes): Likewise.
9660 (node_has_low_bound): Likewise.
9661 (node_has_high_bound): Likewise.
9662 (node_is_bounded): Likewise.
9663 (case_values_threshold): Likewise.
9664 (reset_out_edges_aux): Likewise.
9665 (compute_cases_per_edge): Likewise.
9666 (add_case_node): Likewise.
9667 (dump_case_nodes): Likewise.
9668 (balance_case_nodes): Likewise.
9669 (expand_switch_as_decision_tree_p): Likewise.
9670 (emit_jump): Likewise.
9671 (emit_case_decision_tree): Likewise.
9672 (try_switch_expansion): Likewise.
9673 (do_jump_if_equal): Likewise.
9674 (emit_cmp_and_jump_insns): Likewise.
9675 (fix_phi_operands_for_edge): New function.
9676 (record_phi_operand_mapping): Likewise.
9677 (class pass_lower_switch): New pass.
9678 (pass_lower_switch::execute): New function.
9679 (make_pass_lower_switch): Likewise.
9680 (conditional_probability):
9681 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9682 * tree-pass.h: Add make_pass_lower_switch.
9684 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9687 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9690 2017-08-29 Richard Biener <rguenther@suse.de>
9692 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9693 we do not add a DW_AT_inline attribute twice.
9694 (gen_subprogram_die): Remove code setting DW_AT_inline on
9695 DECL_ABSTRACT_P nodes.
9697 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9699 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9700 calls to internal functions.
9701 (gimplify_modify_expr): Likewise.
9702 * tree-call-cdce.c (use_internal_fn): Likewise.
9703 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9704 (convert_to_divmod): Set the nothrow flag.
9705 * tree-if-conv.c (predicate_mem_writes): Likewise.
9706 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9707 (vectorizable_call): Likewise.
9708 (vectorizable_store): Likewise.
9709 (vectorizable_load): Likewise.
9710 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9711 (vect_recog_mask_conversion_pattern): Likewise.
9713 2017-08-29 Martin Liska <mliska@suse.cz>
9716 * gcc.c (driver_handle_option): Add new argument.
9717 * opts-common.c (handle_option): Pass
9718 target_option_override_hook.
9719 * opts-global.c (lang_handle_option): Add new option.
9720 (set_default_handlers): Add new argument.
9721 (decode_options): Likewise.
9722 * opts.c (target_handle_option): Likewise.
9723 (common_handle_option): Call target_option_override_hook.
9724 * opts.h (struct cl_option_handler_func): Add hook for
9725 target option override.
9726 (struct cl_option_handlers): Likewise.
9727 (set_default_handlers): Add new argument.
9728 (decode_options): Likewise.
9729 (common_handle_option): Likewise.
9730 (target_handle_option): Likewise.
9731 * toplev.c (toplev::main): Pass targetm.target_option.override
9734 2017-08-29 Richard Biener <rguenther@suse.de>
9735 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9737 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9738 life to the active subtree.
9740 2017-08-28 Jeff Law <law@redhat.com>
9742 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9743 derive_equivalences.
9744 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9746 (edge_info::derive_equivalences): New private member function
9748 * tree-ssa-dom.c (class edge_info): Changed from a struct
9749 to a class. Add ctor/dtor, methods and data members.
9750 (edge_info::edge_info): Renamed from allocate_edge_info.
9751 Initialize additional members.
9752 (edge_info::~edge_info): New.
9753 (free_dom_edge_info): Delete the edge info.
9754 (record_edge_info): Use new class & associated member functions.
9755 Tighten forms for testing for edge equivalences.
9756 (record_temporary_equivalences): Iterate over the simple
9757 equivalences rather than assuming there's only one per edge.
9758 (cprop_into_successor_phis): Iterate over the simple
9759 equivalences rather than assuming there's only one per edge.
9760 (optimize_stmt): Use operand_equal_p rather than pointer
9761 equality for mini-DSE code.
9763 2017-08-28 Nathan Sidwell <nathan@acm.org>
9765 * gcc.c (execute): Fold SIGPIPE handling into switch
9766 statement. Adjust internal error message.
9768 2017-08-28 Richard Biener <rguenther@suse.de>
9771 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9772 Do nothing for removed DIEs.
9774 2017-08-28 Richard Biener <rguenther@suse.de>
9776 PR tree-optimization/81977
9777 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9780 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9783 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9784 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9785 using targetm.gen_mem_thread_fence.
9787 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9790 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9791 predicate to register_operand. Reorder operands.
9792 (*btr<mode>): Ditto.
9793 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9794 (*btr<mode>_mask): Ditto.
9796 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9798 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9799 * config/rs6000/xmmintrin.h: New file.
9800 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9802 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9805 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9806 parameter and_insn and return it.
9807 (recombine_lvx_pattern): Insert a copy to ensure availability of
9808 the base register of the copied masking operation at the point of
9809 the instruction replacement.
9810 (recombine_stvx_pattern): Likewise.
9812 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9814 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9815 undocumented switches.
9816 (-mpower9-dform-vector): Likewise.
9817 (-mpower9-dform): Likewise.
9818 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9819 comments to delete references to -mpower9-dform* switches.
9820 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9821 Delete reference to -mpower9-dform* switches, test for
9822 -mpower9-vector instead.
9823 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9824 (OTHER_P9_VECTOR_MASKS): Likewise.
9825 (POWERPC_MASKS): Likewise.
9826 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9827 tests against -mpower9-dform* to -mpower9-vector. Delete code
9828 that checked for -mpower9-dform* consistancy with other options.
9829 Add test for -mpower9-misc to enable other power9 switches.
9830 (rs6000_init_hard_regno_mode_ok): Likewise.
9831 (rs6000_option_override_internal): Likewise.
9832 (rs6000_emit_prologue): Likewise.
9833 (rs6000_emit_epilogue): Likewise.
9834 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9835 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9836 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9838 (emit_fusion_p9_store): Likewise.
9839 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9840 resetting these macros if the assembler does not support ISA 3.0
9842 (TARGET_P9_DFORM_VECTOR): Likewise.
9843 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9844 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9846 2017-08-25 Alan Modra <amodra@gmail.com>
9848 PR rtl-optimization/81747
9849 * cse.c (cse_extended_basic_block): Don't attempt to record
9850 equivalences for degenerate conditional branches that branch
9851 to their fall-through.
9853 2017-08-24 Martin Sebor <msebor@redhat.com>
9856 * gimple-fold.c (size_must_be_zero_p): New function.
9857 (gimple_fold_builtin_memory_op): Call it.
9859 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9861 * config/rs6000/mm_malloc.h: New file.
9863 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9865 PR tree-optimization/81913
9866 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9867 analysis when either IVs in condition can wrap.
9869 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9871 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9872 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9874 2017-08-24 Richard Biener <rguenther@suse.de>
9877 * targhooks.c (default_target_can_inline_p): Properly
9878 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9879 is present and always compare.
9880 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9881 infer -mfpmath=sse from TARGET_SSE_P.
9882 (ix86_can_inline_p): Properly use target_option_default_node when
9883 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9885 2017-08-24 Richard Biener <rguenther@suse.de>
9888 * dwarf2out.c (output_die): Handle flag_generate_offload like
9890 (output_comp_unit): Likewise.
9891 (gen_array_type_die): Likewise.
9892 (dwarf2out_early_finish): Likewise.
9893 (note_variable_value_in_expr): Likewise.
9894 (dwarf2out_finish): Likewise. Adjust assert.
9895 * cgraphunit.c (symbol_table::compile): Move setting of
9896 flag_generate_offload earlier ...
9897 (symbol_table::finalize_compilation_unit): ... here, before
9898 early debug finalization.
9900 2017-08-24 Richard Biener <rguenther@suse.de>
9902 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9903 and ipa-fnsummary.h.
9904 (ix86_can_inline_p): When ix86_fpmath flags do not match
9905 check whether the callee uses FP math at all.
9907 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9910 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9911 instead of TYPE_PRECISION.
9913 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9914 Alan Hayward <alan.hayward@arm.com>
9915 David Sherwood <david.sherwood@arm.com>
9917 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9919 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9921 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9923 (ix86_frame::stack_realign_allocate): New field.
9924 (struct machine_frame_state): Modify comments.
9925 (machine_frame_state::sp_realigned_fp_end): New field.
9926 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9928 (sp_valid_at): Add assertion to assure no attempt to access invalid
9929 offset of a realigned stack.
9930 (fp_valid_at): Likewise.
9931 (choose_baseaddr): Modify comments.
9932 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9933 ix86_expand_prologue.
9934 (ix86_expand_prologue): Modify stack realignment and allocation.
9935 (ix86_expand_epilogue): Modify comments.
9936 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9937 avx2_runtime, avx512f, and avx512f_runtime.
9939 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9941 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9942 (mstackrealign): Do not init to -1.
9943 * config/i386/i386.c (ix86_option_override_internal):
9944 Check opts_set, not opts when setting default value of
9945 opts->x_ix86_force_align_arg_pointer.
9947 2017-08-23 Richard Biener <rguenther@suse.de>
9949 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9950 lang_hooks.decl_printable_name.
9951 * print-rtl-function.c (print_rtx_function): Likewise.
9952 * tree-pretty-print.c (dump_function_header): Likewise.
9954 2017-08-23 Richard Biener <rguenther@suse.de>
9957 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9958 -g0 at compile-time.
9960 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9963 * doc/sourcebuild.texi (Other hardware attributes):
9966 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9969 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9970 Fix single-use check.
9972 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9974 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9976 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9978 * doc/install.texi: Modify to add more details on running selected
9981 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9983 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9984 is combined with -mabi=ms.
9985 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9988 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9990 PR tree-optimization/81488
9991 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9992 and cached_basis fields.
9993 (MAX_SPREAD): New constant.
9994 (alloc_cand_and_find_basis): Initialize new fields.
9995 (clear_visited): New function.
9996 (create_phi_basis_1): Rename from create_phi_basis, set visited
9997 and cached_basis fields.
9998 (create_phi_basis): New wrapper function.
9999 (phi_add_costs_1): Rename from phi_add_costs, add spread
10000 parameter, set visited field, short-circuit when limits reached.
10001 (phi_add_costs): New wrapper function.
10002 (record_phi_increments_1): Rename from record_phi_increments, set
10004 (record_phi_increments): New wrapper function.
10005 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10006 (phi_incr_cost): New wrapper function.
10007 (all_phi_incrs_profitable_1): Rename from
10008 all_phi_incrs_profitable, set visited field.
10009 (all_phi_incrs_profitable): New wrapper function.
10011 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10012 Alan Hayward <alan.hayward@arm.com>
10013 David Sherwood <david.sherwood@arm.com>
10015 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10016 that takes the outer and inner modes.
10017 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10018 comparison as the canonical test for a paradoxical subreg.
10019 * combine.c (simplify_set): Use paradoxical_subreg_p.
10020 (make_extraction): Likewise.
10021 (force_to_mode): Likewise.
10022 (rtx_equal_for_field_assignment_p): Likewise.
10023 (gen_lowpart_for_combine): Likewise.
10024 (simplify_comparison): Likewise.
10025 * cse.c (equiv_constant): Likewise.
10026 * expmed.c (store_bit_field_1): Likewise.
10027 * final.c (alter_subreg): Likewise.
10028 * fwprop.c (propagate_rtx): Likewise.
10029 (forward_propagate_subreg): Likewise.
10030 * ira-conflicts.c (ira_build_conflicts): Likewise.
10031 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10032 * lra-constraints.c (curr_insn_transform): Likewise.
10033 (split_reg): Likewise.
10034 * lra-eliminations.c (move_plus_up): Likewise.
10035 (lra_eliminate_regs_1): Likewise.
10036 * recog.c (general_operand): Likewise.
10037 * ree.c (combine_reaching_defs): Likewise.
10038 * reload.c (push_reload): Likewise.
10039 (find_reloads): Likewise.
10040 * reload1.c (elimination_effects): Likewise.
10041 (compute_reload_subreg_offset): Likewise.
10042 (choose_reload_regs): Likewise.
10043 * rtlanal.c (subreg_lsb_1): Likewise.
10044 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10045 (simplify_subreg): Likewise.
10046 * var-tracking.c (track_loc_p): Likewise.
10047 * emit-rtl.c (byte_lowpart_offset): Likewise.
10048 (paradoxical_subreg_p): Delete out-of-line definition.
10050 2017-08-22 Jeff Law <law@redhat.com>
10052 PR tree-optimization/81741
10053 PR tree-optimization/71947
10054 * tree-ssa-dom.c: Include tree-inline.h.
10055 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10056 equivalences if one is more expensive to compute than the other.
10057 * tree-ssa-scopedtables.h (class const_or_copies): Make
10058 record_const_or_copy_raw method private.
10059 (class avail_exprs_stack): New method simplify_binary_operation.
10060 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10061 avail_exprs_stack::simplify_binary_operation as needed.
10062 (avail_exprs_stack::simplify_binary_operation): New function.
10064 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10066 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10067 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10068 (DOT_SYMBOLS): Likewise.
10069 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10070 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10071 (RS6000_ABI_NAME): Likewise.
10072 (TARGET_CMODEL): Likewise.
10073 (TOC_SECTION_ASM_OP): Likewise.
10074 (SET_CMODEL): New macro.
10075 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10077 2017-08-22 Richard Biener <rguenther@suse.de>
10079 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10080 to allow for free-lang-data replacements similar to verify_type_variant.
10082 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10083 Alan Hayward <alan.hayward@arm.com>
10084 David Sherwood <david.sherwood@arm.com>
10086 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10087 VOIDmode for the LABEL_REF.
10089 2017-08-22 Richard Biener <rguenther@suse.de>
10091 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10093 2017-08-22 Richard Biener <rguenther@suse.de>
10095 * common.opt (feliminate-dwarf2-dups): Ignore.
10096 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10097 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10098 same_die_p_wrap, compute_section_prefix,
10099 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10100 (comdat_symbol_id, comdat_symbol_number): Likewise.
10101 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10103 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10104 (output_die): Mark unreachable path unreachable.
10105 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10106 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10107 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10108 (dwarf2out_early_finish): Likewise.
10110 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10112 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10114 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10117 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10118 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10119 (avr_attribute_table) <io, io_low, address>: Initialize
10120 .decl_required with true.
10122 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10124 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10125 undocumented debugging options.
10126 (-mvsx-scalar-double): Likewise.
10127 (-mallow-df-permute): Likewise.
10128 (-mvectorize-builtins): Likewise.
10129 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10130 (rs6000_builtin_vectorized_function): Likewise.
10131 (rs6000_builtin_md_vectorized_function): Likewise.
10132 (rs6000_opt_vars): Likewise.
10134 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10137 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10138 (*btrq_imm): Rename from *btrq.
10139 (*btcq_imm): Rename from *btcq.
10140 (btsc): New code attribute.
10141 (*<btsc><mode>): New insn pattern.
10142 (*btr<mode>): Ditto.
10143 (*<btsc><mode>_mask): New insn_and_split pattern.
10144 (*btr<mode>_mask): Ditto.
10146 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10147 Alan Hayward <alan.hayward@arm.com>
10148 David Sherwood <david.sherwood@arm.com>
10150 * function.c (pad_below): Simplify padding calculation.
10152 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10153 Alan Hayward <alan.hayward@arm.com>
10154 David Sherwood <david.sherwood@arm.com>
10156 * target.def (function_prologue): Remove frame size argument.
10157 (function_epilogue): Likewise.
10158 * doc/tm.texi: Regenerate.
10159 * final.c (final_start_function): Update call to function_prologue.
10160 (final_end_function): Update call to function_epilogue.
10161 (default_function_pro_epilogue): Remove frame size argument.
10162 * output.h (default_function_pro_epilogue): Likewise.
10163 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10164 (arm_output_function_prologue): Likewise.
10165 * config/frv/frv.c (frv_function_prologue): Likewise.
10166 (frv_function_epilogue): Likewise.
10167 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10168 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10169 (ia64_output_function_epilogue): Likewise.
10170 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10171 (m32r_output_function_epilogue): Likewise.
10172 * config/microblaze/microblaze.c (microblaze_function_prologue)
10173 (microblaze_function_epilogue): Likewise.
10174 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10175 (mips_output_function_epilogue): Likewise.
10176 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10177 (mmix_target_asm_function_epilogue): Likewise.
10178 * config/msp430/msp430.c (msp430_start_function): Likewise.
10179 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10180 (nds32_asm_function_epilogue): Likewise.
10181 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10182 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10183 (pa_output_function_epilogue): Likewise.
10184 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10185 (rs6000_output_function_epilogue): Likewise.
10186 * config/rl78/rl78.c (rl78_start_function): Likewise.
10187 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10188 (rs6000_output_function_epilogue): Likewise.
10189 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10190 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10191 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10192 (sparc_asm_function_epilogue): Likewise.
10194 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10196 * tree.h (type_has_mode_precision_p): New function.
10197 * convert.c (convert_to_integer_1): Use it.
10198 * expr.c (expand_expr_real_2): Likewise.
10199 (expand_expr_real_1): Likewise.
10200 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10201 * match.pd: Likewise.
10202 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10203 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10204 * tree-tailcall.c (process_assignment): Likewise.
10205 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10206 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10207 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10208 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10209 (vectorizable_assignment): Likewise.
10210 (vectorizable_shift): Likewise.
10211 (vectorizable_operation): Likewise.
10212 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10214 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10216 * match.pd: Add pow (C, x) simplification.
10218 2017-08-21 Richard Biener <rguenther@suse.de>
10220 PR tree-optimization/81900
10221 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10222 for blocks with abnormal predecessors.
10223 (compute_antic): Do not set visited flag prematurely.
10225 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10228 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10230 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10232 * stor-layout.h (vector_type_mode): Move to...
10233 * tree.h (vector_type_mode): ...here.
10234 * stor-layout.c (vector_type_mode): Move to...
10235 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10237 2017-08-21 Richard Biener <rguenther@suse.de>
10239 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10240 register_external_die hooks.
10241 (debug_false_tree_charstarstar_uhwistar): Declare.
10242 (debug_nothing_tree_charstar_uhwi): Likewise.
10243 * debug.c (do_nothing_debug_hooks): Adjust.
10244 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10245 (debug_nothing_tree_charstar_uhwi): Likewise.
10246 * dbxout.c (dbx_debug_hooks): Adjust.
10247 (xcoff_debug_hooks): Likewise.
10248 * sdbout.c (sdb_debug_hooks): Likewise.
10249 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10250 * dwarf2out.c (macinfo_label_base): New global.
10251 (dwarf2out_register_external_die): New function for the
10252 register_external_die hook.
10253 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10254 (dwarf2_debug_hooks): Use them.
10255 (dwarf2_lineno_debug_hooks): Adjust.
10256 (struct die_struct): Add with_offset flag.
10257 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10258 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10259 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10260 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10261 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10262 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10263 defining section names for the early LTO debug variants.
10264 (reset_indirect_string): New helper.
10265 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10266 (print_dw_val): Add support for offsetted symbol references.
10267 (get_ultimate_context): Split out from is_cxx.
10268 (is_cxx): Use get_ultimate_context.
10269 (is_fortran): Add decl overload.
10270 (compute_comp_unit_symbol): Split out worker from
10271 compute_section_prefix.
10272 (compute_section_prefix): Call compute_comp_unit_symbol and
10273 set comdat_type_p here.
10274 (output_die): Skip DIE symbol output for the LTO added one.
10275 Handle DIE symbol references with offset.
10276 (output_comp_unit): Guard section name mangling properly.
10277 For LTO debug sections emit a symbol at the section beginning
10278 which we use to refer to its DIEs.
10279 (add_abstract_origin_attribute): For DIEs registered via
10280 dwarf2out_register_external_die directly refer to the early
10281 DIE rather than indirectly through the shadow one we created.
10282 Remove obsolete call to dwarf2out_abstract_function for
10283 non-function/block origins.
10284 (gen_array_type_die): When generating early LTO debug do
10285 not emit DW_AT_string_length.
10286 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10287 late when in LTO. As suggested place a gcc_unreachable for
10288 the DECL_ABSTRACT_P case.
10289 (gen_subprogram_die): Avoid another specification DIE
10290 for early built declarations/definitions for the late LTO case.
10291 (gen_variable_die): Add type references for late duplicated VLA dies
10293 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10294 we have the abstract instance already.
10295 (process_scope_var): Adjust decl DIE contexts in LTO which
10296 first puts them in limbo.
10297 (gen_decl_die): Do not generate type DIEs late apart from
10298 types for VLAs or for decls we do not yet have a DIE. Do not
10299 call dwarf2out_abstract_function late.
10300 (dwarf2out_early_global_decl): Make sure to create DIEs
10301 for abstract instances of a decl first.
10302 (dwarf2out_late_global_decl): Adjust comment.
10303 (output_macinfo_op): With multiple macro sections use
10304 macinfo_label_base to distinguish labels.
10305 (output_macinfo): Likewise. Update macinfo_label_base.
10306 Pass in the line info label.
10307 (note_variable_value_in_expr): When generating LTO resolve
10308 all variable values here by generating DIEs as needed.
10309 (init_sections_and_labels): Add early LTO debug flag parameter
10310 and generate different sections and names if set. Add generation
10311 counter for the labels so we can have multiple of them.
10312 (reset_dies): Helper to allow DIEs to be output multiple times.
10313 (dwarf2out_finish): When outputting DIEs to the fat part of an
10314 LTO object first reset DIEs.
10315 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10316 (modified_type_die): Check for decl_ultimate_origin being self
10318 (gen_type_die_with_usage): Likewise.
10319 (gen_typedef_die): Allow decl_ultimate_origin being self.
10320 (set_decl_abstract_flags): Remove.
10321 (set_block_abstract_flags): Likewise.
10322 (dwarf2out_abstract_function): Treat the early generated DIEs
10323 as the abstract copy and only add DW_AT_inline and
10324 DW_AT_artificial here and call set_decl_origin_self.
10325 If the DIE has an abstract origin don't do anything.
10326 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10327 if we have none yet (Go fails to build one, PR78628).
10328 (variably_modified_type_p): Prevent endless recursion for Ada
10329 cyclic pointer types.
10330 * lto-streamer-in.c: Include debug.h.
10331 (dref_queue): New global.
10332 (lto_read_tree_1): Stream in DIE references.
10333 (lto_input_tree): Register DIE references.
10334 (input_function): Stream DECL_DEBUG_ARGS.
10335 * lto-streamer-out.c: Include debug.h.
10336 (lto_write_tree_1): Output DIE references.
10337 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10338 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10339 (output_function): Stream DECL_DEBUG_ARGS.
10340 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10341 Stream DECL_ABSTRACT_ORIGIN.
10342 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10343 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10344 DECL_CONTEXT for file-scope decls.
10345 * lto-streamer.h (struct dref_entry): Declare.
10346 (dref_queue): Likewise.
10347 * cfgexpand.c (pass_expand::execute): Do not call the
10348 outlining_inline_function hook here.
10349 * lto-wrapper.c (debug_obj): New global.
10350 (tool_cleanup): Unlink it if required.
10351 (debug_objcopy): New function.
10352 (run_gcc): Handle early debug sections in the IL files by
10353 extracting them to separate files, partially linkin them and
10354 feeding the result back as result to the linker.
10355 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10356 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10357 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10358 sections into a separate segment.
10359 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10361 (darwin_asm_dwarf_section): Likewise.
10362 (darwin_asm_output_dwarf_offset): Likewise.
10363 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10365 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10366 Alan Hayward <alan.hayward@arm.com>
10367 David Sherwood <david.sherwood@arm.com>
10369 * read-md.h (md_reader::record_potential_iterator_use): Replace
10370 pointer argument with an rtx and an index.
10371 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10372 (apply_mode_iterator): Likewise.
10373 (apply_code_iterator): Likewise.
10374 (apply_int_iterator): Likewise.
10375 (apply_subst_iterator): Likewise.
10376 (record_iterator_use): Likewise.
10377 (record_attribute_use): Likewise.
10378 (md_reader::record_potential_iterator_use): Likewise. Update calls
10379 to record_iterator_use and apply_iterator.
10380 (iterator_use): Replace ptr with x and index.
10381 (attribute_use): Likewise.
10382 (apply_attribute_uses): Update calls to apply_iterator.
10383 (apply_iterators): Likewise. Update initialization of iterator_use.
10384 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10385 and record_potential_iterator_use.
10386 (rtx_reader::read_rtx_operand): Likewise.
10388 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10389 Alan Hayward <alan.hayward@arm.com>
10390 David Sherwood <david.sherwood@arm.com>
10392 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10395 2017-08-21 Richard Biener <rguenther@suse.de>
10397 PR middle-end/81884
10398 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10399 at struct end conservatively when comparing common bases.
10401 2017-08-21 Richard Biener <rguenther@suse.de>
10403 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10404 (mem_ref_in_stmt): Remove.
10405 (determine_max_movement): Use ref index to get at the reference.
10406 (invariantness_dom_walker::before_dom_children): Deal with
10407 lim data already initialized.
10408 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10410 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10412 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10413 (TARGET_ISA_ROUND): Ditto.
10414 (TARGET_ROUND): Ditto.
10415 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10416 * config/i386/i386.md: Ditto.
10417 * config/i386/sse.md: Ditto.
10418 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10419 with OPTION_MASK_ISA_SSE4_1.
10421 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10424 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10425 __builtin_ia32_lzcnt_u16.
10427 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10430 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10431 (rs6000_set_current_function): Rewrite function to use it.
10433 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10436 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10437 and TYPE_WARN_IF_NOT_ALIGN.
10438 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10439 (handle_warn_if_not_align): New.
10440 (place_union_field): Call handle_warn_if_not_align.
10441 (place_field): Call handle_warn_if_not_align.
10442 Copy TYPE_WARN_IF_NOT_ALIGN.
10443 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10444 (layout_type): Likewise.
10445 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10447 (tree_decl_common): Add warn_if_not_align.
10448 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10449 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10450 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10451 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10452 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10453 * doc/extend.texi: Document warn_if_not_aligned attribute.
10454 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10456 2017-08-17 Martin Liska <mliska@suse.cz>
10459 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10460 (get_data_dependence): Use it as pointer type.
10461 (distribute_loop): Likewise.
10463 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10465 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10466 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10467 (p8_vmrgow_<mode>_direct): New define_insn.
10468 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10469 handle endianness for vmrgew and vmrgow permute patterns.
10471 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10473 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10474 * config/rs6000/rs6000-cpus.def: Remove comment.
10475 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10476 (POWERPC_MASKS): Likewise.
10477 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10478 use of TARGET_VSX_TIMODE.
10479 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10480 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10481 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10482 (rs6000_option_override_internal): Remove dead code.
10483 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10484 (rs6000_legitimize_reload_address): Likewise.
10485 (rs6000_legitimate_address_p): Likewise.
10486 (rs6000_opt_masks): Delete "vsx-timode".
10487 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10488 from function comment.
10489 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10490 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10491 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10493 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10494 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10495 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10496 (VSX_TI): Likewise.
10498 (define_peephole2): Likewise.
10500 2017-08-17 Martin Sebor <msebor@redhat.com>
10503 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10504 past the end of an array.
10505 (test_pp_format): Add test cases.
10507 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10509 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10510 missing ECF_NOTHROW flags.
10512 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10515 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10516 operands residing in integer registers.
10517 (*vsx_le_perm_load_<mode>): Likewise.
10518 (*vsx_le_perm_store_<mode>): Likewise.
10519 (define_peephole2): Add peepholes to optimize the above.
10521 2017-08-17 Marek Polacek <polacek@redhat.com>
10523 PR middle-end/81814
10524 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10525 to mimic what shorten_compare did. Change the return type to bool.
10526 (fold_cond_expr_with_comparison): Update call to
10527 operand_equal_for_comparison_p.
10528 (fold_ternary_loc): Likewise.
10530 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10532 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10534 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10536 2017-08-17 Richard Biener <rguenther@suse.de>
10538 * tree-ssa-structalias.c (solve_graph): When propagating
10539 to successors update the graphs succ edges and avoid duplicate work.
10541 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10544 * config/i386/i386.c (ix86_option_override_internal): Save target
10545 specific options after ix86_stack_protector_guard_reg was changed.
10547 2017-08-17 Richard Biener <rguenther@suse.de>
10549 PR tree-optimization/81827
10550 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10551 (new_var_info): Initialize it conservatively.
10552 (get_call_vi): Mark register vars.
10553 (new_scalar_tmp_constraint_exp): Likewise.
10554 (handle_rhs_call): Likewise.
10555 (handle_const_call): Likewise.
10556 (create_function_info_for): Likewise.
10557 (solve_constraints): Sort varinfos to separate register from
10558 non-register vars to pack points-to solution bitmaps during
10561 2017-08-17 Marek Polacek <polacek@redhat.com>
10563 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10565 2017-08-17 Richard Biener <rguenther@suse.de>
10567 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10568 to true when overflow is undefined and we saturated the result.
10570 2017-08-17 Alan Modra <amodra@gmail.com>
10573 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10574 Don't use store multiple if only one reg needs saving.
10575 (interesting_frame_related_regno): New function.
10576 (rs6000_frame_related): Don't emit frame info for regs that
10578 (rs6000_emit_epilogue): Likewise.
10580 2017-08-16 Nathan Sidwell <nathan@acm.org>
10582 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10583 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10584 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10585 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10586 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10587 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10588 (verify_type): Adjust for TYPE_BINFO move.
10589 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10590 process TYPE_BINFO directly.
10591 (hash_tree): Likewise.
10592 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10594 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10597 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10599 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10601 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10604 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10605 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10606 Add ix86_binary_operator_ok to insn constraint.
10607 (*iordi_1_bts): Ditto.
10608 (*xordi_1_btc): Ditto.
10609 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10610 Update corresponding peephole2 pattern.
10614 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10616 PR tree-optimization/81832
10617 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10618 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10620 2017-08-16 Marek Polacek <polacek@redhat.com>
10623 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10624 regarding size_zero_node.
10626 2017-08-16 Martin Liska <mliska@suse.cz>
10629 * config.gcc: Respect previously set extra_objs in case
10632 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10634 PR tree-optimization/81835
10635 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10636 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10637 not depend on the phi.
10639 2017-08-16 Alan Modra <amodra@gmail.com>
10641 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10644 2017-08-16 Alan Modra <amodra@gmail.com>
10646 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10647 (save_reg_p): ..into this. Update all callers.
10648 (first_reg_to_save): Simplify.
10650 2017-08-16 Alan Modra <amodra@gmail.com>
10652 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10655 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10659 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10660 constant count if that count is less than 32.
10662 2017-08-15 Nathan Sidwell <nathan@acm.org>
10664 * gcc.c (execute): Emit friendlier message if inferior is killed
10665 by an external cause.
10667 2017-08-15 Richard Biener <rguenther@suse.de>
10669 PR tree-optimization/81790
10670 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10671 CONSTRUCTORs from simplifying and VN.
10673 2017-08-14 Martin Sebor <msebor@redhat.com>
10675 * builtin-attrs.def: Add comments.
10677 2017-08-14 Martin Sebor <msebor@redhat.com>
10680 * doc/extend.texi (attribute nonstring): Document new attribute.
10682 2017-08-14 Martin Sebor <msebor@redhat.com>
10685 * tree-diagnostic.c (default_tree_printer): Handle %G.
10686 * gimple-pretty-print.h (percent_G_format): Declare new function.
10687 * gimple-pretty-print.c (percent_G_format): Define.
10688 * tree-pretty-print.c (percent_K_format): Add argument.
10690 2017-08-14 Martin Sebor <msebor@redhat.com>
10692 PR translation/79998
10693 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10694 Remove a stray space.
10696 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10699 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10700 (*iordi_1_bts): Ditto.
10701 (*xordi_1_btc): Ditto.
10703 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10706 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10707 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10709 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10710 quoted strings, and make more translator-friendly.
10711 (darwin_rs6000_override_options): Likewise.
10712 (rs6000_option_override_internal): Likewise.
10713 (rs6000_return_in_memory): Fix overlong line.
10714 (init_cmulative_args): Use quoted strings, and make more
10715 translator-friendly.
10716 (rs6000_pass_by_reference): Fix overlong line.
10717 (def_builtin): Use quoted strings.
10718 (altivec_expand_predicate_builtin): Use quoted strings, and make
10719 more translator-friendly.
10720 (htm_expand_builtin): Use quoted strings.
10721 (cpu_expand_builtin): Use quoted strings, and make more
10722 translator-friendly.
10723 (altivec_expand_builtin): Likewise.
10724 (paired_expand_predicate_builtin): Likewise.
10725 (rs6000_invalid_builtin): Likewise.
10726 (builtin_function_type): Use quoted strings.
10727 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10728 more translator-friendly.
10729 (rs6000_trampoline_init): Likewise.
10730 (rs6000_handle_altivec_attribute): Likewise.
10731 (rs6000_inner_target_options): Use quoted strings.
10732 (rs6000_disable_incompatible_switches): Likewise.
10733 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10734 strings, and make more translator-friendly.
10735 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10737 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10739 PR tree-optimization/81799
10740 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10741 cond_expr to simple gimple operand.
10743 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10745 PR middle-end/46932
10746 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10748 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10752 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10753 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10754 TARGET_GASISR_PROLOGUES.
10755 * config/avr/avr.c (avr_option_override): Same.
10756 (avr_pass_pre_proep::execute): Same.
10758 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10761 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10762 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10764 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10766 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10767 define_insn_and_split. Split to a memory load from 0 in
10768 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10769 using PTR mode iterator.
10770 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10771 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10772 (*add_tp_<mode>): Redefine as define_insn_and_split.
10773 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10774 address space. Merge with *add_tp_x32 using PTR mode iterator.
10775 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10776 Split to an add with a memory load from 0 in
10777 DEFAULT_TLS_SEG_REG address space.
10779 2017-08-12 Andrew Pinski <apinski@cavium.com>
10781 * config/aarch64/aarch64-option-extensions.def (rdma):
10782 Fix feature string to what Linux prints out in /proc/cpuinfo.
10784 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10787 * dwarf2out.c (modified_type_die): For C typedef types that have
10788 an ultimate origin, process the ultimate origin instead of the
10790 (gen_typedef_die): Assert that input DECLs have no ultimate
10792 (gen_type_die_with_usage): For typedef variants that have an
10793 ultimate origin, just call gen_decl_die on the original DECL.
10794 (process_scope_var): Avoid creating DIEs for local typedefs and
10795 concrete static variables.
10797 2017-08-12 Alan Modra <amodra@gmail.com>
10801 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10802 match gnu-user.h startfile.
10803 (ENDFILE_LINUX_SPEC): Similarly.
10805 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10808 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10810 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10812 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10813 * config/aarch64/aarch64.md (mov<mode>): Change.
10814 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10815 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10816 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10818 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10820 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10821 for storage order barriers.
10823 2017-08-11 Martin Liska <mliska@suse.cz>
10826 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10827 variables of void type.
10829 2017-08-11 Martin Liska <mliska@suse.cz>
10831 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10832 TARGET_SUPPORTS_ALIASES.
10833 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10834 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10835 (optimize_weakref): Likewise.
10836 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10837 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10838 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10840 2017-08-11 Martin Liska <mliska@suse.cz>
10843 * config/i386/i386.c (make_resolver_func): Do complete
10844 refactoring of the function.
10846 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10849 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10850 * config/i386/i386.c (ix86_stack_protect_guard): Use
10851 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10852 * doc/invoke.texi (x86 Options): Document
10853 -mstack-protector-guard-symbol= option.
10855 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10857 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10858 * config/i386/i386.c (ix86_split_stack_guard): New function.
10859 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10860 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10861 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10862 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10863 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10864 (split_stack_space_check): Call ix86_split_stack_guard.
10866 2017-08-10 Martin Sebor <msebor@redhat.com>
10868 * print-tree.c (print_node): Print location using the established
10870 Replace spaces with colons.
10871 (debug_raw, debug): Ditto.
10873 2017-08-10 Martin Sebor <msebor@redhat.com>
10876 * pretty-print.c (pp_format): Correct the handling of %s precision.
10878 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10881 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10883 (ix86_finalize_stack_frame_flags): This. Also clear
10884 frame_pointer_needed if -fno-omit-frame-pointer is used without
10886 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10887 with ix86_finalize_stack_frame_flags.
10888 (ix86_expand_epilogue): Likewise.
10889 (ix86_expand_split_stack_prologue): Likewise.
10890 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10892 2017-08-10 Martin Liska <mliska@suse.cz>
10895 * c-attribs.c (handle_target_attribute):
10896 Report warning for an empty string argument of target attribute.
10898 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10901 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10903 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10904 or DECL_NONLOCAL labels.
10905 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10906 or DECL_NONLOCAL labels here.
10908 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10910 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10911 to indicate when early gimple folding has been disabled.
10912 (rs6000_gimple_fold_builtin): Add debug content.
10913 (rs6000_invalid_builtin): Fix whitespace.
10914 (rs6000_expand_builtin): Fix whitespace.
10915 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10917 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10920 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10921 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10923 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10925 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10926 (qdf24xx): Likewise.
10927 * config/aarch64/aarch64.md: Include falkor.md.
10928 * config/aarch64/falkor.md: New.
10930 2017-08-09 Marek Polacek <polacek@redhat.com>
10933 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10934 * diagnostic.c (emit_diagnostic): Add a comment.
10935 (emit_diagnostic_valist): New function.
10937 2017-08-09 Marek Polacek <polacek@redhat.com>
10940 * input.c (make_location): New overload.
10941 * input.h (make_location): Declare.
10943 2017-08-08 Alan Modra <amodra@gmail.com>
10944 H.J. Lu <hongjiu.lu@intel.com>
10947 * gcc.c (NO_PIE_SPEC): Delete.
10948 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10950 (LINK_PIE_SPEC): ..to here.
10951 (LINK_COMMAND_SPEC): Support -no-pie.
10952 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10953 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10955 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10956 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10957 (ENDFILE_CRTEND_SPEC): Similarly.
10959 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10962 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10963 (mstack-protector-guard-offset=): Ditto.
10964 * config/i386/i386.c (ix86_option_override): Handle
10965 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10967 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10968 ix86_stack_protect_guard_offset variables.
10969 (TARGET_STACK_PROTECT_GUARD): Always define.
10970 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10971 and -mstack-protector-guard-offset= options.
10973 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10975 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10976 boundary case for the last candidate.
10978 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10980 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10981 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10983 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10985 PR middle-end/19706
10986 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10987 * config/aarch64/aarch64-builtins.c
10988 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10989 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10990 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10992 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10993 Andrew Pinski <pinskia@gmail.com>
10995 PR middle-end/19706
10996 * internal-fn.def (XORSIGN): New.
10997 * optabs.def (xorsign_optab): New.
10998 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10999 (convert_expand_mult_copysign): New.
11000 (pass_optimize_widening_mul::execute): Call
11001 convert_expand_mult_copysign.
11003 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11005 PR tree-optimization/81354
11006 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11007 Insert on edges rather than explicitly creating landing pads.
11008 (analyze_candidates_and_replace): Commit edge inserts.
11010 2017-08-08 Richard Biener <rguenther@suse.de>
11012 PR middle-end/81719
11013 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11014 (expand_simple_operations): Also look through ADDR_EXPRs with
11015 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11017 2017-08-08 Richard Biener <rguenther@suse.de>
11019 PR tree-optimization/81723
11020 * tree-vect-slp.c (struct bst_traits): New hash traits.
11021 (bst_fail): New global.
11022 (vect_build_slp_tree_2): New worker, split out from ...
11023 (vect_build_slp_tree): ... this now wrapping it with using
11024 bst_fail set to cache SLP tree build fails. Properly handle
11026 (vect_analyze_slp_instance): Allocate and free bst_fail.
11028 2017-08-08 Martin Liska <mliska@suse.cz>
11031 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11032 LABEL_DECLs that can be from a different function.
11034 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11036 PR tree-optimization/81744
11037 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11038 loop's number of iterations.
11040 2017-08-08 Martin Liska <mliska@suse.cz>
11042 * asan.c: Include header files.
11043 * attribs.c (build_decl_attribute_variant): New function moved
11045 (build_type_attribute_qual_variant): Likewise.
11046 (cmp_attrib_identifiers): Likewise.
11047 (simple_cst_list_equal): Likewise.
11048 (omp_declare_simd_clauses_equal): Likewise.
11049 (attribute_value_equal): Likewise.
11050 (comp_type_attributes): Likewise.
11051 (build_type_attribute_variant): Likewise.
11052 (lookup_ident_attribute): Likewise.
11053 (remove_attribute): Likewise.
11054 (merge_attributes): Likewise.
11055 (merge_type_attributes): Likewise.
11056 (merge_decl_attributes): Likewise.
11057 (merge_dllimport_decl_attributes): Likewise.
11058 (handle_dll_attribute): Likewise.
11059 (attribute_list_equal): Likewise.
11060 (attribute_list_contained): Likewise.
11061 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11062 (lookup_attribute_by_prefix): Likewise.
11063 * bb-reorder.c: Include header files.
11064 * builtins.c: Likewise.
11065 * calls.c: Likewise.
11066 * cfgexpand.c: Likewise.
11067 * cgraph.c: Likewise.
11068 * cgraphunit.c: Likewise.
11069 * convert.c: Likewise.
11070 * dwarf2out.c: Likewise.
11071 * final.c: Likewise.
11072 * fold-const.c: Likewise.
11073 * function.c: Likewise.
11074 * gimple-expr.c: Likewise.
11075 * gimple-fold.c: Likewise.
11076 * gimple-pretty-print.c: Likewise.
11077 * gimple.c: Likewise.
11078 * gimplify.c: Likewise.
11079 * hsa-common.c: Likewise.
11080 * hsa-gen.c: Likewise.
11081 * internal-fn.c: Likewise.
11082 * ipa-chkp.c: Likewise.
11083 * ipa-cp.c: Likewise.
11084 * ipa-devirt.c: Likewise.
11085 * ipa-fnsummary.c: Likewise.
11086 * ipa-inline.c: Likewise.
11087 * ipa-visibility.c: Likewise.
11089 * lto-cgraph.c: Likewise.
11090 * omp-expand.c: Likewise.
11091 * omp-general.c: Likewise.
11092 * omp-low.c: Likewise.
11093 * omp-offload.c: Likewise.
11094 * omp-simd-clone.c: Likewise.
11095 * opts-global.c: Likewise.
11096 * passes.c: Likewise.
11097 * predict.c: Likewise.
11098 * sancov.c: Likewise.
11099 * sanopt.c: Likewise.
11100 * symtab.c: Likewise.
11101 * toplev.c: Likewise.
11102 * trans-mem.c: Likewise.
11103 * tree-chkp.c: Likewise.
11104 * tree-eh.c: Likewise.
11105 * tree-into-ssa.c: Likewise.
11106 * tree-object-size.c: Likewise.
11107 * tree-parloops.c: Likewise.
11108 * tree-profile.c: Likewise.
11109 * tree-ssa-ccp.c: Likewise.
11110 * tree-ssa-live.c: Likewise.
11111 * tree-ssa-loop.c: Likewise.
11112 * tree-ssa-sccvn.c: Likewise.
11113 * tree-ssa-structalias.c: Likewise.
11114 * tree-ssa.c: Likewise.
11115 * tree-streamer-in.c: Likewise.
11116 * tree-vectorizer.c: Likewise.
11117 * tree-vrp.c: Likewise.
11118 * tsan.c: Likewise.
11119 * ubsan.c: Likewise.
11120 * varasm.c: Likewise.
11121 * varpool.c: Likewise.
11122 * tree.c: Remove functions moved to attribs.[ch].
11123 * tree.h: Likewise.
11124 * config/aarch64/aarch64.c: Add attrs.h header file.
11125 * config/alpha/alpha.c: Likewise.
11126 * config/arc/arc.c: Likewise.
11127 * config/arm/arm.c: Likewise.
11128 * config/avr/avr.c: Likewise.
11129 * config/bfin/bfin.c: Likewise.
11130 * config/c6x/c6x.c: Likewise.
11131 * config/cr16/cr16.c: Likewise.
11132 * config/cris/cris.c: Likewise.
11133 * config/darwin.c: Likewise.
11134 * config/epiphany/epiphany.c: Likewise.
11135 * config/fr30/fr30.c: Likewise.
11136 * config/frv/frv.c: Likewise.
11137 * config/ft32/ft32.c: Likewise.
11138 * config/h8300/h8300.c: Likewise.
11139 * config/i386/winnt.c: Likewise.
11140 * config/ia64/ia64.c: Likewise.
11141 * config/iq2000/iq2000.c: Likewise.
11142 * config/lm32/lm32.c: Likewise.
11143 * config/m32c/m32c.c: Likewise.
11144 * config/m32r/m32r.c: Likewise.
11145 * config/m68k/m68k.c: Likewise.
11146 * config/mcore/mcore.c: Likewise.
11147 * config/microblaze/microblaze.c: Likewise.
11148 * config/mips/mips.c: Likewise.
11149 * config/mmix/mmix.c: Likewise.
11150 * config/mn10300/mn10300.c: Likewise.
11151 * config/moxie/moxie.c: Likewise.
11152 * config/msp430/msp430.c: Likewise.
11153 * config/nds32/nds32-isr.c: Likewise.
11154 * config/nds32/nds32.c: Likewise.
11155 * config/nios2/nios2.c: Likewise.
11156 * config/nvptx/nvptx.c: Likewise.
11157 * config/pa/pa.c: Likewise.
11158 * config/pdp11/pdp11.c: Likewise.
11159 * config/powerpcspe/powerpcspe.c: Likewise.
11160 * config/riscv/riscv.c: Likewise.
11161 * config/rl78/rl78.c: Likewise.
11162 * config/rx/rx.c: Likewise.
11163 * config/s390/s390.c: Likewise.
11164 * config/sh/sh.c: Likewise.
11165 * config/sol2.c: Likewise.
11166 * config/sparc/sparc.c: Likewise.
11167 * config/spu/spu.c: Likewise.
11168 * config/stormy16/stormy16.c: Likewise.
11169 * config/tilegx/tilegx.c: Likewise.
11170 * config/tilepro/tilepro.c: Likewise.
11171 * config/v850/v850.c: Likewise.
11172 * config/vax/vax.c: Likewise.
11173 * config/visium/visium.c: Likewise.
11174 * config/xtensa/xtensa.c: Likewise.
11176 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11179 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11180 constraints since the -mupper-regs-* switches have been
11182 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11183 into a vector from a double word element that was extracted from
11184 another vector, and eliminate extra XXPERMDI instructions.
11185 (vsx_concat_<mode>_2): Likewise.
11186 (vsx_concat_<mode>_3): Likewise.
11187 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11188 concat to allow optimizing inserts from previous extracts.
11190 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11192 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11193 memory reference to a SSP offset in TLS address space.
11194 (ix86_print_operand) <case '@'>: Remove.
11195 (ix86_print_operand_punct_valid_p): Remove '@' code.
11196 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11197 UNSPEC_SP_TLS_TEST.
11198 (stack_tls_protect_set_<mode>): Remove.
11199 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11200 (stack_tls_protect_test_<mode>): Remove.
11201 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11203 2017-08-07 Olivier Hainque <hainque@adacore.com>
11206 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11208 2017-08-07 Douglas Rupp <rupp@adacore.com>
11210 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11211 variable was referenced as multidir in command.
11213 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11216 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11219 2017-08-07 Martin Liska <mliska@suse.cz>
11221 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11222 * config/rl78/rl78.c: Add include of attribs.h.
11223 * config/sh/sh.c: Likewise.
11224 * config/v850/v850.c: Likewise.
11226 2017-08-07 Tom de Vries <tom@codesourcery.com>
11228 PR middle-end/78266
11229 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11231 2017-08-07 Martin Liska <mliska@suse.cz>
11233 * config/mips/mips.c: Include attribs.h.
11235 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11238 * doc/invoke.texi: Document change in behvaior for -Ofast for
11241 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11243 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11245 (aarch64_pop_regs): Likewise.
11246 (aarch64_gen_load_pair): Likewise.
11247 (aarch64_save_callee_saves): Likewise.
11248 (aarch64_restore_callee_saves): Likewise.
11250 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11252 * config/i386/i386.c: Revert the last change.
11254 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11257 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11259 (ix86_finalize_stack_frame_flags): This. Also clear
11260 frame_pointer_needed if -fno-omit-frame-pointer is used without
11262 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11263 with ix86_finalize_stack_frame_flags.
11264 (ix86_expand_epilogue): Likewise.
11265 (ix86_expand_split_stack_prologue): Likewise.
11267 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11270 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11271 to P_AES for Westmere.
11273 2017-08-07 Jonathan Yong <10walls@gmail.com>
11275 * config/i386/mingw.opt (fset-stack-executable): Removed.
11276 * config/i386/cygming.opt (fset-stack-executable): Moved
11278 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11280 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11282 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11284 2017-08-07 Marek Polacek <polacek@redhat.com>
11286 PR middle-end/81737
11287 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11289 2017-08-07 Martin Liska <mliska@suse.cz>
11291 * attribs.h (canonicalize_attr_name): New function.
11292 (cmp_attribs): Move from c-format.c and adjusted.
11293 (is_attribute_p): Moved from tree.h.
11294 * tree-inline.c: Add new includes.
11295 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11296 (private_is_attribute_p): Remove.
11297 (private_lookup_attribute): Likewise.
11298 (private_lookup_attribute_by_prefix): Simplify.
11299 (remove_attribute): Use is_attribute_p.
11300 * tree.h: Remove removed declarations.
11302 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11304 PR middle-end/81698
11305 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11306 instead of computing it in the function. Formatting fix.
11307 (expand_case): Don't rely on default_edge being the first edge,
11308 clear it if removing it, pass default_edge to
11309 emit_case_dispatch_table.
11310 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11313 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11315 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11316 insn in the function, emit NOP after the insn.
11318 2017-08-06 Tom de Vries <tom@codesourcery.com>
11320 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11323 2017-08-06 Tom de Vries <tom@codesourcery.com>
11325 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11328 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11330 PR tree-optimization/57371
11331 * match.pd: New pattern.
11333 2017-08-04 Marek Polacek <polacek@redhat.com>
11335 PR middle-end/81695
11336 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11337 perform the computation in offset_int.
11339 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11341 PR tree-optimization/81136
11342 * tree-vectorizer.h: Include tree-hash-traits.h.
11343 (vec_base_alignments): New typedef.
11344 (vec_info): Add a base_alignments field.
11345 (vect_record_base_alignments): Declare.
11346 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11348 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11349 (create_data_ref): Add an is_conditional_in_stmt argument.
11350 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11351 the is_conditional_in_stmt field.
11352 (data_ref_loc): Add an is_conditional_in_stmt field.
11353 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11354 (find_data_references_in_stmt): Update call to create_data_ref.
11355 (graphite_find_data_references_in_stmt): Likewise.
11356 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11357 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11358 (vect_record_base_alignment): New function.
11359 (vect_record_base_alignments): Likewise.
11360 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11361 for nested statements even if we fail to compute a misalignment.
11362 Use pooled base alignments for unconditional references.
11363 (vect_find_same_alignment_drs): Compare base addresses instead
11365 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11366 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11368 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11370 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11371 Add an explicit name for the enum. Use auto_vec for slp_instances
11372 and grouped_stores.
11373 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11375 (_bb_vec_info): Add a constructor and destructor.
11376 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11377 (destroy_loop_vec_info): Delete.
11378 (vect_destroy_datarefs): Likewise.
11379 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11380 (vec_info::vec_info): New function.
11381 (vec_info::~vec_info): Likewise.
11382 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11383 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11384 destroy_loop_vec_info.
11385 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11386 (_loop_vec_info::_loop_vec_info): ...this.
11387 (destroy_loop_vec_info): Replace with...
11388 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11389 the stmt_vec_infos. Leave handling of vec_info information to its
11390 destructor. Remove explicit vector releases.
11391 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11392 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11393 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11394 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11395 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11396 (destroy_bb_vec_info): Replace with...
11397 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11398 information to its destructor.
11399 (vect_slp_analyze_bb_1): Use new and delete instead of
11400 new_bb_vec_info and destroy_bb_vec_info.
11401 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11404 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11406 * tree-data-ref.h (subscript): Add access_fn field.
11407 (data_dependence_relation): Add could_be_independent_p.
11408 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11409 (same_access_functions): Move to tree-data-ref.c.
11410 * tree-data-ref.c (ref_contains_union_access_p): New function.
11411 (access_fn_component_p): Likewise.
11412 (access_fn_components_comparable_p): Likewise.
11413 (dr_analyze_indices): Add a reference to access_fn_component_p.
11414 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11416 (constant_access_functions): Likewise.
11417 (add_other_self_distances): Likewise.
11418 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11419 (initialize_data_dependence_relation): Use XCNEW and remove
11420 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11421 of access functions that have the same type. Allow the
11422 subsequence to end with different bases in some circumstances.
11423 Record the chosen access functions in SUB_ACCESS_FN.
11424 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11425 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11426 (subscript_dependence_tester_1): Likewise dra and drb.
11427 (build_classic_dist_vector): Update calls accordingly.
11428 (subscript_dependence_tester): Likewise.
11429 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11430 DDR_COULD_BE_INDEPENDENT_P.
11431 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11432 comp_alias_ddrs instead of may_alias_ddrs.
11433 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11435 (vect_analyze_data_ref_dependence): Use it if
11436 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11437 distance vectors if that fails.
11438 (dependence_distance_ge_vf): New function.
11439 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11440 LOOP_VINFO_MAY_ALIAS_DDRS.
11442 2017-08-04 Richard Biener <rguenther@suse.de>
11444 PR middle-end/81705
11445 * fold-const.c (fold_binary_loc): Properly restrict
11446 minus_var0 && minus_var1 case when associating undefined overflow
11449 2017-08-04 Tom de Vries <tom@codesourcery.com>
11451 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11453 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11455 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11456 Don't start diagnostic messages with a capital letter.
11457 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11459 (rs6000_invalid_builtin): Likewise.
11460 (rs6000_trampoline_init): Likewise.
11462 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11465 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11466 after setting changeable df flags.
11468 2017-08-03 Richard Biener <rguenther@suse.de>
11470 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11471 up if the use is in USE - X.
11473 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11475 * toplev.c (dumpfile.h): New include.
11476 (internal_error_reentered): New static function. Use it...
11477 (internal_error_function): ...here to handle reentered internal_error.
11479 2017-08-03 Richard Biener <rguenther@suse.de>
11481 PR middle-end/81148
11482 * fold-const.c (split_tree): Add minus_var and minus_con
11483 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11484 here but always use minus_*.
11485 (associate_trees): Assert we never associate with MINUS_EXPR
11486 and NULL first operand. Do not recurse for PLUS_EXPR operands
11487 when associating as MINUS_EXPR either.
11488 (fold_binary_loc): Track minus_var and minus_con.
11490 2017-08-03 Tom de Vries <tom@codesourcery.com>
11493 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11494 ACCEL_COMPILER, apply finish_options on
11495 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11497 2017-08-03 Tom de Vries <tom@codesourcery.com>
11500 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11501 function_entry_patch_area_size > 0.
11503 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11506 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11507 instead of 10??LU, perform unit multiplication in wide_int,
11508 don't change alloc_object_size_limit if the limit is larger
11511 PR tree-optimization/81655
11512 PR tree-optimization/81588
11513 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11514 the case when ranges[i].low and high are 1 for unsigned type with
11517 PR middle-end/81052
11518 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11519 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11521 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11523 * tree-vrp.h: Add include guard.
11525 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11528 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11529 (ud2): New insn pattern.
11530 * config/i386/i386.c (ix86_expand_epilogue):
11531 For naked functions, generate ud2 instead of trap insn.
11533 2017-08-02 Marek Polacek <polacek@redhat.com>
11536 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11538 2017-08-02 Tom de Vries <tom@codesourcery.com>
11539 Cesar Philippidis <cesar@codesourcery.com>
11541 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11542 Add missing edge probabilities.
11544 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11546 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11547 Correct endianness.
11549 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11551 PR middle-end/79499
11552 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11553 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11554 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11555 prologue_seq sequences - if any.
11557 2017-08-02 Richard Biener <rguenther@suse.de>
11559 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11560 via vectors if supported, integer extracts via punning if supported
11561 or otherwise vector extracts.
11563 2017-08-02 Richard Biener <rguenther@suse.de>
11565 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11567 (bitmap_insert_into_set): ... this.
11569 2017-08-02 Richard Biener <rguenther@suse.de>
11571 PR tree-optimization/81633
11573 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11575 PR tree-optimization/71752
11576 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11578 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11580 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11581 (machine_function::call_ms2sysv_pad_out): Remove field.
11582 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11583 (ix86_compute_frame_layout): Likewise.
11585 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11588 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11589 attribute with interrupt attribute.
11591 2017-08-01 Andrew Pinski <apinski@cavium.com>
11593 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11594 BIT_INSERT_EXPR's operand 1
11595 to see if the types precision matches.
11597 2017-08-01 Martin Liska <mliska@suse.cz>
11599 PR middle-end/70140
11600 * builtins.c (expand_builtin_memcpy_args): Remove.
11601 (expand_builtin_memcpy): Call newly added function
11602 expand_builtin_memory_copy_args.
11603 (expand_builtin_memcpy_with_bounds): Likewise.
11604 (expand_builtin_mempcpy): Remove last argument.
11605 (expand_builtin_mempcpy_with_bounds): Likewise.
11606 (expand_builtin_memory_copy_args): New function created from
11607 expand_builtin_mempcpy_args with small modifications.
11608 (expand_builtin_mempcpy_args): Remove.
11609 (expand_builtin_stpcpy): Remove unused argument.
11610 (expand_builtin): Likewise.
11611 (expand_builtin_with_bounds): Likewise.
11613 2017-08-01 Martin Liska <mliska@suse.cz>
11616 Make mempcpy more optimal (PR middle-end/70140).
11618 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11622 __builtin_vec_cmpne verify both arguments are compatible vectors
11623 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11624 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11625 move computation of aligned to after checking the argument types.
11629 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11630 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11633 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11634 Doug Rupp <rupp@adacore.com>
11635 Olivier Hainque <hainque@adacore.com>
11637 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11638 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11640 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11641 for TARGET_OS_CPP_BUILTIN.
11642 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11643 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11645 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11646 passing required abi options to the assembler for EABI configurations.
11647 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11648 of .text.hot and .text.unlikely sections for kernel modules when
11649 using ARM style exceptions.
11650 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11651 options. Add EXTRA_CC1_SPEC.
11652 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11654 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11656 (ARM_TARGET2_DWARF_FORMAT): Define.
11657 * config/arm/t-vxworks: Adjust multilib control to removal of the
11658 Diab command line options.
11660 2017-08-01 Martin Liska <mliska@suse.cz>
11662 PR gcov-profile/81561
11663 * gcov.c (unblock): Make unblocking safe as we need to preserve
11664 index correspondence of blocks and block_lists.
11666 2017-08-01 Richard Biener <rguenther@suse.de>
11668 PR tree-optimization/81181
11669 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11670 (compute_antic): ... end of iteration here.
11672 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11674 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11675 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11676 (ftree-slp-vectorize): Likewise.
11677 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11678 can no longer be set independent of flag_tree_loop_vectorize.
11679 * omp-general.c (emp_max_vf): Likewise.
11680 * opts.c (enable_fdo_optimizations): Remove references to
11681 flag_tree_vectorize, these are now implicit.
11682 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11683 and leave it for the options machinery.
11685 2017-08-01 Martin Liska <mliska@suse.cz>
11687 PR middle-end/70140
11688 * builtins.c (expand_builtin_memcpy_args): Remove.
11689 (expand_builtin_memcpy): Call newly added function
11690 expand_builtin_memory_copy_args.
11691 (expand_builtin_memcpy_with_bounds): Likewise.
11692 (expand_builtin_mempcpy): Remove last argument.
11693 (expand_builtin_mempcpy_with_bounds): Likewise.
11694 (expand_builtin_memory_copy_args): New function created from
11695 expand_builtin_mempcpy_args with small modifications.
11696 (expand_builtin_mempcpy_args): Remove.
11697 (expand_builtin_stpcpy): Remove unused argument.
11698 (expand_builtin): Likewise.
11699 (expand_builtin_with_bounds): Likewise.
11701 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11704 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11705 print "ds:" only for immediates in generic address space.
11707 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11710 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11711 (ix86_function_ok_for_sibcall): Return false for naked functions.
11713 2017-08-01 Richard Biener <rguenther@suse.de>
11715 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11716 (compute_antic): Seed worklist with exit block predecessors.
11717 * cfganal.c (dfs_find_deadend): For a cycle return the source
11718 of the edge closing it.
11720 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11722 * config/aarch64/aarch64.c
11723 (aarch64_can_const_movi_rtx_p): Move 0 check.
11725 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11727 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11728 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11730 * match.pd: Ditto in address comparison pattern.
11732 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11734 PR tree-optimization/81627
11735 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11736 closed ssa form for store-store chain.
11738 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11740 PR tree-optimization/81620
11741 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11742 for store-store chain.
11744 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11746 PR tree-optimization/81588
11747 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11748 ranges[i].in_p, invert comparison code ccode. For >/>=,
11749 swap rhs1 and rhs2 and comparison code unconditionally,
11750 for </<= don't do that. Don't swap rhs1/rhs2 again if
11751 ranges[i].in_p, instead invert comparison code ccode if
11752 opcode or oe->rank is BIT_IOR_EXPR.
11755 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11756 a direct optab to conversion optab.
11757 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11758 with GET_MODE_INNER as last argument instead of optab_handler.
11759 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11760 vector extraction if possible and optab is available.
11761 * expr.c (store_constructor): Use convert_optab_handler instead
11762 of optab_handler. Use vector initialization from smaller
11763 vectors if possible and optab is available.
11764 * tree-vect-stmts.c (vectorizable_load): Likewise.
11765 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11766 now have two modes.
11767 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11768 of vec_init from half-sized vectors with the same element mode.
11769 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11770 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11771 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11772 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11773 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11774 after mode in gen_vec_extract* calls.
11775 (vec_extract<mode>): Renamed to ...
11776 (vec_extract<mode><ssescalarmodelower>): ... this.
11777 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11778 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11779 element mode after mode in gen_vec_init* calls.
11780 (VEC_INIT_HALF_MODE): New mode iterator.
11781 (vec_init<mode>): Renamed to ...
11782 (vec_init<mode><ssescalarmodelower>): ... this.
11783 (vec_init<mode><ssehalfvecmodelower>): New expander.
11784 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11785 (vec_extractv2sfsf): ... this.
11786 (vec_initv2sf): Renamed to ...
11787 (vec_initv2sfsf): ... this.
11788 (vec_extractv2si): Renamed to ...
11789 (vec_extractv2sisi): ... this.
11790 (vec_initv2si): Renamed to ...
11791 (vec_initv2sisi): ... this.
11792 (vec_extractv4hi): Renamed to ...
11793 (vec_extractv4hihi): ... this.
11794 (vec_initv4hi): Renamed to ...
11795 (vec_initv4hihi): ... this.
11796 (vec_extractv8qi): Renamed to ...
11797 (vec_extractv8qiqi): ... this.
11798 (vec_initv8qi): Renamed to ...
11799 (vec_initv8qiqi): ... this.
11800 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11801 (vec_init<mode>): Renamed to ...
11802 (vec_init<mode><VEC_base_l>): ... this.
11803 (vec_extract<mode>): Renamed to ...
11804 (vec_extract<mode><VEC_base_l>): ... this.
11805 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11806 (vec_initv2sfsf): ... this.
11807 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11808 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11809 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11810 element mode after mode in gen_vec_init* calls.
11811 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11812 (vec_init<mode><Vel>): ... this.
11813 (vec_extract<mode>): Renamed to ...
11814 (vec_extract<mode><Vel>): ... this.
11815 * config/aarch64/iterators.md (Vel): New mode attribute.
11816 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11817 Add element mode after mode in gen_vec_extract* calls.
11818 * config/s390/vector.md (non_vec_l): New mode attribute.
11819 (vec_extract<mode>): Renamed to ...
11820 (vec_extract<mode><non_vec_l>): ... this.
11821 (vec_init<mode>): Renamed to ...
11822 (vec_init<mode><non_vec_l>): ... this.
11823 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11824 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11826 * config/arm/iterators.md (V_elem_l): New mode attribute.
11827 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11828 (vec_extract<mode><V_elem_l>): ... this.
11829 (vec_extractv2di): Renamed to ...
11830 (vec_extractv2didi): ... this.
11831 (vec_init<mode>): Renamed to ...
11832 (vec_init<mode><V_elem_l>): ... this.
11833 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11834 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11835 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11836 Add element mode after gen_vec_extract* calls.
11837 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11838 (vec_init<mode><unitmode>): ... this.
11839 (vec_extract<mode>): Renamed to ...
11840 (vec_extract<mode><unitmode>): ... this.
11841 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11842 (vec_init<mode><unitmode>): ... this.
11843 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11844 (vec_initv2sfsf): ... this.
11845 (vec_extractv2sf): Renamed to ...
11846 (vec_extractv2sfsf): ... this.
11847 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11848 Add element mode after gen_vec_extract* calls.
11849 * config/mips/mips.md (unitmode): New mode iterator.
11850 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11851 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11852 * config/spu/spu.md (inner_l): New mode attribute.
11853 (vec_init<mode>): Renamed to ...
11854 (vec_init<mode><inner_l>): ... this.
11855 (vec_extract<mode>): Renamed to ...
11856 (vec_extract<mode><inner_l>): ... this.
11857 * config/sparc/sparc.md (veltmode): New mode iterator.
11858 (vec_init<VMALL:mode>): Renamed to ...
11859 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11860 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11861 (vec_initv2sisi): ... this.
11862 (vec_initv2sf): Renamed to ...
11863 (vec_initv2sfsf): ... this.
11864 (vec_extractv2sf): Renamed to ...
11865 (vec_extractv2sfsf): ... this.
11866 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11867 (vec_init<mode>): Renamed to ...
11868 (vec_init<mode><VEC_base_l>): ... this.
11869 (vec_extract<mode>): Renamed to ...
11870 (vec_extract<mode><VEC_base_l>): ... this.
11871 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11872 (vec_initv2sfsf): ... this.
11873 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11874 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11875 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11876 gen_vec_init* calls.
11878 2017-08-01 Richard Biener <rguenther@suse.de>
11880 PR tree-optimization/81297
11881 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11882 TREE_OVERFLOW from INTEGER_CSTs.
11884 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11886 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11888 2017-07-31 Carl Love <cel@us.ibm.com>
11890 * config/rs6000/rs6000-c: Add support for built-in functions
11891 vector signed char vec_xl_be (signed long long, signed char *);
11892 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11893 vector signed int vec_xl_be (signed long long, signed int *);
11894 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11895 vector signed long long vec_xl_be (signed long long, signed long long *);
11896 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11897 vector signed short vec_xl_be (signed long long, signed short *);
11898 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11899 vector double vec_xl_be (signed long long, double *);
11900 vector float vec_xl_be (signed long long, float *);
11901 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11902 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11903 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11905 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11906 (altivec_expand_builtin): Add switch statement to call
11907 altivec_expand_xl_be for each builtin.
11908 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11909 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11910 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11911 __builtin_vsx_le_be_v16qi.
11912 * doc/extend.texi: Update the built-in documentation file for the
11913 new built-in functions.
11915 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11918 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11920 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11922 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11924 * config.gcc: Add z14.
11925 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11926 CPU model numbers for z13s and z14.
11927 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11929 * config/s390/s390-opts.h (enum processor_type): Rename
11930 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11931 * config/s390/s390.c (processor_table): Add field for CPU name to
11932 be passed to Binutils.
11933 (s390_asm_output_machine_for_arch): Use the new field in
11934 processor_table for Binutils.
11935 (s390_expand_builtin): Replace arch12 with z14.
11936 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11937 (s390_get_sched_attrmask): Likewise.
11938 (s390_get_unit_mask): Likewise.
11939 * config/s390/s390.opt: Add z14 to processor_type enum.
11941 2017-07-31 Martin Jambor <mjambor@suse.cz>
11944 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11945 regardless of optimization level.
11947 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11948 Martin Liska <mliska@suse.cz>
11950 * predict.def: Remove old comment and adjust probability.
11951 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11952 PREDICT statements.
11954 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11957 * config/i386/i386.c (ix86_function_naked): New function.
11958 (ix86_can_use_return_insn_p): Return false for naked functions.
11959 (ix86_expand_prologue): Skip prologue for naked functions.
11960 (ix86_expand_epilogue): Skip epilogue for naked functions
11961 and emit trap instruction.
11962 (ix86_warn_func_return): New function.
11963 (ix86_attribute_table): Add "naked" attribute specification.
11964 (TARGET_WARN_FUNC_RETURN): Define.
11965 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11967 2017-07-31 Martin Liska <mliska@suse.cz>
11969 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11970 (dump_gimple_bb_header): Always dump BB info.
11971 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11973 2017-07-31 Martin Liska <mliska@suse.cz>
11976 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11977 also with current_function_decl non-null equality.
11979 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11982 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11983 change type to the element type, instead add eltype variable and
11984 use it where we are interested in the element type.
11986 PR tree-optimization/81603
11987 * ipa-polymorphic-call.c
11988 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11989 offset arithmetic in offset_int, bail out if the resulting bit offset
11990 doesn't fit into shwi.
11992 2017-07-31 Martin Liska <mliska@suse.cz>
11994 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11995 (gimplify_save_expr): Fix comment.
11997 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12000 * config/i386/i386.c (ix86_function_arg): Update arguments for
12002 (ix86_compute_frame_layout): Set the initial stack offset to
12003 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12004 INCOMING_FRAME_SP_OFFSET.
12005 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12006 stack before exception handler returns.
12007 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12008 the 'ERROR_CODE' for exception handler.
12010 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12012 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12013 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12014 (ASM_OUTPUT_REG_POP): Ditto.
12015 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12016 instead of asm_fprintf to output pure string.
12018 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12020 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12021 to imported_module_or_decl hook.
12022 (debug_nothing_tree_tree_tree_bool): Remove.
12023 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12024 * debug.c (do_nothing_debug_hooks): Use
12025 debug_nothing_tree_tree_tree_bool_bool instead of
12026 debug_nothing_tree_tree_tree_bool.
12027 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12028 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12029 * sdbout.c (sdb_debug_hooks): Likewise.
12030 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12031 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12033 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12034 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12035 attribute, don't add anything.
12037 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12039 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12040 (fold_build2_stat_loc): Likewise.
12041 (fold_build3_stat_loc): Likewise.
12042 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12043 (fold_build1_loc): Remove macro.
12044 (fold_build2_loc): Likewise.
12045 (fold_build3_loc): Likewise.
12047 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12049 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12050 (gimple_build_debug_bind_source_stat): Likewise.
12051 * gimple.h (gimple_build_debug_bind): Remove macro.
12052 (gimple_build_debug_bind_source): Likewise.
12054 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12056 * bitmap.c (bitmap_alloc): Adjust.
12057 (bitmap_gc_alloc): Likewise.
12058 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12060 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12062 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12063 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12064 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12065 (bitmap_gc_alloc_stat): Likewise.
12066 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12068 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12070 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12071 * rtl.h (shallow_copy_rtx): Remove macro.
12073 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12075 * emit-rtl.c (gen_raw_REG): Adjust.
12076 * gengenrtl.c (gendef): Likewise.
12077 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12078 * rtl.h (rtx_alloc): Remove macro.
12080 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12082 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12083 (build_tree_list_stat): Likewise.
12084 * tree.h (build_tree_list): Remove macro.
12085 (build_tree_list_vec): Likewise.
12087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12089 * tree.c (make_vector_stat): Remove _stat from name.
12090 (build_vector_stat): Likewise.
12091 * tree.h (make_vector_stat): Remove macro.
12092 (build_vector_stat): Likewise.
12094 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12096 * tree.h (build_var_debug_value): Remove prototype.
12098 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12100 * tree.c (tree_cons_stat): Remove _stat from name.
12101 * tree.h (tree_cons): Remove macro.
12103 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12105 * tree.c (build_vl_exp_stat): Remove _stat from name.
12106 * tree.h (build_vl_exp): Remove macro.
12108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12110 * tree.c (build_decl_stat): Remove _stat from name.
12111 * tree.h (build_decl): Remove macro.
12113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12115 * gimple.c (gimple_build_with_ops_stat): Adjust.
12116 (gimple_alloc_stat): Remove _stat from name.
12117 * gimple.h (gimple_alloc): Remove macro.
12119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12121 * tree.c (make_tree_vec_stat): Remove _stat from name.
12122 (grow_tree_vec_stat): Likewise.
12123 * tree.h (make_tree_vec_stat): Adjust prototype.
12124 (grow_tree_vec_stat): Likewise.
12125 (make_tree_vec): Remove macro.
12126 (grow_tree_vec): Likewise.
12128 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12130 * fold-const.c (fold_build1_stat_loc): Adjust.
12131 (fold_build2_stat_loc): Likewise.
12132 (fold_build3_stat_loc): Likewise.
12133 * tree.c (build0_stat): Remove _stat from name.
12134 (build1_stat): Likewise.
12135 (build2_stat): Likewise.
12136 (build3_stat): Likewise.
12137 (build4_stat): Likewise.
12138 (build5_stat): Likewise.
12139 * tree.h (build1_loc): Remove macro, and rename _stat function
12141 (build2_loc): Likewise.
12142 (build3_loc): Likewise.
12143 (build4_loc): Likewise.
12144 (build5_loc): Likewise.
12146 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12148 * tree.c (make_int_cst_stat): Remove _stat from name.
12149 * tree.h (make_int_cst_stat): Adjust prototype.
12150 (make_int_cst): Remove macro.
12152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12154 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12155 * tree.h (make_tree_binfo_stat): Adjust prototype.
12156 (make_tree_binfo): Remove.
12158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12160 * tree.c (copy_node_stat): Rename to copy_node.
12161 (build_distinct_type_copy): Adjust.
12162 * tree.h (copy_node_stat): Adjust prototype.
12163 (copy_node): Remove macro.
12165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12167 * tree.c (make_node_stat): rename to make_node.
12168 (build_tree_list_stat): Adjust.
12169 (build0_stat): Likewise.
12170 (build2_stat): Likewise.
12171 (build3_stat): Likewise.
12172 (build4_stat): Likewise.
12173 (build5_stat): Likewise.
12174 (build_decl_stat): Likewise.
12175 * tree.h (make_node_stat): Adjust prototype.
12176 (make_node): remove macro.
12178 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12180 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12181 (PPC_FEATURE2_SCV): Likewise.
12182 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12184 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12186 * config/aarch64/aarch64.c
12187 (aarch64_internal_mov_immediate): Add new special pattern.
12188 * config/aarch64/aarch64.md (*movdi_aarch64):
12189 Add reg/32bit const mov case.
12191 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12192 Richard Sandiford <richard.sandiford@linaro.org>
12194 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12195 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12196 Add integer and movi cases.
12197 (movi-split-hf-df-sf split, fp16): New.
12198 (enabled): Added TARGET_FP_F16INST.
12199 * config/aarch64/iterators.md (GPF_HF): New.
12200 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12202 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12204 * config/aarch64/aarch64.c
12205 (aarch64_simd_container_mode): Add prototype.
12206 (aarch64_expand_mov_immediate): Add HI support.
12207 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12208 (aarch64_can_const_movi_rtx_p): New.
12209 (aarch64_preferred_reload_class):
12210 Remove restrictions of using FP registers for certain SIMD operations.
12211 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12212 (aarch64_valid_floating_const): Add integer move validation.
12213 (aarch64_simd_imm_scalar_p): Remove.
12214 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12215 (aarch64_legitimate_constant_p): Expand list of supported cases.
12216 * config/aarch64/aarch64-protos.h
12217 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12218 (aarch64_reinterpret_float_as_int): New.
12219 (aarch64_simd_imm_scalar_p): Remove.
12220 * config/aarch64/constraints.md (Uvi): New.
12221 (Dd): Split into Ds and new Dd.
12222 * config/aarch64/aarch64.md (*movsi_aarch64):
12224 (*movdi_aarch64): Add SIMD mov case.
12226 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12228 * tree-predcom.c: (struct chain): Handle store-store chain in which
12229 stores for elimination only store loop invariant values.
12230 (execute_pred_commoning_chain): Ditto.
12231 (prepare_initializers_chain_store_elim): Ditto.
12232 (prepare_finalizers): Ditto.
12233 (is_inv_store_elimination_chain): New function.
12234 (initialize_root_vars_store_elim_1): New function.
12236 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12238 * tree-predcom.c: Revise general description of the pass.
12239 (enum chain_type): New enum type for store elimination.
12240 (struct chain): New field supporting store elimination.
12241 (struct component): Ditto.
12242 (dump_chain): Dump store-stores chain.
12243 (release_chain): Release resources.
12244 (split_data_refs_to_components): Compute and create component
12245 contains only stores for elimination.
12246 (get_chain_last_ref_at): New function.
12247 (make_invariant_chain): Initialization.
12248 (make_rooted_chain): Specify chain type in parameter and record it.
12249 (add_looparound_copies): Skip for store-stores chain.
12250 (determine_roots_comp): Compute type of chain and pass it to
12252 (initialize_root_vars_store_elim_2): New function.
12253 (finalize_eliminated_stores): New function.
12254 (remove_stmt): Handle store for elimination.
12255 (execute_pred_commoning_chain): Execute predictive commoning on
12256 store-store chains.
12257 (determine_unroll_factor): Skip unroll for store-stores chain.
12258 (prepare_initializers_chain_store_elim): New function.
12259 (prepare_initializers_chain): Hanlde store-store chain.
12260 (prepare_finalizers_chain, prepare_finalizers): New function.
12261 (tree_predictive_commoning_loop): Return integer value indicating
12262 if loop is unrolled or lcssa form is corrupted.
12263 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12265 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12267 * tree-predcom.c (initialize_root): Delete.
12268 (execute_pred_commoning_chain): Initialize root vars and replace
12269 reference of non-combined chain directly, rather than call above
12272 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12274 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12275 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12277 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12279 * tree-predcom.c (struct chain): New field init_seq.
12280 (release_chain): Release init_seq.
12281 (prepare_initializers_chain): Record intialization stmts in above
12283 (insert_init_seqs): New function.
12284 (tree_predictive_commoning_loop): Call insert_init_seqs.
12286 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12288 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12290 2017-07-28 Richard Biener <rguenther@suse.de>
12292 * match.pd: Remove superfluous :c.
12293 * genmatch.c (simplify::id): Add member.
12294 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12296 (current_id): New global.
12297 (dt_node::parent): Move from ...
12298 (dt_operand::parent): ... here. Add for_id member.
12299 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12300 (decision_tree::find_node): Relax order requirement when
12301 merging DT_TRUE nodes to ones inbetween the current simplify
12302 and the one we try to merge with. Add diagnostic whenever
12303 we need to enforce pattern order by not merging.
12304 (decision_tree::insert): Set current_id.
12305 (decision_tree::print_node): Dump parent node and for_id.
12306 (parser::last_id): Add member.
12307 (parser::push_simplify): Assign unique id.
12308 (parser::parser): Initialize last_id.
12310 2017-07-28 Martin Liska <mliska@suse.cz>
12313 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12314 gimple_build_debug_bind.
12316 2017-07-28 Richard Biener <rguenther@suse.de>
12318 PR tree-optimization/81502
12319 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12321 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12323 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12324 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12325 for BIT_FIELD_REF args.
12326 * fold-const.c (make_bit_field_ref): Likewise.
12327 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12329 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12332 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12333 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12334 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12335 Or it into SANITIZER_UNDEFINED.
12336 * ubsan.c: Include gimple-fold.h and varasm.h.
12337 (ubsan_expand_ptr_ifn): New function.
12338 (instrument_pointer_overflow): New function.
12339 (maybe_instrument_pointer_overflow): New function.
12340 (instrument_object_size): Formatting fix.
12341 (pass_ubsan::execute): Call instrument_pointer_overflow
12342 and maybe_instrument_pointer_overflow.
12343 * internal-fn.c (expand_UBSAN_PTR): New function.
12344 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12345 * sanitizer.def (__ubsan_handle_pointer_overflow,
12346 __ubsan_handle_pointer_overflow_abort): New builtins.
12347 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12348 * internal-fn.def (UBSAN_PTR): New internal function.
12349 * opts.c (sanitizer_opts): Add pointer-overflow.
12350 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12351 * fold-const.c (build_range_check): Compute pointer range check in
12352 integral type if pointer arithmetics would be needed. Formatting
12355 2017-07-28 Martin Liska <mliska@suse.cz>
12358 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12359 parameters that are of a variable-length.
12361 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12363 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12365 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12366 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12367 (CRT_CALL_STATIC_FUNCTION): Likewise.
12368 (ASM_DEFAULT_SPEC): New define.
12369 (ASM_SPEC32): Likewise.
12370 (ASM_SPEC64): Likewise.
12371 (ASM_SPEC_COMMON): Likewise.
12372 (ASM_SPEC): Likewise.
12373 (INVALID_64BIT): Likewise.
12374 (LINK_OS_DEFAULT_SPEC): Likewise.
12375 (LINK_OS_SPEC32): Likewise.
12376 (LINK_OS_SPEC64): Likewise.
12377 (POWERPC_LINUX): Likewise.
12378 (PTRDIFF_TYPE): Likewise.
12379 (RESTORE_FP_PREFIX): Likewise.
12380 (RESTORE_FP_SUFFIX): Likewise.
12381 (SAVE_FP_PREFIX): Likewise.
12382 (SAVE_FP_SUFFIX): Likewise.
12383 (SIZE_TYPE): Likewise.
12384 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12385 (TARGET_64BIT): Likewise.
12386 (TARGET_64BIT): Likewise.
12387 (TARGET_AIX): Likewise.
12388 (WCHAR_TYPE_SIZE): Likewise.
12389 (WCHAR_TYPE): Undefine.
12390 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12391 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12392 (CPP_OS_RTEMS_SPEC): Delete.
12393 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12394 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12396 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12398 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12400 PR tree-optimization/81578
12401 * tree-parloops.c (build_new_reduction): Bail out if
12402 reduction_code isn't one of the standard OpenMP reductions.
12403 Move the details printing after that decision.
12405 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12407 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12408 related to reload_in_progress.
12409 (splat_input_operand): Likewise.
12410 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12412 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12414 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12415 (TARGET_INSTANTIATE_DECLS): Likewise.
12416 (legitimate_indexed_address_p): Delete reload_in_progress code.
12417 (rs6000_debug_legitimate_address_p): Likewise.
12418 (rs6000_eliminate_indexed_memrefs): Likewise.
12419 (rs6000_emit_le_vsx_store): Likewise.
12420 (rs6000_emit_move_si_sf_subreg): Likewise.
12421 (rs6000_emit_move): Likewise.
12422 (register_to_reg_type): Likewise.
12423 (rs6000_pre_atomic_barrier): Likewise.
12424 (rs6000_machopic_legitimize_pic_address): Likewise.
12425 (rs6000_allocate_stack_temp): Likewise.
12426 (rs6000_address_for_fpconvert): Likewise.
12427 (rs6000_address_for_altivec): Likewise.
12428 (rs6000_secondary_memory_needed_rtx): Delete function.
12429 (rs6000_check_sdmode): Likewise.
12430 (rs6000_alloc_sdmode_stack_slot): Likewise.
12431 (rs6000_instantiate_decls): Likewise.
12432 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12433 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12434 Delete reload_in_progress.
12435 (*vec_reload_and_plus_<mptrsize>): Likewise.
12436 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12437 (vsx_div_v2di): Likewise.
12438 (vsx_udiv_v2di): Likewise.
12440 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12442 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12443 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12444 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12445 (rs6000_debug_reg_global): Delete print of LRA status.
12446 (rs6000_option_override_internal): Delete dead LRA related code.
12447 (rs6000_lra_p): Delete function.
12448 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12450 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12452 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12453 * config/riscv/rtems.h: New file.
12455 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12456 Sudakshina Das <sudi.das@arm.com>
12458 * config/aarch64/aarch64.md
12459 (define_split for and<mode>3nr_compare): Move
12460 non aarch64_logical_operand to a register.
12461 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12462 register immediate operand to a register.
12463 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12465 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12467 PR middle-end/81564
12468 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12470 2017-07-27 Richard Biener <rguenther@suse.de>
12472 PR tree-optimization/81573
12473 PR tree-optimization/81494
12474 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12475 multi defuse cycle case.
12477 2017-07-27 Richard Biener <rguenther@suse.de>
12479 PR tree-optimization/81571
12480 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12483 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12485 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12486 earlier and only if MASK_FPU is set. Adjust formatting.
12488 2017-07-27 Martin Liska <mliska@suse.cz>
12490 * opt-functions.awk: Add validation of value of Init.
12491 * optc-gen.awk: Pass new argument.
12493 2017-07-27 Martin Liska <mliska@suse.cz>
12495 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12496 Fix wrong condition.
12498 2017-07-27 Martin Liska <mliska@suse.cz>
12500 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12501 BBs and edges seen by autoFDO.
12503 2017-07-27 Richard Biener <rguenther@suse.de>
12505 PR tree-optimization/81502
12506 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12507 with incompatible but same sized type.
12508 (execute_update_addresses_taken): Likewise.
12510 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12512 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12513 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12515 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12518 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12519 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12520 Change s_operand to memory_operand.
12522 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12524 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12525 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12526 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12527 Emit instructions rather than returning an expression. Handle TFmode
12528 and KFmode by casting to TImode.
12529 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12530 (rs6000_emit_le_vsx_store): Likewise.
12531 * config/rs6000/vsx.md (VSX_TI): New iterator.
12532 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12533 (*vsx_le_undo_permute_<mode>): Likewise.
12534 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12535 emit the split sequence.
12536 (*vsx_le_perm_store_<mode>): Likewise.
12538 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12540 PR tree-optimization/81555
12541 PR tree-optimization/81556
12542 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12543 if true, force CHANGED for the recursive invocation.
12544 (reassociate_bb): Remember original length of ops array, pass
12545 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12547 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12548 attributes for noipa attribute. For naked attribute use
12549 lookup_attribute first before lookup_attribute_spec.
12550 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12552 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12553 for functions with noipa attribute.
12554 (cgraph_externally_visible_p): Return true for functions with noipa
12556 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12557 for functions with noipa attribute.
12558 * doc/extend.texi: Document noipa function attribute.
12559 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12560 also for functions with noipa attribute.
12561 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12563 2017-07-26 Andrew Pinski <apinski@cavium.com>
12565 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12566 vec_unalign_load_cost and vec_unalign_store_cost.
12568 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12570 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12571 -mvsx-small-integer option.
12572 (ISA_3_0_MASKS_IEEE): Likewise.
12573 (OTHER_VSX_VECTOR_MASKS): Likewise.
12574 (POWERPC_MASKS): Likewise.
12575 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12576 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12577 code, only testing for DImode being allowed in non-VSX floating
12579 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12580 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12582 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12583 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12584 TARGET_P8_VECTOR test.
12585 (rs6000_secondary_reload_simple_move): Likewise.
12586 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12587 since TARGET_P9_VECTOR was already tested.
12588 (rs6000_opt_masks): Remove -mvsx-small-integer.
12589 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12590 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12592 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12593 test for TARGET_VEXTRACTUB was used, and that uses
12595 (p9 extract splitter): Likewise.
12596 (vsx_extract_<mode>_di_p9): Likewise.
12597 (vsx_extract_<mode>_store_p9): Likewise.
12598 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12599 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12600 the elimination of TARGET_VSX_SMALL_INTEGER.
12601 (vsx_extract_<mode>_p8): Likewise.
12602 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12603 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12604 (vsx_set_<mode>_p9): Likewise.
12605 (vsx_set_v4sf_p9): Likewise.
12606 (vsx_set_v4sf_p9_zero): Likewise.
12607 (vsx_insert_extract_v4sf_p9): Likewise.
12608 (vsx_insert_extract_v4sf_p9_2): Likewise.
12609 * config/rs6000/rs6000.md (sign extend splitter): Change
12610 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12611 (floatsi<mode>2_lfiwax_mem): Likewise.
12612 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12613 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12614 since a test for TARGET_P9_VECTOR was used.
12615 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12616 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12617 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12618 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12619 TARGET_P8_VECTOR test.
12620 (fix_trunc<mode>si2_stfiwx): Likewise.
12621 (fix_trunc<mode>si2_internal): Likewise.
12622 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12623 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12625 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12626 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12627 TARGET_P8_VECTOR test.
12628 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12629 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12630 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12632 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12633 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12634 since a test for TARGET_P9_VECTOR was used.
12635 (splitter for loading small constants): Likewise.
12637 2017-07-26 Andrew Pinski <apinski@cavium.com>
12639 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12642 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12645 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12646 (fp_valid_at): Likewise.
12648 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12650 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12651 (qdf24xx_addrcost_table): Likewise.
12652 (cortexa57_tunings): Update to use generic_branch_cost.
12653 (cortexa72_tunings): Likewise.
12654 (cortexa73_tunings): Likewise.
12655 (qdf24xx_tunings): Likewise.
12657 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12659 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12660 (thunderx2t99_branch_cost): Likewise.
12661 (cortexa35_tunings): Update to use generic_branch_cost.
12662 (cortexa53_tunings): Likewise.
12663 (cortexa57_tunings): Likewise.
12664 (cortexa72_tunings): Likewise.
12665 (cortexa73_tunings): Likewise.
12666 (thunderx2t99_tunings): Likewise.
12668 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12670 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12671 (sparc_option_override): Honour MASK_FSMULD.
12672 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12673 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12674 * config/sparc/sparc.opt (mfsmuld): New option.
12675 * doc/invoke.texi (mfsmuld): Document option.
12677 2017-07-26 Marek Polacek <polacek@redhat.com>
12679 PR middle-end/70992
12680 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12682 2017-07-26 Richard Biener <rguenther@suse.de>
12684 * gimple-match-head.c (do_valueize): Return OP if valueize
12686 (get_def): New helper to get at the def stmt of a SSA name
12687 if valueize allows.
12688 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12689 do_valueize to get at the def stmt.
12690 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12692 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12694 PR middle-end/46932
12695 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12697 2017-07-26 Martin Liska <mliska@suse.cz>
12700 * function.c (expand_function_start): Make expansion of
12701 nonlocal_goto_save_area after parm_birth_insn.
12703 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12705 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12706 from all CPU target flags enable members.
12708 2017-07-26 Richard Biener <rguenther@suse.de>
12710 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12711 (decision_tree::gen): Make 'type' const.
12712 (write_predicate): Likewise.
12714 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12716 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12717 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12718 (rs6000_option_override_internal): Likewise.
12719 (rs6000_expand_vector_set): Likewise.
12720 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12721 (TARGET_UPPER_REGS_SF): Likewise.
12722 (TARGET_UPPER_REGS_DI): Likewise.
12723 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12724 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12725 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12726 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12727 (Splitters for DI constants in Altivec registers): Likewise.
12728 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12729 (vsx_set_v4sf_p9): Likewise.
12730 (vsx_set_v4sf_p9_zero): Likewise.
12731 (vsx_insert_extract_v4sf_p9): Likewise.
12732 (vsx_insert_extract_v4sf_p9_2): Likewise.
12734 2017-07-25 Carl Love <cel@us.ibm.com>
12736 * doc/extend.texi: Update the built-in documentation file for the
12737 existing built-in functions
12738 vector signed char vec_cnttz (vector signed char);
12739 vector unsigned char vec_cnttz (vector unsigned char);
12740 vector signed short vec_cnttz (vector signed short);
12741 vector unsigned short vec_cnttz (vector unsigned short);
12742 vector signed int vec_cnttz (vector signed int);
12743 vector unsigned int vec_cnttz (vector unsigned int);
12744 vector signed long long vec_cnttz (vector signed long long);
12745 vector unsigned long long vec_cnttz (vector unsigned long long);
12747 2017-07-25 Andrew Pinski <apinski@cavium.com>
12749 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12750 accesses where the use is for the first operand of a BIT_INSERT.
12752 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12755 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12756 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12758 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12760 * config/i386/gstabs.h: Delete.
12761 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12763 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12765 * config/i386/i386.c (ix86_decompose_address): Do not check for
12766 register RTX when looking at index_reg or base_reg.
12767 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12769 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12771 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12772 to update EH info here.
12774 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12776 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12778 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12780 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12782 2017-07-25 Torsten Duwe <duwe@suse.de>
12784 * common.opt: Introduce -fpatchable-function-entry
12785 command line option, and its variables function_entry_patch_area_size
12786 and function_entry_patch_area_start.
12787 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12788 including a two-value parser.
12789 * target.def (print_patchable_function_entry): New target hook.
12790 * targhooks.h (default_print_patchable_function_entry): New function.
12791 * targhooks.c (default_print_patchable_function_entry): Likewise.
12792 * toplev.c (process_options): Switch off IPA-RA if
12793 patchable function entries are being generated.
12794 * varasm.c (assemble_start_function): Look at the
12795 patchable-function-entry command line switch and current
12796 function attributes and maybe generate NOP instructions by
12797 calling the print_patchable_function_entry hook.
12798 * doc/extend.texi: Document patchable_function_entry attribute.
12799 * doc/invoke.texi: Document -fpatchable_function_entry
12800 command line option.
12801 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12803 * doc/tm.texi: Re-generate.
12805 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12808 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12809 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12811 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12813 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12815 2017-07-25 Richard Biener <rguenther@suse.de>
12817 PR tree-optimization/81455
12818 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12819 not walk in cycles when looking for guards.
12821 2017-07-25 Richard Biener <rguenther@suse.de>
12823 PR tree-optimization/81529
12824 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12825 when optimizing backedge uses.
12827 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12829 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12831 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12832 to dl_section_ref. On AIX, append an expression to subtract
12833 the size of the section length to dl_section_ref.
12835 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12837 * configure.ac: If any of the config.* scripts fail, exit 1.
12838 * configure: Regenerate.
12840 2017-07-25 Richard Biener <rguenther@suse.de>
12842 PR middle-end/81546
12843 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12844 of immediate uses, be more verbose on errors.
12846 2017-07-25 Richard Biener <rguenther@suse.de>
12848 PR tree-optimization/81510
12849 * tree-vect-loop.c (vect_is_simple_reduction): When the
12850 reduction stmt is not inside the loop bail out.
12852 2017-07-25 Richard Biener <rguenther@suse.de>
12854 PR tree-optimization/81303
12855 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12856 profitability check against LOOP_VINFO_NITERSM1.
12858 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12860 * domwalk.c (cmp_bb_postorder): Simplify.
12861 (sort_bbs_postorder): New function. Use it...
12862 (dom_walker::walk): ...here to optimize common cases.
12864 2017-07-25 Martin Liska <mliska@suse.cz>
12867 * ipa-visibility.c (function_and_variable_visibility): Make the
12868 redirection just on target that supports aliasing.
12869 Fix GNU coding style.
12871 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12874 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12876 * config/arm/rtems.h: Likewise.
12877 * config/bfin/rtems.h: Likewise.
12878 * config/i386/rtemself.h: Likewise.
12879 * config/lm32/rtems.h: Likewise.
12880 * config/m32c/rtems.h: Likewise.
12881 * config/m68k/rtemself.h: Likewise.
12882 * config/microblaze/rtems.h: Likewise.
12883 * config/mips/rtems.h: Likewise.
12884 * config/moxie/rtems.h: Likewise.
12885 * config/nios2/rtems.h: Likewise.
12886 * config/powerpcspe/rtems.h: Likewise.
12887 * config/rs6000/rtems.h: Likewise.
12888 * config/rtems.h: Likewise.
12889 * config/sh/rtems.h: Likewise.
12890 * config/sh/rtemself.h: Likewise.
12891 * config/sparc/rtemself.h: Likewise.
12893 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12896 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12897 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12898 * tree-ssa-structalias.c (alias_get_name): Same.
12900 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12903 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12904 instructions if no du chain is found.
12906 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12908 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12910 2017-07-25 Richard Biener <rguenther@suse.de>
12912 PR middle-end/81505
12913 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12916 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12918 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12919 upper-regs options.
12920 (ISA_2_7_MASKS_SERVER): Likewise.
12921 (ISA_3_0_MASKS_IEEE): Likewise.
12922 (OTHER_P8_VECTOR_MASKS): Likewise.
12923 (OTHER_VSX_VECTOR_MASKS): Likewise.
12924 (POWERPC_MASKS): Likewise.
12925 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12926 duplicate list of options.
12927 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12928 explicit -mupper-regs options.
12929 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12930 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12931 alias for -mupper-regs-df.
12932 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12933 (rs6000_init_hard_regno_mode_ok): Likewise.
12934 (rs6000_option_override_internal): Likewise.
12935 (rs6000_opt_masks): Likewise.
12936 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12937 options in terms of whether -mvsx or -mpower8-vector was used.
12938 (TARGET_UPPER_REGS_DI): Likewise.
12939 (TARGET_UPPER_REGS_SF): Likewise.
12940 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12941 -mupper-regs-* options.
12943 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12945 * passes.c (emergency_dump_function): Print some empty lines and a
12946 header before the RTL dump.
12948 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12950 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12952 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12955 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12956 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12958 2017-07-24 Carl Love <cel@us.ibm.com>
12960 * config/rs6000/rs6000-c.c: Add support for built-in functions
12961 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12962 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12963 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12964 vec_extract_fp_from_shortl): Add defines for the two builtins.
12965 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12966 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12968 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12969 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12970 * doc/extend.texi: Update the built-in documentation file for the
12971 new built-in function.
12973 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12976 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12978 * doc/generic.texi: Likewise.
12979 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12980 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12982 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12984 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12985 (aarch64_mls_elt_merge<mode>): Likewise.
12987 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12989 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12990 having __cxa_atexit.
12992 2017-07-23 Michael Collison <michael.collison@arm.com>
12994 * config/arm/arm.c (arm_option_override): Deprecate
12995 use of -mstructure-size-boundary.
12996 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12997 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12999 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13002 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13003 Reduce cost estimate for direct moves.
13005 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13008 * config/i386/i386.c (ix86_option_override_internal): Disable
13009 BMI, BMI2 and TBM instructions for -m16.
13011 2017-07-21 Carl Love <cel@us.ibm.com>
13013 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13014 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13015 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13016 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13017 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13018 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13019 VMULOSW): New enum "unspec" values.
13020 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13021 altivec_vmulosw): New patterns.
13022 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13023 VMULOSW): Add definitions.
13025 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13027 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13028 (qdf24xx): Likewise.
13029 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13030 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13031 (AARCH64_FL_V8_1): Renumber.
13032 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13033 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13034 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13035 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13036 rdma to feature modifiers list.
13038 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13040 PR middle-end/56727
13041 * ipa-visibility (function_and_variable_visibility): Convert
13042 recursive PLT call to direct call if appropriate.
13044 2017-07-21 Andrew Pinski <apinski@cavium.com>
13046 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13047 operand 1 to see if the types precision matches.
13048 * fold-const.c (operand_equal_p): Likewise.
13050 2017-07-21 Richard Biener <rguenther@suse.de>
13052 PR tree-optimization/81303
13053 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13054 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13055 (vect_peeling_hash_get_lowest_cost): Adjust.
13056 (vect_enhance_data_refs_alignment): Likewise. Use
13057 vect_get_peeling_costs_all_drs to compute the penalty for no
13058 peeling to match up costs.
13060 2017-07-21 Richard Biener <rguenther@suse.de>
13062 PR tree-optimization/81500
13063 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13064 we didn't identify a reduction path.
13066 2017-07-21 Tom de Vries <tom@codesourcery.com>
13067 Cesar Philippidis <cesar@codesourcery.com>
13069 PR gcov-profile/81442
13070 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13073 2017-07-21 Tom de Vries <tom@codesourcery.com>
13076 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13078 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13079 nvptx_override_options_after_change.
13081 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13083 * dwarf2out.c (output_file_names): Avoid double testing for
13084 dwarf_version >= 5.
13086 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13088 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13090 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13092 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13094 (try_crossjump_to_edge): Do not punt on partitioned functions.
13096 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13098 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13099 Put all BBs reachable only via paths crossing cold region to cold
13101 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13103 2016-07-21 Richard Biener <rguenther@suse.de>
13105 PR tree-optimization/81303
13106 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13107 into account prologue and epilogue iterations when raising
13108 min_profitable_iters to sth at least covering one vector iteration.
13110 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13112 * config/arm/arm.c (arm_test_cpu_arch_dat):
13115 2017-07-20 Nathan Sidwell <nathan@acm.org>
13117 Remove TYPE_METHODS.
13118 * tree.h (TYPE_METHODS): Delete.
13119 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13120 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13121 (dbxout_type_methods): Scan TYPE_FIELDS.
13122 (dbxout_type): Don't check TYPE_METHODS here.
13123 * function.c (use_register_for_decl): Always ignore register for
13124 class types when not optimizing.
13125 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13126 * tree.c (free_lang_data_in_type): Stitch out member functions and
13127 templates from TYPE_FIELDS.
13128 (build_distinct_type_copy, verify_type_variant,
13129 verify_type): Member fns are on TYPE_FIELDS.
13130 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13131 * tree-pretty-print.c (dump_generic_node): Likewise.
13133 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13136 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13137 V2TImode and V4TImode.
13138 (ix86_expand_vector_extract): Likewise.
13139 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13140 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13141 (ssescalarmode): Handle V4TImode and V2TImode.
13142 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13143 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13144 (VEXTRACTI128_MODE): New mode iterator.
13145 (splitter for *vec_extractv?ti first element): New.
13146 (VEC_INIT_MODE): New mode iterator.
13147 (vec_init<mode>): Consolidate 3 expanders into one using
13148 VEC_INIT_MODE mode iterator.
13150 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13152 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13153 non_spilled_static_chain_regno_p.
13155 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13157 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13159 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13161 * bb-reorder.c (connect_traces): Allow copying of blocks within
13164 2017-07-20 Richard Biener <rguenther@suse.de>
13166 * gimple.h (gimple_phi_result): Add gphi * overload.
13167 (gimple_phi_result_ptr): Likewise.
13168 (gimple_phi_arg): Likewise. Adjust index assert to only
13169 allow actual argument accesses rather than all slots available
13171 (gimple_phi_arg_def): Add gphi * overload.
13172 * tree-phinodes.c (make_phi_node): Initialize only actual
13174 (resize_phi_node): Clear memory not covered by old node,
13175 do not initialize excess argument slots.
13176 (reserve_phi_args_for_new_edge): Initialize new argument slot
13179 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13181 PR tree-optimization/81388
13183 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13185 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13186 by removing computation of may_be_zero.
13188 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13189 Tom de Vries <tom@codesourcery.com>
13191 PR middle-end/81030
13192 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13193 when gimple level profile disagrees with what RTL expander did.
13195 2017-07-20 Richard Biener <rguenther@suse.de>
13197 PR tree-optimization/61171
13198 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13199 (vect_analyze_stmt): Add slp instance parameter.
13200 (vectorizable_reduction): Likewise.
13201 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13202 (vect_is_simple_reduction): Deal with chains not detected
13203 as SLP reduction chain, specifically not properly associated
13204 chains containing a mix of plus/minus.
13205 (get_reduction_op): Remove.
13206 (get_initial_defs_for_reduction): Simplify, pass in whether
13207 this is a reduction chain, pass in the SLP node for the PHIs.
13208 (vect_create_epilog_for_reduction): Get the SLP instance as
13210 (vectorizable_reduction): Get the SLP instance as arg.
13211 During analysis remember the SLP node with the PHIs in the
13212 instance. Simplify getting at the vectorized reduction PHIs.
13213 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13214 through SLP instance.
13215 (vect_slp_analyze_operations): Likewise.
13216 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13217 (vect_transform_stmt): Likewise.
13219 2017-07-20 Tom de Vries <tom@codesourcery.com>
13221 PR tree-optimization/81489
13222 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13223 read of phi arg location to before loop that modifies phi.
13225 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13227 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13230 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13232 PR middle-end/81331
13233 * except.c (execute): Fix ordering issue.
13235 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13237 PR rtl-optimization/81423
13238 * combine.c (make_compound_operation_int): Don't try to optimize
13239 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13241 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13243 PR rtl-optimization/81423
13244 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13245 with a constant that is -1 in the truncated to mode.
13247 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13249 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13250 (determine_unlikely_bbs): ... here.
13251 * predict.h (propagate_unlikely_bbs_forward): Declare.
13252 * cfgexpand.c (pass_expand::execute): Use it.
13253 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13255 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13256 propagate_unlikely_bbs_forward.
13258 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13260 PR middle-end/81331
13261 * except.c (maybe_add_nop_after_section_switch): New function.
13264 2017-07-19 Tom de Vries <tom@codesourcery.com>
13266 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13268 2017-07-19 Tom de Vries <tom@codesourcery.com>
13270 * gimple.h (gimple_phi_arg): Make assert more strict.
13272 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13274 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13275 * config/rs6000/mmintrin.h: New file.
13276 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13278 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13280 PR tree-optimization/81346
13281 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13283 2017-07-19 Tom de Vries <tom@codesourcery.com>
13285 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13287 2017-07-19 Tom de Vries <tom@codesourcery.com>
13289 * config/nvptx/nvptx-modes.def: Add V2DImode.
13290 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13291 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13292 (nvptx_output_mov_insn): Handle lack of mov.b128.
13293 (nvptx_print_operand): Handle 'H' and 'L' codes.
13294 (nvptx_vector_mode_supported): Allow V2DImode.
13295 (nvptx_preferred_simd_mode): New function.
13296 (nvptx_data_alignment): New function.
13297 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13298 nvptx_preferred_simd_mode.
13299 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13301 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13303 2017-07-19 Tom de Vries <tom@codesourcery.com>
13305 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13306 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13307 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13308 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13309 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13310 (mov<VECIM>_insn): New define_insn.
13311 (define_expand "mov<VECIM>): New define_expand.
13313 2017-07-19 Tom de Vries <tom@codesourcery.com>
13315 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13317 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13319 PR tree-optimization/81346
13320 * fold-const.h (fold_div_compare, range_check_type): Declare.
13321 * fold-const.c (range_check_type): New function.
13322 (build_range_check): Use range_check_type.
13323 (fold_div_compare): No longer static, rewritten into
13324 a match.pd helper function.
13325 (fold_comparison): Don't call fold_div_compare here.
13326 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13327 as helper function.
13329 2017-07-19 Nathan Sidwell <nathan@acm.org>
13331 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13332 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13333 * tree.c (find_decls_types_r, verify_type): Use
13334 TYPE_{MIN,MAX}_VALUE_RAW.
13335 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13336 (hash_tree): Likewise.
13337 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13339 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13342 2017-07-18 Tom de Vries <tom@codesourcery.com>
13344 PR middle-end/81464
13345 * omp-expand.c (expand_omp_for_static_chunk): Handle
13346 equal-argument loop exit phi.
13348 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13351 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13352 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13353 operand 2 predicate.
13354 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13355 operand 2 predicate.
13356 (ror,rol -> rorx splitters): Use const_int_operand as
13357 operand 2 predicate.
13359 2017-06-18 Richard Biener <rguenther@suse.de>
13361 PR tree-optimization/81410
13362 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13363 the gap in the ! slp_perm SLP case after each group.
13365 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13367 PR middle-end/81463
13368 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13371 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13373 PR middle-end/81462
13374 * predict.c (set_even_probabilities): Cleanup; do not affect
13375 probabilities that are already known.
13376 (combine_predictions_for_bb): Call even when count is set.
13378 2017-07-18 Nathan Sidwell <nathan@acm.org>
13380 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13383 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13386 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13387 optimization for loop niter analysis.
13389 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13392 * config/avr/avr.c (avr_optimize_casesi): Don't use
13393 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13395 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13397 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13398 body_cost_vec from _vect_peel_extended_info.
13399 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13400 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13403 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13405 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13407 2017-07-18 Richard Biener <rguenther@suse.de>
13409 PR tree-optimization/80620
13410 PR tree-optimization/81403
13411 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13412 info when re-using a VN table entry.
13414 2017-07-18 Richard Biener <rguenther@suse.de>
13416 PR tree-optimization/81418
13417 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13418 vectype_in. Verify that with lane-reducing reduction operations
13419 we have a single def-use cycle.
13421 2017-07-17 Carl Love <cel@us.ibm.com>
13423 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13426 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13427 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13429 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13430 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13431 VMULOSW): New enum "unspec" values.
13432 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13433 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13434 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13435 altivec_vmulosw): New patterns.
13436 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13437 VMULOSW): Add definitions.
13438 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13440 * config/alpha/alpha.c: Include predict.h.
13442 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13444 * tree-vrp.c (compare_assert_loc): Fix comparison function
13445 to return predictable results.
13447 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13449 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13451 (subdi3): Likewise.
13452 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13453 * doc/invoke.texi (mexpand-adddi): Update text.
13455 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13457 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13458 that also clobbers the CC register. The old expand code is moved
13460 (*arc_clzsi2): ... here.
13461 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13462 the CC register. The old expand code is moved to ...
13463 (arc_ctzsi2): ... here.
13465 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13467 * config/arc/arc.opt (mindexed-loads): Use initial value
13468 TARGET_INDEXED_LOADS_DEFAULT.
13469 (mauto-modify-reg): Use initial value
13470 TARGET_AUTO_MODIFY_REG_DEFAULT.
13471 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13472 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13473 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13474 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13476 2017-07-17 Martin Liska <mliska@suse.cz>
13479 * opts.c (finish_options): Do not allow -fgnu-tm
13480 w/ -fsanitize={kernel-,}address. Say sorry.
13482 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13485 * tree-loop-distribution.c (classify_partition): Only assert on
13486 numer of iterations.
13487 (merge_dep_scc_partitions): Delete prameter. Update function call.
13488 (distribute_loop): Remove code handling loop with unknown niters.
13489 (pass_loop_distribution::execute): Skip loop with unknown niters.
13491 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13494 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13495 function sort_partitions_by_post_order.
13497 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13499 PR tree-optimization/81374
13500 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13501 the max index of basic blocks, rather than number of basic blocks.
13503 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13505 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13507 (arc_legitimate_pic_operand_p): Likewise.
13508 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13510 (arc_needs_pcl_p): Likewise.
13511 (arc_legitimate_pc_offset_p): Likewise.
13512 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13513 function is also used in constrains.md.
13514 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13515 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13516 PLUS. Only return true/false in known cases, otherwise assert.
13517 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13518 is already called in arc_legitimate_constant_p.
13519 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13521 (LEGITIMATE_PIC_OPERAND_P): Use
13522 arc_raw_symbolic_reference_mentioned_p function.
13523 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13528 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13529 Andrew Burgess <andrew.burgess@embecosm.com>
13531 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13532 (arc_return_address_register): New function.
13533 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13534 (arc_handle_fndecl_attribute): Add naked attribute.
13535 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13536 (TARGET_WARN_FUNC_RETURN): Likewise.
13537 (arc_allocate_stack_slots_for_args): New function.
13538 (arc_warn_func_return): Likewise.
13539 (machine_function): Change type fn_type.
13540 (arc_compute_function_type): Consider new naked function type,
13541 change function return type.
13542 (arc_must_save_register): Adapt to handle new
13543 arc_compute_function_type's return type.
13544 (arc_expand_prologue): Likewise.
13545 (arc_expand_epilogue): Likewise.
13546 (arc_return_address_regs): Delete.
13547 (arc_return_address_register): New function.
13548 (arc_epilogue_uses): Use above function.
13549 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13550 (arc_function_type): Change encoding, add naked type.
13551 (ARC_INTERRUPT_P): Change to handle the new encoding.
13552 (ARC_FAST_INTERRUPT_P): Likewise.
13553 (ARC_NORMAL_P): Define.
13554 (ARC_NAKED_P): Likewise.
13555 (arc_compute_function_type): Delete prototype.
13556 * config/arc/arc.md (in_ret_delay_slot): Use
13557 arc_return_address_register function.
13558 (simple_return): Likewise.
13559 (p_return_i): Likewise.
13561 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13563 PR tree-optimization/81428
13564 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13565 can't be built for those types.
13567 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13569 Remove stuff dead since r239246.
13571 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13572 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13573 (avr_inform_devices): Remove dead stuff.
13575 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13577 * config/arm/arm_neon.h: Fix softp typo.
13579 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13581 PR tree-optimization/81365
13582 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13583 aggregate moves onto bb predecessor edges, make sure there are no
13584 loads that could alias the lhs in between the start of bb and the
13587 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13590 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13591 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13592 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13594 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13596 PR tree-optimization/81396
13597 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13598 (init_symbolic_number): Initialize it to 1.
13599 (perform_symbolic_merge): Add n_ops from both operands into the new
13601 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13602 without base_addr as useless if they need more than one operation.
13603 (bswap_replace): Handle !bswap case for NULL base_addr.
13605 2017-07-17 Tom de Vries <tom@codesourcery.com>
13608 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13611 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13613 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13614 conditional builtin define __FIX_LEON3FT_B2BST.
13616 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13618 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13619 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13622 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13624 PR rtl-optimization/81424
13625 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13626 to remove potential trapping from operands if -fnon-call-exceptions.
13628 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13631 profile_proability for scalling.
13632 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13634 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13636 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13638 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13640 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13641 fixpoint arithmetics.
13643 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13645 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13646 fixpoint arithmetics.
13648 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13650 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13651 fixpoint arithmetics.
13653 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13655 * profile-count.h (profile_probability::from_reg_br_prob_note,
13656 profile_probability::to_reg_br_prob_note): New functions.
13657 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13658 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13659 * predict.c (probability_reliable_p): Update.
13660 (edge_probability_reliable_p): Update.
13661 (br_prob_note_reliable_p): Update.
13662 (invert_br_probabilities): Update.
13663 (add_reg_br_prob_note): New function.
13664 (combine_predictions_for_insn): Update.
13665 * asan.c (asan_clear_shadow): Update.
13666 * cfgbuild.c (compute_outgoing_frequencies): Update.
13667 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13668 (update_br_prob_note): Update.
13669 (rtl_verify_edges): Update.
13670 (purge_dead_edges): Update.
13671 (fixup_reorder_chain): Update.
13672 * emit-rtl.c (try_split): Update.
13673 * ifcvt.c (cond_exec_process_insns): Update.
13674 (cond_exec_process_if_block): Update.
13675 (dead_or_predicable): Update.
13676 * internal-fn.c (expand_addsub_overflow): Update.
13677 (expand_neg_overflow): Update.
13678 (expand_mul_overflow): Update.
13679 * loop-doloop.c (doloop_modify): Update.
13680 * loop-unroll.c (compare_and_jump_seq): Update.
13681 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13682 * predict.h: Update.
13683 * reorg.c (mostly_true_jump): Update.
13685 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13686 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13687 * config/arc/arc.c: (emit_unlikely_jump): Update.
13688 * config/arm/arm.c: (emit_unlikely_jump): Update.
13689 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13690 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13691 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13692 (ix86_print_operand): Update.
13693 (ix86_split_fp_branch): Update.
13694 (predict_jump): Update.
13695 * config/ia64/ia64.c (ia64_print_operand): Update.
13696 * config/mmix/mmix.c (mmix_print_operand): Update.
13697 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13698 (rs6000_expand_split_stack_prologue): Update.
13699 * config/rs6000/rs6000.c: Update.
13700 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13701 (s390_expand_vec_movstr): Update.
13702 (s390_expand_cs_tdsi): Update.
13703 (s390_expand_split_stack_prologue): Update.
13704 * config/sh/sh.c (sh_print_operand): Update.
13705 (expand_cbranchsi4): Update.
13706 (expand_cbranchdi4): Update.
13707 * config/sparc/sparc.c (output_v9branch): Update.
13708 * config/spu/spu.c (get_branch_target): Update.
13709 (ea_load_store_inline): Update.
13710 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13711 * config/tilepro/tilepro.c: Update.
13713 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13715 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13716 (gimplify_save_expr): Likewise.
13718 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13720 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13722 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13724 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13726 * timevar.def (TV_IPA_FNSUMMARY): Define.
13728 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13730 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13731 to back store errata sensitive sequence from being generated.
13732 (sqrtdf2_fix): Likewise.
13734 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13736 * tree-ssa-threadupdate.c (compute_path_counts,
13737 update_joiner_offpath_counts): Use profile_probability.
13739 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13742 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13744 * config/arm/arm-c.c (arm_cpu_builtins): Define
13745 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13747 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13749 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13750 array entries to represent __ieee128 versions of the
13751 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13752 scalar_extract_sig, and scalar_insert_exp built-in functions.
13753 (altivec_resolve_overloaded_builtin): Add special case handling
13754 for the __builtin_scalar_insert_exp function, as represented by
13755 the P9V_BUILTIN_VEC_VSIEDP constant.
13756 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13757 exponent support for __ieee128 argument.
13758 (VSESQP): Add scalar extract signature support for __ieee128
13760 (VSTDCNQP): Add scalar test negative support for __ieee128
13762 (VSIEQP): Add scalar insert exponent support for __int128 argument
13763 with __ieee128 result.
13764 (VSIEQPF): Add scalar insert exponent support for __ieee128
13765 argument with __ieee128 result.
13766 (VSTDCQP): Add scalar test data class support for __ieee128
13768 (VSTDCNQP): Add overload support for scalar test negative with
13769 __ieee128 argument.
13770 (VSTDCQP): Add overload support for scalar test data class
13771 __ieee128 argument.
13772 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13773 UNSPEC_VSX_SXSIGDP.
13774 (UNSPEC_VSX_SIEXPQP): New constant.
13775 (xsxexpqp): New insn for VSX scalar extract exponent quad
13777 (xsxsigqp): New insn for VSX scalar extract significand quad
13779 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13780 precision with floating point argument.
13781 (xststdcqp): New expand for VSX scalar test data class quad
13783 (xststdcnegqp): New expand for VSX scalar test negative quad
13785 (xststdcqp): New insn to match expansions for VSX scalar test data
13786 class quad precision and VSX scalar test negative quad precision.
13787 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13788 special case operand checking to enforce that second operand of
13789 VSX scalar test data class with quad precision argument is a 7-bit
13791 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13792 prototypes and descriptions of __ieee128 versions of
13793 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13794 scalar_test_data_class, and scalar_test_neg built-in functions.
13796 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13798 PR tree-optimization/81162
13799 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13800 replace a negate with an add.
13802 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13804 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13806 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13808 * config/arm/arm-c.c (arm_cpu_builtins): Define
13809 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13811 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13813 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13814 (armv8-r): Set ARM Cortex-R52 as default CPU.
13815 * config/arm/arm-tables.opt: Regenerate.
13816 * config/arm/arm-tune.md: Regenerate.
13817 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13819 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13820 extension for -mcpu=cortex-r52.
13822 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13824 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13825 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13826 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13827 (fp-armv8): Define it as FP_ARMv8 only.
13828 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13829 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13831 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13832 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13833 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13834 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13835 than TARGET_FPU_ARMV8.
13836 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13837 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13838 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13840 * config/arm/neon.md (neon_vrint): Likewise.
13841 (neon_vcvt): Likewise.
13842 (neon_<fmaxmin_op><mode>): Likewise.
13843 (<fmaxmin><mode>3): Likewise.
13844 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13845 * config/arm/predicates.md (arm_cond_move_operator): Check against
13846 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13848 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13850 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13851 to top of function.
13853 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13855 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13856 loop in comment with memset.
13858 2017-07-14 Martin Liska <mliska@suse.cz>
13860 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13861 * dwarf2out.c (is_java): Remove the function.
13862 (output_pubname): Remove usage of the function.
13863 (lower_bound_default): Remove usage of DW_LANG_Java.
13864 (gen_compile_unit_die): Likewise.
13865 * gcc.c: Remove compiler defaults for .java and .zip files.
13866 * gimple-expr.c (remove_suffix): Change as there's no longer
13867 extension than 4-letter one.
13868 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13869 (gimplify_save_expr): Likewise.
13870 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13871 as it's possible even for other languages than Java.
13872 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13873 * lto-opts.c (lto_write_options): Remove reference to Java.
13874 * opts.c (strip_off_ending): Update file extension handling.
13875 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13876 * tree-eh.c (lower_resx): Likewise.
13877 * tree.c (free_lang_data_in_type): Remove dead code.
13878 (find_decls_types_r): Likewise.
13879 (build_common_builtin_nodes): Remove Java from a comment.
13880 (verify_type): Remove dead code.
13881 * varasm.c (assemble_external): Remove Java from a comment.
13883 2017-07-14 Martin Liska <mliska@suse.cz>
13885 * opts.c (finish_options): Add quotes.
13886 (common_handle_option): Likewise.
13888 2017-07-14 Martin Liska <mliska@suse.cz>
13890 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13891 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13892 Remove N_SO_PASCAL.
13893 * dwarf2out.c (lower_bound_default): Do not handle
13895 (gen_compile_unit_die): Likewise.
13896 * gcc.c: Remove default extension binding for GNU Pascal.
13897 * stmt.c: Remove Pascal language from a comment.
13898 * xcoffout.c: Likewise.
13900 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13903 * diagnostic-show-locus.c (fixit_cmp): New function.
13904 (layout::layout): Sort m_fixit_hints.
13905 (column_range::column_range): Assert that the values are valid.
13906 (struct char_span): New struct.
13907 (correction::overwrite): New method.
13908 (struct source_line): New struct.
13909 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13910 calls in terms of classes source_line and char_span, and
13911 correction::overwrite.
13912 (selftest::test_overlapped_fixit_printing_2): New function.
13913 (selftest::diagnostic_show_locus_c_tests): Call it.
13915 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13917 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13918 early if there is no lhs.
13920 2017-07-13 Martin Liska <mliska@suse.cz>
13922 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13923 (gen_reference_type_die): Likewise.
13924 * stor-layout.c: Remove Pascal-related comment.
13926 2017-07-13 Martin Liska <mliska@suse.cz>
13928 * opts.c (finish_options): Add quotes to error messages.
13929 (parse_sanitizer_options): Likewise.
13931 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13933 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13935 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13937 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13939 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13941 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13943 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13945 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13948 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13949 provides the hardware capability bits, define the macro
13950 __BUILTIN_CPU_SUPPORTS__.
13951 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13952 if GLIBC does not provide the hardware capability bits. Add a
13953 gcc_unreachable call if the built-in cpu function is neither
13954 __builtin_cpu_is nor __builtin_cpu_supports.
13955 (rs6000_get_function_versions_dispatcher): Change the warning
13956 that an old GLIBC is used which does not export the capability
13957 bits to be an error.
13958 * doc/extend.texi (target_clones attribute): Document the
13959 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13960 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13961 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13962 the macros defined by GCC if the newer GLIBC is available.
13964 2017-07-12 Jeff Law <law@redhat.com>
13966 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13967 remaining includes slightly.
13968 * config/riscv/riscv-builtins.c: Include profile-count.h.
13970 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13973 * config/avr/avr.c (avr_set_current_function): In diagnostic
13974 messages: Quote keywords and (parts of) identifiers.
13975 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13978 2017-07-12 Carl Love <cel@us.ibm.com>
13980 * config/rs6000/rs6000-c.c: Add support for built-in functions
13981 vector bool char vec_revb (vector bool char);
13982 vector bool short vec_revb (vector short char);
13983 vector bool int vec_revb (vector bool int);
13984 vector bool long long vec_revb (vector bool long long);
13985 * doc/extend.texi: Update the built-in documentation file for the
13986 new built-in functions.
13988 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13990 * config/s390/s390.md: Remove movcc splitter.
13992 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13994 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13995 load/store on condition.
13997 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14000 * config/avr/avr.c (avr_encode_section_info)
14001 [progmem && !TREE_READONLY]: Error if progmem object needs
14004 2017-07-11 Michael Collison <michael.collison@arm.com>
14006 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14009 2017-07-11 Carl Love <cel@us.ibm.com>
14011 * config/rs6000/rs6000-c.c: Add support for builtins
14012 vector unsigned int vec_parity_lsbb (vector signed int);
14013 vector unsigned int vec_parity_lsbb (vector unsigned int);
14014 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14015 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14016 vector unsigned long long vec_parity_lsbb (vector signed long long);
14017 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14018 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14019 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14020 * doc/extend.texi: Update the built-in documentation file for the
14021 new built-in functions.
14023 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14025 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14026 (layout::m_primary_loc): New field.
14027 (layout::layout): Initialize new field. Move location filtering
14028 logic from here to...
14029 (layout::maybe_add_location_range): ...this new method. Add
14030 support for filtering to just the lines already specified by other
14032 (layout::will_show_line_p): New method.
14033 (gcc_rich_location::add_location_if_nearby): New method.
14034 (selftest::test_add_location_if_nearby): New test function.
14035 (selftest::diagnostic_show_locus_c_tests): Call it.
14036 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14039 2017-07-11 Tom de Vries <tom@codesourcery.com>
14041 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14042 (bb_first_real_insn): New function.
14043 (nvptx_single): Add extra initialization of broadcasted condition
14046 2017-07-11 Nathan Sidwell <nathan@acm.org>
14048 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14050 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14052 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14053 Binutils doc as TEXI will mess them up.
14054 * doc/invoke.texi (AVR Options): Same here.
14056 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14058 * config/sparc/sparc.opt (mfix-ut700): New option.
14059 (mfix-gr712rc): Likewise.
14060 (sparc_fix_b2bst): New variable.
14061 * doc/invoke.texi (SPARC options): Document them.
14062 (ARM options): Fix warnings.
14063 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14064 instructions to prevent sequences that can trigger the store-store
14065 errata for certain LEON3FT processors.
14066 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14067 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14068 * config/sparc/sparc.md (fix_b2bst): New attribute.
14069 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14071 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14074 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14076 (*rsqrtsf2_sse): Ditto.
14078 (div<mode>3): Macroize insn from divdf3 and divsf3
14079 using MODEF mode iterator.
14081 2017-07-10 Martin Sebor <msebor@redhat.com>
14083 PR tree-optimization/80397
14084 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14085 instead of testing for equality to INTEGER_TYPE.
14087 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14089 * config.gcc: Remove uclibc from arc target spec.
14091 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14093 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14095 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14098 * lto-wrapper.c (remove_option): New function.
14099 (merge_and_complain): Merge PIC/PIE options more realistically.
14101 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14103 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14107 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14108 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14109 * config.in: Regenerate.
14110 * configure: Regenerate.
14111 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14112 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14113 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14114 (TARGET_GASISR_PROLOGUES): ...target mask.
14115 * common/config/avr/avr-common.c
14116 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14117 Set -mgas-isr-prologues.
14118 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14119 INSERT_PASS_BEFORE for it.
14120 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14121 * config/avr/avr.c (avr_option_override)
14122 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14123 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14124 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14125 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14126 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14128 (make_avr_pass_pre_proep): New function.
14129 (emit_push_sfr) <treg>: Add argument to function and use it
14130 instead of TMP_REG.
14131 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14132 and set machine->gasisr.yes.
14133 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14134 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14135 __gcc_isr.n_pushed to .L__stack_usage.
14136 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14137 (avr_asm_final_postscan_insn): ...this new static function.
14138 * config/avr/avr.h (machine_function)
14139 <is_no_gccisr, use_L__stack_usage>: New fields.
14140 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14141 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14142 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14143 (gasisr, *gasisr): New expander and insn.
14144 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14145 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14146 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14148 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14150 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14153 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14155 Move jump-tables out of .text again.
14158 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14159 (ASM_OUTPUT_ADDR_VEC): New function.
14160 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14161 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14162 INSN_ADDRESSes as asm comment.
14163 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14164 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14165 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14166 * config/avr/avr.md (*tablejump): Adjust comment.
14167 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14168 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14170 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14171 (avr_output_addr_vec): New proto.
14172 (avr_log_t) <insn_addresses>: New field.
14174 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14177 * config/i386/i386.c (ix86_function_arg_advance): Set
14178 outgoing_args_on_stack to true if there are outgoing arguments
14180 (ix86_function_arg): Likewise.
14181 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14182 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14183 * config/i386/i386.h (machine_function): Add
14184 outgoing_args_on_stack.
14186 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14188 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14189 supporting pthreds.
14190 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14192 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14194 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14195 (REAL_H): Remove $(MACHMODE_H).
14196 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14198 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14199 $(MACHMODE_H) and double-int.h.
14200 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14202 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14205 2017-07-07 Andrew Pinski <apinski@cavium.com>
14207 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14208 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14210 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14212 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14213 Add warning if GCC was not configured to link against a GLIBC that
14214 exports the hardware capability bits.
14215 (make_resolver_func): Make resolver function private and not a
14216 COMDAT function. Create the name with clone_function_name instead
14217 of make_unique_name.
14220 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14221 correct operand in doing the split.
14223 2017-07-07 Carl Love <cel@us.ibm.com>
14225 * config/rs6000/rs6000-c: Add support for built-in function
14226 vector unsigned short vec_pack_to_short_fp32 (vector float,
14228 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14229 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14230 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14231 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14232 (convert_4f32_8i16): Add define_expand.
14233 * doc/extend.texi: Update the built-in documentation file for the
14234 new built-in function.
14236 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14238 * config/sparc/m8.md: New file.
14239 * config/sparc/sparc.md: Include m8.md.
14241 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14243 * config/sparc/sparc.opt: New option -mvis4b.
14244 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14245 (sparc_option_override): Handle VIS4B.
14246 (enum sparc_builtins): Define
14247 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14248 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14249 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14250 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14251 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14252 (check_constant_argument): New function.
14253 (sparc_vis_init_builtins): Define builtins
14254 __builtin_vis_dictunpack{8,16,32},
14255 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14256 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14257 __builtin_vis_fpcmpde{8,16,32}shl and
14258 __builtin_vis_fpcmpur{8,16,32}shl.
14259 (sparc_expand_builtin): Check that the constant operands to
14260 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14261 constant and in range.
14262 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14264 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14265 (SPARC_IMM5_P): Likewise.
14266 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14267 (enabled): Handle vis4b.
14268 (UNSPEC_DICTUNPACK): New unspec.
14269 (UNSPEC_FPCMPSHL): Likewise.
14270 (UNSPEC_FPUCMPSHL): Likewise.
14271 (UNSPEC_FPCMPDESHL): Likewise.
14272 (UNSPEC_FPCMPURSHL): Likewise.
14273 (cpu_feature): New CPU feature `vis4b'.
14274 (dictunpack{8,16,32}): New insns.
14275 (FPCSMODE): New mode iterator.
14276 (fpcscond): New code iterator.
14277 (fpcsucond): Likewise.
14278 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14279 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14280 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14281 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14282 * config/sparc/constraints.md: Define constraints `q' for unsigned
14283 2-bit integer constants and `t' for unsigned 5-bit integer
14285 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14287 (imm5_operand_dictunpack16): Likewise.
14288 (imm5_operand_dictunpack32): Likewise.
14289 (imm2_operand): Likewise.
14290 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14291 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14292 ditunpack* and fpcmp*shl builtins.
14294 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14296 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14297 * config.in: Add HAVE_AS_SPARC6 define.
14298 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14300 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14302 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14303 (CPP_CPU_SPEC): Handle m8.
14304 (ASM_CPU_SPEC): Likewise.
14305 * config/sparc/sparc-opts.h (enum processor_type): Add
14307 * config/sparc/sparc.c (m8_costs): New struct.
14308 (sparc_option_override): Handle TARGET_CPU_m8.
14309 (sparc32_initialize_trampoline): Likewise.
14310 (sparc64_initialize_trampoline): Likewise.
14311 (sparc_issue_rate): Likewise.
14312 (sparc_register_move_cost): Likewise.
14313 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14314 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14315 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14316 (CPP_CPU_SPEC): Handle M8.
14317 (ASM_CPU_SPEC): Likewise.
14318 (AS_M8_FLAG): Define.
14319 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14320 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14321 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14323 * configure: Regenerate.
14324 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14327 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14329 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14331 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14332 ("*movdi_insn_sp32"): Do not set v3pipe.
14333 ("*movsi_insn"): Likewise.
14334 ("*movdi_insn_sp64"): Likewise.
14335 ("*movsf_insn"): Likewise.
14336 ("*movdf_insn_sp32"): Likewise.
14337 ("*movdf_insn_sp64"): Likewise.
14338 ("*zero_extendsidi2_insn_sp64"): Likewise.
14339 ("*sign_extendsidi2_insn"): Likewise.
14340 ("*mov<VM32:mode>_insn"): Likewise.
14341 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14342 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14343 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14344 ("<vlop:code><VL:mode>3"): Likewise.
14345 ("*not_<vlop:code><VL:mode>3"): Likewise.
14346 ("*nand<VL:mode>_vis"): Likewise.
14347 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14348 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14349 ("one_cmpl<VL:mode>2"): Likewise.
14350 ("faligndata<VM64:mode>_vis"): Likewise.
14351 ("alignaddrsi_vis"): Likewise.
14352 ("alignaddrdi_vis"): Likweise.
14353 ("alignaddrlsi_vis"): Likewise.
14354 ("alignaddrldi_vis"): Likewise.
14355 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14356 ("bmaskdi_vis"): Likewise.
14357 ("bmasksi_vis"): Likewise.
14358 ("bshuffle<VM64:mode>_vis"): Likewise.
14359 ("cmask8<P:mode>_vis"): Likewise.
14360 ("cmask16<P:mode>_vis"): Likewise.
14361 ("cmask32<P:mode>_vis"): Likewise.
14362 ("pdistn<P:mode>_vis"): Likewise.
14363 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14365 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14367 * config/sparc/sparc.md ("subtype"): New insn attribute.
14368 ("*wrgsr_sp64"): Set insn subtype.
14369 ("*rdgsr_sp64"): Likewise.
14370 ("alignaddrsi_vis"): Likewise.
14371 ("alignaddrdi_vis"): Likewise.
14372 ("alignaddrlsi_vis"): Likewise.
14373 ("alignaddrldi_vis"): Likewise.
14374 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14375 ("fexpand_vis"): Likewise.
14376 ("fpmerge_vis"): Likewise.
14377 ("faligndata<VM64:mode>_vis"): Likewise.
14378 ("bshuffle<VM64:mode>_vis"): Likewise.
14379 ("cmask8<P:mode>_vis"): Likewise.
14380 ("cmask16<P:mode>_vis"): Likewise.
14381 ("cmask32<P:mode>_vis"): Likewise.
14382 ("fchksm16_vis"): Likewise.
14383 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14384 ("fmean16_vis"): Likewise.
14385 ("fp<plusminus_insn>64_vis"): Likewise.
14386 ("<plusminus_insn>v8qi3"): Likewise.
14387 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14388 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14389 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14390 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14391 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14392 ("*movqi_insn"): Likewise.
14393 ("*movhi_insn"): Likewise.
14394 ("*movsi_insn"): Likewise.
14395 ("movsi_pic_gotdata_op"): Likewise.
14396 ("*movdi_insn_sp32"): Likewise.
14397 ("*movdi_insn_sp64"): Likewise.
14398 ("movdi_pic_gotdata_op"): Likewise.
14399 ("*movsf_insn"): Likewise.
14400 ("*movdf_insn_sp32"): Likewise.
14401 ("*movdf_insn_sp64"): Likewise.
14402 ("*zero_extendhisi2_insn"): Likewise.
14403 ("*zero_extendqihi2_insn"): Likewise.
14404 ("*zero_extendqisi2_insn"): Likewise.
14405 ("*zero_extendqidi2_insn"): Likewise.
14406 ("*zero_extendhidi2_insn"): Likewise.
14407 ("*zero_extendsidi2_insn_sp64"): Likewise.
14408 ("ldfsr"): Likewise.
14409 ("prefetch_64"): Likewise.
14410 ("prefetch_32"): Likewise.
14411 ("tie_ld32"): Likewise.
14412 ("tie_ld64"): Likewise.
14413 ("*tldo_ldub_sp32"): Likewise.
14414 ("*tldo_ldub1_sp32"): Likewise.
14415 ("*tldo_ldub2_sp32"): Likewise.
14416 ("*tldo_ldub_sp64"): Likewise.
14417 ("*tldo_ldub1_sp64"): Likewise.
14418 ("*tldo_ldub2_sp64"): Likewise.
14419 ("*tldo_ldub3_sp64"): Likewise.
14420 ("*tldo_lduh_sp32"): Likewise.
14421 ("*tldo_lduh1_sp32"): Likewise.
14422 ("*tldo_lduh_sp64"): Likewise.
14423 ("*tldo_lduh1_sp64"): Likewise.
14424 ("*tldo_lduh2_sp64"): Likewise.
14425 ("*tldo_lduw_sp32"): Likewise.
14426 ("*tldo_lduw_sp64"): Likewise.
14427 ("*tldo_lduw1_sp64"): Likewise.
14428 ("*tldo_ldx_sp64"): Likewise.
14429 ("*mov<VM32:mode>_insn"): Likewise.
14430 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14431 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14433 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14435 * config/sparc/sparc.md ("type"): New insn type viscmp.
14436 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14438 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14439 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14440 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14441 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14443 ("n7_vis_logical_11cycle"): Likewise.
14444 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14445 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14446 * config/sparc/niagara.md ("niag_vis"): Likewise.
14447 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14448 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14450 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14452 * config/sparc/sparc.md: New instruction type `bmask'.
14453 (bmaskdi_vis): Use the `bmask' type.
14454 (bmasksi_vis): Likewise.
14455 * config/sparc/ultra3.md (us3_array): Likewise.
14456 * config/sparc/niagara7.md (n7_array): Likewise.
14457 * config/sparc/niagara4.md (n4_array): Likewise.
14458 * config/sparc/niagara2.md (niag2_vis): Likewise.
14459 (niag3_vis): Likewise.
14460 * config/sparc/niagara.md (niag_vis): Likewise.
14462 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14464 * ipa-comdats.c: Remove optimize check from gate.
14465 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14466 for functions not optimized.
14467 (ipa_fn_summary_read): Skip optimize check.
14468 (ipa_fn_summary_write): Likewise.
14469 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14471 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14473 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14474 (check_callers): Give up on caller which is not optimized.
14475 (inline_small_functions): Likewise.
14476 (ipa_inline): Do not give up when not optimizing.
14477 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14478 away unoptimizes cdtors.
14479 (whole_program_function_and_variable_visibility): Do
14480 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14481 * ipa.c (process_references): Do not check optimize.
14482 (symbol_table::remove_unreachable_nodes): Update optimize check.
14483 (set_writeonly_bit): Update optimize check.
14484 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14485 (pass_ipa_single_use::gate): Remove.
14487 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14489 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14490 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14491 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14492 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14493 permute_load, permute_store, adjust_extract, adjust_splat,
14494 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14495 replace_swap_with_copy, dump_swap_insn_table,
14496 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14497 recombine_lvx_pattern, recombine_stvx_pattern,
14498 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14499 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14500 to file rs6000-p8swap.c.
14501 * config/rs6000/rs6000-p8swap.c: New file.
14502 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14503 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14504 and rs6000*-*-* targets.
14506 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14508 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14510 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14512 * lto-wrapper.c (merge_and_complain): Do not merge
14513 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14514 fsigned_zeros, ftrapping_math, fwrapv.
14515 (append_compiler_options): Do not track these options.
14516 (append_linker_options): Likewie
14518 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14520 * cgraphunit.c (cgraph_node::finalize_function): When
14521 !flag_toplevel_reorde set no_reorder flag.
14522 (varpool_node::finalize_decl): Likewise.
14523 (symbol_table::compile): Drop no toplevel reorder path.
14525 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14527 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14528 edges; zero probability is not better than uninitialized.
14530 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14532 * asan.h (asan_sanitize_allocas_p): Declare.
14533 * asan.c (asan_sanitize_allocas_p): New function.
14534 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14535 (handle_builtin_alloca): Likewise.
14536 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14537 if !asan_sanitize_allocas_p.
14538 * params.def (asan-instrument-allocas): Add new option.
14539 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14540 * opts.c (common_handle_option): Disable allocas sanitization for
14543 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14545 * asan.c: Include gimple-fold.h.
14546 (get_last_alloca_addr): New function.
14547 (handle_builtin_stackrestore): Likewise.
14548 (handle_builtin_alloca): Likewise.
14549 (asan_emit_allocas_unpoison): Likewise.
14550 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14551 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14552 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14553 (instrument_builtin_call): Pass gimple iterator to
14554 get_mem_refs_of_builtin_call.
14555 (last_alloca_addr): New global.
14556 * asan.h (asan_emit_allocas_unpoison): Declare.
14557 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14558 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14559 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14560 if function calls alloca.
14561 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14562 * gimple-fold.h (replace_call_with_value): Declare.
14563 * internal-fn.c: Include asan.h.
14564 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14565 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14567 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14569 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14570 (C_SELFTEST_FLAGS): New.
14571 (CPP_SELFTEST_FLAGS): New.
14572 (SELFTEST_DEPS): New, from deps of s-selftest.
14573 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14574 (CPP_SELFTEST_DEPS): New.
14575 (selftest): Add dependency on s-selftest-c++.
14576 (s-selftest): Rename to...
14577 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14578 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14579 than SELFTEST_FLAGS.
14580 (selftest-gdb): Rename to...
14581 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14583 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14584 (selftest-valgrind): Rename to...
14585 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14587 (selftest-valgrind): Reintroduce as an alias for
14588 selftest-c-valgrind.
14589 (s-selftest-c++): New.
14590 (selftest-c++-gdb): New.
14591 (selftest-c++-valgrind): New.
14593 2017-07-06 Olivier Hainque <hainque@adacore.com>
14595 * gcc.c (process_command): When deciding if undefined variables
14596 should be ignored when processing specs, accept "gcc -v" as well.
14598 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14600 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14601 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14603 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14605 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14606 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14607 * config/arm/arm-tables.opt: Regenerate.
14608 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14610 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14612 2017-07-06 Carl Love <cel@us.ibm.com>
14614 * ChangeLog: Clean up from mid air collision
14616 2017-07-06 Carl Love <cel@us.ibm.com>
14618 * config/rs6000/rs6000-c.c: Add support for built-in functions
14619 vector signed int vec_subc (vector signed int, vector signed int);
14620 vector signed __int128 vec_subc (vector signed __int128,
14621 vector signed __int128);
14622 vector unsigned __int128 vec_subc (vector unsigned __int128,
14623 vector unsigned __int128);
14624 vector signed int vec_sube (vector signed int, vector signed int,
14625 vector signed int);
14626 vector unsigned int vec_sube (vector unsigned int,
14627 vector unsigned int,
14628 vector unsigned int);
14629 vector signed __int128 vec_sube (vector signed __int128,
14630 vector signed __int128,
14631 vector signed__int128);
14632 vector unsigned __int128 vec_sube (vector unsigned __int128,
14633 vector unsigned __int128,
14634 vector unsigned __int128);
14635 vector signed int vec_subec (vector signed int, vector signed int,
14636 vector signed int);
14637 vector unsigned int vec_subec (vector unsigned int,
14638 vector unsigned int,
14639 vector unsigned int);
14640 vector signed __int128 vec_subec (vector signed __int128,
14641 vector signed __int128,
14642 vector signed__int128);
14643 vector unsigned __int128 vec_subec (vector unsigned __int128,
14644 vector unsigned __int128,
14645 vector unsigned __int128);
14646 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14647 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14648 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14649 BU_ALTIVEC_OVERLOAD_X definitions.
14650 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14651 * doc/extend.texi: Update the built-in documentation file for the new
14652 built-in functions.
14654 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14657 * diagnostic-show-locus.c (layout::layout): Use start and finish
14658 spelling location for the start and finish of each range.
14659 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14660 Add unused aspect param.
14661 * input.c (expand_location_1): Add "aspect" param, and use it
14662 to access the correct part of the location.
14663 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14665 (expand_location_to_spelling_point): Likewise.
14666 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14667 param, and pass it to expand_location_1.
14669 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14671 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14672 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14673 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14674 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14675 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14676 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14677 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14678 _mm_maskz_getmant_ss): New intrinsics.
14679 (__builtin_ia32_getexpss128_mask): Changed to ...
14680 __builtin_ia32_getexpss128_round ... this.
14681 (__builtin_ia32_getexpsd128_mask): Changed to ...
14682 __builtin_ia32_getexpsd128_round ... this.
14683 * config/i386/i386-builtin-types.def
14684 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14685 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14686 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14687 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14688 __builtin_ia32_getmantss_mask_round): New builtins.
14689 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14690 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14691 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14692 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14693 * config/i386/sse.md
14694 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14695 avx512f_sgetexp<mode><mask_scalar_name>
14696 <round_saeonly_scalar_name> ... this.
14697 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14698 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14699 vgetexp<ssescalarmodesuffix>
14700 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14701 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14702 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14703 avx512f_vgetmant<mode><mask_scalar_name>
14704 <round_saeonly_scalar_name> ... this.
14705 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14706 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14707 vgetmant<ssescalarmodesuffix>
14708 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14709 %0<mask_scalar_operand4>, %1, %2
14710 <round_saeonly_scalar_mask_op4>, %3} ... this.
14711 * config/i386/subst.md (mask_scalar_operand4,
14712 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14713 round_saeonly_scalar_nimm_predicate): New subst attributes.
14715 2017-07-06 Julia Koval <julia.koval@intel.com>
14717 * config/i386/i386.c (ix86_erase_embedded_rounding):
14718 Remove code for old rounding pattern.
14720 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14722 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14724 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14726 * doc/sourcebuild.texi (Test Directives, Variants of
14727 dg-require-support): Add documentation for dg-require-stack-check.
14729 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14731 * config/i386/subst.md (mask_scalar, round_scalar,
14732 round_saeonly_scalar): New meta-templates.
14733 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14734 round_scalar_mask_operand3, round_scalar_mask_op3,
14735 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14736 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14737 round_saeonly_scalar_constraint,
14738 round_saeonly_scalar_prefix): New subst attribute.
14739 * config/i386/sse.md
14740 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14741 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14742 <round_scalar_name> ... this.
14743 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14744 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14745 <round_scalar_name> ... this.
14746 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14747 <sse>_vm<code><mode>3<mask_scalar_name>
14748 <round_saeonly_scalar_name> ... this.
14749 (v<plusminus_mnemonic><ssescalarmodesuffix>
14750 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14751 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14752 v<plusminus_mnemonic><ssescalarmodesuffix>
14753 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14754 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14755 (v<multdiv_mnemonic><ssescalarmodesuffix>
14756 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14757 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14758 v<multdiv_mnemonic><ssescalarmodesuffix>
14759 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14760 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14761 (v<maxmin_float><ssescalarmodesuffix>
14762 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14763 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14764 v<maxmin_float><ssescalarmodesuffix>
14765 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14766 %0<mask_scalar_operand3>, %1, %<iptr>2
14767 <round_saeonly_scalar_mask_op3>} ... this.
14769 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14771 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14772 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14774 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14775 Alan Hayward <alan.hayward@arm.com>
14776 David Sherwood <david.sherwood@arm.com>
14778 * combine.c (simplify_if_then_else): Remove "enum" before
14780 * compare-elim.c (can_eliminate_compare): Likewise.
14781 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14783 (aarch64_lookup_simd_builtin_type): Likewise.
14784 (aarch64_simd_builtin_type): Likewise.
14785 (aarch64_init_simd_builtin_types): Likewise.
14786 (aarch64_simd_expand_args): Likewise.
14787 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14789 (aarch64_reverse_mask): Likewise.
14790 (aarch64_simd_emit_reg_reg_move): Likewise.
14791 (aarch64_gen_adjusted_ldpstp): Likewise.
14792 (aarch64_ccmp_mode_to_code): Likewise.
14793 (aarch64_operands_ok_for_ldpstp): Likewise.
14794 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14795 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14797 (aarch64_min_divisions_for_recip_mul): Likewise.
14798 (aarch64_reassociation_width): Likewise.
14799 (aarch64_get_condition_code_1): Likewise.
14800 (aarch64_simd_emit_reg_reg_move): Likewise.
14801 (aarch64_simd_attr_length_rglist): Likewise.
14802 (aarch64_reverse_mask): Likewise.
14803 (aarch64_operands_ok_for_ldpstp): Likewise.
14804 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14805 (aarch64_gen_adjusted_ldpstp): Likewise.
14806 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14808 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14809 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14810 (arm_lookup_simd_builtin_type): Likewise.
14811 (arm_simd_builtin_type): Likewise.
14812 (arm_init_simd_builtin_types): Likewise.
14813 (arm_expand_builtin_args): Likewise.
14814 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14815 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14816 (ft32_setup_incoming_varargs): Likewise.
14817 (ft32_function_arg): Likewise.
14818 (ft32_function_arg_advance): Likewise.
14819 (ft32_pass_by_reference): Likewise.
14820 (ft32_arg_partial_bytes): Likewise.
14821 (ft32_valid_pointer_mode): Likewise.
14822 (ft32_addr_space_pointer_mode): Likewise.
14823 (ft32_addr_space_legitimate_address_p): Likewise.
14824 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14826 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14827 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14828 (iamcu_alignment): Likewise.
14829 (canonicalize_vector_int_perm): Likewise.
14830 (ix86_noce_conversion_profitable_p): Likewise.
14831 (ix86_mpx_bound_mode): Likewise.
14832 (ix86_operands_ok_for_move_multiple): Likewise.
14833 * config/microblaze/microblaze-protos.h
14834 (microblaze_expand_conditional_branch_reg): Likewise.
14835 * config/microblaze/microblaze.c
14836 (microblaze_expand_conditional_branch_reg): Likewise.
14837 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14839 (rs6000_reassociation_width): Likewise.
14840 (rs6000_invalid_binary_op): Likewise.
14841 (fusion_p9_p): Likewise.
14842 (emit_fusion_p9_load): Likewise.
14843 (emit_fusion_p9_store): Likewise.
14844 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14846 (riscv_hard_regno_mode_ok_p): Likewise.
14847 (riscv_address_insns): Likewise.
14848 (riscv_split_symbol): Likewise.
14849 (riscv_legitimize_move): Likewise.
14850 (riscv_function_value): Likewise.
14851 (riscv_hard_regno_nregs): Likewise.
14852 (riscv_expand_builtin): Likewise.
14853 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14854 (riscv_build_integer): Likewise.
14855 (riscv_split_integer): Likewise.
14856 (riscv_legitimate_constant_p): Likewise.
14857 (riscv_cannot_force_const_mem): Likewise.
14858 (riscv_regno_mode_ok_for_base_p): Likewise.
14859 (riscv_valid_base_register_p): Likewise.
14860 (riscv_valid_offset_p): Likewise.
14861 (riscv_valid_lo_sum_p): Likewise.
14862 (riscv_classify_address): Likewise.
14863 (riscv_legitimate_address_p): Likewise.
14864 (riscv_address_insns): Likewise.
14865 (riscv_load_store_insns): Likewise.
14866 (riscv_force_binary): Likewise.
14867 (riscv_split_symbol): Likewise.
14868 (riscv_force_address): Likewise.
14869 (riscv_legitimize_address): Likewise.
14870 (riscv_move_integer): Likewise.
14871 (riscv_legitimize_const_move): Likewise.
14872 (riscv_legitimize_move): Likewise.
14873 (riscv_address_cost): Likewise.
14874 (riscv_subword): Likewise.
14875 (riscv_output_move): Likewise.
14876 (riscv_canonicalize_int_order_test): Likewise.
14877 (riscv_emit_int_order_test): Likewise.
14878 (riscv_function_arg_boundary): Likewise.
14879 (riscv_pass_mode_in_fpr_p): Likewise.
14880 (riscv_pass_fpr_single): Likewise.
14881 (riscv_pass_fpr_pair): Likewise.
14882 (riscv_get_arg_info): Likewise.
14883 (riscv_function_arg): Likewise.
14884 (riscv_function_arg_advance): Likewise.
14885 (riscv_arg_partial_bytes): Likewise.
14886 (riscv_function_value): Likewise.
14887 (riscv_pass_by_reference): Likewise.
14888 (riscv_setup_incoming_varargs): Likewise.
14889 (riscv_print_operand): Likewise.
14890 (riscv_elf_select_rtx_section): Likewise.
14891 (riscv_save_restore_reg): Likewise.
14892 (riscv_for_each_saved_reg): Likewise.
14893 (riscv_register_move_cost): Likewise.
14894 (riscv_hard_regno_mode_ok_p): Likewise.
14895 (riscv_hard_regno_nregs): Likewise.
14896 (riscv_class_max_nregs): Likewise.
14897 (riscv_memory_move_cost): Likewise.
14898 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14899 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14900 (rl78_addr_space_address_mode): Likewise.
14901 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14903 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14904 (rs6000_reassociation_width): Likewise.
14905 (rs6000_invalid_binary_op): Likewise.
14906 (fusion_p9_p): Likewise.
14907 (emit_fusion_p9_load): Likewise.
14908 (emit_fusion_p9_store): Likewise.
14909 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14910 (ok_for_simple_move_operands): Likewise.
14911 (ok_for_simple_move_strict_operands): Likewise.
14912 (ok_for_simple_arith_logic_operands): Likewise.
14913 (visium_legitimize_reload_address): Likewise.
14914 (visium_select_cc_mode): Likewise.
14915 (output_cbranch): Likewise.
14916 (visium_split_double_move): Likewise.
14917 (visium_expand_copysign): Likewise.
14918 (visium_expand_int_cstore): Likewise.
14919 (visium_expand_fp_cstore): Likewise.
14920 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14921 (visium_function_arg): Likewise.
14922 (visium_function_arg_advance): Likewise.
14923 (visium_libcall_value): Likewise.
14924 (visium_setup_incoming_varargs): Likewise.
14925 (visium_legitimate_constant_p): Likewise.
14926 (visium_legitimate_address_p): Likewise.
14927 (visium_legitimize_address): Likewise.
14928 (visium_secondary_reload): Likewise.
14929 (visium_register_move_cost): Likewise.
14930 (visium_memory_move_cost): Likewise.
14931 (prepare_move_operands): Likewise.
14932 (ok_for_simple_move_operands): Likewise.
14933 (ok_for_simple_move_strict_operands): Likewise.
14934 (ok_for_simple_arith_logic_operands): Likewise.
14935 (visium_function_value_1): Likewise.
14936 (rtx_ok_for_offset_p): Likewise.
14937 (visium_legitimize_reload_address): Likewise.
14938 (visium_split_double_move): Likewise.
14939 (visium_expand_copysign): Likewise.
14940 (visium_expand_int_cstore): Likewise.
14941 (visium_expand_fp_cstore): Likewise.
14942 (visium_split_cstore): Likewise.
14943 (visium_select_cc_mode): Likewise.
14944 (visium_split_cbranch): Likewise.
14945 (output_cbranch): Likewise.
14946 (visium_print_operand_address): Likewise.
14947 * expmed.c (flip_storage_order): Likewise.
14948 * expmed.h (emit_cstore): Likewise.
14949 (flip_storage_order): Likewise.
14950 * genrecog.c (validate_pattern): Likewise.
14951 * hsa-gen.c (gen_hsa_addr): Likewise.
14952 * internal-fn.c (expand_arith_overflow): Likewise.
14953 * ira-color.c (allocno_copy_cost_saving): Likewise.
14954 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14955 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14956 (process_invariant_for_inheritance): Likewise.
14957 * lra-eliminations.c (move_plus_up): Likewise.
14958 * omp-low.c (lower_oacc_reductions): Likewise.
14959 * simplify-rtx.c (simplify_subreg): Likewise.
14960 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14961 (TARGET_CHKP_BOUND_MODE): Likewise..
14962 * targhooks.c (default_chkp_bound_mode): Likewise.
14963 (default_setup_incoming_vararg_bounds): Likewise.
14964 * targhooks.h (default_chkp_bound_mode): Likewise.
14965 (default_setup_incoming_vararg_bounds): Likewise.
14966 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14967 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14968 (have_whole_vector_shift): Likewise.
14969 * tree-vect-stmts.c (vectorizable_load): Likewise.
14970 * doc/tm.texi: Regenerate.
14972 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14974 Graceful degrade if Binutils PR21472 is not available.
14977 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14978 .rodata in flash test fails.
14979 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14980 * confgure: Regenerate.
14981 * config.in: Regenerate.
14982 * config/avr/avr.c (avr_asm_named_section)
14983 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14984 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14985 (avr_asm_init_sections): Same.
14987 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14989 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14990 (fma<VH:mode>4_intrinsic): Likewise.
14991 (*fmsub<VCVTF:mode>4): Likewise.
14992 (*fmsub<VH:mode>4_intrinsic): Likewise.
14994 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14997 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14998 Don't depend on "optimize > 0".
14999 (out_movhi_r_mr, out_movqi_mr_r): Same.
15000 (out_movhi_mr_r, out_movqi_r_mr): Same.
15001 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15002 io_address_operand on "optimize > 0".
15004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15006 * tree-loop-distribution.c: Add general explanantion on the pass.
15007 (generate_loops_for_partition): Mark distributed loop.
15008 (pg_add_dependence_edges): New parameter. Handle alias data
15009 dependence specially and record it in the parameter if asked.
15010 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15011 (init_partition_graph_vertices, add_partition_graph_edge): New.
15012 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15013 (free_partition_graph_vdata, build_partition_graph): New.
15014 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15015 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15016 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15017 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15018 (version_for_distribution_p, finalize_partitions): New.
15019 (distribute_loop): Handle alias data dependence specially. Factor
15020 out loop fusion code as functions and call these functions.
15022 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15024 * tree-loop-distribution.c (classify_partition): New parameter and
15025 better handle reduction statement.
15026 (rdg_build_partitions): Revise comment.
15027 (distribute_loop): Compute statements in all partitions and pass it
15028 to classify_partition.
15030 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15032 * tree-loop-distribution.c (enum partition_type): New.
15033 (struct partition): New field type.
15034 (partition_merge_into): Add parameter. Update partition type.
15035 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15036 (build_rdg_partition_for_vertex): Compute partition type.
15037 (rdg_build_partitions): Dump partition type.
15038 (distribute_loop): Update calls to partition_merge_into.
15040 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15042 * tree-loop-distribution.c (struct ddr_hasher): New.
15043 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15045 (classify_partition): Call get_data_dependence.
15046 (pg_add_dependence_edges): Ditto.
15047 (distribute_loop): Release data dependence hash table.
15049 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15051 * tree-loop-distribution.c (ref_base_address): Delete.
15052 (similar_memory_accesses): Rename ...
15053 (share_memory_accesses): ... to this. Check if partitions access
15054 the same memory reference.
15055 (distribute_loop): Call share_memory_accesses.
15057 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15059 * tree-loop-distribution.c (struct partition): New field recording
15060 its data reference.
15061 (partition_alloc, partition_free): Init and release data refs.
15062 (partition_merge_into): Merge data refs.
15063 (build_rdg_partition_for_vertex): Collect data refs for partition.
15064 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15066 (distribute_loop): Remve data refs from vertice data of partition
15069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15071 * tree-loop-distribution.c (params.h): Include header file.
15072 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15073 (datarefs_vec): New global var.
15074 (create_rdg_vertices): Use datarefs_vec directly.
15075 (free_rdg): Don't free data references.
15076 (build_rdg): Update use. Don't free data references.
15077 (distribute_loop): Compute global variable for data references.
15078 Bail out if there are too many data references.
15080 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15082 * tree-loop-distribution.c (loop_nest): New global var.
15083 (build_rdg): Use loop directly, rather than loop nest.
15084 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15086 (distribute_loop): Compute global variable loop nest. Update use.
15088 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15090 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15091 (partition_merge_into): New parameter. Dump reason for fusion.
15092 (distribute_loop): Update use of partition_merge_into.
15094 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15096 * tree-loop-distribution.c (bb_top_order_index): New.
15097 (bb_top_order_index_size, bb_top_order_cmp): New.
15098 (stmts_from_loop): Use topological order.
15099 (pass_loop_distribution::execute): Compute and release topological
15100 order for basic blocks.
15102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15104 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15107 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15109 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15110 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15111 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15112 * internal-fn.def (LOOP_DIST_ALIAS): New.
15113 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15114 (fold_loop_internal_call): ... this.
15115 (vect_loop_dist_alias_call): New function.
15116 (set_uid_loop_bbs): Call fold_loop_internal_call.
15117 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15120 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15123 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15124 Require dead FLAGS_REG at the beginning of a peephole.
15126 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15129 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15130 arguments in the call to __builtin_ia32_sbb_u32.
15131 (_subborrow_u64): Swap _X and _Y arguments in the call to
15132 __builtin_ia32_sbb_u64.
15134 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15137 * tree-vrp.c (compare_assert_loc): Turn into a function template
15138 with stable template parameter. Only test if a->e is NULL,
15139 !a->e == !b->e has been verified already. Use e == NULL or
15140 e != NULL instead of e or ! e tests. If stable is true, don't use
15141 iterative_hash_expr, on the other side allow a or b or both NULL
15142 and sort the NULLs last.
15143 (process_assert_insertions): Sort using compare_assert_loc<false>
15144 instead of compare_assert_loc, later sort using
15145 compare_assert_loc<true> before calling process_assert_insertions_for
15146 in a loop. Use break instead of continue once seen NULL pointer.
15148 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15150 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15151 Cortex-R7 and Cortex-R8 processors.
15153 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15155 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15156 uninitialized while src is not.
15158 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15160 * common/config/arm/arm-common.c: Adjust include path for
15162 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15163 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15164 (arm-cpu-data.h): Likewise.
15165 (arm-cpu-cdata.h): Likewise.
15166 * config/arm/arm-cpu.h: Delete.
15167 * config/arm/arm-cpu-cdata.h: Delete.
15168 * config/arm/arm-cpu-data.h: Delete.
15170 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15172 * config/arm/arm-cpus.in (cortex-a55): New.
15173 (cortex-a75): Likewise.
15174 (cortex-a75.cortex-a55): Likewise.
15175 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15177 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15178 * config/arm/arm-cpu-cdata.h: Regenerate.
15179 * config/arm/arm-cpu-data.h: Regenerate.
15180 * config/arm/arm-cpu.h: Regenerate.
15181 * config/arm/arm-tables.opt: Regenerate.
15182 * config/arm/arm-tune.md: Regenerate.
15184 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15186 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15188 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15190 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15193 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15195 PR tree-optimization/81292
15196 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15197 full_string_p, also call adjust_related_strinfos if the adjustment
15198 is simple, otherwise invalidate related strinfos.
15200 2017-07-04 Martin Liska <mliska@suse.cz>
15203 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15204 newly created variable as DECL_IGNORED_P.
15206 2017-07-04 Martin Liska <mliska@suse.cz>
15209 * ipa-inline.c (inline_small_functions):
15210 Use xstrdup_for_dump.
15212 2017-07-04 Tom de Vries <tom@codesourcery.com>
15214 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15216 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15219 * config/darwin.c (darwin_function_switched_text_sections):
15222 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15224 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15226 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15228 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15230 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15232 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15233 min_profitable_iters, and th as inclusive lower bounds.
15234 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15235 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15236 for min_profitable_iters and min_profitable_estimate.
15237 (vect_transform_loop): Treat th as an inclusive lower bound.
15238 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15240 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15243 * config/darwin.c (darwin_function_switched_text_sections):
15244 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15245 in two pieces, and suppress the use of buf.
15247 2017-07-03 Nathan Sidwell <nathan@acm.org>
15249 * hash-table.h (hash_table_mod1): Fix indentation.
15251 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15253 PR middle-end/81290
15254 * predict.c (force_edge_cold): Be more careful about propagation
15256 * profile-count.h (profile_probability::guessed,
15257 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15259 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15261 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15263 * doc/invoke.texi (rcpc architecture extension): Document it.
15265 2017-07-03 Richard Biener <rguenther@suse.de>
15267 PR tree-optimization/60510
15268 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15269 the scalar reduction PHI and use it.
15270 (vectorizable_reduction): Properly guard the single_defuse_cycle
15271 path for non-SLP reduction chains where we cannot use it.
15272 Rework reduc_def/index and vector type deduction. Rework
15273 vector operand gathering during reduction op code-gen.
15274 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15275 chains dissolve the chain and leave it to non-SLP reduction
15278 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15280 * tree-data-ref.h (dr_alignment): Declare.
15281 * tree-data-ref.c (dr_alignment): New function.
15282 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15283 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15285 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15287 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15289 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15290 and base_misalignment fields.
15291 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15292 * tree-data-ref.c: Include builtins.h.
15293 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15294 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15295 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15296 * tree-vect-data-refs.c: Include tree-cfg.h.
15297 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15298 fields instead of calculating an alignment here.
15299 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15300 innermost_loop_behavior fields.
15302 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15304 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15306 (DR_STEP_ALIGNMENT): New macro.
15307 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15308 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15309 (create_data_ref): Print it.
15310 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15311 to tell whether the step preserves vector (mis)alignment.
15312 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15313 Move the check for an integer step and generalise to all INTEGER_CST.
15314 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15315 Print the outer step alignment.
15317 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15319 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15320 with offset_alignment.
15321 (DR_ALIGNED_TO): Delete.
15322 (DR_OFFSET_ALIGNMENT): New macro.
15323 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15324 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15325 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15326 (create_data_ref): Likewise.
15327 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15328 (vect_analyze_data_refs): Likewise.
15329 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15330 creating dummy innermost behavior.
15332 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15334 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15335 with a "innermost_loop_behavior *" and refeence tree.
15336 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15337 (create_data_ref): Update call accordingly.
15338 * tree-predcom.c (find_looparound_phi): Likewise.
15340 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15342 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15343 fields with dr_wrt_vec_loop.
15344 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15345 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15346 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15347 (vect_dr_behavior): New function.
15348 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15349 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15350 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15351 track whether the step preserves the misalignment.
15352 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15353 Use vect_dr_behavior.
15354 (vect_setup_realignment): Update call accordingly.
15355 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15356 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15357 call to vect_create_addr_base_for_vector_ref.
15358 (vect_create_cond_for_align_checks): Likewise.
15359 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15360 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15361 (vect_recog_mask_conversion_pattern): Likewise.
15362 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15363 (new_stmt_vec_info): Remove redundant zeroing.
15365 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15367 * common/config/arm/arm-common.c (arm_be8_option): New function.
15368 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15369 (ISA_ARMv6): Add isa_bit_be8.
15370 * config/arm/arm.h (arm_be8_option): Add prototype.
15371 (BE8_SPEC_FUNCTION): New define.
15372 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15373 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15374 (mlittle-endian): Similarly.
15375 (mbe8, mbe32): New options.
15376 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15377 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15379 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15381 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15385 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15386 (cleanup_tree_cfg_bb): Use it.
15387 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15389 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15391 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15394 * loop-doloop.c (add_test): Update profile.
15396 2017-07-03 Martin Liska <mliska@suse.cz>
15399 * sanopt.c (rewrite_usage_of_param): New function.
15400 (sanitize_rewrite_addressable_params): Likewise.
15401 (pass_sanopt::execute): Call rewrite_usage_of_param.
15403 2017-07-03 Richard Biener <rguenther@suse.de>
15405 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15406 back to using VIEW_CONVERT_EXPR.
15408 2017-07-03 Martin Liska <mliska@suse.cz>
15411 * doc/extend.texi: Document when a resolver function is
15412 generated for target_clones.
15414 2017-07-03 Martin Liska <mliska@suse.cz>
15416 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15417 and shadow memory of auto variables which are subject of
15418 use-after-scope sanitization.
15419 (asan_expand_mark_ifn): Add do set only when is_poison.
15421 2016-07-03 Richard Biener <rguenther@suse.de>
15423 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15425 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15426 (vectorizable_reduction): Perform reduction PHI creation when
15427 visiting a reduction PHI and adjust and simplify code generation
15428 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15429 (vect_transform_loop): Visit reduction PHIs.
15430 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15431 defs into the SLP tree.
15432 (vect_build_slp_tree): Reduction defs terminate the recursion.
15433 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15435 (vect_get_vec_defs_for_stmt_copy): Export.
15436 (vect_get_vec_defs): Likewise.
15437 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15439 (vect_get_vec_defs_for_stmt_copy): Declare.
15440 (vect_get_vec_defs): Likewise.
15442 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15444 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15445 parameter with a "loop" parameter and use it instead of the
15446 loop containing DR_STMT. Don't check simple_iv when doing
15447 BB analysis. Describe the two analysis modes in the comment.
15449 2017-07-03 Tom de Vries <tom@codesourcery.com>
15451 PR tree-optimization/69468
15452 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15453 (find_same_succ_bb): Handle ignore_edge_flags.
15455 2017-07-03 Tom de Vries <tom@codesourcery.com>
15457 PR tree-optimization/81192
15458 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15460 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15462 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15464 2017-07-03 Tom de Vries <tom@codesourcery.com>
15466 PR tree-optimization/81192
15467 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15468 BB_SAME_SUCC (bb) == NULL.
15470 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15472 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15475 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15477 * dumpfile.c: Include profile-count.h
15478 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15480 (insert_cond_bb): Update profile.
15481 * tree-cfg.h (insert_cond_bb): Update prototype.
15482 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15483 * tree-dump.c: Do not include tree-cfg.
15485 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15487 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15489 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15491 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15494 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15496 * tree-complex.c (expand_complex_div_wide): update profile.
15498 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15499 Alan Hayward <alan.hayward@arm.com>
15500 David Sherwood <david.sherwood@arm.com>
15502 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15503 (CORETYPES_H): New define.
15504 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15505 (insn-modes-inline.h, s-modes-inline-h): New rules.
15506 (generated_files): Add insn-modes-inline.h.
15507 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15508 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15509 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15510 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15511 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15512 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15513 (build/gencodes.o, build/genconditions.o): Likewise.
15514 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15515 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15516 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15517 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15518 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15519 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15520 * coretypes.h: Include everything up to real.h for generators.
15521 Include insn-modes.h first. Include wide-int-print.h after
15522 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15523 * machmode.h: Don't include insn-modes.h here.
15524 * function-tests.c: Remove includes of signop.h, machmode.h,
15525 double-int.h and wide-int.h.
15527 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15529 * optc-save-gen.awk: Likewise.
15530 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15531 * godump.c: Remove include of wide-int-print.h.
15532 * pretty-print.h: Likewise.
15533 * wide-int-print.cc: Likewise.
15534 * wide-int.cc: Likewise.
15535 * hash-map-tests.c: Remove include of signop.h.
15536 * hash-set-tests.c: Likewise.
15537 * rtl-tests.c: Likewise.
15538 * mkconfig.sh: Remove include of machmode.h.
15539 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15541 (emit_insn_modes_inline_h): ...this new function. Emit the code
15542 into an insn-modes-inline.h header file, adding appropriate
15543 include guards and end comments.
15544 (emit_insn_modes_c_header): Remove include of machmode.h.
15545 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15547 (main): Handle -i flag and call emit_insn_modes_inline_h when
15550 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15552 * tree-ssa-strlen.c (strinfo): Rename the length field to
15553 nonzero_chars. Add a full_string_p field.
15554 (compare_nonzero_chars, zero_length_string_p): New functions.
15555 (get_addr_stridx): Add an offset_out parameter.
15556 Use compare_nonzero_chars.
15557 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15558 (new_strinfo): Update after above changes to strinfo.
15559 (set_endptr_and_length): Set full_string_p.
15560 (get_string_length): Update after above changes to strinfo.
15561 (unshare_strinfo): Update call to new_strinfo.
15562 (maybe_invalidate): Likewise.
15563 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15564 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15565 as a uhwi instead of an shwi. Update after above changes to
15566 strinfo and new_strinfo.
15567 (zero_length_string): Assert that chainsi contains full strings.
15568 Use zero_length_string_p. Update call to new_strinfo.
15569 (adjust_related_strinfos): Update after above changes to strinfo.
15570 Copy full_string_p from origsi.
15571 (adjust_last_stmt): Use zero_length_string_p.
15572 (handle_builtin_strlen): Update after above changes to strinfo and
15573 new_strinfo. Install the lhs as the string length if the previous
15574 entry didn't describe a full string.
15575 (handle_builtin_strchr): Update after above changes to strinfo
15577 (handle_builtin_strcpy): Likewise.
15578 (handle_builtin_strcat): Likewise.
15579 (handle_builtin_malloc): Likewise.
15580 (handle_pointer_plus): Likewise.
15581 (handle_builtin_memcpy): Likewise. Track nonzero characters
15582 that aren't necessarily followed by a nul terminator.
15583 (handle_char_store): Likewise.
15585 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15587 PR tree-optimization/80769
15588 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15589 for malloc and calloc. Document the new invariant that all related
15590 strinfos have delayed lengths or none do.
15591 (verify_related_strinfos): Move earlier in file.
15592 (set_endptr_and_length): New function, split out from...
15593 (get_string_length): ...here. Also set the lengths of related
15595 (zero_length_string): Assert that chainsi has known (rather than
15597 (adjust_related_strinfos): Likewise.
15599 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15601 PR tree-optimization/81136
15602 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15603 assert that two references with the same misalignment have the same
15604 compile-time misalignment if those compile-time misalignments
15607 2017-07-01 Andi Kleen <ak@linux.intel.com>
15609 * print-tree.c (print_node): Print all attributes.
15611 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15613 * cfg.c (scale_bbs_frequencies): New function.
15614 * cfg.h (scale_bbs_frequencies): Declare it.
15615 * cfgloopanal.c (single_likely_exit): Cleanup.
15616 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15618 (scale_loop_profile): Likewise.
15619 (loop_version): Likewise.
15620 (create_empty_loop_on_edge): Update.
15621 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15622 scale_loop_frequencies, scale_loop_profile, loopify,
15623 loop_version): Update prototypes.
15624 * modulo-sched.c (sms_schedule): Update.
15625 * predict.c (unlikely_executed_edge_p): Also check probability.
15626 (probably_never_executed_edge_p): Fix typo.
15627 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15628 * tree-parloops.c (gen_parallel_loop): Update.
15629 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15631 * tree-ssa-loop-split.c (split_loop): Update.
15632 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15633 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15634 (vect_loop_versioning): Update.
15635 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15637 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15639 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15641 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15643 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15644 to keep profile consistent.
15646 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15648 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15649 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15650 * profile-count.h (max_safe_multiplier): Make unsigned.
15651 (profile_count::guessed_zero): New.
15653 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15655 * bb-reorder.c (fix_up_crossing_landing_pad,
15656 fix_crossing_conditional_branches): Use make_single_succ_edge
15657 to keep profile consistent.
15659 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15661 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15664 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15667 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15668 the right scopes, make sure cond_jump isn't preserved between multiple
15669 iterations. Search for fallthru edge whenever there are 3+ edges and
15670 use find_fallthru_edge for it.
15672 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15674 Patch by Alexander Monakov <amonakov@ispras.ru>
15675 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15676 probabilities consistently.
15678 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15680 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15681 emit_cmp_and_jump_insns.
15683 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15686 * tree-inline.c (expand_call_inline): Combine profile statuses.
15688 2017-06-30 Andrew Pinski <apinski@cavium.com>
15690 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15691 fold_stmt returned true.
15693 2017-06-30 Nathan Sidwell <nathan@acm.org>
15695 * ggc.h (empty_string): Delete.
15696 * cfgexpand.c (expand_asm_stmt): Use plain "".
15697 * optabs.c (expand_asm_memory_barrier): Likewise.
15698 * stringpool.c (empty_string): Delete.
15699 (digit_vector, digit_string): Delete.
15700 (ggc_alloc_string): Use plain "", don't optimize single digit
15701 strings. Use ggc_alloc_atomic.
15703 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15705 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15706 comparison set and one other set, use the cost of the non-comparison
15709 2017-06-30 Nathan Sidwell <nathan@acm.org>
15711 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15714 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15716 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15717 loops. Remove now unneeded calls to gimple_switch_set_label() that
15718 just set removed labels to NULL_TREE.
15720 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15722 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15723 (set_range_info): ...here. Only call set_range_info_raw if domain
15725 (set_nonzero_bits): Call set_range_info_raw.
15726 * tree-ssanames.h (set_range_info_raw): New.
15728 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15731 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15732 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15733 of nonimmediate_operand and <store_mask_constraint> instead of m
15734 for the input operand. For V8FI iterator, always split if input
15735 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15736 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15737 <store_mask_predicate> instead of register_operand and
15738 <store_mask_constraint> instead of v for the input operand. Make
15739 sure both operands aren't MEMs for if not <mask_applied>.
15741 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15743 * lto-wrapper.c (copy_file) Close both file descriptors before
15746 2017-06-30 Martin Liska <mliska@suse.cz>
15749 * multiple_target.c (create_dispatcher_calls): Make ifunc
15750 also for function that don't have calls or are not referenced.
15752 2017-06-30 Richard Biener <rguenther@suse.de>
15754 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15755 analyze the first scalar stmt. Move vector type computation
15756 for the BB case here from ...
15757 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15758 live operation processing in the SLP case properly.
15760 2017-06-30 Richard Biener <rguenther@suse.de>
15762 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15764 2017-06-30 Martin Liska <mliska@suse.cz>
15767 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15768 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15770 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15772 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15775 2017-06-30 Richard Biener <rguenther@suse.de>
15777 PR tree-optimization/81249
15778 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15779 condition reduction result to original scalar type.
15781 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15783 * profile-count.h (enum profile_quality): Fix typos and whitespace
15786 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15788 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15789 type for branch probabilities.
15791 2017-06-29 Julian Brown <julian@codesourcery.com>
15792 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15794 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15795 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15796 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15797 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15799 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15801 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15802 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15803 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15804 CC usage from generic code to here.
15805 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15806 CC usage into the target macros.
15808 2017-06-29 Maya Rashish <coypu@sdf.org>
15810 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15813 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15815 * arm/arm-builtins.c: Include profile-count.h
15816 * except.c (sjlj_emit_function_enter): Use
15817 profile_probability::unlikely.
15819 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15821 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15822 and tocrel_offset be pointer args rather than implicitly using
15824 (legitimate_constant_pool_address_p, rs6000_emit_move,
15825 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15826 tocrel_offset and use in toc_relative_expr_p call.
15827 (print_operand, print_operand_address): Use static tocrel_base_oac
15828 and tocrel_offset_oac.
15829 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15832 2017-06-29 Maya Rashish <coypu@sdf.org>
15834 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15836 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15838 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15839 objects, take into account only the alignment of 'op0' and 'mode1' if
15842 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15844 * ccmp.c (ccmp_tree_comparison_p): New function.
15845 (ccmp_candidate_p): Update to use above function.
15846 (get_compare_parts): New function.
15847 (expand_ccmp_next): Update to use new functions.
15848 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15850 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15851 take mode as argument.
15852 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15853 * expr.c (expand_expr_real_1): Pass mode as argument.
15855 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15857 * combine.c (combine_instructions): Print insns to dump_file, together
15860 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15862 * asan.c (asan_emit_stack_protection): Update.
15863 (create_cond_insert_point): Update.
15864 * auto-profile.c (afdo_propagate_circuit): Update.
15865 * basic-block.h (struct edge_def): Turn probability to
15866 profile_probability.
15867 (EDGE_FREQUENCY): Update.
15868 * bb-reorder.c (find_traces_1_round): Update.
15869 (better_edge_p): Update.
15870 (sanitize_hot_paths): Update.
15871 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15872 (make_single_succ_edge): Update.
15873 (check_bb_profile): Update.
15874 (dump_edge_info): Update.
15875 (update_bb_profile_for_threading): Update.
15876 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15877 probabilitycount to 0.
15878 * cfgbuild.c (compute_outgoing_frequencies): Update.
15879 * cfgcleanup.c (try_forward_edges): Update.
15880 (outgoing_edges_match): Update.
15881 (try_crossjump_to_edge): Update.
15882 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15883 (expand_gimple_tailcall): Update.
15884 (construct_init_block): Use make_single_succ_edge.
15885 (construct_exit_block): Use make_single_succ_edge.
15886 * cfghooks.c (verify_flow_info): Update.
15887 (redirect_edge_succ_nodup): Update.
15888 (split_edge): Update.
15889 (account_profile_record): Update.
15890 * cfgloopanal.c (single_likely_exit): Update.
15891 * cfgloopmanip.c (scale_loop_profile): Update.
15892 (set_zero_probability): Remove.
15893 (duplicate_loop_to_header_edge): Update.
15894 * cfgloopmanip.h (loop_version): Update prototype.
15895 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15896 (force_nonfallthru_and_redirect): Update.
15897 (update_br_prob_note): Update.
15898 (rtl_verify_edges): Update.
15899 (purge_dead_edges): Update.
15900 (rtl_lv_add_condition_to_bb): Update.
15901 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15902 * cgraphunit.c (init_lowered_empty_function): Update.
15903 (cgraph_node::expand_thunk): Update.
15904 * cilk-common.c: Include profile-count.h
15905 * dojump.c (inv): Remove.
15906 (jumpifnot): Update.
15907 (jumpifnot_1): Update.
15908 (do_jump_1): Update.
15910 (do_jump_by_parts_greater_rtx): Update.
15911 (do_compare_rtx_and_jump): Update.
15912 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15913 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15914 * dwarf2cfi.c: Include profile-count.h
15915 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15916 (sjlj_emit_dispatch_table): Likewise.
15917 * explow.c: Include profile-count.h
15918 * expmed.c (emit_store_flag_force): Update.
15919 (do_cmp_and_jump): Update.
15920 * expr.c (compare_by_pieces_d::generate): Update.
15921 (compare_by_pieces_d::finish_mode): Update.
15922 (emit_block_move_via_loop): Update.
15923 (store_expr_with_bounds): Update.
15924 (store_constructor): Update.
15925 (expand_expr_real_2): Update.
15926 (expand_expr_real_1): Update.
15927 * expr.h (try_casesi, try_tablejump): Update prototypes.
15928 * gimple-pretty-print.c (dump_probability): Update.
15929 (dump_profile): New.
15930 (dump_gimple_label): Update.
15931 (dump_gimple_bb_header): Update.
15932 * graph.c (draw_cfg_node_succ_edges): Update.
15933 * hsa-gen.c (convert_switch_statements): Update.
15934 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15935 (find_if_case_1): Update.
15936 (find_if_case_2): Update.
15937 * internal-fn.c (expand_arith_overflow_result_store): Update.
15938 (expand_addsub_overflow): Update.
15939 (expand_neg_overflow): Update.
15940 (expand_mul_overflow): Update.
15941 (expand_vector_ubsan_overflow): Update.
15942 * ipa-cp.c (good_cloning_opportunity_p): Update.
15943 * ipa-split.c (split_function): Use make_single_succ_edge.
15944 * ipa-utils.c (ipa_merge_profiles): Update.
15945 * loop-doloop.c (add_test): Update.
15946 (doloop_modify): Update.
15947 * loop-unroll.c (compare_and_jump_seq): Update.
15948 (unroll_loop_runtime_iterations): Update.
15949 * lra-constraints.c (lra_inheritance): Update.
15950 * lto-streamer-in.c (input_cfg): Update.
15951 * lto-streamer-out.c (output_cfg): Update.
15952 * mcf.c (adjust_cfg_counts): Update.
15953 * modulo-sched.c (sms_schedule): Update.
15954 * omp-expand.c (expand_omp_for_init_counts): Update.
15955 (extract_omp_for_update_vars): Update.
15956 (expand_omp_ordered_sink): Update.
15957 (expand_omp_for_ordered_loops): Update.
15958 (expand_omp_for_generic): Update.
15959 (expand_omp_for_static_nochunk): Update.
15960 (expand_omp_for_static_chunk): Update.
15961 (expand_cilk_for): Update.
15962 (expand_omp_simd): Update.
15963 (expand_omp_taskloop_for_outer): Update.
15964 (expand_omp_taskloop_for_inner): Update.
15965 * omp-simd-clone.c (simd_clone_adjust): Update.
15966 * optabs.c (expand_doubleword_shift): Update.
15967 (expand_abs): Update.
15968 (emit_cmp_and_jump_insn_1): Update.
15969 (expand_compare_and_swap_loop): Update.
15970 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15971 * predict.c (predictable_edge_p): Update.
15972 (edge_probability_reliable_p): Update.
15973 (set_even_probabilities): Update.
15974 (combine_predictions_for_insn): Update.
15975 (combine_predictions_for_bb): Update.
15976 (propagate_freq): Update.
15977 (estimate_bb_frequencies): Update.
15978 (force_edge_cold): Update.
15979 * profile-count.c (profile_count::dump): Add missing space into dump.
15980 (profile_count::debug): Add newline.
15981 (profile_count::differs_from_p): Explicitly convert to unsigned.
15982 (profile_count::stream_in): Update.
15983 (profile_probability::dump): New member function.
15984 (profile_probability::debug): New member function.
15985 (profile_probability::differs_from_p): New member function.
15986 (profile_probability::differs_lot_from_p): New member function.
15987 (profile_probability::stream_in): New member function.
15988 (profile_probability::stream_out): New member function.
15989 * profile-count.h (profile_count_quality): Rename to ...
15990 (profile_quality): ... this one.
15991 (profile_probability): New.
15992 (profile_count): Update.
15993 * profile.c (compute_branch_probabilities): Update.
15994 * recog.c (peep2_attempt): Update.
15995 * sched-ebb.c (schedule_ebbs): Update.
15996 * sched-rgn.c (find_single_block_region): Update.
15997 (compute_dom_prob_ps): Update.
15998 (schedule_region): Update.
15999 * sel-sched-ir.c (compute_succs_info): Update.
16000 * stmt.c (struct case_node): Update.
16001 (do_jump_if_equal): Update.
16002 (get_outgoing_edge_probs): Update.
16003 (conditional_probability): Update.
16004 (emit_case_dispatch_table): Update.
16005 (expand_case): Update.
16006 (expand_sjlj_dispatch_table): Update.
16007 (emit_case_nodes): Update.
16008 * targhooks.c: Update.
16009 * tracer.c (better_p): Update.
16010 (find_best_successor): Update.
16011 * trans-mem.c (expand_transaction): Update.
16012 * tree-call-cdce.c: Update.
16013 * tree-cfg.c (gimple_split_edge): Upate.
16014 (move_sese_region_to_fn): Upate.
16015 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16016 * tree-eh.c (lower_resx): Upate.
16017 (cleanup_empty_eh_move_lp): Upate.
16018 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16019 * tree-inline.c (copy_edges_for_bb): Update.
16020 (copy_cfg_body): Update.
16021 * tree-parloops.c (gen_parallel_loop): Update.
16022 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16023 (gimple_gen_time_profiler): Update.
16024 * tree-ssa-dce.c (remove_dead_stmt): Update.
16025 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16026 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16027 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16028 (unloop_loops): Update.
16029 (try_peel_loop): Update.
16030 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16031 * tree-ssa-loop-split.c (connect_loops): Update.
16032 (split_loop): Update.
16033 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16034 (hoist_guard): Update.
16035 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16036 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16037 (value_replacement): Update.
16038 * tree-ssa-reassoc.c (branch_fixup): Update.
16039 * tree-ssa-tail-merge.c (replace_block_by): Update.
16040 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16041 (create_edge_and_update_destination_phis): Update.
16042 (compute_path_counts): Update.
16043 (recompute_probabilities): Update.
16044 (update_joiner_offpath_counts): Update.
16045 (freqs_to_counts_path): Update.
16046 (duplicate_thread_path): Update.
16047 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16048 (struct switch_conv_info): Update.
16049 (gen_inbound_check): Update.
16050 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16051 (vect_do_peeling): Update.
16052 (vect_loop_versioning): Update.
16053 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16054 (optimize_mask_stores): Update.
16055 * ubsan.c (ubsan_expand_null_ifn): Update.
16056 * value-prof.c (gimple_divmod_fixed_value): Update.
16057 (gimple_divmod_fixed_value_transform): Update.
16058 (gimple_mod_pow2): Update.
16059 (gimple_mod_pow2_value_transform): Update.
16060 (gimple_mod_subtract): Update.
16061 (gimple_mod_subtract_transform): Update.
16062 (gimple_ic): Update.
16063 (gimple_stringop_fixed_value): Update.
16064 (gimple_stringops_transform): Update.
16065 * value-prof.h: Update.
16067 2017-06-29 Carl Love <cel@us.ibm.com>
16069 * config/rs6000/rs6000-c.c: Add support for built-in functions
16070 vector signed int vec_signed (vector float);
16071 vector signed long long vec_signed (vector double);
16072 vector signed int vec_signed2 (vector double, vector double);
16073 vector signed int vec_signede (vector double);
16074 vector signed int vec_signedo (vector double);
16075 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16076 instruction generator.
16077 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16078 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16079 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16081 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16082 vunsignede_v2df): Add define_expands.
16083 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16084 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16085 VEC_UNSIGNEDO): Add definitions.
16086 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16087 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16088 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16089 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16090 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16091 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16092 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16093 vec_unsignede, vec_unsignedo): Add builtin defines.
16094 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16096 * doc/extend.texi: Update the built-in documentation file for the
16097 new built-in functions.
16099 2017-06-29 Richard Biener <rguenther@suse.de>
16101 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16102 reduction chains to LOOP_VINFO_REDUCTIONS.
16103 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16104 SLP reductions after processing reduction chains.
16106 2017-06-29 Nathan Sidwell <nathan@acm.org>
16108 * builtins.c (fold_builtin_FUNCTION): Use
16109 lang_hooks.decl_printable_name.
16111 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16113 PR middle-end/81194
16114 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16115 with only one label.
16116 * stmt.c (expand_case): Assert NCASES is greater than one.
16118 2017-06-29 Richard Biener <rguenther@suse.de>
16120 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16122 (group_case_labels): Likewise.
16123 (find_taken_edge): Push sanity checking on val to workers...
16124 (find_taken_edge_cond_expr): ... here
16125 (find_taken_edge_switch_expr): ... and here, handle cases
16126 with just a default label.
16127 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16128 (group_case_labels): Likewise.
16129 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16130 group_case_labels does anything cleanup the CFG again.
16132 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16134 PR tree-optimization/81196
16135 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16136 exit condition comparing two IVs.
16138 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16140 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16141 profile to the dummy entry at the end of the list of architectures.
16142 * config/arm/arm-cpu-cdata.h: Regenerated.
16144 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16145 Michael Collison <michael.collison@arm.com>
16148 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16150 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16151 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16152 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16153 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16154 of shift when the shift amount is masked with constant equal to
16155 the size of the mode.
16156 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16159 2017-06-29 Martin Liska <mliska@suse.cz>
16161 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16163 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16166 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16169 (edge_growth_cache_entry::edge_growth_cache_entry): New
16171 (reset_edge_growth_cache): Update to use constructor.
16173 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16175 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16176 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16177 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16179 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16181 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16182 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16184 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16186 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16187 (*-linux-uclibc*): Add t-uclibc tmake_file.
16188 * config/t-musl: New.
16189 * config/t-uclibc: New.
16191 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16193 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16195 (gen_comm_data): Emit architectural setting of arch_prof.
16196 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16198 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16199 (armv8-m.base, armv8-m.main): Likewise.
16200 * arm-protos.h (arm_build_target): Add profile field.
16201 (arch_option): Likewise.
16202 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16204 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16205 arm_active_target.profile.
16207 2017-06-28 Richard Biener <rguenther@suse.de>
16209 PR middle-end/81227
16210 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16211 TYPE_OVERFLOW_WRAPS.
16212 * match.pd (negate_expr_p): Likewise.
16213 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16214 fold_build2, not fold_binary.
16216 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16218 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16219 Convert memory address to Pmode.
16220 (aarch64_print_operand): Assert MEM operands are always Pmode.
16222 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16225 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16226 Remove redundant if.
16227 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16228 * config/arm/aarch-common-protos.h
16229 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16230 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16232 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16235 * multiple_target.c (create_dispatcher_calls): Set the default
16236 clone to be static, not public.
16238 2017-06-28 Richard Biener <rguenther@suse.de>
16240 * tree-vect-loop.c (vectorizable_reduction): Move special
16241 cond reduction IV var creation ...
16242 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16243 parameter. Use STMT_VINFO_VECTYPE.
16244 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16247 2017-06-28 Martin Liska <mliska@suse.cz>
16250 * ipa-visibility.c (non_local_p): Handle visibility.
16252 2017-06-28 Martin Liska <mliska@suse.cz>
16255 * common.opt: Add IntegerRange to various options.
16256 * opt-functions.awk (integer_range_info): New function.
16257 * optc-gen.awk: Add integer_range_info to cl_options struct.
16258 * opts-common.c (decode_cmdline_option): Handle
16259 CL_ERR_INT_RANGE_ARG.
16260 (cmdline_handle_error): Likewise.
16261 * opts.c (print_filtered_help): Show valid interval in
16262 when --help is provided.
16263 * opts.h (struct cl_option): Add range_min and range_max fields.
16264 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16266 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16268 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16269 (x * C EQ/NE y * C): New transformation.
16271 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16273 * genmultilib (combination_space): Accept '+' in option names.
16275 2017-06-28 Martin Liska <mliska@suse.cz>
16278 * asan.c (instrument_derefs): Bail out inner references
16279 that are hard register variables.
16281 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16284 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16285 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16287 2017-06-28 Richard Biener <rguenther@suse.de>
16289 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16290 (vect_get_slp_defs): Adjust.
16291 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16293 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16295 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16296 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16297 (vectorizable_reduction): Adjust.
16298 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16300 (vect_get_slp_defs): Likewise.
16301 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16302 (vectorizable_bswap): Adjust.
16303 (vectorizable_call): Likewise.
16304 (vectorizable_conversion): Likewise.
16305 (vectorizable_assignment): Likewise.
16306 (vectorizable_shift): Likewise.
16307 (vectorizable_operation): Likewise.
16308 (vectorizable_store): Likewise.
16309 (vectorizable_condition): Likewise.
16310 (vectorizable_comparison): Likewise.
16312 2017-06-28 Michael Collison <michael.collison@arm.com>
16315 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16317 (arm_gen_movmemqi): Removed unused variable 'i'.
16318 Convert 'for' loop into 'while' loop.
16319 (arm_expand_prologue): Remove last unnecessary set of insn.
16320 (thumb_pop): Remove unused variable 'pushed_words'.
16321 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16323 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16325 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16326 * config/s390/s390-protos.h: Add prototype of
16327 s390_rel_address_ok_p.
16328 * config/s390/s390.c (s390_got_symbol): New function.
16329 (s390_rel_address_ok_p): New function.
16330 (legitimize_pic_address): Use s390_rel_address_ok_p.
16331 (s390_load_got): Use s390_got_symbol.
16332 (s390_option_override): Issue error if
16333 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16334 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16336 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16338 2017-06-27 Andrew Pinski <apinski@cavium.com>
16340 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16341 (X * copysign (1.0, X)): New pattern.
16342 (X * copysign (1.0, -X)): New pattern.
16343 (copysign (-1.0, CST)): New pattern.
16345 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16347 * genmultilib (combination_space): Remove variable.
16348 Validate reuse rules against regular expression for any sequence
16349 of multilib options in any order.
16351 2017-06-27 Michael Collison <michael.collison@arm.com>
16353 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16354 call aarch64_split_simd_combine.
16355 * (aarch64_combine_internal<mode>): Delete pattern.
16356 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16357 Allow register and subreg operands.
16359 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16361 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16362 specific need, just fallback on defaults.
16363 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16365 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16366 Olivier Hainque <hainque@adacore.com>
16368 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16370 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16371 targets. Pick a default if no particular attempt applied.
16372 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16375 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16377 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16378 (x86_64-wrs-vxworks7): Likewise.
16380 2017-06-27 Marek Polacek <polacek@redhat.com>
16383 * ubsan.c (instrument_null): Check get_base_address's result for null.
16385 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16387 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16389 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16391 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16392 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16393 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16394 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16395 New function types.
16396 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16397 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16398 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16399 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16400 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16401 BUILT_IN_FEUPDATEENV): New builtins.
16402 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16403 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16404 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16405 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16407 (builtin_structptr_types): Adjust size.
16408 * tree.c (builtin_structptr_types): Add four entries.
16410 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16411 Olivier Hainque <hainque@adacore.com>
16413 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16414 (TLS_SYM): New local macro, forcing reference to __tls__ on
16415 link command lines for VxWorks 7 RTPs, triggering initialization
16417 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16418 OS features TLS support, true for RTPs on VxWorks 7.
16419 * config/vxworks.c (vxworks_override_options): Setup emutls
16422 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16424 * predict.c (test_prediction_value_range): Use -1U instead of -1
16425 to avoid narrowing conversion warning.
16426 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16427 to avoid narrowing conversion warning.
16428 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16430 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16432 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16434 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16435 64bit configurations.
16436 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16437 (SIZE_TYPE): Likewise.
16438 * config/vxworks.c (vxworks_emutls_var_fields): Use
16439 long_unsigned_type_node instead of unsigned_type_node as the offset
16440 field type, which is "pointer" mode in emutls.c.
16442 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16445 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16447 PR middle-end/81207
16448 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16449 gimple_vuse copying separately from gimple_vdef copying.
16451 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16453 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16456 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16457 Olivier Hainque <hainque@adacore.com>
16459 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16460 for all vxworks7 targets.
16461 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16462 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16463 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16464 variations for VX6/VX7 and 32/64bits later on in ...
16465 (VXWORKS_LIB_SPEC): Leverage new macros.
16466 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16467 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16469 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16471 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16472 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16474 2017-06-26 Carl Love <cel@us.ibm.com>
16476 * config/rs6000/rs6000-c.c: Add support for built-in functions
16477 vector bool char vec_reve (vector bool char);
16478 vector signed char vec_reve (vector signed char);
16479 vector unsigned char vec_reve (vector unsigned char);
16480 vector bool int vec_reve (vector bool int);
16481 vector signed int vec_reve (vector signed int);
16482 vector unsigned int vec_reve (vector unsigned int);
16483 vector bool long long vec_reve (vector bool long long);
16484 vector signed long long vec_reve (vector signed long long);
16485 vector unsigned long long vec_reve (vector unsigned long long);
16486 vector bool short vec_reve (vector bool short);
16487 vector signed short vec_reve (vector signed short);
16488 vector double vec_reve (vector double);
16489 vector float vec_reve (vector float);
16490 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16491 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16492 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16493 (altivec_vreve): New pattern.
16494 * config/rs6000/altivec.h (vec_reve): New define.
16495 * doc/extend.texi (vec_rev): Update the built-in documentation file
16496 for the new built-in functions.
16498 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16500 PR tree-optimization/71815
16501 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16503 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16505 (slsr_process_phi): Likewise.
16506 (replace_uncond_cands_and_profitable_phis): Don't replace a
16507 multiply candidate with a stride of 1 (copy or cast).
16508 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16510 (lowest_cost_path): Likewise.
16511 (total_savings): Likewise.
16513 2017-06-26 Richard Biener <rguenther@suse.de>
16516 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16517 Use def_builtin_pure for all gather builtins.
16519 2017-06-26 Richard Biener <rguenther@suse.de>
16521 PR tree-optimization/81203
16522 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16523 non-dominating BBs.
16525 2017-06-26 Marek Polacek <polacek@redhat.com>
16528 * doc/invoke.texi: Document -Wmultistatement-macros.
16530 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16532 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16533 arm_neon_ok_no_float_abi effective target.
16535 2017-06-26 Richard Biener <rguenther@suse.de>
16537 PR tree-optimization/80928
16538 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16539 (copy_bbs): Set BB_DUPLICATED flag early.
16540 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16542 (execute_on_shrinking_pred): Likewise.
16543 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16544 BB_DUPLICATED blocks.
16545 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16546 iterate over all PHIs considering removal of *gsi.
16548 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16550 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16553 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16555 * config/rs6000/rs6000-string.c: (expand_block_clear,
16556 do_load_for_compare, select_block_compare_mode,
16557 compute_current_alignment, expand_block_compare,
16558 expand_strncmp_align_check, expand_strn_compare,
16559 expand_block_move, rs6000_output_load_multiple)
16560 Move functions related to string/block move/compare
16561 to a separate file.
16562 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16563 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16564 for this function which is now used in two files.
16565 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16566 * config.gcc: Add rs6000-string.o to extra_objs for
16567 targets powerpc*-*-* and rs6000*-*-*.
16569 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16572 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16573 32-bit, since indexed is not valid for DImode.
16574 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16575 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16576 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16577 (define_peephole2 for Altivec d-form store): Likewise.
16580 * multiple_target.c (create_dispatcher_calls): Only create the
16581 dispatcher call if the function is the default clone of a
16582 versioned function.
16584 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16586 PR middle-end/80902
16587 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16588 a call, force the call to not be a tail call.
16590 2017-06-23 Jeff Law <law@redhat.com>
16592 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16595 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16597 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16598 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16599 handling for early expansion of vector shifts (sl,sr,sra,rl).
16600 (builtin_function_type): Add vector shift right instructions
16601 to the unsigned argument list.
16603 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16605 rtl-optimizatoin/79286
16606 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16607 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16608 trap. PIC register plus a const unspec without offset can never trap.
16610 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16612 * tree.h (builtin_structptr_type): New type.
16613 (builtin_structptr_types): Declare new array.
16614 * tree.c (builtin_structptr_types): New array.
16615 (free_lang_data, build_common_tree_nodes): Use it.
16617 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16620 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16623 2017-06-22 Matt Turner <mattst88@gmail.com>
16625 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16626 Lake models to skylake case. Assume skylake for unknown
16627 models with clflushopt.
16629 2017-06-22 Jeff Law <law@redhat.com>
16631 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16632 frame sizes that do not satisfy aarch64_uimm12_shift.
16634 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16636 * profile-count.h (apply_probability,
16637 apply_scale, probability_in): Fix checks for zero.
16639 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16641 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16642 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16644 2016-06-22 Richard Biener <rguenther@suse.de>
16646 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16647 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16648 REDUC_MAX_EXPR support.
16649 (vectorizable_reduction): Likewise.
16650 (vect_create_epilog_for_reduction): Likewise.
16652 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16654 * match.pd (A / (1 << B) -> A >> B): New.
16655 * generic-match-head.c: Include optabs-tree.h.
16656 * gimple-match-head.c: Likewise.
16657 * optabs-tree.h (target_supports_op_p): New.
16658 * optabs-tree.c (target_supports_op_p): New.
16660 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16662 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16663 $gcc_cv_ld --help output.
16664 (gcc_cv_ld_demangle): Likewise.
16665 (gcc_cv_ld_eh_frame_hdr): Likewise.
16666 (gcc_cv_ld_pie): Likewise.
16667 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16668 (gcc_cv_ld_buildid): Likewise.
16669 (gcc_cv_ld_sysroot): Likewise.
16670 (ld_bndplt_support): Likewise.
16671 (ld_pushpopstate_support): Likewise.
16672 * configure: Regenerate.
16673 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16675 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16678 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16679 operands indexes to avoid gap between operands and match_dups.
16681 2017-06-21 Andrew Pinski <apinski@cavium.com>
16683 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16684 Increment Arith_shift and Arith_shift_reg by 1.
16685 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16687 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16688 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16689 (aarch64_strip_extend): Add new argument and test for it.
16690 (aarch64_cheap_mult_shift_p): New function.
16691 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16692 add a cost if it is true.
16693 Update calls to aarch64_strip_extend.
16694 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16696 2017-06-21 Andrew Pinski <apinski@cavium.com>
16698 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16700 (thunderxt88): Likewise.
16701 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16702 (thunderx_prefetch_tune): New variable.
16703 (thunderx2t99_prefetch_tune): Update for the correct values.
16704 (thunderxt88_tunings): New variable.
16705 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16706 generic_prefetch_tune.
16707 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16709 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16711 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16712 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16713 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16714 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16715 (aarch64_atomic_cas<mode>, GPI): Likewise.
16717 2017-06-21 Martin Liska <mliska@suse.cz>
16719 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16720 statements on cold and hot labels.
16721 * predict.c (tree_estimate_probability_bb): Remove the
16722 prediction from this place.
16724 2017-06-21 Martin Liska <mliska@suse.cz>
16726 PR tree-optimization/79489
16727 * gimplify.c (maybe_add_early_return_predict_stmt): New
16729 (gimplify_return_expr): Call the function.
16730 * predict.c (tree_estimate_probability_bb): Remove handling
16732 * predict.def: Update comment about early return predictor.
16733 * gimple-predict.h (is_gimple_predict): New function.
16734 * predict.def: Change default value of early return to 66.
16735 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16737 * passes.def: Put pass_strip_predict_hints to the beginning of
16740 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16742 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16743 FUNCTION_DECL declarations.
16744 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16746 (dwaf2out_decl): Likewise.
16747 * godump.c (go_early_global_decl): Skip call to the real debug hook
16748 for FUNCTION_DECL declarations.
16749 * passes.c (rest_of_decl_compilation): Skip call to the
16750 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16751 -fdump-go-spec is passed.
16753 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16755 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16756 Reorder for compactness.
16757 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16758 (def_builtin_pure, def_builtin_pure2): New functions.
16759 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16761 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16763 * match.pd (nop_convert): New predicate.
16764 ((A +- CST1) +- CST2): Allow some NOP conversions.
16766 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16769 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16770 with ctors/dtors if GOVD_SHARED is set.
16772 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16774 * config/aarch64/aarch64.md (movti_aarch64):
16775 Emit mov rather than orr.
16776 (movtf_aarch64): Likewise.
16777 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16778 Emit mov rather than orr.
16780 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16782 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16783 Swap alternatives, make integer dup more expensive.
16785 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16787 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16788 Return true for non-tls symbols.
16790 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16792 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16793 (cortex-a75): Likewise.
16794 (cortex-a75.cortex-a55): Likewise.
16795 * config/aarch64/aarch64-tune.md: Regenerate.
16796 * doc/invoke.texi (-mtune): Document new values for -mtune.
16798 2017-06-21 Tom de Vries <tom@codesourcery.com>
16800 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16801 stack_size feature.
16802 (Effective-Target Keywords, Other attributes): Suggest using
16803 dg-add-options stack_size feature to get stack limit in stack_size
16804 effective target documentation.
16806 2017-06-21 Julian Brown <julian@codesourcery.com>
16807 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16809 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16810 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16811 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16813 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16814 attribute type list for neon_multiply.
16815 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16816 attribute type list for neon_multiply.
16817 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16818 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16819 attribute type list for neon_multiply.
16820 * config/arm/types.md (crypto_pmull): Add.
16821 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16822 attribute type list.
16824 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16826 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16829 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16831 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16832 to make sure not to dereference a NULL cost_classes_ptr pointer.
16834 2017-06-20 Carl Love <cel@us.ibm.com>
16836 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16837 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16838 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16840 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16841 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16842 VMULOSW): New enum "unspec" values.
16843 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16844 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16845 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16846 altivec_vmulosw): New patterns.
16847 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16848 VMULOSW): Add definitions.
16850 2017-06-20 Julia Koval <julia.koval@intel.com>
16852 * config/i386/i386.c: Fix rounding expand for new pattern.
16853 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16855 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16857 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16858 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16860 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16862 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16865 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16867 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16868 architecture, then by implementer ID.
16869 * config/aarch64/aarch64-tune.md: Regenerate.
16871 2017-06-20 Richard Biener <rguenther@suse.de>
16873 PR middle-end/81097
16874 * fold-const.c (split_tree): Fold to type before negating.
16876 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16878 * diagnostic-show-locus.c
16879 (selftest::test_fixit_deletion_affecting_newline): New function.
16880 (selftest::diagnostic_show_locus_c_tests): Call it.
16882 2017-06-20 Andreas Schwab <schwab@suse.de>
16885 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16888 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16890 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16891 __ARM_FEATURE_COPROC according to support.
16893 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16895 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16896 Rewritten to avoid overflow for > 32-bit pointers.
16899 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16900 by removing enum keyword.
16901 (ubsan_type_descriptor): Likewise. Formatting fix.
16904 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16905 splitter): Require TARGET_SSE2 in the condition.
16907 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16910 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16911 for doing vector set of SFmode on ISA 3.0.
16912 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16913 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16915 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16916 SFmode value into a V4SF variable that was extracted from another
16917 V4SF variable without converting the element to double precision
16918 and back to single precision vector format.
16919 (vsx_insert_extract_v4sf_p9_2): Likewise.
16921 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16923 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16924 in UWHI to avoid undefined overflow.
16927 * ubsan.h (enum ubsan_encode_value_phase): New.
16928 (ubsan_encode_value): Change second argument to
16929 enum ubsan_encode_value_phase with default value of
16930 UBSAN_ENCODE_VALUE_GENERIC.
16931 * ubsan.c (ubsan_encode_value): Change second argument to
16932 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16933 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16934 create_tmp_var_raw instead of create_tmp_var and use a
16936 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16937 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16938 ubsan_encode_value callers.
16941 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16942 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16943 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16945 2017-06-19 Richard Biener <rguenther@suse.de>
16947 PR middle-end/81118
16948 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16949 estimates if we changed anything.
16951 2017-06-19 Richard Biener <rguenther@suse.de>
16953 PR tree-optimization/80887
16954 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16955 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16956 simplified lookups, then reset mprts_hook.
16957 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16959 (try_to_simplify): Likewise.
16961 2017-06-19 Martin Liska <mliska@suse.cz>
16964 * gimplify.c (gimplify_switch_expr):
16965 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16967 2017-06-19 Martin Liska <mliska@suse.cz>
16969 * doc/install.texi: Document that PGO runs in 4 stages.
16971 2017-06-19 Martin Liska <mliska@suse.cz>
16974 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16975 to dispatcher function name.
16976 * multiple_target.c (replace_function_decl): New function.
16977 (create_dispatcher_calls): Redirect both edges and references.
16979 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16981 * profile-count.c (profile_count::dump): Dump quality.
16982 (profile_count::differs_from_p): Update for unsigned val.
16983 * profile-count.h (profile_count_quality): New enum.
16984 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16986 2017-06-19 Richard Biener <rguenther@suse.de>
16988 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16989 struct function as arg.
16990 (estimate_numbers_of_iterations): Export overload with loop arg.
16991 (free_numbers_of_iterations_estimates_loop): Use an overload of
16992 free_numbers_of_iterations_estimates instead.
16993 * tree-cfg.c (remove_bb): Adjust.
16994 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16995 * tree-parloops.c (gen_parallel_loop): Likewise.
16996 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16998 (tree_unroll_loops_completely): Likewise.
16999 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17000 Use an overload instead and export.
17001 (estimated_loop_iterations): Adjust.
17002 (max_loop_iterations): Likewise.
17003 (likely_max_loop_iterations): Likewise.
17004 (estimate_numbers_of_iterations): Take struct function as arg
17006 (loop_exits_before_overflow): Adjust.
17007 (free_numbers_of_iterations_estimates_loop): Use an overload.
17008 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17009 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17011 2017-06-19 Richard Biener <rguenther@suse.de>
17014 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17015 RANGE_EXPR conservatively.
17017 2017-06-16 Carl Love <cel@us.ibm.com>
17019 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17020 definitions for vec_float, vec_float2, vec_floato,
17021 vec_floate built-ins.
17022 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17023 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17025 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17026 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17027 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17028 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17029 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17030 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17031 vec_floato): Add builtin defines.
17032 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17033 Update the built-in documentation file for the new built-in
17036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17038 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17039 (mthumb): Mark as the negative of -marm.
17041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17043 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17045 (ARM Options, -mtune): Document that this accepts the same
17046 extension options as -mcpu.
17047 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17049 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17051 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17052 permitted extensions.
17054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17056 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17057 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17058 (armv8-m.main): Add option +nodsp.
17059 * config/arm/arm-cpu-cdata.h: Regenerated.
17061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17063 * config/arm/t-fuchsia: New file.
17064 * config.gcc (arm*-*-fuchsia*): Use it.
17066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17068 * config/arm/t-symbian: Rewrite for new option infrastructure.
17070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17072 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17073 (MULTILIB_REQUIRED): Likewise.
17075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17077 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17078 (MULTILIB_RESUE): Likewise.
17079 (MULTILIB_MATCHES): Likewise.
17080 (MULTLIB_REQUIRED): Likewise.
17082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17084 * config/arm/t-rtems: Rewrite for new option framework.
17086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17088 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17089 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17090 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17091 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17092 * config/arm/t-multilib: ... here.
17093 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17094 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17095 armv7-a and armv8*-a when A-profile libraries have not been built.
17096 * config/arm/t-rmprofile: Rewrite.
17098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17100 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17101 with a backslash. Remove the backslash after substituting unescaped
17103 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17107 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17108 the driver to pass through the default setting of -mfloat-abi.
17109 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17111 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17112 (all_feat_combs): New rule.
17113 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17115 * config/arm/t-aprofile: Rewrite.
17117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17119 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17120 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17121 fpu is not specified by the user/command-line.
17122 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17123 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17124 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17125 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17126 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17127 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17131 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17132 * config/arm/t-arm-elf: Rewritten.
17134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17136 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17137 have some floating-point instructions.
17138 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17139 (TARGET_MAYBE_HARD_FLOAT): New macro.
17140 * config/arm/arm-builtins.c (arm_init_builtins): Use
17141 TARGET_MAYBE_HARD_FLOAT.
17142 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17146 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17147 (configargs.h): Include it.
17148 (arm_print_hint_for_fpu_option): New function.
17149 (arm_parse_fpu_option): New function.
17150 (candidate_extension): New class.
17151 (arm_canon_for_multilib): New function.
17152 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17153 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17154 (ARCH_CANONICAL_SPECS): New macro.
17155 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17159 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17160 are set after handling multilib fragments. Set target_cpu_default2
17163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17165 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17167 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17168 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17170 (check_arch): Likewise.
17171 * config/arm/arm.c (arm_configure_build_target): Handle
17172 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17173 options in the default.
17175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17177 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17178 when an option is an alias of another.
17179 * config/arm/parsecpu.awk (optalias): New parser token.
17180 (gen_comm_data): Mark non-alias options as such. Emit entries
17181 for extension aliases.
17182 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17183 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17184 (armv6kz, armv6zk, armv6t2): Likewise.
17185 (armv7): Make vfpv3-d16 an alias.
17186 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17188 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17189 Sort in canonical order.
17190 (armv8-a): Sort in canonical order.
17191 (armv8.1-a, armv8.2-a): Likewise.
17192 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17194 (cortex-a9): Sort in canonical order.
17195 * config/arm/arm.c (selftests.h): Include it.
17196 (arm_test_cpu_arch_data): New function.
17197 (arm_run_self_tests): New function.
17198 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17199 (targetm): Move declaration to the end of the file.
17200 * arm-cpu-cdata.h: Regenerated.
17202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17204 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17205 call to target_mode_check describing the type of option passed.
17206 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17207 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17208 arm_parse_cpu_option_name to match parameters against list of
17210 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17211 arm_arch_core_flags data structure.
17212 * config/arm/arm-cpu_cdata.h: Regenerated.
17214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17216 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17218 (arm_print_hint_for_cpu_option): Likewise.
17219 (arm_print_hint_for_arch_option): Likewise.
17220 (arm_parse_cpu_option_name): Likewise.
17221 (arm_parse_arch_option_name): Likewise.
17222 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17223 of entries in the all_fpus list.
17224 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17225 (arm_parse_cpu_option_name): Declare.
17226 (arm_parse_arch_option_name): Declare.
17227 (arm_parse_option_features): Declare.
17228 (arm_intialize_isa): Declare.
17229 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17231 (gen_comm_data): ... here. Make definitions non-static.
17232 * config/arm/arm-cpu-data.h: Regenerated.
17233 * config/arm/arm-cpu-cdata.h: Regenerated.
17235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17237 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17238 (cpu_arch_extension): New structure.
17239 (cpu_arch_option, arch_option, cpu_option): New structures.
17240 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17241 architecture types.
17242 (gen_data): Generate new format data tables.
17243 * config/arm/arm.c (cpu_tune): New structure.
17244 (cpu_option, processors): Delete.
17245 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17246 (arm_print_hint_for_cpu_option): ... this and ...
17247 (arm_print_hint_for_arch_option): ... this.
17248 (arm_parse_arch_cpu_name): Delete. Replace with ...
17249 (arm_parse_cpu_option_name): ... this and ...
17250 (arm_parse_arch_option_name): ... this.
17251 (arm_unrecognized_feature): Change type of target parameter to
17253 (arm_parse_arch_cpu_features): Delete. Replace with ...
17254 (arm_parse_option_features): ... this.
17255 (arm_configure_build_target): Rework to use new configuration data
17257 (arm_print_tune_info): Rework for new configuration data tables.
17258 * config/arm/arm-cpu-data.h: Regenerated.
17259 * config/arm/arm-cpu.h: Regenerated.
17261 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17263 * Makefile.in (OBJS): Move sbitmap.o from here ...
17264 (OBJS-libcommon): ... to here.
17266 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17268 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17269 (ISA_ALL_CRYPTO): New macro.
17270 (ISA_ALL_SIMD): New macro
17271 (ISA_ALL_FP): New macro.
17272 * config/arm/arm.c (fpu_bitlist): Update initializer.
17273 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17275 (arm9e): Add fpu. Add option for nofp
17276 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17277 (arm926ej-s, arm1026ej-s): Likewise.
17278 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17279 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17280 neon-fp16, neon-vfpv4, nofp and nosimd.
17281 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17282 (cortex-a8): Add fpu. Add option for nofp.
17283 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17284 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17285 (cortex-r4f): Add fpu.
17286 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17287 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17289 (cortex-r8): Likewise.
17290 (cortex-m4): Add fpu. Add option for nofp.
17291 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17292 (cortex-a17.cortex-a7): Likewise.
17293 (cortex-a32): Add fpu. Add options for crypto and nofp.
17294 (cortex-a35, cortex-a53): Likewise.
17295 (cortex-a57): Add fpu. Add option for crypto.
17296 (cortex-a72, cortex-a73): Likewise.
17297 (exynos-m1): Likewise.
17298 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17299 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17300 (cortex-m33): Add fpu. Add option for nofp.
17301 * config/arm/arm-cpu-cdata.h: Regenerated
17302 * config/arm/arm-cpu-data.h: Regenerated.
17304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17306 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17307 (armv5te, armv5tej): Likewise.
17308 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17309 (armv7): Add options fp and vfpv3-d16.
17310 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17311 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17313 (armv7ve): Likewise.
17314 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17315 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17316 (armv8-a): Add nocrypto option.
17317 (armv8.1-a, armv8.2-a): Likewise.
17318 (armv8-m.main): add options fp, fp.dp and nofp.
17320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17322 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17324 (armv8-a+crc): Delete.
17325 (armv8.1-a): Add options simd, crypto and nofp.
17326 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17327 (armv8.2-a+fp16): Delete.
17328 (armv8-m.main): Add option dsp.
17329 (armv8-m.main+dsp): Delete.
17330 (cortex-a8): Add fpu. Add nofp option.
17331 (cortex-a9): Add fpu. Add nofp and nosimd options.
17332 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17333 link to main cpu and architecture data structures.
17334 (gen_comm_data): Only put isa attributes from the main architecture
17336 (option): New statement for architecture and CPU entries.
17337 * arm.c (struct cpu_option): New structure.
17338 (struct processors): Add entry for options.
17339 (arm_unrecognized_feature): New function.
17340 (arm_parse_arch_cpu_name): Ignore any characters after the first
17342 (arm_parse_arch_cpu_feature): New function.
17343 (arm_configure_build_target): Separate out any CPU and architecture
17344 features and parse separately. Don't error out if -mfpu=auto is
17345 used with only an architecture string.
17346 (arm_print_asm_arch_directives): New function.
17347 (arm_file_start): Call it.
17348 * config/arm/arm-cpu-cdata.h: Regenerated.
17349 * config/arm/arm-cpu-data.h: Likewise.
17350 * config/arm/arm-tables.opt: Likewise.
17352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17354 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17355 assembler when it is not -mfpu=auto.
17357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17359 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17360 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17361 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17362 (ASM_CPU_SPEC): Rewrite.
17363 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17364 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17365 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17366 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17367 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17368 copied string is NUL-terminated. Also strip any characters prefixed
17370 (arm_rewrite_selected_arch): New function.
17371 (arm_rewrite_march): New function.
17373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17375 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17376 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17377 (march, mcpu, mtune): Convert to string-based options.
17378 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17379 (arm_parse_arch_cpu_name): New function.
17380 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17381 identify selected architecture or CPU.
17382 (arm_option_save): New function.
17383 (TARGET_OPTION_SAVE): Redefine.
17384 (arm_option_restore): Restore string options.
17385 (arm_option_print): Print string options.
17387 2017-06-16 Martin Sebor <msebor@redhat.com>
17389 PR tree-optimization/80933
17390 PR tree-optimization/80934
17391 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17392 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17393 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17394 (gimple_fold_builtin): Call them.
17396 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17398 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17399 as unlikely; update profile.
17401 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17403 * predict.c (force_edge_cold): Handle declaring edges impossible
17406 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17408 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17410 (try_unroll_loop_completely): Fix reporting.
17412 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17414 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17416 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17419 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17420 if given a non-constant argument for an intrinsic which requires a
17423 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17425 * profile.c (compare_freqs): New function.
17426 (branch_prob): Sort edge list.
17427 (find_spanning_tree): Assume that the list is priority sorted.
17429 2017-06-16 Richard Biener <rguenther@suse.de>
17431 PR tree-optimization/81090
17432 * passes.def (pass_record_bounds): Remove.
17433 * tree-pass.h (make_pass_record_bounds): Likewise.
17434 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17435 make_pass_record_bounds): Likewise.
17436 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17437 not free niter estimates at the beginning but at the end.
17438 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17440 2017-06-16 Richard Biener <rguenther@suse.de>
17442 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17443 initializer to workaround ICE in host GCC 4.8.
17445 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17447 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17449 (clone_inlined_nodes): Update.
17451 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17453 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17454 prefetch settings, and enable prefetching by default at -O3.
17456 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17458 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17459 Set flag_prefetch_loop_arrays according to tuning data.
17461 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17463 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17464 New tune structure.
17465 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17466 [Unrelated to main purpose of the patch] Place the pointer field last
17467 to enable type checking errors when tune structure are wrongly merged.
17468 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17469 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17470 (thunderx2t99_prefetch_tune): New tune constants.
17471 (tune_params *_tunings): Update all tunings (no functional change).
17472 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17473 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17474 from tunings structures.
17476 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17479 * ubsan.c (instrument_null): Add T argument, use it instead
17480 of computing it based on IS_LHS.
17481 (instrument_object_size): Likewise.
17482 (pass_ubsan::execute): Adjust instrument_null and
17483 instrument_object_size callers to pass gimple_get_lhs or
17484 gimple_assign_rhs1 result to it. Use instrument_null instead of
17485 calling get_base_address and instrument_mem_ref. Handle
17486 aggregate call arguments for object-size sanitization.
17488 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17490 PR tree-optimization/81089
17491 * tree-vrp.c (is_masked_range_test): Validate operands of
17494 2017-06-15 Martin Sebor <msebor@redhat.com>
17497 * dumpfile.c (dump_register): Avoid calling memset to initialize
17498 a class with a default ctor.
17499 * gcc.c (struct compiler): Remove const qualification.
17500 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17501 * hash-table.h: Ditto.
17502 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17504 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17505 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17507 * params.h (struct param_info): Make struct members non-const.
17508 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17509 with default initialization.
17510 * vec.h (vec_copy_construct, vec_default_construct): New helper
17512 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17513 with vec_copy_construct.
17514 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17515 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17516 * doc/invoke.texi (-Wclass-memaccess): Document.
17518 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17520 * emit-rtl.h (is_leaf): Update comment about local
17521 register allocator.
17523 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17526 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17527 for a variable to have a section before checking if the section has a
17529 Set section to.persistent if persistent attribute is set.
17530 Warn if .persistent attribute is used on an automatic variable.
17532 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17534 PR rtl-optimization/80474
17535 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17537 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17539 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17542 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17544 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17545 (casesi): Emit bounds checking as RTL.
17546 (casesi_internal_mips16_<mode>): Remove bounds checking.
17548 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17550 * config/xtensa/xtensa.c (xtensa_option_override): Append
17551 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17552 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17553 xtensa_doloop_hooks): Define unconditionally.
17554 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17556 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17557 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17558 for it in xtensa_option_override.
17559 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17560 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17562 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17564 * doc/cppopts.texi: Document '-' special value to -MF.
17566 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17568 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17569 (cortex_a53_fconst): Likewise.
17570 (cortex_a53_fpmul): Likewise.
17571 (cortex_a53_f_load_64): Likewise.
17572 (cortex_a53_f_load_many): Likewise.
17573 (cortex_a53_advsimd_alu): Likewise.
17574 (cortex_a53_advsimd_alu_q): Likewise.
17575 (cortex_a53_advsimd_mul): Likewise.
17576 (cortex_a53_advsimd_mul_q): Likewise.
17577 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17578 Add missing fmul, r2f_cvt and fconst cases.
17580 2017-06-14 Richard Biener <rguenther@suse.de>
17582 PR middle-end/81088
17583 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17585 (fold_binary_loc): When associating do not treat pre-existing
17586 TREE_OVERFLOW on literal constants as a reason to allow
17587 TREE_OVERFLOW on associated literal constants.
17589 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17591 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17592 (MASK_FEATURES): New macro.
17593 * config/sparc/sparc.c (sparc_option_override): Remove the special
17594 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17596 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17598 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17599 a division of 0 if non-call exceptions are enabled.
17601 2017-06-14 Andrew Pinski <apinski@cavium.com>
17602 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17605 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17606 Improve vector initialization code gen for only variable case.
17608 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17610 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17612 2017-06-14 Richard Biener <rguenther@suse.de>
17614 PR tree-optimization/81083
17615 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17618 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17620 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17621 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17622 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17623 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17624 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17625 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17627 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17629 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17630 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17632 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17634 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17636 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17638 * config/rs6000/t-rtems: Don't handle SPE.
17640 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17642 * config/rs6000/t-linux: Don't handle SPE.
17644 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17646 * config/rs6000/eabispe.h: Delete file.
17648 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17650 * config/rs6000/t-spe: Delete file.
17652 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17654 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17655 (rs6000_legitimate_offset_address_p): Return false for anything in
17656 V2SImode or V2SFmode.
17658 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17660 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17661 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17662 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17664 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17665 (rs6000_legitimate_offset_address_p): Ditto.
17666 (rs6000_emit_move): Ditto.
17667 (rs6000_init_builtins): Remove V4HI_type_node.
17669 2017-06-13 Martin Liska <mliska@suse.cz>
17672 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17673 (gate_asan): Likewise.
17674 * asan.h (asan_no_sanitize_address_p): Remove the function.
17675 (sanitize_flags_p): New function.
17676 * builtins.def: Fix coding style.
17677 * common.opt: Use renamed enum value.
17678 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17679 * doc/extend.texi: Document no_sanitize attribute.
17680 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17681 to SANITIZE_UNDEFINED_NONDEFAULT.
17682 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17683 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17684 Use sanitize_flags_p.
17685 * gimplify.c (gimplify_function_tree): Likewise.
17686 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17687 * opts.c (parse_no_sanitize_attribute): New function.
17688 (common_handle_option): Use renamed enum value.
17689 * opts.h (parse_no_sanitize_attribute): Declare.
17690 * tree.c (sanitize_flags_p): New function.
17691 * tree.h: Declared here.
17692 * tsan.c: Use sanitize_flags_p.
17693 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17694 (instrument_mem_ref): Likewise.
17695 (instrument_bool_enum_load): Likewise.
17696 (do_ubsan_in_current_function): Remove the function.
17697 (pass_ubsan::execute): Use sanitize_flags_p.
17698 * ubsan.h: Remove do_ubsan_in_current_function
17699 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17700 (dump_function_to_file): Use it here.
17702 2017-06-13 Martin Jambor <mjambor@suse.cz>
17704 PR tree-optimization/80803
17705 PR tree-optimization/81063
17706 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17707 (propagate_subaccesses_across_link): Enqueue subtree whenever
17708 necessary instead of relying on the caller.
17710 2017-06-13 Martin Jambor <mjambor@suse.cz>
17712 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17713 that have a first_link.
17714 (sort_and_splice_var_accesses): Do not check first_link before
17716 (subtree_mark_written_and_enqueue): Likewise.
17717 (propagate_all_subaccesses): Likewise and do not stop at first
17718 parent with a first_link.
17720 2017-06-13 Martin Jambor <mjambor@suse.cz>
17722 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17725 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17727 * match.pd: New pattern.
17729 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17731 * tree-vrp.c (is_masked_range_test): New function.
17732 (register_edge_assert_for): Determine ranges for
17735 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17737 PR tree-optimization/67328
17738 * fold-const.c (maskable_range_p): New function.
17739 (build_range_check): Generate bittests if possible.
17741 2017-06-13 Martin Liska <mliska@suse.cz>
17743 * gimple-pretty-print.c (dump_probability): Add new argument.
17744 (dump_edge_probability): Dump both probability and count.
17745 (dump_gimple_label): Likewise.
17746 (dump_gimple_bb_header): Likewise.
17748 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17751 * config/avr/avr-devices.c: Fix indentation.
17752 * config/avr/gen-avr-mmcu-specs.c: Dito.
17754 2017-06-13 Richard Biener <rguenther@suse.de>
17756 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17757 instead get vector type from stmt_info.
17758 (vectorizable_reduction): Adjust. Remove dead code.
17760 2017-06-13 Richard Biener <rguenther@suse.de>
17762 PR middle-end/81065
17763 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17764 case of C * (x * C2 + C3).
17765 (fold_addr_of_array_ref_difference): Properly fold index difference.
17767 2017-06-12 David S. Miller <davem@davemloft.net>
17770 * config/sparc/sparc.md (return expander): Emit frame blockage if
17771 function uses alloca.
17773 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17775 * combine.c (make_field_assignment): Check len rather than the mode
17776 precision when calling force_to_mode.
17778 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17780 Support multilibs and devices that see flash in RAM address range.
17783 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17784 (avr_mcu_t) <flash_pm_offset>: New field.
17785 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17786 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17787 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17788 (AVR_TINY_PM_OFFSET): Remove macro.
17789 * config/avr/avr.opt (-mshort-calls): New option.
17790 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17791 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17792 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17793 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17794 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17795 instead of avr_arch->have_jmp_call.
17796 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17797 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17798 avr_arch->flash_pm_offset to define.
17799 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17800 new field flash_pm_offset. Add entry for avrxmega3.
17801 (avr_texinfo): Add entry for avrxmega3.
17802 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17803 attiny212, attiny214,
17804 attiny412, attiny414, attiny416, attiny417,
17805 attiny814, attiny816, attiny817,
17806 attiny1614, attiny1616, attiny1617,
17807 attiny3214, attiny3216, attiny3217.
17808 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17809 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17810 (avr_print_operand_address) [AVR_TINY]: Same.
17811 (avr_asm_init_sections) <readonly_data_section>: Only patch
17812 callback if avr_arch->flash_pm_offset = 0.
17813 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17814 for rodata if avr_arch->flash_pm_offset != 0.
17815 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17816 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17817 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17818 (m_options): Append opt_rcall.
17819 (m_dirnames): Append dir_rcall.
17820 * config/avr/t-multilib: Regenerate.
17822 * configure.ac [target=avr]: Check whether avrxmega3 default
17823 linker description file works as needed.
17824 * configure: Regenerate.
17825 * doc/avr-mmcu.texi: Regenerate.
17826 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17827 <__AVR_ARCH__>: Document avrxmega3 and 103.
17828 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17829 <__AVR_SHORT_CALLS__>: Document it.
17830 <__AVR_PM_BASE_ADDRESS__>: Document it.
17831 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17832 (AVR Variable Attributes) <progmem>: Document this is
17833 not needed for avrxmega3.
17834 (AVR Named Address Spaces) <__flash>: Dito.
17836 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17838 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17840 2017-06-12 Doug Rupp <rupp@adacore.com>
17842 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17843 Append vxworks-stdint.h to the tm_file list.
17844 * config/vxworks-stdint.h: New file.
17846 2017-06-12 Martin Liska <mliska@suse.cz>
17848 PR tree-optimization/81041
17849 * tree-profile.c (gimple_gen_ic_func_profiler):
17850 Create an extra BB in profile-generate
17851 (gimple_gen_time_profiler): Likewise.
17853 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17855 PR tree-optimization/81003
17856 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17857 (update_range_test): Use it instead of force_gimple_operand_gsi.
17859 2017-06-12 Richard Biener <rguenther@suse.de>
17861 PR tree-optimization/81053
17862 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17863 with backedge value not defined in loop. Simplify def stmt
17866 2017-06-11 Tom de Vries <tom@codesourcery.com>
17869 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17871 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17872 nvptx_cannot_force_const_mem.
17874 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17876 * opts.c (finish_options): Move test for flag_split_stack after
17877 it has been initialized.
17879 2017-06-11 Jason Merrill <jason@redhat.com>
17881 * tree.h (id_equal): New.
17882 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17883 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17884 instead of strcmp of IDENTIFIER_POINTER.
17886 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17888 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17889 (mark_all_inlined_calls_cdtor): Fix formating.
17890 (inline_transform): Rescale profile before inlining.
17892 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17894 * cgraph.h (cgraph_edge::clone): Update prototype.
17895 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17896 (cgraph_node::create_clone): Update.
17897 (cgraph_node::create_version_clone): Update.
17898 * tree-inline.c (copy_bb): Update.
17899 (expand_call_inline): Update.
17901 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17903 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17904 factored out from ...
17905 (rs6000_emit_prologue): ... here.
17907 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17909 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17910 factored out from ...
17911 (rs6000_emit_prologue): ... here.
17913 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17915 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17917 (handle_missing_profiles): Fix computation of tp_first_run.
17918 (counts_to_freqs): Do not touch freqs when count is 0.
17920 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17922 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17925 2017-06-10 Tom de Vries <tom@codesourcery.com>
17927 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17928 attributes): Document signal effective target.
17930 2017-06-10 Tom de Vries <tom@codesourcery.com>
17932 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17933 Document effective target stack_size.
17935 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17937 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17938 to the edit_context if they can be auto-applied.
17940 2017-06-9 Ian Lance Taylor <iant@golang.org>
17942 * opts.c (finish_options): If -fsplit-stack, disable implicit
17943 -forder-blocks-and-partition.
17944 * doc/invoke.texi (Optimize Options): Document that when using
17945 -fsplit-stack -forder-blocks-and-partition is not implicitly
17948 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17950 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17951 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17952 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17953 * builtins.def (abort, trap, unreachable): Declare cold.
17954 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17955 * tree-core.h (ECF_COLD): New.
17956 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17957 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17959 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17961 * predict.c (unlikely_executed_stmt_p): Cleanup.
17963 2017-06-09 Richard Biener <rguenther@suse.de>
17965 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17966 model if the ref is always written to.
17968 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17970 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17972 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17974 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17977 2017-06-09 Tom de Vries <tom@codesourcery.com>
17980 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17981 "target cannot support label values" when encountering LABEL_REF.
17983 2017-06-09 Martin Liska <mliska@suse.cz>
17985 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17986 (gimple_gen_ic_func_profiler): Emit direct comparison
17987 of __gcov_indirect_call_callee with NULL.
17988 (gimple_gen_time_profiler): Change probability from
17989 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17991 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17993 * profile.c (edge_gcov_counts): Turn to pointer.
17994 (compute_branch_probabilities, compute_branch_probabilities): Update.
17995 (branch_prob): Do not clear edge_gcov_count.
17996 * profile.h (edge_gcov_counts): Turn to pointer.
17997 (edge_gcov_count): Update.
17999 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18001 * gimple.h (gimple_check_failed): Mark cold.
18003 2017-06-09 Richard Biener <rguenther@suse.de>
18005 PR tree-optimization/66623
18006 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18007 refactor check_reduction into two parts, properly computing
18008 whether we have to check reduction validity for outer loop
18011 2017-06-09 Richard Biener <rguenther@suse.de>
18013 PR tree-optimization/79483
18014 * graphite-scop-detection.c (order): New global.
18015 (get_order): Compute bb to order mapping that satisfies code
18016 generation constraints.
18017 (cmp_pbbs): New helper.
18018 (build_scops): Start domwalk at entry block, sort generated
18021 2017-06-09 Richard Biener <rguenther@suse.de>
18023 PR middle-end/81007
18024 * ipa-polymorphic-call.c
18025 (ipa_polymorphic_call_context::restrict_to_inner_class):
18026 Skip FIELD_DECLs with error_mark_node type.
18027 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18030 2017-06-09 Martin Liska <mliska@suse.cz>
18032 * predict.c (struct branch_predictor): New struct.
18033 (test_prediction_value_range): New test.
18034 (predict_c_tests): New function.
18035 * selftest-run-tests.c (selftest::run_tests): Run the function.
18036 * selftest.h: Declare new tests.
18038 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18041 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18042 gen_add3_insn did not fail.
18043 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18044 r0, construct that number in a temporary reg and add that reg to r0.
18045 If asked to put the result in r0 as well, fail.
18047 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18050 for early expansion of vec_eqv.
18052 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18054 PR middle-end/81005
18055 * ubsan.c (instrument_null): Avoid pointless code temporary.
18056 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18058 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18062 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18065 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18068 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18070 * predict.c (unlikely_executed_edge_p): Move ahead.
18071 (probably_never_executed_edge_p): Use it.
18073 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18075 PR middle-end/79988
18076 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18077 gimple_call_builtin_p call.
18079 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18081 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18082 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18083 rtl_check_failed_type2, rtl_check_failed_code1,
18084 rtl_check_failed_code2, rtl_check_failed_code_mode,
18085 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18086 rtvec_check_failed_bounds, rtl_check_failed_flag,
18087 _fatal_insn_not_found, _fatal_insn): Likewise.
18088 * tree.h (tree_contains_struct_check_failed,
18089 tree_check_failed, tree_not_check_failed,
18090 tree_class_check_failed, tree_range_check_failed,
18091 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18092 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18093 tree_operand_check_failed, omp_clause_check_failed,
18094 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18097 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18099 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18100 flag_branch_probabilities.
18101 * ipa-inline.c (edge_badness): Likewise.
18102 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18103 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18104 * predict.c (maybe_hot_frequency_p): Likewise.
18105 (probably_never_executed): Likewise.
18106 * sched-ebb.c (schedule_ebbs): Likewise.
18107 * sched-rgn.c (find_single_block_region): Likewise.
18108 * tracer.c (tail_duplicate): Likewise.
18110 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18112 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18113 longer requires x_flag_profile_use.
18115 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18117 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18118 instead of flag_reorder_blocks_and_partition.
18119 * dbxout.c (dbxout_function_end): Likewise.
18120 * dwarf2out.c (gen_subprogram_die): Likewise.
18121 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18122 * hw-doloop.c (reorg_loops): Likewise.
18123 * varasm.c (assemble_start_function,
18124 assemble_end_function): Likewise.
18125 (decide_function_section): Do not check for
18126 flag_reorder_blocks_and_partition.
18128 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18130 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18132 (chkp_get_hard_register_fake_addr_expr): Ditto.
18133 (chkp_build_addr_expr): Add check for hard reg case.
18134 (chkp_parse_array_and_component_ref): Ditto.
18135 (chkp_find_bounds_1): Ditto.
18136 (chkp_process_stmt): Don't generate bounds store for
18139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18141 * predict.c (maybe_hot_bb_p): Do not check profile status.
18142 (maybe_hot_edge_p): Likewise.
18143 (probably_never_executed): Check for zero counts even if profile
18145 (unlikely_executed_edge_p): New function.
18146 (unlikely_executed_stmt_p): New function.
18147 (unlikely_executed_bb_p): New function.
18148 (set_even_probabilities): Use unlikely predicates.
18149 (combine_predictions_for_bb): Likewise.
18150 (predict_paths_for_bb): Likewise.
18151 (predict_paths_leading_to_edge): Likewise.
18152 (determine_unlikely_bbs): New function.
18153 (estimate_bb_frequencies): Use it.
18154 (compute_function_frequency): Use zero counts even if profile is
18156 * profile-count.h: Fix typo.
18158 2017-08-08 Julia Koval <julia.koval@intel.com>
18160 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18161 _mm512_mask_cvtsepi16_storeu_epi8,
18162 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18163 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18164 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18165 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18166 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18167 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18168 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18169 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18170 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18171 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18172 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18173 __builtin_ia32_pmovuswb256mem_mask,
18174 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18175 __builtin_ia32_pmovwb512mem_mask): New builtins.
18177 2017-08-08 Julia Koval <julia.koval@intel.com>
18179 PR target/73350,80862
18180 * config/i386/subst.md (round): Fix round pattern.
18181 * config/i386/i386.c (ix86_erase_embedded_rounding):
18182 Fix erasing rounding for the fixed pattern.
18184 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18186 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18188 2017-06-08 Martin Liska <mliska@suse.cz>
18190 PR gcov-profile/80911
18191 * gcov.c (block_info::block_info): New constructor.
18193 2017-06-07 Carl Love <cel@us.ibm.com>
18195 * config/rs6000/rs6000-c: The return type of the following
18196 built-in functions was implemented as int not long long. Fix sign
18197 of return value for the unsigned version of vec_mulo and vec_mule.
18198 vector unsigned long long vec_bperm (vector unsigned long long,
18199 vector unsigned char)
18200 vector signed long long vec_mule (vector signed int,
18202 vector unsigned long long vec_mule (vector unsigned int,
18203 vector unsigned int)
18204 vector signed long long vec_mulo (vector signed int,
18206 vector unsigned long long vec_mulo (vector unsigned int,
18207 vector unsigned int)
18208 * doc/extend.texi: Fix the documentation for the built-in
18211 2017-06-07 Carl Love <cel@us.ibm.com>
18214 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18217 2017-06-07 Carl Love <cel@us.ibm.com>
18219 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18220 support, Generate doublehv for signed int/float for BE case only.
18222 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18224 * doc/invoke.texi (mcx16): Rewrite.
18226 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18228 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18229 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18230 *mov<mode>_softfloat, and an anonymous splitter): Use
18231 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18235 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18237 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18238 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18239 (rs6000_debug_reg_global): Adjust.
18240 (rs6000_init_hard_regno_mode_ok): Adjust.
18241 (rs6000_dbx_register_number): Adjust.
18242 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18243 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18244 Remove SPE_ACC and SPEFSCR.
18245 (REG_ALLOC_ORDER): Ditto.
18246 (FRAME_POINTER_REGNUM): Change to 111.
18247 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18248 (REG_CLASS_NAMES): Ditto.
18249 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18250 (REGISTER_NAMES): Ditto.
18251 (ADDITIONAL_REG_NAMES): Ditto.
18252 (rs6000_reg_names): Ditto.
18253 * config/rs6000/rs6000.md: Renumber some register number
18256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18258 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18260 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18261 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18263 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18264 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18265 Delete the SPE high registers.
18266 (REG_ALLOC_ORDER): Ditto.
18267 (enum reg_class): Remove SPE_HIGH_REGS.
18268 (REG_CLASS_NAMES): Ditto.
18269 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18270 (REGISTER_NAMES): Ditto.
18271 (rs6000_reg_names): Ditto.
18272 * doc/tm.texi.in: Remove SPE as example.
18273 * doc/tm.texi: Regenerate.
18275 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18277 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18278 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18279 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18280 * config/rs6000/rs6000.md (type): Remove "brinc".
18282 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18284 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18285 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18286 * config/rs6000/linuxspe.h: Delete file.
18287 * config/rs6000/rs6000.md: Don't include spe.md.
18288 * config/rs6000/spe.h: Delete file.
18289 * config/rs6000/spe.md: Delete file.
18290 * config/rs6000/t-rs6000: Remove spe.md.
18292 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18294 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18295 (reg_or_none500mem_operand): Delete.
18296 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18297 instead of reg_or_none500mem_operand.
18299 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18302 handling of SPE flags.
18303 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18307 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18309 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18310 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18311 paired_divv2sf3): Similar.
18312 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18313 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18314 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18316 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18317 Rename the paired_* instruction patterns.
18318 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18320 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18321 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18322 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18323 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18324 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18325 PAIRED_VECTOR_MODE.
18326 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18327 (spe_func_has_64bit_regs_p): Delete.
18328 (spe_expand_predicate_builtin): Delete.
18329 (spe_expand_evsel_builtin): Delete.
18330 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18331 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18332 (invalid_e500_subreg): Delete.
18333 (rs6000_legitimize_address): Always force_reg op2 as well, for
18334 paired single memory accesses.
18335 (rs6000_member_type_forces_blk): Delete.
18336 (rs6000_spe_function_arg): Delete.
18337 (rs6000_expand_unop_builtin): Delete SPE handling.
18338 (rs6000_expand_binop_builtin): Ditto.
18339 (spe_expand_stv_builtin): Delete.
18340 (bdesc_2arg_spe): Delete.
18341 (spe_expand_builtin): Delete.
18342 (spe_expand_predicate_builtin): Delete.
18343 (spe_expand_evsel_builtin): Delete.
18344 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18345 (spe_init_builtins): Delete.
18346 (spe_func_has_64bit_regs_p): Delete.
18347 (savres_routine_name): Delete "info" parameter. Adjust callers.
18348 (rs6000_emit_stack_reset): Ditto.
18349 (rs6000_dwarf_register_span): Delete.
18350 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18351 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18352 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18354 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18356 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18357 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18358 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18359 mulv2sf3, divv2sf3): Delete expanders.
18361 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18363 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18365 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18367 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18368 * config/rs6000/rs6000.c: Ditto.
18370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18372 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18373 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18374 comparison_operator.
18376 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18378 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18379 * config/rs6000/rs6000.opt: Ditto.
18380 * config/rs6000/t-rtems: Ditto.
18382 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18384 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18385 TARGET_E500_SINGLE by 0, simplify.
18386 * config/rs6000/rs6000.c: Ditto.
18387 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18388 (spe_build_register_parallel): Delete.
18389 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18390 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18391 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18392 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18393 (E500_CONVERT): Delete.
18394 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18398 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18399 * config/rs6000/dfp.md: Ditto.
18400 (negdd2, *negdd2_fpr): Merge.
18401 (absdd2, *absdd2_fpr): Merge.
18402 (negtd2, *negtd2_fpr): Merge.
18403 (abstd2, *abstd2_fpr): Merge.
18404 * config/rs6000/e500.h: Delete file.
18405 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18406 TARGET_FPRS by 1 and simplify.
18407 * config/rs6000/rs6000-c.c: Ditto.
18408 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18409 TARGET_DF_SPE by 0.
18410 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18412 * config/rs6000/rs6000.md: Ditto.
18413 (floatdidf2, *floatdidf2_fpr): Merge.
18414 (move_from_CR_gt_bit): Delete.
18415 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18416 (E500_CR_IOR_COMPARE): Delete.
18417 (All patterns that require !TARGET_FPRS): Delete.
18418 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18420 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18422 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18424 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18426 * graphds.c (add_edge): Intitialize edge's attached data.
18427 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18428 pointer parameter. Call pointed function on each edge during
18429 graph traversing. Skip traversing the edge when the function
18431 (graphds_dfs, graphds_scc): Ditto.
18432 (for_each_edge): New parameter. Pass the new parameter to callback
18434 * graphds.h (skip_edge_callback): New function pointer type.
18435 (graphds_dfs, graphds_scc): New function pointer parameter.
18436 (graphds_edge_callback, for_each_edge): New parameter.
18438 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18440 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18441 out code checking if runtime alias check is possible to below ...
18442 Call the new function.
18443 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18444 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18446 2017-06-07 Marek Polacek <polacek@redhat.com>
18449 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18450 TYPE_OVERFLOW_WRAPS check.
18452 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18454 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18455 if versioning is required.
18456 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18457 peeling with the check for versioning.
18459 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18461 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18462 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18463 Set true to new parameter if new ssa variable is defined.
18464 (vect_gen_vector_loop_niters): Refactor. Set range information
18465 for the new vector loop bound variable.
18466 (vect_do_peeling): Ditto.
18468 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18470 * tree-affine.c (ssa.h): Include header file.
18471 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18472 has wrapping overflow behavior.
18474 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18476 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18478 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18480 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18481 (tree_to_aff_combination): ... here.
18483 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18485 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18486 reg_pressure model function.
18487 (ivopts_global_cost_for_size): Delete.
18488 (determine_set_costs, iv_ca_recount_cost): Call new model function
18489 ivopts_estimate_reg_pressure.
18491 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18493 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18494 expensive than udiv. Remove floating point cases from mod.
18496 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18498 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18499 Increase idiv cost.
18501 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18503 * config/aarch64/aarch64.md
18504 (copysignsf3): Fix mask generation.
18506 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18508 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18510 (class dump_manager): Add register_dumps method.
18511 * dumpfile.c: Include langhooks.h.
18512 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18513 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18514 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18515 (dump_manager::dump_register): Start with 512 entries instead of 32.
18516 (dump_manager::register_dumps): New method.
18517 * toplev.c (general_init): Instead of invoking register_dumps
18518 langhook, invoke register_dumps method on the dump manager.
18519 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18522 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18524 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18525 Say that # requires an associated define_split to exist, and that
18526 the define_split must be suitable for use after register allocation.
18528 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18530 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18531 (compute_outgoing_frequencies): Also initialize zero counts.
18532 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18533 around loops; preserve more of profile when nothing changes.
18535 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18537 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18539 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18540 * config/arm/arm-cpu-cdata.h: Regenerate.
18541 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18542 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18543 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18544 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18545 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18547 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18549 * config/arm/t-rmprofile: Likewise.
18550 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18552 2017-06-06 David S. Miller <davem@davemloft.net>
18555 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18556 blockage if function uses alloca.
18558 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18560 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18561 New "uid" fields to hold pretty-print IDs of group and ref.
18562 Memory references are now identified as <group_id>:<ref_id>
18563 instead of using [random] addresses.
18564 (dump_mem_details): Simplify, no functional change.
18565 (dump_mem_ref): Simplify and make output more concise.
18566 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18567 (find_or_create_group): Initialize group uid.
18568 (record_ref): Initialize ref uid. Improve debug output.
18569 (prune_group_by_reuse, should_issue_prefetch_p,)
18570 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18571 (mark_nontemporal_store, determine_loop_nest_reuse):
18572 Improve debug output.
18574 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18576 * dbgcnt.def (prefetch): New debug counter.
18577 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18578 (schedule_prefetches): Stop issueing prefetches if debug counter
18581 2017-06-06 Tom de Vries <tom@codesourcery.com>
18583 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18584 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18586 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18588 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18589 Use aarch64_reg_or_zero predicate for operand 4.
18590 (aarch64_compare_and_swap<mode> define_insn_and_split):
18591 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18592 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18594 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18596 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18597 (arm_compute_save_core_reg_mask): This.
18598 (thumb1_compute_save_reg_mask): Rename into ...
18599 (thumb1_compute_save_core_reg_mask): This.
18600 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18601 (arm_compute_frame_layout): Likewise.
18603 2017-06-06 Richard Biener <rguenther@suse.de>
18605 PR tree-optimization/80974
18606 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18607 keep or clear leaders SSA info.
18609 2017-06-06 Tom de Vries <tom@codesourcery.com>
18611 * config/nvptx/nvptx.c (split_mode_p): New function.
18612 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18614 2017-06-06 Tom de Vries <tom@codesourcery.com>
18616 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18618 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18621 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18624 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18626 * shrink-wrap.c (handle_simple_exit): Update profile.
18627 (try_shrink_wrapping): Upate profile.
18629 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18631 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18632 (tree_guess_outgoing_edge_probabilities): New.
18633 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18634 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18636 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18638 * ipa-split.c (split_function): Initialize return bb profile.
18640 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18642 * profile.c (compute_branch_probabilities): Also initialize
18643 EXIT_BLOCK profile.
18645 2017-06-06 Richard Biener <rguenther@suse.de>
18647 PR tree-optimization/80928
18648 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18649 (vect_analyze_loop_operations): Properly guard analysis for
18651 (vect_transform_loop): Likewise.
18652 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18653 (vect_model_induction_cost): Do not cost for pure SLP.
18654 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18655 of induction in inner loop vectorization.
18656 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18657 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18658 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18660 (vect_analyze_slp_cost_1): Cost induction.
18661 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18662 (vect_get_slp_vect_defs): Likewise.
18663 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18664 (vect_transform_stmt): Handle SLP reductions.
18665 * tree-vectorizer.h (vectorizable_induction): Adjust.
18667 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18669 * config/rs6000/rs6000.c (make_resolver_func): Update
18670 init_lowered_empty_function call.
18672 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18674 * doc/invoke.texi: Document the -fprofile-abs-path option.
18675 * common.opt (fprofile-abs-path): New option.
18676 * gcov-io.h (gcov_write_filename): Declare.
18677 * gcov-io.c (gcov_write_filename): New function.
18678 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18679 * profile.c (output_location): Likewise.
18681 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18683 * shring-wrap.c: Revert accidental commit.
18685 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18687 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18689 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18691 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18693 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18694 profile in callgraph edge.
18695 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18696 (apply_scale): Likewise.
18697 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18698 Also scale profile when inlining function with zero profile.
18699 (initialize_cfun): Update exit block profile even when it is zero.
18700 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18701 when profile is read.
18703 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18705 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18706 (CLONE_*): New constants to define the processors we can generate
18707 code for with the target_clone attribute.
18708 (rs6000_clone_map): New array to identify which clone processors
18709 the current program is running on.
18710 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18711 target_clone attribute.
18712 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18713 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18714 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18715 (cpu_expand_builtin): Add support for target_clone attribute.
18716 (rs6000_valid_attribute_p): Allow "default" attribute.
18717 (get_decl_name): New debug function to simplify printing the
18718 current function name in debugging statements.
18719 (rs6000_clone_priority): New functions to support the target_clone
18720 attribute, and be able to generate code to switch between ISA 2.05
18721 through ISA 3.0 (power6 through power9).
18722 (rs6000_compare_version_priority): Likewise.
18723 (rs6000_get_function_versions_dispatcher): Likewise.
18724 (make_resolver_func): Likewise.
18725 (add_condition_to_bb): Likewise.
18726 (dispatch_function_versions): Likewise.
18727 (rs6000_generate_version_dispatcher_body): Likewise.
18728 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18729 (fusion_gpr_load_p): Fix a spacing issue.
18730 * doc/extend.texi (Common Function Attributes): Document that the
18731 PowerPC supports the target_clone attribute.
18733 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18735 * config/arm/arm.h: explain F symbol found in description of ARM
18736 register allocation in its legend.
18738 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18740 * config/mips/frame-header-opt.c: Include profile-count.h.
18741 * config/riscv/riscv.c: Include profile-count.h
18743 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18745 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18747 (sm_set_flag_if_changed): Add bbs field.
18748 (execute_sm_if_changed_flag_set): Pass BBS.
18749 (execute_sm): Update.
18751 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18753 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18756 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18758 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18759 (peephole2): New peephole2 to emit the above.
18760 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18762 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18764 * config/aarch64/aarch64.c (define_peephole2 above
18765 *sub_<shift>_<mode>): New peephole.
18767 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18769 * config/i386/i386.c (make_resolver_func): Update.
18770 * Makefile.in: Add profile-count.h and profile-count.o
18771 * auto-profile.c (afdo_indirect_call): Update to new API.
18772 (afdo_set_bb_count): Update.
18773 (afdo_propagate_edge): Update.
18774 (afdo_propagate_circuit): Update.
18775 (afdo_calculate_branch_prob): Update.
18776 (afdo_annotate_cfg): Update.
18777 * basic-block.h: Include profile-count.h
18778 (struct edge_def): Turn count to profile_count.
18779 (struct basic_block_def): Likewie.
18780 (REG_BR_PROB_BASE): Move to profile-count.h
18781 (RDIV): Move to profile-count.h
18782 * bb-reorder.c (max_entry_count): Turn to profile_count.
18783 (find_traces): Update.
18784 (rotate_loop):Update.
18785 (connect_traces):Update.
18786 (sanitize_hot_paths):Update.
18787 * bt-load.c (migrate_btr_defs): Update.
18788 * cfg.c (RDIV): Remove.
18789 (init_flow): Use alloc_block.
18790 (alloc_block): Uninitialize count.
18791 (unchecked_make_edge): Uninitialize count.
18792 (check_bb_profile): Update.
18793 (dump_edge_info): Update.
18794 (dump_bb_info): Update.
18795 (update_bb_profile_for_threading): Update.
18796 (scale_bbs_frequencies_int): Update.
18797 (scale_bbs_frequencies_gcov_type): Update.
18798 (scale_bbs_frequencies_profile_count): New.
18799 * cfg.h (update_bb_profile_for_threading): Update.
18800 (scale_bbs_frequencies_profile_count): Declare.
18801 * cfgbuild.c (compute_outgoing_frequencies): Update.
18802 (find_many_sub_basic_blocks): Update.
18803 * cfgcleanup.c (try_forward_edges): Update.
18804 (try_crossjump_to_edge): Update.
18805 * cfgexpand.c (expand_gimple_tailcall): Update.
18806 (construct_exit_block): Update.
18807 * cfghooks.c (verify_flow_info): Update.
18808 (dump_bb_for_graph): Update.
18809 (split_edge): Update.
18810 (make_forwarder_block): Update.
18811 (duplicate_block): Update.
18812 (account_profile_record): Update.
18813 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18814 (get_estimated_loop_iterations): Update.
18815 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18816 (single_likely_exit): Update.
18817 * cfgloopmanip.c (scale_loop_profile): Update.
18819 (set_zero_probability): Update.
18820 (lv_adjust_loop_entry_edge): Update.
18821 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18822 (purge_dead_edges): Update.
18823 (rtl_account_profile_record): Update.
18824 * cgraph.c (cgraph_node::create): Uninitialize count.
18825 (symbol_table::create_edge): Uninitialize count.
18826 (cgraph_update_edges_for_call_stmt_node): Update.
18827 (cgraph_edge::dump_edge_flags): Update.
18828 (cgraph_node::dump): Update.
18829 (cgraph_edge::maybe_hot_p): Update.
18830 * cgraph.h: Include profile-count.h
18831 (create_clone), create_edge, create_indirect_edge): Update.
18832 (cgraph_node): Turn count to profile_count.
18833 (cgraph_edge0: Likewise.
18834 (make_speculative, clone): Update.
18835 (create_edge): Update.
18836 (init_lowered_empty_function): Update.
18837 * cgraphclones.c (cgraph_edge::clone): Update.
18838 (duplicate_thunk_for_node): Update.
18839 (cgraph_node::create_clone): Update.
18840 * cgraphunit.c (cgraph_node::analyze): Update.
18841 (cgraph_node::expand_thunk): Update.
18842 * final.c (dump_basic_block_info): Update.
18843 * gimple-streamer-in.c (input_bb): Update.
18844 * gimple-streamer-out.c (output_bb): Update.
18845 * graphite.c (print_global_statistics): Update.
18846 (print_graphite_scop_statistics): Update.
18847 * hsa-brig.c: Include basic-block.h.
18848 * hsa-dump.c: Include basic-block.h.
18849 * hsa-gen.c (T sum_slice): Update.
18850 (convert_switch_statements):Update.
18851 * hsa-regalloc.c: Include basic-block.h.
18852 * ipa-chkp.c (chkp_produce_thunks): Update.
18853 * ipa-cp.c (struct caller_statistics): Update.
18854 (init_caller_stats): Update.
18855 (gather_caller_stats): Update.
18856 (ipcp_cloning_candidate_p): Update.
18857 (good_cloning_opportunity_p): Update.
18858 (get_info_about_necessary_edges): Update.
18859 (dump_profile_updates): Update.
18860 (update_profiling_info): Update.
18861 (update_specialized_profile): Update.
18862 (perhaps_add_new_callers): Update.
18863 (decide_about_value): Update.
18864 (ipa_cp_c_finalize): Update.
18865 * ipa-devirt.c (struct odr_type_warn_count): Update.
18866 (struct decl_warn_count): Update.
18867 (struct final_warning_record): Update.
18868 (possible_polymorphic_call_targets): Update.
18869 (ipa_devirt): Update.
18870 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18871 * ipa-icf.c (sem_function::merge): Update.
18872 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18873 * ipa-inline.c (compute_uninlined_call_time): Update.
18874 (compute_inlined_call_time): Update.
18875 (want_inline_small_function_p): Update.
18876 (want_inline_self_recursive_call_p): Update.
18877 (edge_badness): Update.
18878 (lookup_recursive_calls): Update.
18879 (recursive_inlining): Update.
18880 (inline_small_functions): Update.
18881 (dump_overall_stats): Update.
18882 (dump_inline_stats): Update.
18883 * ipa-profile.c (ipa_profile_generate_summary): Update.
18884 (ipa_propagate_frequency): Update.
18885 (ipa_profile): Update.
18886 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18887 * ipa-utils.c (ipa_merge_profiles): Update.
18888 * loop-doloop.c (doloop_modify): Update.
18889 * loop-unroll.c (report_unroll): Update.
18890 (unroll_loop_runtime_iterations): Update.
18891 * lto-cgraph.c (lto_output_edge): Update.
18892 (lto_output_node): Update.
18893 (input_node): Update.
18894 (input_edge): Update.
18895 (merge_profile_summaries): Update.
18896 * lto-streamer-in.c (input_cfg): Update.
18897 * lto-streamer-out.c (output_cfg): Update.
18898 * mcf.c (create_fixup_graph): Update.
18899 (adjust_cfg_counts): Update.
18900 (sum_edge_counts): Update.
18901 * modulo-sched.c (sms_schedule): Update.
18902 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18903 * predict.c (maybe_hot_count_p): Update.
18904 (probably_never_executed): Update.
18905 (dump_prediction): Update.
18906 (combine_predictions_for_bb): Update.
18907 (propagate_freq): Update.
18908 (handle_missing_profiles): Update.
18909 (counts_to_freqs): Update.
18910 (rebuild_frequencies): Update.
18911 (force_edge_cold): Update.
18912 * predict.h: Include profile-count.h
18913 (maybe_hot_count_p, counts_to_freqs): UPdate.
18914 * print-rtl-function.c: Do not include cfg.h
18915 * print-rtl.c: Include basic-block.h
18916 * profile-count.c: New file.
18917 * profile-count.h: New file.
18918 * profile.c (is_edge_inconsistent): Update.
18919 (correct_negative_edge_counts): Update.
18920 (is_inconsistent): Update.
18921 (set_bb_counts): Update.
18922 (read_profile_edge_counts): Update.
18923 (compute_frequency_overlap): Update.
18924 (compute_branch_probabilities): Update; Initialize and deinitialize
18926 (branch_prob): Update.
18927 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18928 (edge_gcov_count): New.
18929 (bb_gcov_count): New.
18930 * shrink-wrap.c (try_shrink_wrapping): Update.
18931 * tracer.c (better_p): Update.
18932 * trans-mem.c (expand_transaction): Update.
18933 (ipa_tm_insert_irr_call): Update.
18934 (ipa_tm_insert_gettmclone_call): Update.
18935 * tree-call-cdce.c: Update.
18936 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18937 (gimple_duplicate_sese_tail): Update.
18938 (gimple_account_profile_record): Update.
18939 (execute_fixup_cfg): Update.
18940 * tree-inline.c (copy_bb): Update.
18941 (copy_edges_for_bb): Update.
18942 (initialize_cfun): Update.
18943 (freqs_to_counts): Update.
18944 (copy_cfg_body): Update.
18945 (expand_call_inline): Update.
18946 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18947 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18948 (try_unroll_loop_completely): Update.
18949 (try_peel_loop): Update.
18950 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18951 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18952 * tree-ssa-loop-split.c (connect_loops): Update.
18953 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18954 * tree-ssa-reassoc.c (branch_fixup): Update.
18955 * tree-ssa-tail-merge.c (replace_block_by): Update.
18956 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18957 (compute_path_counts): Update.
18958 (update_profile): Update.
18959 (recompute_probabilities): Update.
18960 (update_joiner_offpath_counts): Update.
18961 (estimated_freqs_path): Update.
18962 (freqs_to_counts_path): Update.
18963 (clear_counts_path): Update.
18964 (ssa_fix_duplicate_block_edges): Update.
18965 (duplicate_thread_path): Update.
18966 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18967 (struct switch_conv_info): Update.
18968 * tree-tailcall.c (decrease_profile): Update.
18969 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18970 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18971 * value-prof.c (check_counter): Update.
18972 (gimple_divmod_fixed_value): Update.
18973 (gimple_mod_pow2): Update.
18974 (gimple_mod_subtract): Update.
18975 (gimple_ic_transform): Update.
18976 (gimple_stringop_fixed_value): Update.
18977 * value-prof.h (gimple_ic): Update.
18979 2017-06-02 Carl Love <cel@us.ibm.com>
18981 * config/rs6000/rs6000-c: Add support for built-in functions
18982 vector double vec_doublee (vector signed int);
18983 vector double vec_doublee (vector unsigned int);
18984 vector double vec_doublee (vector float);
18985 vector double vec_doubleh (vector signed int);
18986 vector double vec_doubleh (vector unsigned int);
18987 vector double vec_doubleh (vector float);
18988 vector double vec_doublel (vector signed int);
18989 vector double vec_doublel (vector unsigned int);
18990 vector double vec_doublel (vector float);
18991 vector double vec_doubleo (vector signed int);
18992 vector double vec_doubleo (vector unsigned int);
18993 vector double vec_doubleo (vector float);.
18994 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18995 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18997 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18998 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18999 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19001 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19002 vec_doublel, vec_doubleh.
19003 * doc/extend.texi: Update the built-in documentation file for the
19004 new built-in functions.
19006 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19009 * ipa-inline-analysis.c (free_growth_caches): Set
19010 edge_removal_hook_holder to NULL after removing it.
19012 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19014 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19015 comparision with zero.
19017 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19018 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19019 for early expansion of vec_min and vec_max builtins.
19020 (builtin_function_type): Add min/max unsigned variants to those
19021 identified as having unsigned arguments.
19023 2017-06-02 Olivier Hainque <hainque@adacore.com>
19025 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19027 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19029 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19030 Use VALL_F16 iterator rather than VALL.
19032 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19034 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19035 Emit CBNZ inside loop when doing a strong exchange and comparing
19036 against zero. Generate the CC flags after the loop.
19038 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19040 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19041 (dl_section_ref): New.
19042 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19043 On AIX, append an expression to subtract the size of the
19044 section length to dl_section_ref.
19046 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19048 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19049 for early expansion of vector absolute builtins.
19051 2017-06-02 Richard Biener <rguenther@suse.de>
19053 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19054 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19056 2017-06-02 Richard Biener <rguenther@suse.de>
19058 PR tree-optimization/80948
19059 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19060 stmt order as well.
19062 2017-06-02 Richard Biener <rguenther@suse.de>
19064 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19066 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19067 for inductions as relevant.
19069 2017-06-02 Richard Biener <rguenther@suse.de>
19071 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19072 (vectorizable_induction): ... this. Remove dead code.
19074 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19076 * builtins. (expand_builtin_alloca): Remove second parameter and
19077 infer its value from the first parameter instead.
19078 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19080 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19082 PR rtl-optimization/80903
19083 * loop-doloop.c (add_test): Unshare sequence.
19085 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19087 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19089 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19091 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19093 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19094 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19095 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19096 (xlogue_layout::get_stub_rtx): Make static.
19097 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19098 (xlogue_layout::compute_stub_managed_regs): Rename to...
19099 (xlogue_layout::count_stub_managed_regs): ...this.
19100 (xlogue_layout::is_stub_managed_reg): New function.
19101 (xlogue_layout::m_stub_names): Rename to...
19102 (xlogue_layout::s_stub_names): ...this, make static.
19103 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19104 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19105 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19106 xlogue_layout::s_stub_names): Instantiate statics.
19107 (stub_managed_regs): Remove.
19108 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19109 (disable_call_ms2sysv_xlogues): Rename to...
19110 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19111 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19113 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19114 change after reload_completed.
19115 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19117 (ix86_expand_prologue): Likewise.
19118 (ix86_expand_epilogue): Likewise.
19119 (ix86_expand_split_stack_prologue): Likewise.
19120 (ix86_compute_frame_layout): Remove frame parameter ...
19121 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19122 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19124 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19125 (ix86_frame): Move from here ...
19126 * config/i386/i386.h (ix86_frame): ... to here.
19127 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19128 complete ix86_frame data structure instead. Remove some_ld_name.
19130 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19132 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19133 symbols that hold a DECL_VALUE_EXPR.
19135 2017-06-01 Martin Jambor <mjambor@suse.cz>
19137 PR tree-optimization/80898
19138 * tree-sra.c (process_subtree_disqualification): Removed.
19139 (disqualify_candidate): Do not acll
19140 process_subtree_disqualification.
19141 (subtree_mark_written_and_enqueue): New function.
19142 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19143 RHS has been disqualified and re-queue LHS if necessary. Apart
19144 from that, ignore disqualified RHS.
19146 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19148 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19149 address fetch for z10 or later.
19151 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19153 * config/arc/arc.md (tst_movb): Add guard when splitting.
19155 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19157 * config/arc/arc.c (arc_can_eliminate): Test against
19158 arc_frame_pointer_needed.
19160 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19162 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19163 to prevent store reordering.
19164 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19165 (type): Add block type.
19166 (stack_tie): Define special instruction to be used in
19169 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19171 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19172 constraint. It is not valid for the pattern.
19173 (noncommutative_binary_comparison): Likewise.
19175 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19177 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19180 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19182 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19183 be used by the reg-alloc.
19185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19187 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19188 reg-alloc when having mul64 or mul32x16 instructions.
19189 (mulsidi3): Likewise.
19190 (umulsidi3): Likewise.
19191 (mulsi32x16): New pattern.
19192 (mulsi64): Likewise.
19193 (mulsidi64): Likewise.
19194 (umulsidi64): Likewise.
19195 (MUL32x16_REG): Define.
19196 (mul64_600): Use MUL32x16_REG.
19197 (mac64_600): Likewise.
19198 (umul64_600): Likewise.
19199 (umac64_600): Likewise.
19201 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19203 * config/arc/arc.md (mulsi3_700): Make it commutative.
19205 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19207 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19209 (*sign_extendsidi2_insn): Likewise for movstosw.
19211 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19213 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19214 the type of the input discriminant value. Convert the
19215 discriminant value of signedness vary.
19217 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19219 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19220 Add to -Wall section.
19222 2017-06-01 Richard Biener <rguenther@suse.de>
19224 PR middle-end/66313
19225 * fold-const.c (fold_plusminus_mult_expr): If the factored
19226 factor may be zero use a wrapping type for the inner operation.
19227 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19228 and handle moved defs.
19229 (process_assignment): Properly guard the unary op case. Return a
19230 tri-state indicating that moving the stmt before the call may allow
19231 to continue. Pass through to_move.
19232 (find_tail_calls): Handle moving unrelated defs before
19235 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19238 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19239 splitter result in the canonical way.
19241 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19243 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19244 also for 32bit target. Update insn attributes.
19245 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19247 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19249 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19250 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19251 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19252 (_mm_maskz_min_ss): New intrinsics.
19254 2017-05-31 Martin Liska <mliska@suse.cz>
19256 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19257 Change comment style to one we normally use.
19258 (vectorizable_reduction): Likewise.
19259 (vectorizable_induction): Likewise.
19260 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19261 (vectorizable_call): Likewise.
19262 (vectorizable_simd_clone_call): Likewise.
19263 (vectorizable_conversion): Likewise.
19264 (vectorizable_assignment): Likewise.
19265 (vectorizable_shift): Likewise.
19266 (vectorizable_operation): Likewise.
19267 (vectorizable_store): Likewise.
19268 (vectorizable_load): Likewise.
19269 * tree-vectorizer.h: Likewise.
19271 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19273 * passes.c (emergency_dump_function): New.
19274 * tree-pass.h (emergency_dump_function): Declare.
19275 * plugin.c (plugins_internal_error_function): Remove.
19276 * plugin.h (plugins_internal_error_function): Remove declaration.
19277 * toplev.c (internal_error_function): New static function. Use it...
19278 (general_init): ...here.
19280 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19282 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19283 (arc_rtx_costs): Add costs for new patterns.
19284 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19285 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19287 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19289 * tree-ssa-strlen.c (get_next_strinfo): New function.
19290 (get_stridx_plus_constant): Use it.
19291 (zero_length_string): Likewise.
19292 (adjust_related_strinfos): Likewise.
19293 (adjust_last_stmt): Likewise.
19295 2017-05-31 Richard Biener <rguenther@suse.de>
19298 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19299 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19301 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19303 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19304 loop_vinfo argument and use of dependence distance vectors.
19305 Check instead whether the two references differ only in their
19306 initial value and assume that they have the same alignment if the
19307 difference is a multiple of the vector alignment.
19308 (vect_analyze_data_refs_alignment): Update call accordingly.
19310 2017-05-31 Martin Liska <mliska@suse.cz>
19313 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19315 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19317 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19318 (create_intersect_range_checks): Move from ...
19319 * tree-data-ref.c (create_intersect_range_checks_index)
19320 (create_intersect_range_checks): ... to here.
19321 (create_runtime_alias_checks): New function factored from ...
19322 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19323 here. Call above function.
19324 * tree-data-ref.h (create_runtime_alias_checks): New function.
19326 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19328 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19329 segment length for dr_b and compute it in wide_int.
19331 2017-05-31 Richard Biener <rguenther@suse.de>
19333 PR tree-optimization/80906
19334 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19335 and pass through iv_map.
19336 (copy_bb_and_scalar_dependences): Adjust.
19337 (translate_pending_phi_nodes): Likewise.
19338 (copy_loop_close_phi_args): Handle code-generating IVs instead
19341 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19343 * diagnostic-color.c (color_dict): Add "type-diff".
19344 (parse_gcc_colors): Update comment.
19345 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19346 -fdiagnostics-show-template-tree and -fno-elide-type.
19347 (GCC_COLORS): Add type-diff to example.
19349 (-fdiagnostics-show-template-tree): New.
19350 (-fno-elide-type): New.
19351 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19352 the pp_format_decoder callback. Call any m_format_postprocessor's
19354 (pretty_printer::pretty_printer): Initialize
19355 m_format_postprocessor.
19356 (pretty_printer::~pretty_printer): Delete any
19357 m_format_postprocessor.
19358 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19359 (class format_postprocessor): New class.
19360 (struct pretty_printer::format_decoder): Document the new parameters.
19361 (struct pretty_printer::m_format_postprocessor): New field.
19362 * tree-diagnostic.c (default_tree_printer): Update for new
19363 bool and const char ** params.
19364 * tree-diagnostic.h (default_tree_printer): Likewise.
19366 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19368 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19369 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19370 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19371 rs6000_gen_cell_microcode code.
19372 (rs6000_final_prescan_insn): Delete.
19373 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19374 "warn-cell-microcode" entries.
19375 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19376 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19377 throughout. Change cc_reg_not_micro_cr0_operand to
19378 cc_reg_not_cr0_operand throughout.
19379 (*extendhi<mode>2_noload): Delete.
19380 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19381 (mwarn-cell-microcode): Delete.
19382 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19383 -mgen-cell-microcode and -mwarn-cell-microcode.
19385 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19388 * config/i386/constraints.md (Yd): New constraint.
19390 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19391 and (?Yd, r) alternatives. Update insn attributes.
19392 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19393 and (?*Yd, r) alternatives. Update insn attributes.
19394 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19396 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19398 * gimplify.c (gimplify_modify_expr): Don't create a
19399 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19402 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19404 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19406 2017-05-30 Richard Biener <rguenther@suse.de>
19408 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19409 and reduc_def fields.
19410 (STMT_VINFO_REDUC_TYPE): New define.
19411 (STMT_VINFO_REDUC_DEF): Likewise.
19412 (vect_force_simple_reduction): Adjust prototype.
19413 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19414 (vect_is_simple_reduction): Remove check_reduction argument.
19415 (vect_force_simple_reduction): Adjust and set
19416 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19417 (vectorizable_reduction): Do not re-do reduction analysis
19418 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19419 * tree-parloops.c (gather_scalar_reductions): Adjust.
19421 2017-05-30 Richard Biener <rguenther@suse.de>
19423 PR middle-end/80901
19424 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19427 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19429 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19430 Introduce unknown_misalignment parameter and remove vf.
19431 (vect_peeling_hash_get_lowest_cost):
19432 Pass unknown_misalignment parameter.
19433 (vect_enhance_data_refs_alignment):
19434 Fix unsupportable data ref treatment.
19436 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19438 * tree-vect-data-refs.c (vect_get_data_access_cost):
19439 Workaround for SLP handling.
19440 (vect_enhance_data_refs_alignment):
19441 Compute costs for doing no peeling at all, compare to the best
19442 peeling costs so far and avoid peeling if cheaper.
19444 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19446 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19447 Return peeling info and set costs to zero for unlimited cost
19449 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19450 with unknown misalignment. Compute and costs for unknown
19451 misalignment, compare them to the costs for known misalignment
19452 and choose the cheapest for peeling.
19454 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19456 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19457 (vect_get_peeling_costs_all_drs): Create function.
19458 (vect_peeling_hash_get_lowest_cost):
19459 Use vect_get_peeling_costs_all_drs.
19460 (vect_peeling_supportable): Create function.
19461 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19463 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19465 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19466 DR_HAS_NEGATIVE_STEP.
19467 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19468 (vect_enhance_data_refs_alignment): Use.
19469 (vect_duplicate_ssa_name_ptr_info): Use.
19470 * tree-vectorizer.h (dr_misalignment): Use.
19471 (known_alignment_for_access_p): Use.
19473 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19476 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19478 (has_section_name): New function.
19480 2017-05-30 Martin Liska <mliska@suse.cz>
19483 * auto-profile.c (get_function_decl_from_block): Fix
19486 2017-05-30 Richard Biener <rguenther@suse.de>
19488 PR middle-end/80876
19489 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19491 2017-05-30 Martin Liska <mliska@suse.cz>
19493 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19494 * dumpfile.h (struct dump_file_info): Remove ctors.
19496 2017-05-30 Martin Liska <mliska@suse.cz>
19498 * predict.def: Fix GNU coding style.
19500 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19502 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19503 Mark 'to' argument with ATTRIBUTE_UNUSED.
19505 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19507 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19508 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19509 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19512 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19514 * doc/install.texi (Options specification): Restore entry of
19515 --enable-sjlj-exceptions.
19517 2017-05-27 Michael Eager <eager@eagercon.com>
19520 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19522 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19524 * config/microblaze/microblaze.h
19525 (FIXED_REGISTERS): Update in macro.
19526 (CALL_USED_REGISTERS): Update in macro.
19528 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19530 * doc/install.texi: Add links to macOS binary distributions.
19532 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19536 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19538 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19540 2017-05-26 Martin Liska <mliska@suse.cz>
19542 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19544 2017-05-26 Martin Liska <mliska@suse.cz>
19546 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19547 always leading ';; '.
19548 (dump_bb_info): Likewise.
19549 (brief_dump_cfg): Likewise.
19550 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19551 * dumpfile.c: Remove usage of TDF_VERBOSE.
19552 * dumpfile.h (enum dump_kind): Likewise.
19553 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19554 * print-tree.c (debug_verbose): Remove.
19555 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19556 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19557 * tree-diagnostic.c (default_tree_printer): Replace
19558 TDF_DIAGNOSTIC with TDF_SLIM.
19560 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19562 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19563 in parameter loop, rather than loop_vinfo.
19564 (create_intersect_range_checks): Ditto.
19565 (vect_create_cond_for_alias_checks): Update call to above functions.
19567 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19569 PR tree-optimization/80815
19570 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19571 for merging runtime alias checks. Handle negative DR_STEPs.
19573 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19575 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19577 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19578 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19579 out code pruning runtime alias checks.
19580 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19581 factored out from above.
19582 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19584 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19586 (prune_runtime_alias_test_list): New decalaration.
19588 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19590 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19591 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19592 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19593 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19594 (operator==, comp_dr_with_seg_len_pair): Ditto.
19595 (vect_prune_runtime_alias_test_list): Ditto.
19597 2017-05-26 Martin Liska <mliska@suse.cz>
19600 * params.def: Bound partial-inlining-entry-probability param.
19602 2017-05-26 Marek Polacek <polacek@redhat.com>
19605 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19608 2017-05-26 Richard Biener <rguenther@suse.de>
19610 PR tree-optimization/80842
19611 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19614 2017-05-26 Richard Biener <rguenther@suse.de>
19616 PR tree-optimization/80844
19617 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19619 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19621 * doc/md.texi (Machine Constraints): Update x86 family
19622 machine constraints section to match 'config/i386/constraints.md'.
19624 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19626 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19628 2017-05-25 Nathan Sidwell <nathan@acm.org>
19630 * doc/invoke.texi (--enable-languages): Update documentation.
19632 2017-05-25 Martin Liska <mliska@suse.cz>
19634 * dumpfile.c: Add TDF_FOLDING.
19635 * dumpfile.h (enum dump_kind): Likewise.
19636 * genmatch.c (dt_simplify::gen_1): Use it.
19638 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19640 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19642 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19644 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19645 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19647 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19649 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19650 * match.pd (X == C): Rewrite it here.
19651 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19652 with_certain_nonzero_bits2): New predicates.
19653 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19655 2017-05-24 Nathan Sidwell <nathan@acm.org>
19657 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19660 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19663 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19665 * config/powerpcspe: New port. Files are copied from the rs6000
19666 port, with "rs6000" in filenames replaced by "powerpcspe".
19668 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19670 PR rtl-optimization/80754
19671 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19673 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19675 * config/sparc/sparc.md (length): Return the correct value for -mflat
19676 sibcalls to match output_sibcall.
19678 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19682 * config/rs6000/rs6000.c (struct machine_function): Add new field
19684 (rs6000_get_separate_components): Init that field, use it.
19685 (rs6000_components_for_bb): Use the field.
19687 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19689 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19691 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19693 PR middle-end/80823
19694 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19696 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19699 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19700 address against address_operand predicate.
19701 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19703 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19705 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19706 parameters passed indirectly.
19708 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19710 * config/i386/i386.md (*movdi_internal): Remove SSE4
19711 alternative 18 (?r, *v). Update insn attributes.
19712 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19713 Update insn attributes.
19714 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19715 Update insn attributes.
19716 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19717 alternative 1 (r, v). Remove isa attribute.
19718 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19719 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19720 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19722 2017-05-23 Tom de Vries <tom@codesourcery.com>
19724 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19727 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19729 * cgraphunit.c (symbol_table::process_new_functions): Update.
19730 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19731 (inline_generate_summary): Rename to ...
19732 (ipa_fn_summary_generate): ... this one.
19733 (inline_read_summary): Rename to ...
19734 (ipa_fn_summary_read): ... this one.
19735 (inline_write_summary): Rename to ...
19736 (ipa_fn_summary_write): ... this one.
19737 (inline_free_summary): Rename to ...
19738 (ipa_free_fn_summary): ... this one.
19739 (pass_data_local_fn_summary, pass_local_fn_summary,
19740 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19741 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19742 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19743 make_pass_ipa_fn_summary): New.
19744 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19745 inline_write_summary, inline_free_summary): Remove.
19746 (ipa_free_fn_summary) : New.
19747 * ipa-inline.c (ipa_inline): Update.
19748 (pass_ipa_inline): Do not generate summaries.
19749 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19751 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19752 and add pass_ipa_fn_summary.
19753 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19755 (make_pass_inline_parameters): Remove.
19757 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19759 * omp-low.c (struct omp_context): Remove "default_kind" member.
19762 * omp-offload.c (execute_oacc_device_lower): Remove the
19763 parallelism dimensions function attributes for unparallelized
19764 OpenACC kernels constructs.
19766 2017-05-23 Martin Liska <mliska@suse.cz>
19768 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19770 (cgraph_edge::make_speculative): Likewise.
19771 (cgraph_edge::resolve_speculation): Likewise.
19772 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19773 (cgraph_node::dump): Likewise.
19774 * cgraph.h: Likewise.
19775 * cgraphunit.c (analyze_functions): Likewise.
19776 (symbol_table::compile): Likewise.
19777 * ipa-cp.c (print_all_lattices): Likewise.
19778 (determine_versionability): Likewise.
19779 (initialize_node_lattices): Likewise.
19780 (ipcp_verify_propagated_values): Likewise.
19781 (estimate_local_effects): Likewise.
19782 (update_profiling_info): Likewise.
19783 (create_specialized_node): Likewise.
19784 (perhaps_add_new_callers): Likewise.
19785 (decide_about_value): Likewise.
19786 (decide_whether_version_node): Likewise.
19787 (identify_dead_nodes): Likewise.
19788 (ipcp_store_bits_results): Likewise.
19789 * ipa-devirt.c (dump_targets): Likewise.
19790 (ipa_devirt): Likewise.
19791 * ipa-icf.c (sem_item::dump): Likewise.
19792 (sem_function::equals): Likewise.
19793 (sem_variable::equals): Likewise.
19794 (sem_item_optimizer::read_section): Likewise.
19795 (sem_item_optimizer::execute): Likewise.
19796 (congruence_class::dump): Likewise.
19797 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19798 (dump_inline_summary): Likewise.
19799 (estimate_node_size_and_time): Likewise.
19800 (inline_analyze_function): Likewise.
19801 * ipa-inline-transform.c (inline_call): Likewise.
19802 * ipa-inline.c (report_inline_failed_reason): Likewise.
19803 (want_early_inline_function_p): Likewise.
19804 (edge_badness): Likewise.
19805 (update_edge_key): Likewise.
19806 (inline_small_functions): Likewise.
19807 * ipa-profile.c (ipa_profile): Likewise.
19808 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19809 (ipa_make_edge_direct_to_target): Likewise.
19810 (remove_described_reference): Likewise.
19811 (ipa_impossible_devirt_target): Likewise.
19812 (propagate_controlled_uses): Likewise.
19813 (ipa_print_node_params): Likewise.
19814 (ipcp_transform_function): Likewise.
19815 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19816 (propagate_pure_const): Likewise.
19817 * ipa-reference.c (generate_summary): Likewise.
19818 (read_write_all_from_decl): Likewise.
19819 (propagate): Likewise.
19820 (ipa_reference_read_optimization_summary): Likewise.
19821 * ipa-utils.c (ipa_merge_profiles): Likewise.
19822 * ipa.c (walk_polymorphic_call_targets): Likewise.
19823 (symbol_table::remove_unreachable_nodes): Likewise.
19824 (ipa_single_use): Likewise.
19825 * passes.c (execute_todo): Likewise.
19826 * predict.c (drop_profile): Likewise.
19827 * symtab.c (symtab_node::get_dump_name): New function.
19828 (symtab_node::dump_name): Likewise.
19829 (symtab_node::dump_asm_name): Likewise.
19830 (symtab_node::dump_references): Likewise.
19831 (symtab_node::dump_referring): Likewise.
19832 (symtab_node::dump_base): Likewise.
19833 (symtab_node::debug_symtab): Likewise.
19834 * tree-sra.c (convert_callers_for_node): Likewise.
19835 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19836 * value-prof.c (init_node_map): Likewise.
19838 2017-05-23 Martin Liska <mliska@suse.cz>
19840 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19841 and symtab_node::debug_symtab to symbol_table::debug.
19842 * cgraphunit.c (analyze_functions): Use the renamed function.
19843 (symbol_table::compile): Likewise.
19844 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19845 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19846 * passes.c (execute_todo): Likewise.
19847 * symtab.c (symbol_table::dump): New function.
19848 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19850 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19852 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19853 that nonconst implies exec.
19855 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19857 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19858 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19859 (inline_edge_summary_vec): Turn into ...
19860 (ipa_call_summaries): ... this one.
19861 (redirect_to_unreachable, edge_set_predicate,
19862 evaluate_properties_for_edge, inline_summary_alloc,
19863 reset_ipa_call_summary, reset_inline_summary,
19864 inline_summary_t::duplicate): Update.
19865 (inline_edge_duplication_hook): Turn to ...
19866 (ipa_call_summary_t::duplicate): ... this one.
19867 (inline_edge_removal_hook): Turn to ...
19868 (ipa_call_summary_t::remove): ... this one.
19869 (dump_inline_edge_summary): Turn to ...
19870 (dump_ipa_call_summary): ... this one.
19871 (estimate_function_body_sizes): Update.
19872 (inline_update_callee_summaries): Update.
19873 (remap_edge_change_prob): Update.
19874 (remap_edge_summaries): Update.
19875 (inline_merge_summary): Update.
19876 (do_estimate_edge_time): Update.
19877 (inline_generate_summary): Update.
19878 (inline_read_section): Update.
19879 (inline_read_summary): Update.
19880 (inline_free_summary): Update.
19881 * ipa-inline.c (can_inline_edge_p): Update.
19882 (compute_inlined_call_time): Update.
19883 (want_inline_small_function_p): Update.
19884 (edge_badness): Update.
19885 (early_inliner): Update.
19886 * ipa-inline.h (inline_edge_summary): Turn to ...
19887 (ipa_call_summary): ... this one.
19888 (ipa_call_summary_t): New class.
19889 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19890 (ipa_call_summaries): New.
19891 (inline_edge_summary): Remove.
19892 (estimate_edge_growth): Update.
19893 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19894 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19895 * ipa-split.c (execute_split_functions): Update.
19896 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19898 2017-05-23 Tom de Vries <tom@codesourcery.com>
19900 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19901 attributes): Document rdrand effective target.
19903 2017-05-23 Tom de Vries <tom@codesourcery.com>
19905 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19906 attributes): Sort alphabetically.
19908 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19910 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19912 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19915 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19916 V2DF/V2DI splat into two separate patterns, one that handles
19917 registers, and the other that only handles memory. Drop support
19918 for splatting from a GPR on ISA 2.07 and then splitting the
19919 splat into direct move and splat.
19920 (vsx_splat_<mode>_reg): Likewise.
19921 (vsx_splat_<mode>_mem): Likewise.
19923 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19925 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19927 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19929 PR middle-end/80809
19930 * omp-low.c (finish_taskreg_remap): New function.
19931 (finish_taskreg_scan): If unit size of ctx->record_type
19932 is non-constant, unshare the size expression and replace
19933 decls in it with possible outer var refs.
19935 PR middle-end/80809
19936 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19937 GOVD_SHARED rather than GOVD_PRIVATE with it.
19938 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19939 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19941 PR middle-end/80853
19942 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19943 as last argument to build_outer_var_ref for pointer bases of array
19944 section reductions.
19946 2017-05-19 Martin Sebor <msebor@redhat.com>
19948 * print-tree.c (print_node): Print DECL_READ_P flag.
19950 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19952 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19953 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19954 * cgraph.c: Likewise.
19955 * cgraphunit.c: Likewise.
19956 * gengtype.c: Likewise.
19957 * ipa-cp.c: Likewise.
19958 * ipa-devirt.c: Likewise.
19959 * ipa-icf.c: Likewise.
19960 * ipa-predicate.c: Likewise.
19961 * ipa-profile.c: Likewise.
19962 * ipa-prop.c: Likewise.
19963 * ipa-split.c: Likewise.
19965 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19966 edge_predicate_pool, dump_inline_hints,
19967 inline_summary::account_size_time, redirect_to_unreachable,
19968 edge_set_predicate, set_hint_predicate,
19969 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19970 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19971 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19972 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19973 ipa_call_summary_t::remove, initialize_growth_caches,
19974 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19975 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19976 mark_modified, unmodified_parm_1, unmodified_parm,
19977 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19978 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19979 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19980 will_be_nonconstant_predicate, record_modified_bb_info,
19981 get_minimal_bb, record_modified, param_change_prob,
19982 phi_result_unknown_predicate, predicate_for_phi_result,
19983 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19984 estimate_function_body_sizes, compute_inline_parameters,
19985 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19986 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19987 inline_update_callee_summaries, remap_edge_change_prob,
19988 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19989 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19990 inline_analyze_function, inline_summary_t::insert,
19991 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19992 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19993 inline_free_summary): Move to ipa-fnsummary.h
19994 (predicate_t): Remove.
19995 * ipa-fnsummary.c: New file.
19996 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19997 (enum inline_hints_vals, inline_hints, agg_position_info,
19998 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19999 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20000 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20001 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20002 inline_read_summary, inline_write_summary, inline_free_summary,
20003 inline_analyze_function, initialize_inline_failed,
20004 inline_merge_summary, inline_update_overall_summary,
20005 compute_inline_parameters): Move to ipa-fnsummary.h
20006 * ipa-fnsummary.h: New file.
20007 * ipa-inline-transform.h: Include ipa-inline.h.
20008 * ipa-inline.c: LIkewise.
20010 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20012 * ipa-inline.c (edge_badness): Use inlined_time instead of
20013 inline_summaries->get.
20015 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20017 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20019 2017-05-22 Nathan Sidwell <nathan@acm.org>
20021 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20022 (fdump-lang): Document 'raw' option.
20023 * dumpfile.h (TDI_tu): Delete.
20024 * dumpfile.c (dump_files): Remove translation-unit.
20025 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20027 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20029 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20030 command option from $(AWK) call.
20031 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20032 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20033 [FORMAT]: Remove handling of variable.
20034 * config/avr/t-multilib: Regenerate.
20036 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20038 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20040 (dump_inline_summary): Do not print self_time.
20041 (estimate_function_body_sizes): Do not set self_time.
20042 (compute_inline_parameters): Likewise.
20043 (inline_read_section, inline_write_summary): Do not stream self_time.
20044 * ipa-inline.h (inline_summary): Drop self_time.
20046 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20048 * ipa-inline-analysis.c (account_size_time): Rename to ...
20049 (inline_summary::account_size_time): ... this one.
20050 (reset_ipa_call_summary): Turn to ...
20051 (ipa_call_summary::reset): ... this one.
20052 (reset_inline_summary): Turn to ...
20053 (inline_summary::reset): ... this one.
20054 (inline_summary_t::remove): Update.
20055 (inline_summary_t::duplicate): Update.
20056 (ipa_call_summary_t::remove): Update.
20057 (dump_inline_summary): Update.
20058 (estimate_function_body_sizes): Update.
20059 (compute_inline_parameters): Update.
20060 (estimate_node_size_and_time): Update.
20061 (inline_merge_summary): Update.
20062 (inline_update_overall_summary): Update.
20063 (inline_read_section): Update.
20064 (inline_write_summary): Update.
20065 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20066 add account_size_time and reset member functions.
20067 (ipa_call_summary): Add reset function.
20068 * ipa-predicate.h (predicate::operator &): Constify.
20070 2017-05-22 Richard Biener <rguenther@suse.de>
20072 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20074 2017-05-19 Jason Merrill <jason@redhat.com>
20076 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20078 2017-05-19 Marek Polacek <polacek@redhat.com>
20081 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20082 TYPE_OVERFLOW_WRAPS checks.
20084 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20086 * tree-core.h (enum omp_clause_default_kind): Add
20087 "OMP_CLAUSE_DEFAULT_PRESENT".
20088 * tree-pretty-print.c (dump_omp_clause): Handle it.
20089 * gimplify.c (enum gimplify_omp_var_data): Add
20090 "GOVD_MAP_FORCE_PRESENT".
20091 (gimplify_adjust_omp_clauses_1): Map it to
20092 "GOMP_MAP_FORCE_PRESENT".
20093 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20095 * gimplify.c (oacc_default_clause): Clarify.
20097 2017-05-19 Nathan Sidwell <nathan@acm.org>
20099 LANG_HOOK_REGISTER_DUMPS
20100 * toplev.c (general_init): Call register dump lang hook.
20101 * doc/invoke.texi: Document -fdump-lang option family.
20102 * dumpfile.c (dump_files): Remove class dump here.
20103 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20104 * dumpfile.h (tree_dump_index): Remove TDI_class.
20105 * langhooks-def.h (lhd_register_dumps): Declare.
20106 (LANG_HOOKS_REGISTER_DUMPS): Define.
20107 (LANG_HOOKS_INITIALIZER): Add it.
20108 * langhooks.c (lhd_register_dumps): Define.
20109 * langhooks.h (struct lang_hooks): Add register_dumps.
20111 2017-05-19 Nathan Sidwell <nathan@acm.org>
20113 * context.h (context::set_passes): New.
20114 * context.c (context::context): Do not create pass manager.
20115 * toplev.c (general_init): Create pass manager here.
20117 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20119 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20120 use this splitter if two add or or instructions would also work for
20121 the constant we want to generate.
20123 2017-05-19 Richard Biener <rguenther@suse.de>
20126 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20127 predicate evaluation.
20129 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20131 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20133 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20136 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20138 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20139 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20140 (inline_edge_summary_vec): Turn into ...
20141 (ipa_call_summaries): ... this one.
20142 (redirect_to_unreachable, edge_set_predicate,
20143 evaluate_properties_for_edge, inline_summary_alloc,
20144 reset_ipa_call_summary, reset_inline_summary,
20145 inline_summary_t::duplicate): Update.
20146 (inline_edge_duplication_hook): Turn to ...
20147 (ipa_call_summary_t::duplicate): ... this one.
20148 (inline_edge_removal_hook): Turn to ...
20149 (ipa_call_summary_t::remove): ... this one.
20150 (dump_inline_edge_summary): Turn to ...
20151 (dump_ipa_call_summary): ... this one.
20152 (estimate_function_body_sizes): Update.
20153 (inline_update_callee_summaries): Update.
20154 (remap_edge_change_prob): Update.
20155 (remap_edge_summaries): Update.
20156 (inline_merge_summary): Update.
20157 (do_estimate_edge_time): Update.
20158 (inline_generate_summary): Update.
20159 (inline_read_section): Update.
20160 (inline_read_summary): Update.
20161 (inline_free_summary): Update.
20162 * ipa-inline.c (can_inline_edge_p): Update.
20163 (compute_inlined_call_time): Update.
20164 (want_inline_small_function_p): Update.
20165 (edge_badness): Update.
20166 (early_inliner): Update.
20167 * ipa-inline.h (inline_edge_summary): Turn to ...
20168 (ipa_call_summary): ... this one.
20169 (ipa_call_summary_t): New class.
20170 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20171 (ipa_call_summaries): New.
20172 (inline_edge_summary): Remove.
20173 (estimate_edge_growth): Update.
20174 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20175 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20176 * ipa-split.c (execute_split_functions): Update.
20177 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20179 2017-05-19 Richard Biener <rguenther@suse.de>
20181 PR middle-end/80764
20182 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20184 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20186 * config/rs6000/rs6000.c (struct machine_function): Add field
20187 fpr_is_wrapped_separately.
20188 (rs6000_get_separate_components): Use 64 components. Handle the
20189 new FPR components.
20190 (rs6000_components_for_bb): Handle the FPR components.
20191 (rs6000_emit_prologue_components): Handle the FPR components.
20192 (rs6000_emit_epilogue_components): Handle the FPR components.
20193 (rs6000_set_handled_components): Handle the FPR components.
20194 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20195 that are already separately shrink-wrapped.
20196 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20197 that are already separately shrink-wrapped.
20199 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20202 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20205 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20206 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20207 cases where the register allocator uses a move and an offsettable
20208 memory operation to/from a FPR register on ISA 2.06/2.07.
20209 (define_peephole2 for Altivec d-form store): Likewise.
20211 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20214 * config/i386/mmx.md (*mov<mode>_internal): Enable
20215 alternatives 11, 12, 13 and 14 also for 32bit targets.
20216 Remove alternatives 15, 16, 17 and 18.
20217 * config/i386/sse.md (vec_concatv2di): Change
20218 alternative (!x, *y) to (x, ?!*Yn).
20220 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20222 * dumpfile.h (enum dump_kind): Remove stray comma.
20224 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20226 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20227 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20228 predicate::num_conditions
20229 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20230 (CHANGED): turn into predicate::changed.
20231 (agg_position_info): Move to ipa-predicate.h
20232 (add_condition, predicate::add_clause, predicate::operator &=,
20233 predicate::or_with, predicate::evaluate, predicate::probability,
20234 dump_condition, dump_clause, predicate::dump,
20235 predicate::remap_after_duplication, predicate::remap_after_inlining,
20236 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20237 (evaluate_conditions_for_known_args): Update.
20238 (set_cond_stmt_execution_predicate): Update.
20239 * ipa-inline.h: Include ipa-predicate.h
20240 (condition, inline_param_summary, conditions, agg_position_info,
20241 predicate): Move to ipa-predicate.h
20242 * ipa-predicate.c: New file.
20243 * ipa-predicate.h: New file.
20245 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20247 * final.c (leaf_function_p): Check we are not in a sequence.
20249 2017-05-18 Martin Liska <mliska@suse.cz>
20251 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20252 * dumpfile.c (dump_register): Use new enum dump_kind.
20253 (get_dump_file_name): Likewise.
20254 (dump_enable_all): Likewise.
20255 (dump_switch_p_1): Likewise.
20256 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20257 * dumpfile.h (enum dump_kind): New enum type.
20258 (struct dump_file_info): Create constructor and
20259 format fields and comments.
20260 * passes.c (pass_manager::register_one_dump_file):
20262 * statistics.c (statistics_early_init): Likewise.
20263 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20264 TDF_TREE with TDF_SLIM.
20265 (gather_memory_references_ref): Likewise.
20267 2017-05-18 Martin Liska <mliska@suse.cz>
20269 * vec.h (struct vnull): Use it.
20271 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20273 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20274 (true_predicate, false_predicate, true_predicate_p,
20275 false_predicate_p): Remove.
20276 (single_cond_predicate, not_inlined_predicate): Turn to member function
20278 (add_condition): Update.
20279 (add_clause): Turn to...
20280 (predicate::add_clause): ... this one; update; allow passing NULL
20282 (and_predicates): Turn to ...
20283 (predicate::operator &=): ... this one.
20284 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20285 (or_predicates): Turn to ...
20286 (predicate::or_with): ... this one.
20287 (evaluate_predicate): Turn to ...
20288 (predicate::evaluate): ... this one.
20289 (predicate_probability): Turn to ...
20290 (predicate::probability): ... this one.
20291 (dump_condition): Update.
20292 (dump_predicate): Turn to ...
20293 (predicate::dump): ... this one.
20294 (account_size_time): Update.
20295 (edge_set_predicate): Update.
20296 (set_hint_predicate): UPdate.
20297 (evaluate_conditions_for_known_args): Update.
20298 (evaluate_properties_for_edge): Update.
20299 (remap_predicate_after_duplication): Turn to...
20300 (predicate::remap_after_duplication): ... this one.
20301 (remap_hint_predicate_after_duplication): Update.
20302 (inline_summary_t::duplicate): UPdate.
20303 (dump_inline_edge_summary): Update.
20304 (dump_inline_summary): Update.
20305 (set_cond_stmt_execution_predicate): Update.
20306 (set_switch_stmt_execution_predicate): Update.
20307 (compute_bb_predicates): Update.
20308 (will_be_nonconstant_expr_predicate): Update.
20309 (will_be_nonconstant_predicate): Update.
20310 (phi_result_unknown_predicate): Update.
20311 (predicate_for_phi_result): Update.
20312 (array_index_predicate): Update.
20313 (estimate_function_body_sizes): Update.
20314 (estimate_node_size_and_time): Update.
20315 (estimate_ipcp_clone_size_and_time): Update.
20316 (remap_predicate): Rename to ...
20317 (predicate::remap_after_inlining): ... this one.
20318 (remap_hint_predicate): Update.
20319 (inline_merge_summary): Update.
20320 (inline_update_overall_summary): Update.
20321 (estimate_size_after_inlining): Update.
20322 (read_predicate): Rename to ...
20323 (predicate::stream_in): ... this one.
20324 (read_inline_edge_summary): Update.
20325 (write_predicate): Rename to ...
20326 (predicate::stream_out): ... this one.
20327 (write_inline_edge_summary): Update.
20328 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20329 (clause_t): Turn to uint32_t
20330 (predicate): Turn to class; implement constructor and operators
20332 (size_time_entry): Update.
20333 (inline_summary): Update.
20334 (inline_edge_summary): Update.
20336 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20338 * fold-const.c (fold_binary_loc): Move transformation...
20339 * match.pd (C - X CMP X): ... here.
20341 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20343 * config/sparc/sparc.c (sparc_option_override): Set function
20344 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20345 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20347 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20349 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20351 2017-05-18 Marek Polacek <polacek@redhat.com>
20354 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20355 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20356 gimple_assign_load_p.
20358 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20360 PR middle-end/80692
20361 * real.c (do_compare): Give decimal_do_compare preference over
20362 comparing just the signs.
20364 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20366 * doc/md.texi (Canonicalization of Instructions): Describe the
20367 canonical form of instructions that inherently set a condition
20370 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20372 PR middle-end/80775
20373 * tree-cfg.c: Move deletion of unreachable case statements to after
20374 the merging of consecutive case labels.
20376 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20378 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20379 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20380 restoring of callee-saved registers.
20382 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20384 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20385 * config/visium/visium.c (single_set_and_flags): Likewise.
20386 * config/visium/visium.md (Substitutions): Likewise.
20388 2017-05-17 Martin Liska <mliska@suse.cz>
20390 * cfg.c: Introduce dump_flags_t type and
20391 use it instead of int type.
20393 * cfghooks.c: Likewise.
20394 * cfghooks.h (struct cfg_hooks): Likewise.
20395 * cfgrtl.c: Likewise.
20396 * cfgrtl.h: Likewise.
20397 * cgraph.c (cgraph_node::get_body): Likewise.
20398 * coretypes.h: Likewise.
20399 * domwalk.c: Likewise.
20400 * domwalk.h: Likewise.
20401 * dumpfile.c (struct dump_option_value_info): Likewise.
20402 (dump_enable_all): Likewise.
20403 (dump_switch_p_1): Likewise.
20404 (opt_info_switch_p): Likewise.
20405 * dumpfile.h (enum tree_dump_index): Likewise.
20406 (struct dump_file_info): Likewise.
20407 * genemit.c: Likewise.
20408 * generic-match-head.c: Likewise.
20409 * gengtype.c (open_base_files): Likewise.
20410 * gimple-pretty-print.c: Likewise.
20411 * gimple-pretty-print.h: Likewise.
20412 * graph.c (print_graph_cfg): Likewise.
20413 * graphite-scop-detection.c (dot_all_sese): Likewise.
20414 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20415 * loop-unroll.c (report_unroll): Likewise.
20416 * passes.c (pass_manager::register_one_dump_file): Likewise.
20417 * print-tree.c: Likewise.
20418 * statistics.c: Likewise.
20419 * tree-cfg.c: Likewise.
20420 * tree-cfg.h: Likewise.
20421 * tree-dfa.c: Likewise.
20422 * tree-dfa.h: Likewise.
20423 * tree-dump.c (dump_function): Likewise.
20424 * tree-dump.h (struct dump_info): Likewise.
20425 * tree-pretty-print.c: Likewise.
20426 * tree-pretty-print.h: Likewise.
20427 * tree-ssa-live.c: Likewise.
20428 * tree-ssa-live.h: Likewise.
20429 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20430 * tree-vect-loop.c: Likewise.
20431 * tree-vect-slp.c: Likewise.
20433 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20434 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20436 PR tree-optimization/80457
20437 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20438 of all arguments to a statement as scalar_to_vec operations.
20439 (vectorizable_call): Adjust call to vect_model_simple_cost for
20441 (vectorizable_conversion): Likewise.
20442 (vectorizable_assignment): Likewise.
20443 (vectorizable_shift): Likewise.
20444 (vectorizable_operation): Likewise.
20445 (vectorizable_comparison): Likewise.
20446 (vect_is_simple_cond): Record the def types for operands.
20447 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20448 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20449 for statement argument count.
20451 2017-05-16 Carl Love <cel@us.ibm.com>
20453 * config/rs6000/rs6000-c: Add support for built-in functions
20454 vector unsigned long long vec_bperm (vector unsigned long long,
20455 vector unsigned char)
20456 vector signed long long vec_mule (vector signed int,
20458 vector unsigned long long vec_mule (vector unsigned int,
20459 vector unsigned int)
20460 vector signed long long vec_mulo (vector signed int,
20462 vector unsigned long long vec_mulo (vector unsigned int,
20463 vector unsigned int)
20464 vector signed char vec_sldw (vector signed char,
20465 vector signed char,
20467 vector unsigned char vec_sldw (vector unsigned char,
20468 vector unsigned char,
20470 vector signed short vec_sldw (vector signed short,
20471 vector signed short,
20473 vector unsigned short vec_sldw (vector unsigned short,
20474 vector unsigned short,
20476 vector signed int vec_sldw (vector signed int,
20479 vector unsigned int vec_sldw (vector unsigned int,
20480 vector unsigned int,
20482 vector signed long long vec_sldw (vector signed long long,
20483 vector signed long long,
20485 vector unsigned long long vec_sldw (vector unsigned long long,
20486 vector unsigned long long,
20488 * config/rs6000/rs6000-c: Add support for built-in functions
20489 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20490 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20491 * doc/extend.texi: Update the built-in documentation for the
20492 new built-in functions.
20494 2017-05-16 Marek Polacek <polacek@redhat.com>
20498 * tree.c (save_expr): Don't fold the expression.
20500 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20502 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20503 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20504 and (?*y,m). Update insn attributes.
20506 2017-05-16 Martin Liska <mliska@suse.cz>
20508 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20509 flags argument of print_gimple_stmt, print_gimple_expr,
20510 print_generic_stmt and print_generic_expr.
20511 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20512 * coretypes.h: Likewise.
20513 * except.c (dump_eh_tree): Likewise.
20514 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20515 * gimple-pretty-print.h: Likewise.
20516 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20517 (backprop::push_to_worklist): Likewise.
20518 (backprop::pop_from_worklist): Likewise.
20519 (backprop::process_use): Likewise.
20520 (backprop::intersect_uses): Likewise.
20521 (note_replacement): Likewise.
20522 * gimple-ssa-store-merging.c
20523 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20524 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20525 (pass_store_merging::execute): Likewise.
20526 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20527 (ssa_base_cand_dump_callback): Likewise.
20528 (dump_incr_vec): Likewise.
20529 (replace_refs): Likewise.
20530 (replace_mult_candidate): Likewise.
20531 (create_add_on_incoming_edge): Likewise.
20532 (create_phi_basis): Likewise.
20533 (insert_initializers): Likewise.
20534 (all_phi_incrs_profitable): Likewise.
20535 (introduce_cast_before_cand): Likewise.
20536 (replace_one_candidate): Likewise.
20537 * gimplify.c (gimplify_expr): Likewise.
20538 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20539 (set_rename): Likewise.
20540 (rename_uses): Likewise.
20541 (copy_loop_phi_nodes): Likewise.
20542 (add_close_phis_to_merge_points): Likewise.
20543 (copy_loop_close_phi_args): Likewise.
20544 (copy_cond_phi_args): Likewise.
20545 (graphite_copy_stmts_from_block): Likewise.
20546 (translate_pending_phi_nodes): Likewise.
20547 * graphite-poly.c (print_pdr): Likewise.
20548 (dump_gbb_cases): Likewise.
20549 (dump_gbb_conditions): Likewise.
20550 (print_scop_params): Likewise.
20551 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20552 (build_cross_bb_scalars_use): Likewise.
20553 (gather_bbs::before_dom_children): Likewise.
20554 * hsa-dump.c (dump_hsa_immed): Likewise.
20555 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20556 (get_replacement_map): Likewise.
20557 * ipa-inline-analysis.c (dump_condition): Likewise.
20558 (estimate_function_body_sizes): Likewise.
20559 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20560 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20561 * ipa-prop.c (ipa_dump_param): Likewise.
20562 (ipa_print_node_jump_functions_for_edge): Likewise.
20563 (ipa_modify_call_arguments): Likewise.
20564 (ipa_modify_expr): Likewise.
20565 (ipa_dump_param_adjustments): Likewise.
20566 (ipa_dump_agg_replacement_values): Likewise.
20567 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20568 * ipa-pure-const.c (check_stmt): Likewise.
20569 (pass_nothrow::execute): Likewise.
20570 * ipa-split.c (execute_split_functions): Likewise.
20571 * omp-offload.c (dump_oacc_loop_part): Likewise.
20572 (dump_oacc_loop): Likewise.
20573 * trans-mem.c (tm_log_emit): Likewise.
20574 (tm_memopt_accumulate_memops): Likewise.
20575 (dump_tm_memopt_set): Likewise.
20576 (dump_tm_memopt_transform): Likewise.
20577 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20578 (print_loop): Likewise.
20579 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20580 (chkp_gather_checks_info): Likewise.
20581 (chkp_get_check_result): Likewise.
20582 (chkp_remove_check_if_pass): Likewise.
20583 (chkp_use_outer_bounds_if_possible): Likewise.
20584 (chkp_reduce_bounds_lifetime): Likewise.
20585 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20586 (chkp_mark_completed_bounds): Likewise.
20587 (chkp_register_incomplete_bounds): Likewise.
20588 (chkp_mark_invalid_bounds): Likewise.
20589 (chkp_maybe_copy_and_register_bounds): Likewise.
20590 (chkp_build_returned_bound): Likewise.
20591 (chkp_get_bound_for_parm): Likewise.
20592 (chkp_build_bndldx): Likewise.
20593 (chkp_get_bounds_by_definition): Likewise.
20594 (chkp_generate_extern_var_bounds): Likewise.
20595 (chkp_get_bounds_for_decl_addr): Likewise.
20596 * tree-chrec.c (chrec_apply): Likewise.
20597 * tree-data-ref.c (dump_data_reference): Likewise.
20598 (dump_subscript): Likewise.
20599 (dump_data_dependence_relation): Likewise.
20600 (analyze_overlapping_iterations): Likewise.
20601 * tree-inline.c (expand_call_inline): Likewise.
20602 (tree_function_versioning): Likewise.
20603 * tree-into-ssa.c (dump_defs_stack): Likewise.
20604 (dump_currdefs): Likewise.
20605 (dump_names_replaced_by): Likewise.
20606 (dump_update_ssa): Likewise.
20607 (update_ssa): Likewise.
20608 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20609 * tree-parloops.c (build_new_reduction): Likewise.
20610 (try_create_reduction_list): Likewise.
20611 (ref_conflicts_with_region): Likewise.
20612 (oacc_entry_exit_ok_1): Likewise.
20613 (oacc_entry_exit_single_gang): Likewise.
20614 * tree-pretty-print.h: Likewise.
20615 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20616 (get_scalar_evolution): Likewise.
20617 (add_to_evolution): Likewise.
20618 (get_loop_exit_condition): Likewise.
20619 (analyze_evolution_in_loop): Likewise.
20620 (analyze_initial_condition): Likewise.
20621 (analyze_scalar_evolution): Likewise.
20622 (instantiate_scev): Likewise.
20623 (number_of_latch_executions): Likewise.
20624 (gather_chrec_stats): Likewise.
20625 (final_value_replacement_loop): Likewise.
20626 (scev_const_prop): Likewise.
20627 * tree-sra.c (dump_access): Likewise.
20628 (disqualify_candidate): Likewise.
20629 (create_access): Likewise.
20630 (reject): Likewise.
20631 (maybe_add_sra_candidate): Likewise.
20632 (create_access_replacement): Likewise.
20633 (analyze_access_subtree): Likewise.
20634 (analyze_all_variable_accesses): Likewise.
20635 (sra_modify_assign): Likewise.
20636 (initialize_constant_pool_replacements): Likewise.
20637 (find_param_candidates): Likewise.
20638 (decide_one_param_reduction): Likewise.
20639 (replace_removed_params_ssa_names): Likewise.
20640 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20641 * tree-ssa-copy.c (dump_copy_of): Likewise.
20642 (copy_prop_visit_cond_stmt): Likewise.
20643 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20644 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20645 (record_equivalences_from_stmt): Likewise.
20646 * tree-ssa-dse.c (compute_trims): Likewise.
20647 (delete_dead_call): Likewise.
20648 (delete_dead_assignment): Likewise.
20649 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20650 (forward_propagate_into_cond): Likewise.
20651 (pass_forwprop::execute): Likewise.
20652 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20653 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20655 (move_computations_worker): Likewise.
20656 (execute_sm): Likewise.
20657 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20658 (remove_exits_and_undefined_stmts): Likewise.
20659 (remove_redundant_iv_tests): Likewise.
20660 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20661 (adjust_iv_update_pos): Likewise.
20662 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20663 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20664 (value_replacement): Likewise.
20665 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20666 * tree-ssa-pre.c (print_pre_expr): Likewise.
20667 (get_representative_for): Likewise.
20668 (create_expression_by_pieces): Likewise.
20669 (insert_into_preds_of_block): Likewise.
20670 (eliminate_insert): Likewise.
20671 (eliminate_dom_walker::before_dom_children): Likewise.
20672 (eliminate): Likewise.
20673 (remove_dead_inserted_code): Likewise.
20674 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20675 * tree-ssa-reassoc.c (get_rank): Likewise.
20676 (eliminate_duplicate_pair): Likewise.
20677 (eliminate_plus_minus_pair): Likewise.
20678 (eliminate_not_pairs): Likewise.
20679 (undistribute_ops_list): Likewise.
20680 (eliminate_redundant_comparison): Likewise.
20681 (update_range_test): Likewise.
20682 (optimize_range_tests_var_bound): Likewise.
20683 (optimize_vec_cond_expr): Likewise.
20684 (rewrite_expr_tree): Likewise.
20685 (rewrite_expr_tree_parallel): Likewise.
20686 (linearize_expr): Likewise.
20687 (break_up_subtract): Likewise.
20688 (linearize_expr_tree): Likewise.
20689 (attempt_builtin_powi): Likewise.
20690 (attempt_builtin_copysign): Likewise.
20691 (transform_stmt_to_copy): Likewise.
20692 (transform_stmt_to_multiply): Likewise.
20693 (dump_ops_vector): Likewise.
20694 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20695 (print_scc): Likewise.
20696 (set_ssa_val_to): Likewise.
20697 (visit_reference_op_store): Likewise.
20698 (visit_use): Likewise.
20699 (sccvn_dom_walker::before_dom_children): Likewise.
20700 (run_scc_vn): Likewise.
20701 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20703 (expr_hash_elt::print): Likewise.
20704 (const_and_copies::pop_to_marker): Likewise.
20705 (const_and_copies::record_const_or_copy_raw): Likewise.
20706 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20707 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20708 (dump_predicates): Likewise.
20709 (find_uninit_use): Likewise.
20710 (warn_uninitialized_phi): Likewise.
20711 (pass_late_warn_uninitialized::execute): Likewise.
20712 * tree-ssa.c (verify_vssa): Likewise.
20713 (verify_ssa): Likewise.
20714 (maybe_optimize_var): Likewise.
20715 * tree-vrp.c (dump_value_range): Likewise.
20716 (dump_all_value_ranges): Likewise.
20717 (dump_asserts_for): Likewise.
20718 (register_edge_assert_for_2): Likewise.
20719 (vrp_visit_cond_stmt): Likewise.
20720 (vrp_visit_switch_stmt): Likewise.
20721 (vrp_visit_stmt): Likewise.
20722 (vrp_visit_phi_node): Likewise.
20723 (simplify_cond_using_ranges_1): Likewise.
20724 (fold_predicate_in): Likewise.
20725 (evrp_dom_walker::before_dom_children): Likewise.
20726 (evrp_dom_walker::push_value_range): Likewise.
20727 (evrp_dom_walker::pop_value_range): Likewise.
20728 (execute_early_vrp): Likewise.
20730 2017-05-16 Richard Biener <rguenther@suse.de>
20732 * dwarf2out.c (loc_list_from_tree_1): Do not create
20733 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20735 2017-05-16 Richard Biener <rguenther@suse.de>
20737 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20739 (note_variable_value_in_expr): If we resolved the decl ref
20740 do not push to the stack.
20742 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20744 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20745 operations in fast-math mode.
20746 (vaddq_f16): Likewise.
20747 (vmul_f16): Likewise.
20748 (vmulq_f16): Likewise.
20749 (vsub_f16): Likewise.
20750 (vsubq_f16): Likewise.
20751 * config/arm/neon.md (add<mode>3): New.
20753 (fma:<VH:mode>3): New. Also remove outdated comment.
20756 2017-05-16 Martin Liska <mliska@suse.cz>
20760 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20761 (odr_types_equivalent_p): Likewise.
20763 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20765 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20767 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20770 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20771 non-interunit SSE move alternatives with '?'.
20772 (zero-extendsidi peephole2): New peephole to skip intermediate
20773 general register in SSE zero-extend sequence.
20775 2017-05-15 Jeff Law <law@redhat.com>
20777 * reorg.c (relax_delay_slots): Create a new variable to hold
20778 the temporary target rather than clobbering TARGET_LABEL.
20780 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20781 missing argument to extract_bit_field call.
20782 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20784 2017-05-15 Martin Liska <mliska@suse.cz>
20787 * gcc.c (process_command): Do not allow empty argument of -o option.
20789 2017-05-15 Renlin Li <renlin.li@arm.com>
20791 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20792 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20793 * config/aarch64/constraints.md (Usf): Add long call check.
20794 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20795 (call_value): Likewise.
20796 (sibcall): Likewise.
20797 (sibcall_value): Likewise.
20799 (call_value_insn): New.
20800 (sibcall_insn): Update rtx pattern.
20801 (sibcall_value_insn): Likewise.
20802 (call_internal): Remove.
20803 (call_value_internal): Likewise.
20804 (sibcall_internal): Likewise.
20805 (sibcall_value_internal): Likewise.
20806 (call_reg): Likewise.
20807 (call_symbol): Likewise.
20808 (call_value_reg): Likewise.
20809 (call_value_symbol): Likewise.
20811 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20814 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20816 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20818 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20819 compatible with CCGOCmode and with CCZmode.
20821 2017-05-14 Martin Sebor <msebor@redhat.com>
20823 PR middle-end/77671
20824 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20825 (gimple_fold_builtin_snprintf): Same.
20826 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20827 (gimple_fold_builtin_snprintf): Same.
20828 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20829 of character types.
20830 (is_call_safe): New function.
20831 (try_substitute_return_value): Call it.
20832 (try_simplify_call): New function.
20833 (pass_sprintf_length::handle_gimple_call): Call it.
20835 2017-05-14 Martin Sebor <msebor@redhat.com>
20837 PR middle-end/80669
20838 * builtins.c (expand_builtin_stpncpy): Simplify.
20840 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20842 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20843 * config/i386/i386.h
20844 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20845 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20846 (struct machine_function): Add new members call_ms2sysv,
20847 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20848 (struct machine_frame_state): New fields sp_realigned and
20849 sp_realigned_offset.
20850 * config/i386/i386.c
20851 (enum xlogue_stub): New enum.
20852 (enum xlogue_stub_sets): New enum.
20853 (class xlogue_layout): New class.
20854 (struct ix86_frame): New fields stack_realign_allocate_offset,
20855 stack_realign_offset and outlined_save_offset. Modify comments to
20856 detail stack layout when using out-of-line stubs.
20857 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20858 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20859 -mcall-ms2sysv-xlogues.
20860 (stub_managed_regs): New static variable.
20861 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20862 registers managed by out-of-line stub.
20863 (disable_call_ms2sysv_xlogues): New function.
20864 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20865 m->call_ms2sysv when appropriate and compute frame layout for
20867 (sp_valid_at, fp_valid_at): New inline functions.
20868 (choose_basereg): New function.
20869 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20871 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20872 Use align parameter of choose_baseaddr to generated aligned SSE movs
20874 (pro_epilogue_adjust_stack): Modify to track
20875 machine_frame_state::sp_realigned.
20876 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20877 (ix86_nsaved_sseregs): Likewise.
20878 (ix86_emit_save_regs): Likewise.
20879 (ix86_emit_save_regs_using_mov): Likewise.
20880 (ix86_emit_save_sse_regs_using_mov): Likewise.
20881 (get_scratch_register_on_entry): Likewise.
20882 (gen_frame_set): New function.
20883 (gen_frame_load): Likewise.
20884 (gen_frame_store): Likewise.
20885 (emit_outlined_ms2sysv_save): Likewise.
20886 (emit_outlined_ms2sysv_restore): Likewise.
20887 (ix86_expand_prologue): Modify stack re-alignment code and call
20888 emit_outlined_ms2sysv_save when appropriate.
20889 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20890 parameter rtx_insn *insn, which allows the function to be used to only
20891 generate the notes.
20892 (ix86_expand_epilogue): Modify validity checks of frame and stack
20893 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20894 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20895 * config/i386/predicates.md
20896 (save_multiple): New predicate.
20897 (restore_multiple): Likewise.
20898 * config/i386/sse.md
20899 (save_multiple<mode>): New pattern.
20900 (save_multiple_realign<mode>): Likewise.
20901 (restore_multiple<mode>): Likewise.
20902 (restore_multiple_and_return<mode>): Likewise.
20903 (restore_multiple_leave_return<mode>): Likewise.
20904 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20906 2017-05-14 Julia Koval <julia.koval@intel.com>
20908 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20909 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20910 (__builtin_ia32_xsetbv): New builtins.
20911 * config/i386/i386.c (ix86_expand_special_args_builtin):
20913 (ix86_expand_builtin): Special expand for new intrinsics.
20914 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20915 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20916 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20918 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20920 * cfganal.c (inverted_post_order_compute): Change argument type
20922 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20923 * df-core.c (rest_of_handle_df_initialize): Adjust.
20924 (rest_of_handle_df_finish): Likewise.
20925 (df_analyze_1): Likewise.
20926 (df_analyze): Likewise.
20927 (loop_inverted_post_order_compute): Change argument to be a vec *.
20928 (df_analyze_loop): Adjust.
20929 (df_get_n_blocks): Likewise.
20930 (df_get_postorder): Likewise.
20931 * df.h (struct df_d): Change field to be a vec.
20932 * lcm.c (compute_laterin): Adjust.
20933 (compute_available): Likewise.
20934 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20935 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20936 * tree-ssa-pre.c (compute_antic): Likewise.
20938 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20940 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20941 (depth_first_search::depth_first_search): Change structure init
20942 function to this constructor.
20943 (depth_first_search::add_bb): Rename function to this member.
20944 (depth_first_search::execute): Likewise.
20945 (flow_dfs_compute_reverse_finish): Adjust.
20947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20949 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20950 (longest_simple_path): Likewise.
20951 * shrink-wrap.c (spread_components): Likewise.
20952 (disqualify_problematic_components): Likewise.
20953 (emit_common_heads_for_components): Likewise.
20954 (emit_common_tails_for_components): Likewise.
20955 (insert_prologue_epilogue_for_components): Likewise.
20957 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20959 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20962 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20964 * df-core.c (df_set_blocks): Start using auto_bitmap.
20965 (df_compact_blocks): Likewise.
20966 * df-problems.c (df_rd_confluence_n): Likewise.
20967 * df-scan.c (df_insn_rescan_all): Likewise.
20968 (df_process_deferred_rescans): Likewise.
20969 (df_update_entry_block_defs): Likewise.
20970 (df_update_exit_block_uses): Likewise.
20971 (df_entry_block_bitmap_verify): Likewise.
20972 (df_exit_block_bitmap_verify): Likewise.
20973 (df_scan_verify): Likewise.
20974 * lra-constraints.c (lra_constraints): Likewise.
20975 (undo_optional_reloads): Likewise.
20976 (lra_undo_inheritance): Likewise.
20977 * lra-remat.c (calculate_gen_cands): Likewise.
20978 (do_remat): Likewise.
20979 * lra-spills.c (assign_spill_hard_regs): Likewise.
20980 (spill_pseudos): Likewise.
20981 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20982 (bitmap_set_subtract_values): Likewise.
20984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20986 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20987 management with auto_bitmap.
20988 (fix_inter_tick): Likewise.
20989 (fix_recovery_deps): Likewise.
20990 * ira.c (add_store_equivs): Likewise.
20991 (find_moveable_pseudos): Likewise.
20992 (split_live_ranges_for_shrink_wrap): Likewise.
20993 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20994 (rtx_reuse_manager::seen_def_p): Likewise.
20995 (rtx_reuse_manager::set_seen_def): Likewise.
20996 * print-rtl.h (class rtx_reuse_manager): Likewise.
20998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21000 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21002 (migrate_btr_def): Likewise.
21003 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21004 * df-core.c (loop_post_order_compute): Likewise.
21005 (loop_inverted_post_order_compute): Likewise.
21006 * hsa-common.h: Likewise.
21007 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21008 * init-regs.c (initialize_uninitialized_regs): Likewise.
21009 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21010 (inline_small_functions): Likewise.
21011 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21012 * ira.c (combine_and_move_insns): Likewise.
21013 (build_insn_chain): Likewise.
21014 * loop-invariant.c (find_invariants): Likewise.
21015 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21016 * predict.c (tree_predict_by_opcode): Likewise.
21017 (predict_paths_leading_to): Likewise.
21018 (predict_paths_leading_to_edge): Likewise.
21019 (estimate_loops_at_level): Likewise.
21020 (estimate_loops): Likewise.
21021 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21022 (spread_components): Likewise.
21023 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21024 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21025 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21026 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21027 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21028 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21029 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21030 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21031 (mark_threaded_blocks): Likewise.
21032 (thread_through_all_blocks): Likewise.
21033 * tree-ssa.c (verify_ssa): Likewise.
21034 (execute_update_addresses_taken): Likewise.
21035 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21039 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21041 (post_order_compute): Likewise.
21042 (inverted_post_order_compute): Likewise.
21043 (pre_and_rev_post_order_compute_fn): Likewise.
21045 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21047 * genrecog.c (int_set::int_set): Explicitly construct our
21048 auto_vec base class.
21049 * vec.h (auto_vec::auto_vec): New constructor.
21051 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21053 * bitmap.h (class auto_bitmap): New constructor taking
21054 bitmap_obstack * argument.
21056 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21058 * bitmap.h (class auto_bitmap): Change type of m_bits to
21059 bitmap_head, and adjust ctor / dtor and member operators.
21061 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21063 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21064 when returned register mode doesn't match original mode.
21066 2017-05-12 Jeff Law <law@redhat.com>
21067 Jakub Jelinek <jakub@redhat.com>
21069 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21070 we look for cc setter after the compare-elim changes.
21071 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21072 within the vector to match what compare-elim now expects.
21073 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21074 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21076 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21077 after the compare-elim changes.
21078 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21079 the vector to match what compare-elim now expects.
21080 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21081 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21082 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21083 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21084 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21086 * config/visium/visium.c (single_set_and_flags): Fix where
21087 we look for cc setter after the compare-elim changes.
21088 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21089 with the vector to match what compare-elim now expects.
21090 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21091 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21092 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21093 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21094 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21095 (neg<mode>2_insn_set_overflow): Likewise.
21097 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21099 PR middle-end/79794
21100 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21101 maybe_expand_insn call, set ops[0].target. If still set after call,
21102 set alt_rtl. Add extra arg to recursive calls.
21103 (extract_bit_field): Add alt_rtl argument. Pass to
21105 * expmed.h (extract_bit_field): Fix prototype.
21106 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21107 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21108 to extract_bit_field_calls.
21109 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21110 Pass alt_rtl to extract_bit_field calls.
21111 * calls.c (store_unaligned_arguments_into_psuedos)
21112 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21113 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21115 * optabs.h (struct expand_operand): Add target bitfield.
21117 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21119 * compare-elim.c (try_eliminate_compare): Canonicalize
21120 operation with embedded compare to
21121 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21122 (set (reg) (operation)].
21124 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21126 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21129 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21130 cost of adding a carry flag for ADC instruction.
21131 [case MINUS]: Ignore the cost of subtracting a carry flag
21132 for SBB instruction.
21134 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21136 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21138 * config/rs6000/bmiintrin.h: New file.
21139 * config/rs6000/bmi2intrin.h: New file.
21140 * config/rs6000/x86intrin.h: New file.
21142 2017-05-12 Jeff Law <law@redhat.com>
21144 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21147 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21149 PR middle-end/80707
21150 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21152 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21154 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21155 early expansion of vector divide builtins.
21156 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21157 builtins identified as having unsigned arguments.
21159 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21161 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21162 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21163 expansion of vector logical operations (and, andc, or, xor,
21166 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21168 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21169 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21171 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21173 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21174 early GIMPLE expansion of vector multiplies.
21176 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21178 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21179 TARGET_HAVE_MOVT conditional.
21180 (movt splitter): Likewise.
21182 2017-05-12 Richard Biener <rguenther@suse.de>
21184 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21185 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21186 Fold all stmts not inplace.
21188 2017-05-12 Richard Biener <rguenther@suse.de>
21190 PR tree-optimization/80713
21191 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21192 inserted_exprs bit for not removed stmts.
21194 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21196 PR middle-end/69921
21197 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21198 parallelized" attribute for parallelized OpenACC kernels.
21199 * omp-offload.c (execute_oacc_device_lower): Use it.
21201 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21202 Set "oacc kernels" attribute.
21203 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21204 parameter. Adjust all users.
21205 (oacc_fn_attrib_kernels_p): Remove function.
21206 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21207 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21208 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21209 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21210 assert "oacc kernels" attribute is set.
21212 2017-05-11 Carl Love <cel@us.ibm.com>
21214 * config/rs6000/rs6000-c: Add support for built-in functions
21215 vector unsigned char vec_popcnt (vector signed char)
21216 vector unsigned char vec_popcnt (vector unsigned char)
21217 vector unsigned short vec_popcnt (vector signed short)
21218 vector unsigned short vec_popcnt (vector unsigned short)
21219 vector unsigned int vec_popcnt (vector signed int)
21220 vector unsigned int vec_popcnt (vector unsigned int)
21221 vector unsigned long long vec_popcnt (vector signed long long)
21222 vector unsigned long long vec_popcnt (vector unsigned long long)
21223 vector signed long long vec_slo (vector signed long long,
21224 vector signed char)
21225 vector signed long long vec_slo (vector signed long long,
21226 vector unsigned char)
21227 vector unsigned long long vec_slo (vector unsigned long long,
21228 vector signed char)
21229 vector unsigned long long vec_slo (vector unsigned long long,
21230 vector unsigned char)
21231 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21232 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21233 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21234 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21235 * doc/extend.texi: Update the built-in documentation file for the
21236 new built-in functions.
21238 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21240 * attribs.h (sorted_attr_string): Move machine independent
21241 functions for target clone support from the i386 port to common
21242 code. Rename ix86_function_versions to common_function_versions.
21243 Rename make_name to make_unique_name.
21244 (common_function_versions): Likewise.
21245 (make_unique_name): Likewise.
21246 (make_dispatcher_decl): Likewise.
21247 (is_function_default_version): Likewise.
21248 * attribs.c (attr_strcmp): Likewise.
21249 (sorted_attr_string): Likewise.
21250 (common_function_versions): Likewise.
21251 (make_unique_name): Likewise.
21252 (make_dispatcher_decl): Likewise.
21253 (is_function_default_version): Likewise.
21254 * config/i386/i386.c (attr_strcmp): Likewise.
21255 (sorted_attr_string): Likewise.
21256 (ix86_function_versions): Likewise.
21257 (make_name): Likewise.
21258 (make_dispatcher_decl): Likewise.
21259 (is_function_default_version): Likewise.
21260 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21262 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21265 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21266 Account for direct move costs for vec_construct of integer
21269 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21272 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21273 (UNSPEC_STX_ATOMIC): Ditto.
21274 (loaddi_via_sse): New insn.
21275 (storedi_via_sse): Ditto.
21276 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21277 Update corresponding peephole2 patterns.
21278 (atomic_storedi_fpu): Ditto.
21280 2017-05-11 Julia Koval <julia.koval@intel.com>
21282 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21283 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21285 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21286 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21287 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21289 2017-05-11 Nathan Sidwell <nathan@acm.org>
21291 * graphite-poly.c: Include dumpfile.h.
21293 * dumpfle.h (dump_function): Declare here ...
21294 * tree-dump.h (dump_function): ... not here.
21295 * dumpfile.c: #include tree-cfg.h.
21296 (dump_function): Move here from ...
21297 * tree-dump.c (dump_function): ... here.
21298 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21299 * graphite-poly.c: Don't include tree-dump.h.
21300 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21301 * print-tree.c: Likewise.
21302 * stor-layout.c: Likewise.
21303 * tree-nested.c: Likewise.
21305 * dumpfile.c (dump_start): Use TDF_FLAGS.
21306 (dump_enable_all): Fix TDF_KIND check thinko.
21308 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21310 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21311 array entries to represent two legal parameterizations of the
21312 overloaded __builtin_cmpb function, as represented by the
21313 P6_OV_BUILTIN_CMPB constant.
21314 (altivec_resolve_overloaded_builtin): Add special case handling
21315 for the __builtin_cmpb function, as represented by the
21316 P6_OV_BUILTIN_CMPB constant.
21317 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21318 (BU_P6_64BIT_2): New macro.
21319 (BU_P6_OVERLOAD_2): New macro
21320 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21321 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21322 (CMPB): Add overload support to represent both 32-bit and 64-bit
21323 compare-bytes function.
21324 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21325 support for TARGET_CMPB.
21326 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21327 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21328 documentation of the __builtin_cmpb overloaded built-in function.
21330 2017-05-11 Richard Biener <rguenther@suse.de>
21332 PR tree-optimization/80705
21333 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21334 bases are not vectorizable.
21336 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21338 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21339 when counting register pressure.
21341 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21343 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21344 (iv_ca_more_deps): Renamed to ...
21345 (iv_ca_compare_deps): ... this.
21346 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21350 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21352 (determine_group_iv_costs): ... here.
21353 (find_inv_vars_cb): Record inv var if it's not recorded before.
21355 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21357 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21358 (get_shiftadd_cost): Ditto.
21360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21362 * tree-ssa-address.c: Include header file.
21363 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21365 (add_to_parts): Refactor.
21366 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21367 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21372 PR tree-optimization/53090
21373 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21375 (extract_cond_operands): Detect condition with IV on both sides
21376 and return COMP_IV_EXPR_2.
21377 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21378 (rewrite_use_compare): Simplify by removing call to function
21379 extract_cond_operands.
21381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21383 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21384 (extract_cond_operands): Detect condition comparing against non-
21385 invariant bound and return appropriate enum value.
21386 (find_interesting_uses_cond): Update use of extract_cond_operands.
21387 Handle its return value accordingly.
21388 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21390 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21392 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21393 nonlinear iv_use computation in loop invariant sensitive way.
21395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21397 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21398 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21402 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21403 (dump_cand): Support iv_cand.inv_exprs.
21404 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21406 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21409 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21411 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21413 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21414 as local function. Include necessary header files.
21415 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21417 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21419 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21421 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21423 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21424 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21425 RSHIFT_EXPR and BIT_NOT_EXPR.
21427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21429 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21430 (adjust_setup_cost): New parameter supporting round up adjustment.
21431 (struct address_cost_data): Delete.
21432 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21433 (split_address_cost, ptr_difference_cost): Delete.
21434 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21435 (struct ainc_cost_data): New struct.
21436 (get_address_cost_ainc): New function.
21437 (get_address_cost, get_computation_cost): Reimplement.
21438 (determine_group_iv_cost_address): Record inv_expr for all uses of
21440 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21441 (iv_ca_has_deps): Reimplemented to ...
21442 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21444 (iv_ca_extend): Call iv_ca_more_deps.
21446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21448 * tree-ssa-address.c (struct mem_address): Move to header file.
21449 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21450 * tree-ssa-address.h (struct mem_address): Move from C file.
21451 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21455 * tree-affine.h (aff_combination_type): New interface.
21456 (aff_combination_zero_p): Remove static.
21457 (aff_combination_const_p): New interface.
21458 (aff_combination_singleton_var_p): New interfaces.
21460 2017-05-11 Richard Biener <rguenther@suse.de>
21462 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21463 Skip unreachable blocks and destinations.
21464 (eliminate): Move stmt removal and fixup ...
21465 (fini_eliminate): ... here. Skip inserted exprs.
21466 (pass_pre::execute): Move fini_pre after fini_eliminate.
21467 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21468 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21469 PRE to get rid of dead code that has invalid SSA form and
21470 split critical edges again.
21472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21474 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21476 2017-05-11 Richard Biener <rguenther@suse.de>
21478 * passes.c (execute_function_todo): Verify loops if they are
21479 said to be up-to-date.
21480 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21481 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21483 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21486 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21487 handle calling assemble_external ourself.
21490 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21491 modes with zero size. Enhance comment.
21493 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21495 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21496 built-ins for vec_xl and vec_xst with short and char pointer
21499 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21501 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21502 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21503 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21504 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21505 (_mm_maskz_min_round_ss): New intrinsics.
21506 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21507 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21508 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21509 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21510 (__builtin_ia32_minss_mask_round): New builtins.
21511 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21512 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21513 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21515 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21516 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21518 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21521 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21523 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21524 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21525 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21526 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21527 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21528 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21529 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21530 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21531 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21532 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21533 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21534 (__builtin_ia32_mulss_mask_round): New builtins.
21535 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21536 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21537 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21539 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21540 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21542 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21545 2017-05-10 Julia Koval <julia.koval@intel.com>
21547 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21548 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21549 (_mm256_setr_m128i): New intrinsics.
21551 2017-05-10 Julia Koval <julia.koval@intel.com>
21553 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21554 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21555 (_mm_maskz_rcp14_ss): New intrinsics.
21556 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21557 (__builtin_ia32_rcp14ss_mask): New builtins.
21558 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21560 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21562 PR tree-optimization/51513
21563 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21564 (assert_unreachable_fallthru_edge_p): Use it.
21565 (group_case_labels_stmt): Likewise.
21566 * tree-cfg.h: Prototype it.
21567 * stmt.c: Include cfghooks.h and tree-cfg.h.
21568 (emit_case_dispatch_table) <gap_label>: New local variable.
21569 Use it to fill dispatch table gaps.
21570 Test for default_label before updating probabilities.
21571 (expand_case) <default_label>: Remove unneeded initialization.
21572 Test for unreachable default case statement and remove its edge.
21573 Set default_label accordingly.
21574 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21576 2017-05-10 Carl Love <cel@us.ibm.com>
21578 * config/rs6000/rs6000-c: Add support for built-in functions
21579 vector signed char vec_neg (vector signed char)
21580 vector signed short int vec_neg (vector short int)
21581 vector signed int vec_neg (vector signed int)
21582 vector signed long long vec_neg (vector signed long long)
21583 vector float vec_neg (vector float)
21584 vector double vec_neg (vector double)
21585 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21587 * config/rs6000/altivec.h: Add define for vec_neg
21588 * doc/extend.texi: Update the built-in documentation for the
21589 new built-in functions.
21591 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21593 PR tree-optimization/77644
21594 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21596 2017-05-10 Nathan Sidwell <nathan@acm.org>
21598 * dumpfile.h (TDI_lang_all): New.
21599 (TDF_KIND): New. Renumber others
21600 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21602 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21604 (get_dump_file_name): Adjust suffix generation.
21605 (dump_enable_all): Use TDF_KIND.
21606 * doc/invoke.texi (-fdump-lang-all): Document.
21608 * dumpfile.h: Tabify.
21610 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21613 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21614 Move member access before delete.
21616 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21618 * tree-inline.c (expand_call_inline): Split block at stmt
21621 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21624 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21625 are now unused after splitting mov{sf,sd}_hardfloat.
21626 (f32_lr2): Likewise.
21627 (f32_lm): Likewise.
21628 (f32_lm2): Likewise.
21629 (f32_li): Likewise.
21630 (f32_li2): Likewise.
21631 (f32_lv): Likewise.
21632 (f32_sr): Likewise.
21633 (f32_sr2): Likewise.
21634 (f32_sm): Likewise.
21635 (f32_sm2): Likewise.
21636 (f32_si): Likewise.
21637 (f32_si2): Likewise.
21638 (f32_sv): Likewise.
21639 (f32_dm): Likewise.
21640 (f32_vsx): Likewise.
21641 (f32_av): Likewise.
21642 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21643 For movsf, order stores so the VSX stores occur before the GPR
21644 store which encourages the register allocator to use a traditional
21645 FPR instead of a GPR. For movsd, order the stores so that the GPR
21646 store comes before the VSX stores to allow the power6 to work.
21647 This is due to the power6 not having a 32-bit integer store
21648 instruction from a FPR.
21649 (movsf_hardfloat): Likewise.
21650 (movsd_hardfloat): Likewise.
21652 2017-05-09 Martin Sebor <msebor@redhat.com>
21654 PR translation/80280
21655 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21658 PR translation/80280
21659 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21660 data member added in r247778.
21661 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21663 2017-05-09 Nathan Sidwell <nathan@acm.org>
21665 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21667 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21670 2017-05-09 Marek Polacek <polacek@redhat.com>
21672 * doc/invoke.texi: Fix typo.
21674 2017-05-09 Richard Biener <rguenther@suse.de>
21676 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21677 (vrp_val_is_min): Likewise.
21678 (set_value_range_to_value): Likewise.
21679 (set_value_range_to_nonnegative): Likewise.
21680 (gimple_assign_nonzero_p): Likewise.
21681 (gimple_stmt_nonzero_p): Likewise.
21682 (vrp_int_const_binop): Likewise. Remove unreachable case.
21683 (adjust_range_with_scev): Adjust comments.
21684 (compare_range_with_value): Likewise.
21685 (extract_range_from_phi_node): Likewise.
21686 (test_for_singularity): Likewise.
21688 2017-05-09 Richard Biener <rguenther@suse.de>
21690 * tree-vrp.c (get_single_symbol): Add assert that we don't
21691 get overflowed constants as invariant part.
21692 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21693 checks. Use wi::cmp instead of recursing for integer constants.
21694 (compare_values): Just ignore whether we assumed undefined
21695 overflow instead of failing the compare.
21696 (extract_range_for_var_from_comparison_expr): Add comment before the
21697 TREE_NO_WARNING sets.
21698 (test_for_singularity): Likewise.
21699 (extract_range_from_comparison): Do not disable optimization
21700 when we assumed undefined overflow.
21701 (extract_range_basic): Remove init of unused var.
21703 2017-05-09 Richard Biener <rguenther@suse.de>
21705 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21706 (extract_range_from_multiplicative_op_1): Adjust.
21707 (extract_range_from_binary_expr_1): Use int_const_binop.
21709 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21712 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21713 rs6000_store_data_bypass_p in seven define_bypass directives and
21714 in several comments.
21715 * config/rs6000/rs6000-protos.h: Add prototype for
21716 rs6000_store_data_bypass_p function.
21717 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21718 function implements slightly different (rs6000-specific) semantics
21719 than store_data_bypass_p, returning false rather than aborting
21720 with assertion error when arguments do not satisfy the
21721 requirements of store data bypass.
21722 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21723 rs6000_store_data_bypass_p.
21725 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21727 * config/xtensa/xtensa-protos.h
21728 (xtensa_initial_elimination_offset): New declaration.
21729 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21730 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21731 macro definition, add case for FRAME_POINTER_REGNUM when
21732 FRAME_GROWS_DOWNWARD.
21733 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21734 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21735 xtensa_initial_elimination_offset.
21737 2017-05-08 Nathan Sidwell <nathan@acm.org>
21739 * doc/invoke.texi: Alphabetize -fdump options.
21741 2017-05-08 Martin Sebor <msebor@redhat.com>
21743 PR translation/80280
21744 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21746 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21748 * target.def (compute_frame_layout): New optional target hook.
21749 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21750 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21751 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21753 * reload1.c (verify_initial_elim_offsets): Likewise.
21754 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21755 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21756 (arm_get_frame_offsets): Split up into this ...
21757 (arm_compute_frame_layout): ... and this function.
21759 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21761 * config/aarch64/constraints.md (Usa): New constraint.
21762 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21764 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21766 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21767 with_multilib_list after it has been checked.
21769 2017-05-08 Richard Biener <rguenther@suse.de>
21771 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21772 (bitmap_set_subtract_values): Likewise.
21774 2017-05-08 Richard Biener <rguenther@suse.de>
21776 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21777 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21779 (gimple_stmt_nonzero_warnv_p): Rename to ...
21780 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21782 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21783 (extract_range_basic): Adjust, do not disable propagation on
21784 strict overflow sensitive simplification.
21785 (vrp_visit_cond_stmt): Likewise.
21787 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21789 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21790 body size unconditionally.
21792 2017-05-07 Jeff Law <law@redhat.com>
21795 2017-05-06 Jeff Law <law@redhat.com>
21796 PR tree-optimization/78496
21797 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21800 PR tree-optimization/78496
21801 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21802 (simplify_stmt_using_ranges): Call it.
21803 (vrp_dom_walker::before_dom_children): Extract equivalences
21804 from an ASSERT_EXPR with an equality comparison against a
21807 2017-05-06 Jeff Law <law@redhat.com>
21809 PR tree-optimization/78496
21810 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21813 PR tree-optimization/78496
21814 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21815 (simplify_stmt_using_ranges): Call it.
21816 (vrp_dom_walker::before_dom_children): Extract equivalences
21817 from an ASSERT_EXPR with an equality comparison against a
21820 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21822 * lra-constraints.c (lra_copy_reg_equiv): New function.
21823 (split_reg): Use it to copy equivalence information from the
21824 original register to the spill register.
21826 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21828 PR rtl-optimization/75964
21829 * simplify-rtx.c (simplify_const_relational_operation): Remove
21830 invalid handling of comparisons of integer ABS.
21832 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21834 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21835 initialize to zero.
21836 (init_regs): Remove declaration.
21837 (function_arg_advance_32): Initialize error_p as boolean variable.
21839 2017-05-05 Nathan Sidwell <nathan@acm.org>
21841 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21842 lines. Use for (;;).
21844 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21846 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21847 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21848 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21849 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21850 VF=2 that require versioning.
21852 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21854 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21857 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21859 * diagnostic.h (diagnostic_override_option_index): Convert from
21860 macro to inline function.
21862 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21864 * diagnostic.c (last_module_changed_p): New function.
21865 (set_last_module): New function.
21866 (diagnostic_report_current_module): Convert macro usage to
21867 the above functions.
21868 * diagnostic.h (diagnostic_context::last_module): Strengthen
21869 from const line_map * to const line_map_ordinary *.
21870 (diagnostic_last_module_changed): Delete macro.
21871 (diagnostic_set_last_module): Delete macro.
21873 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21875 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21876 with diagnostic_report_diagnostic.
21877 (diagnostic_n_impl_richloc): Likewise.
21878 * diagnostic.h (report_diagnostic): Delete macro.
21879 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21880 with diagnostic_report_diagnostic.
21881 * substring-locations.c (format_warning_va): Likewise.
21883 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21885 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21886 save/restor of format_spec. Move option-printing code to...
21887 (print_option_information): ...this new function, and
21888 reimplement by simply printing to the pretty_printer,
21889 rather than appending to the format string.
21891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21893 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21894 handling logic into...
21895 (update_effective_level_from_pragmas): ...this new function.
21897 2017-05-04 Andrew Waterman <andrew@sifive.com>
21899 * config/riscv/riscv.opt (mstrict-align): New option.
21900 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21901 (SLOW_UNALIGNED_ACCESS): Define.
21902 (riscv_slow_unaligned_access): Declare.
21903 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21905 (riscv_slow_unaligned_access): New variable.
21906 (rocket_tune_info): Set slow_unaligned_access to true.
21907 (optimize_size_tune_info): Set slow_unaligned_access to false.
21908 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21909 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21910 (riscv_option_override): Set riscv_slow_unaligned_access.
21911 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21913 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21915 * config/riscv/riscv.md: Unify indentation.
21917 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21922 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21924 (extendsi<mode>2): Add support for doing sign extension via
21925 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21926 don't have ISA 3.0 instructions.
21927 (extendsi<mode>2 splitter): Likewise.
21928 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21929 generate the normal insns since SImode can now go in vector
21930 registers. Disallow the special UNSPECs needed for previous
21931 machines to hide SImode being used. Add new insns
21932 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21933 (fix_trunc<mode>si2_stfiwx): Likewise.
21934 (fix_trunc<mode>si2_internal): Likewise.
21935 (fixuns_trunc<mode>si2): Likewise.
21936 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21937 (fctiw<u>z_<mode>_smallint): Likewise.
21938 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21939 of floating point to 32-bit integer from doing a direct move to
21940 the GPR registers to do a store.
21941 (fctiwz_<mode>): Break long line.
21943 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21945 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21946 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21947 (addr_list, addr_offset_valid_p): New.
21948 (split_address_groups): Check offset validity with above function.
21949 (gt-tree-ssa-loop-ivopts.h): Include header file.
21951 2017-05-05 Nathan Sidwell <nathan@acm.org>
21953 * config.gcc (arm*-*-*): Add missing 'fi'.
21955 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21957 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21958 included in -fopt-info does not matter.
21959 * doc/optinfo.texi (-fopt-info): Fix description of default
21960 behavour. Explicitly say order of options included in -fopt-info
21963 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21965 * config.gcc: Allow combinations of aprofile and rmprofile values for
21966 --with-multilib-list.
21967 * config/arm/t-multilib: New file.
21968 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21969 variables. Remove setting of ISA and floating-point ABI in
21970 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21971 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21972 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21973 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21975 * config/arm/t-rmprofile: Likewise except for the matches changes.
21976 * doc/install.texi (--with-multilib-list): Document the combination of
21977 aprofile and rmprofile values and warn about pitfalls in doing that.
21979 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21981 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21982 (movdi_aarch64): Likewise.
21984 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21986 PR tree-optimization/80632
21987 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21989 (build_arrays): Initialize it for virtual phis.
21990 (fix_phi_nodes): Use it for virtual phis.
21992 PR tree-optimization/80558
21993 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21994 [x, y] op z into [x op, y op z] for op & or | if conditions
21997 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21998 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22001 * config/arm/arm.md (use_literal_pool): Remove.
22002 (64-bit immediate split): No longer takes cost into consideration
22003 if arm_disable_literal_pool is enabled.
22004 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22005 used when arm_disable_literal_pool is enabled.
22006 (arm_max_const_double_inline_cost): Remove use of
22007 arm_disable_literal_pool.
22008 (push_minipool_fix): Add assert.
22009 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22010 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22011 (no_literal_pool_sf_immediate): New.
22013 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22015 PR tree-optimization/80613
22016 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22017 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22019 2017-05-05 Richard Biener <rguenther@suse.de>
22021 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22023 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22025 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22026 of this flag from insn conditions due to removal from r247495.
22028 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22030 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22032 (arm_early_store_addr_dep_ptr): Likewise.
22033 * config/arm/aarch-common-protos.h
22034 (arm_early_load_addr_dep_ptr): Add prototype.
22035 (arm_early_store_addr_dep_ptr): Likewise.
22036 * config/arm/cortex-a53.md: Add new bypasses.
22038 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22040 * tree.c (next_type_uid): Change type to unsigned.
22041 (type_hash_canon): Decrement back next_type_uid if
22042 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22043 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22046 2017-05-04 Martin Sebor <msebor@redhat.com>
22048 * builtins.c: Fix a trivial typo in a comment.
22050 PR middle-end/79234
22051 * builtins.c (check_sizes): Adjust to handle reading past the end.
22052 Avoid printing excessive upper bound of ranges. Use %E to print
22053 tree nodes instead of converting them to %wu.
22054 (expand_builtin_memchr): New function.
22055 (compute_dest_size): Rename...
22056 (compute_objsize): ...to this.
22057 (expand_builtin_memcpy): Adjust.
22058 (expand_builtin_mempcpy): Adjust.
22059 (expand_builtin_strcat): Adjust.
22060 (expand_builtin_strcpy): Adjust.
22061 (check_strncat_sizes): Adjust.
22062 (expand_builtin_strncat): Adjust.
22063 (expand_builtin_strncpy): Adjust and simplify.
22064 (expand_builtin_memset): Adjust.
22065 (expand_builtin_bzero): Adjust.
22066 (expand_builtin_memcmp): Adjust.
22067 (expand_builtin): Handle memcmp.
22068 (maybe_emit_chk_warning): Check strncat just once.
22070 2017-05-04 Martin Sebor <msebor@redhat.com>
22072 PR preprocessor/79214
22073 PR middle-end/79222
22074 PR middle-end/79223
22075 * builtins.c (check_sizes): Add inlining context and issue
22076 warnings even when -Wno-system-headers is set.
22077 (check_strncat_sizes): Same.
22078 (expand_builtin_strncat): Same.
22079 (expand_builtin_memmove): New function.
22080 (expand_builtin_stpncpy): Same.
22081 (expand_builtin): Handle memmove and stpncpy.
22083 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22085 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22086 which is not used any more.
22088 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22090 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22092 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22094 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22095 (cortexa53_tunings): Likewise.
22096 (cortexa57_tunings): Likewise.
22097 (cortexa72_tunings): Likewise.
22098 (cortexa73_tunings): Likewise.
22100 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22102 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22103 Set loop alignment to 8.
22105 2017-05-04 Martin Sebor <msebor@redhat.com>
22107 PR translation/80280
22108 * builtins.c (expand_builtin_object_size): Add missing quoting to
22109 %D and like directives.
22110 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22111 (hsa_type_for_tree_type): Same.
22112 (verify_function_arguments): Same.
22113 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22114 * varasm.c (get_section): Same.
22117 2017-05-04 Martin Sebor <msebor@redhat.com>
22119 PR translation/80280
22120 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22122 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22124 * config/aarch64/aarch64.c (generic_addrcost_table):
22125 Change HI/TI mode setting.
22127 2017-05-04 Martin Jambor <mjambor@suse.cz>
22129 PR tree-optimization/80622
22130 * tree-sra.c (comes_initialized_p): New function.
22131 (build_accesses_from_assign): Only set write lazily when
22132 comes_initialized_p is false.
22133 (analyze_access_subtree): Use comes_initialized_p.
22134 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22135 instead of testing for PARM_DECL.
22137 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22139 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22140 constraint on operand 0 to allow more general addressing modes.
22141 Adjust output template.
22142 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22144 * config/aarch64/aarch64-protos.h
22145 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22146 * config/aarch64/constraints.md (Dp): New address constraint.
22147 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22150 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22152 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22153 update use of estimate_ipcp_clone_size_and_time.
22154 (estimate_local_effects): Update use of
22155 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22156 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22157 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22158 Return nonspecialized time.
22160 2017-05-04 Richard Biener <rguenther@suse.de>
22162 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22163 for the last VUSE which def dominates the PHI. Directly call
22165 (get_continuation_for_phi_1): Remove.
22167 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22169 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22170 to explain the use of truncating division. Cap the number of
22171 iterations to the maximum given by nb_iterations_upper_bound,
22174 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22176 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22177 * configure: Regenerate.
22178 * config.in: Regenerate.
22179 * config/i386/driver-mingw32.c: new file.
22180 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22181 * config.host: Link driver-mingw32.o on MinGW host.
22182 * doc/install.texi: Document new --enable-mingw-wildcard configure
22185 2017-05-04 Marek Polacek <polacek@redhat.com>
22187 PR tree-optimization/80612
22188 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22190 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22191 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22193 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22194 (movt splitter): Likewise.
22195 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22196 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22197 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22198 block for Thumb-1 with MOVT.
22199 (thumb2_legitimate_address_p): Move code block ...
22200 (can_avoid_literal_pool_for_label_p): ... into this new function.
22201 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22203 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22204 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22205 "M-profile targets with the MOVT instruction".
22207 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22209 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22210 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22211 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22213 2017-05-04 Martin Liska <mliska@suse.cz>
22215 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22216 variable cond_code.
22218 2017-05-04 Richard Biener <rguenther@suse.de>
22220 * tree.c (array_at_struct_end_p): Handle arrays at struct
22221 end with flexarrays more conservatively. Refactor and treat
22222 arrays of arrays or aggregates more strict. Fix
22223 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22224 * tree.c (array_at_struct_end_p): Adjust prototype.
22225 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22226 * gimple-fold.c (get_range_strlen): Likewise.
22227 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22229 2017-05-04 Richard Biener <rguenther@suse.de>
22231 PR tree-optimization/31130
22232 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22234 (supports_overflow_infinity): Likewise.
22235 (is_negative_overflow_infinity): Likewise.
22236 (is_positive_overflow_infinity): Likewise.
22237 (is_overflow_infinity): Likewise.
22238 (stmt_overflow_infinity): Likewise.
22239 (overflow_infinity_range_p): Likewise.
22240 (usable_range_p): Remove as always returning true.
22241 (make_overflow_infinity): Remove.
22242 (negative_overflow_infinity): Likewise.
22243 (positive_overflow_infinity): Likewise.
22244 (avoid_overflow_infinity): Likewise.
22245 (set_value_range): Adjust accordingly.
22246 (set_value_range_to_nonnegative): Likewise, remove now unused
22247 overflow_infinity arg.
22248 (vrp_operand_equal_p): Adjust.
22249 (update_value_range): Likewise.
22250 (range_int_cst_singleton_p): Likewise.
22251 (operand_less_p): Likewise.
22252 (compare_values_warnv): Likewise.
22253 (extract_range_for_var_from_comparison_expr): Likewise.
22254 (vrp_int_const_binop): Likewise.
22255 (zero_nonzero_bits_from_vr): Likewise.
22256 (extract_range_from_multiplicative_op_1): Likewise.
22257 (extract_range_from_binary_expr_1): Likewise.
22258 (extract_range_from_unary_expr): Likewise.
22259 (extract_range_from_comparison): Likewise.
22260 (extract_range_basic): Likewise.
22261 (adjust_range_with_scev): Likewise.
22262 (compare_ranges): Likewise.
22263 (compare_range_with_value): Likewise.
22264 (dump_value_range): Likewise.
22265 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22267 (simplify_cond_using_ranges): Adjust.
22269 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22271 * brig-builtins.def: Added a builtin for class_f64.
22272 * builtin-types.def: Added a builtin type needed by class_f64.
22274 2017-05-03 Jason Merrill <jason@redhat.com>
22276 * timevar.def: Add TV_CONSTEXPR.
22278 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22280 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22282 2017-05-03 Martin Jambor <mjambor@suse.cz>
22284 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22285 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22286 * ipa-cp.c (ipcp_propagate_stage): Do not call
22287 ipa_update_after_lto_read.
22288 * ipa-inline.c (ipa_inline): Likewise.
22290 2017-05-03 Martin Jambor <mjambor@suse.cz>
22292 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22293 tag. Added a default constructor and a destructor.
22294 (ipa_edge_args_sum_t): New class;
22295 (ipa_edge_args_sum): Declare.
22296 (ipa_edge_args_vector): Remove declaration.
22297 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22298 (ipa_free_edge_args_substructures): Remove declaration.
22299 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22300 (ipa_edge_args_info_available_for_edge_p): Likewise.
22301 * ipa-prop.c (ipa_edge_args_vector): Removed.
22302 (edge_removal_hook_holder): Likewise.
22303 (edge_duplication_hook_holder): Likewise.
22304 (ipa_edge_args_sum): New variable.
22305 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22306 ipa_edge_args_vector.
22307 (ipa_free_edge_args_substructures): Likewise.
22308 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22309 ipa_edge_args_vector.
22310 (ipa_edge_removal_hook): Turned into method
22311 ipa_edge_args_sum_t::remove.
22312 (ipa_edge_duplication_hook): Turned into method
22313 ipa_edge_args_sum_t::duplicate.
22314 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22315 registering edge hooks.
22316 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22317 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22318 ipa_edge_args_sum instead of ipa_edge_args_vector.
22319 * ipa-profile.c (ipa_profile): Likewise.
22321 2017-05-03 Martin Jambor <mjambor@suse.cz>
22323 * symbol-summary.h (function_summary): New method exists.
22324 (function_summary::symtab_removal): Deallocate through release.
22325 (call_summary): New class.
22326 (gt_ggc_mx): New overload.
22327 (gt_pch_nx): Likewise.
22328 (gt_pch_nx): Likewise.
22330 2017-05-03 Jeff Law <law@redhat.com>
22332 PR tree-optimization/78496
22333 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22334 from simplify_cond_using_ranges. Split off code to walk
22335 backwards through casts into ...
22336 (simplify_cond_using_ranges_2): New function.
22337 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22338 (execute_vrp): After identifying jump threads, call
22339 simplify_cond_using_ranges_2.
22341 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22344 * ipa-inline.h (inline_summary): Add ctor.
22345 (create_ggc): Do not use ggc_cleared_alloc.
22347 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22348 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22350 * gcc.c (handle_braces): Support escaping in switch matching
22352 * doc/invoke.texi (Spec Files): Document it.
22353 Remove superfluous @code markup in items.
22355 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22357 * diagnostic-show-locus.c (struct column_range): New struct.
22358 (get_affected_columns): New function.
22359 (get_printed_columns): New function.
22360 (struct correction): New struct.
22361 (correction::ensure_capacity): New function.
22362 (correction::ensure_terminated): New function.
22363 (struct line_corrections): New struct.
22364 (line_corrections::~line_corrections): New dtor.
22365 (line_corrections::add_hint): New function.
22366 (layout::print_trailing_fixits): Reimplement in terms of the new
22368 (selftest::test_overlapped_fixit_printing): New function.
22369 (selftest::diagnostic_show_locus_c_tests): Call it.
22371 2017-05-03 Nathan Sidwell <nathan@acm.org>
22373 Canonicalize canonical type hashing
22374 * tree.h (type_hash_canon_hash): Declare.
22375 * tree.c (type_hash_list, attribute_hash_list): Move into
22376 type_hash_canon_hash.
22377 (build_type_attribute_qual_variant): Break out hash code calc into
22378 type_hash_canon_hash.
22379 (type_hash_canon_hash): New. Generic type hash computation.
22380 (build_range_type_1, build_array_type_1, build_function_type,
22381 build_method_type_directly, build_offset_type, build_complex_type,
22382 make_vector_type): Call it.
22384 2017-05-03 Richard Biener <rguenther@suse.de>
22386 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22387 When all DRs have unknown misaligned do not always peel
22388 when there is a store but apply the same costing model as if
22389 there were only loads.
22391 2017-05-03 Richard Biener <rguenther@suse.de>
22394 PR tree-optimization/80492
22395 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22396 compare_base_decls returning dont-know properly.
22398 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22400 * config/arm/iterators.md (CCSI): New mode iterator.
22401 (arch): New mode attribute.
22402 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22403 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22404 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22405 code iterator for success result mode.
22406 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22407 the corresponding new insn generators.
22409 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22412 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22413 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22415 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22417 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22418 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22419 (DDR_A): Wrap DDR argument in brackets.
22420 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22421 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22422 (DDR_REVERSED_P): Likewise.
22424 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22426 PR tree-optimization/79472
22427 * tree-switch-conversion.c (struct switch_conv_info): Add
22428 contiguous_range and default_case_nonstandard fields.
22429 (collect_switch_conv_info): Compute contiguous_range and
22430 default_case_nonstandard fields, don't clear final_bb if
22431 contiguous_range and only the default case doesn't have the required
22433 (check_all_empty_except_final): Set default_case_nonstandard instead
22434 of failing if contiguous_range and the default case doesn't have empty
22436 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22437 and only the default case doesn't have the required constants. Skip
22439 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22440 if default_case_nonstandard.
22441 (build_constructors): Build constant 1 just once. Assert that default
22442 values aren't inserted in between cases if contiguous_range. Skip
22444 (build_arrays): Skip virtual phis.
22445 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22446 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22447 Handle virtual phis.
22448 (gen_inbound_check): Handle default_case_nonstandard case.
22449 (process_switch): Adjust check_final_bb caller. Call
22450 gather_default_values with the first non-default case instead of
22451 default case if default_case_nonstandard.
22453 2017-05-02 Nathan Sidwell <nathan@acm.org>
22455 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22456 check. Fix formatting.
22458 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22460 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22461 errors when comparing specialized and unspecialized times.
22463 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22465 * diagnostic-show-locus.c
22466 (layout::should_print_annotation_line_p): Make private.
22467 (layout::print_annotation_line): Make private.
22468 (layout::annotation_line_showed_range_p): Make private.
22469 (layout::show_ruler): Make private.
22470 (layout::print_source_line): Make private. Pass in line and
22471 line_width, rather than calling location_get_source_line. Drop
22473 (layout::print_leading_fixits): New method.
22474 (layout::print_any_fixits): Rename to...
22475 (layout::print_trailing_fixits): ...this, and make private.
22476 Don't print newline fixits.
22477 (diagnostic_show_locus): Move logic for printing one row into...
22478 (layout::print_line): ...this new function. Move the
22479 location_get_source_line call and error-handling from
22480 print_source_line to here. Call print_leading_fixits, and rename
22481 print_any_fixits to print_trailing_fixits.
22482 (selftest::test_fixit_insert_containing_newline): Update now that
22483 newlines are partially supported.
22484 (selftest::test_fixit_insert_containing_newline_2): New test.
22485 (selftest::test_fixit_replace_containing_newline): Update comments.
22486 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22487 * edit-context.c (class added_line): New class.
22488 (class edited_line): Describe newline handling in comment.
22489 (edited_line::actually_edited_p): New method.
22490 (edited_line::print_content): Delete redundant decl.
22491 (edited_line::m_predecessors): New field.
22492 (edited_file::print_content): Call edited_line::print_content.
22493 (edited_file::print_diff): Update to support newlines.
22494 (edited_file::print_diff_hunk): Likewise.
22495 (edited_file::print_run_of_changed_lines): New function.
22496 (edited_file::print_diff_line): Convert to...
22497 (print_diff_line): ...this.
22498 (edited_file::get_effective_line_count): New function.
22499 (edited_line::edited_line): Initialize new field m_predecessors.
22500 (edited_line::~edited_line): Clean up m_predecessors.
22501 (edited_line::apply_fixit): Handle newlines.
22502 (edited_line::get_effective_line_count): New function.
22503 (edited_line::print_content): New function.
22504 (edited_line::print_diff_lines): New function.
22505 (selftest::test_applying_fixits_insert_containing_newline): New
22507 (selftest::test_applying_fixits_replace_containing_newline): New
22509 (selftest::insert_line): New function.
22510 (selftest::test_applying_fixits_multiple_lines): Add example of
22512 (selftest::edit_context_c_tests): Call the new tests.
22514 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22516 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22517 parameter cand. Update dump information.
22518 (get_computation_cost): Update uses.
22520 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22522 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22523 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22524 (get_computation_at, rewrite_use_address): Update use of
22525 get_computation_aff.
22527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22529 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22530 (get_computation): Delete.
22531 (get_computation_cost): Implement like get_computation_cost_at.
22532 Use get_computation_at.
22533 (get_computation_cost_at): Delete.
22534 (rewrite_use_nonlinear_expr): Use get_computation_at.
22535 (rewrite_use_compare, remove_unused_ivs): Ditto.
22537 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22539 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22541 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22543 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22544 (ivopts_global_cost_for_size): Rename parameter and update uses.
22545 (iv_ca_recount_cost): Update uses.
22546 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22547 candidates seperately in n_invs and n_cands.
22548 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22550 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22552 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22553 (find_inv_vars_cb): New.
22554 (find_depends): Renamed to ...
22555 (find_inv_vars): ... this.
22556 (add_candidate_1, force_var_cost): Call find_inv_vars.
22557 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22559 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22561 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22562 inv_vars. Add inv_exprs.
22563 (struct iv_cand): Rename depends_on to inv_vars.
22564 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22565 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22566 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22567 (dump_cand): Dump inv_vars.
22568 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22569 (record_invariant, find_depends, add_candidate_1): Ditto.
22570 (set_group_iv_cost, force_var_cost): Ditto.
22571 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22572 (get_computation_cost_at, get_computation_cost): Ditto.
22573 (determine_group_iv_cost_generic): Ditto.
22574 (determine_group_iv_cost_address): Ditto.
22575 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22576 (determine_group_iv_costs): Ditto.
22577 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22578 (iv_ca_set_remove_invariants): Renamed to ...
22579 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22580 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22581 (iv_ca_set_add_invariants): Renamed to ...
22582 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22583 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22584 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22585 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22586 (create_new_ivs): Remove useless dump.
22588 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22590 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22592 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22593 (iv_ca_set_no_cp, create_new_iv): Ditto.
22595 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22597 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22599 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22601 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22602 function tree_check2.
22604 2017-05-02 Martin Liska <mliska@suse.cz>
22606 * doc/gcov.texi: Add missing preposition.
22607 * gcov.c (function_info::function_info): Properly fill up
22608 all member variables.
22610 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22612 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22614 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22616 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22618 2017-05-02 Martin Liska <mliska@suse.cz>
22621 * lto-streamer-in.c (lto_read_tree_1): Remove
22622 LTO_STREAMER_DEBUG.
22623 * lto-streamer.c (struct tree_hash_entry): Likewise.
22624 (struct tree_entry_hasher): Likewise.
22625 (tree_entry_hasher::hash): Likewise.
22626 (tree_entry_hasher::equal): Likewise.
22627 (lto_streamer_init): Likewise.
22628 (lto_orig_address_map): Likewise.
22629 (lto_orig_address_get): Likewise.
22630 (lto_orig_address_remove): Likewise.
22631 * lto-streamer.h: Likewise.
22632 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22633 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22635 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22637 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22638 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22639 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22640 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22641 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22642 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22643 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22644 (mm_maskz_sub_ss): New intrinsics.
22645 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22646 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22647 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22648 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22649 (__builtin_ia32_subss_mask_round): New builtins.
22650 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22651 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22652 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22654 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22655 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22657 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22660 2017-05-02 Martin Jambor <mjambor@suse.cz>
22662 PR tree-optimization/78687
22663 * tree-sra.c (access): New field parent.
22664 (process_subtree_disqualification): New function.
22665 (disqualify_candidate): Call it.
22666 (build_accesses_from_assign): Reset write flag if creating an
22668 (build_access_subtree): Fill in parent field and also prpagate
22669 down grp_write flag.
22670 (create_artificial_child_access): New parameter set_grp_write, set
22671 grp_write to its value.
22672 (propagate_subaccesses_across_link): Also propagate grp_write flag
22674 (propagate_all_subaccesses): Push the closest parent back to work
22675 queue if add_access_to_work_queue returned true.
22677 2017-05-02 Richard Biener <rguenther@suse.de>
22679 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22680 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22681 -fstrict-overflow documentation.
22682 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22683 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22684 flag_strict_overflow.
22685 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22686 * lto-opts.c (lto_write_options): Do not stream it.
22687 * lto-wrapper.c (merge_and_complain): Do not handle it.
22688 * opts.c (default_options_table): Do not set -fstrict-overflow.
22689 (finish_options): Likewise do not clear it when sanitizing.
22690 * simplify-rtx.c (simplify_const_relational_operation): Do not
22691 test flag_strict_overflow.
22693 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22695 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22696 using enabled attribute.
22697 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22698 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22699 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22700 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22701 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22702 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22703 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22704 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22705 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22706 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22708 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22710 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22712 2017-05-02 Richard Biener <rguenther@suse.de>
22714 PR tree-optimization/80591
22716 2017-04-10 Richard Biener <rguenther@suse.de>
22718 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22721 2017-05-02 Richard Biener <rguenther@suse.de>
22723 PR tree-optimization/80549
22724 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22725 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22726 headers if they do not have a preheader.
22728 2017-05-02 Martin Liska <mliska@suse.cz>
22731 * common.opt: Fix typo.
22732 * doc/invoke.texi: Likewise.
22734 2017-05-01 Jan Beulich <jbeulich@suse.com>
22736 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22737 swapping, add (x,x,m,x,n) alternative.
22739 2017-05-01 Nathan Sidwell <nathan@acm.org>
22741 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22742 unnecessary unadjusted_alignment check.
22744 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22747 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22749 * gimplify.c (gimplify_cilk_detach): New function.
22750 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22751 * tree-core.h: Document EXPR_CILK_SPAWN.
22752 * tree.h (EXPR_CILK_SPAWN): Define.
22754 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22756 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22757 to use new fixit_hint representation, using the "replace" logic.
22758 (get_line_span_for_fixit_hint): Likewise.
22759 (layout::print_any_fixits): Likewise.
22760 (selftest::test_one_liner_many_fixits): Rename to...
22761 (selftest::test_one_liner_many_fixits_1): ...this, and update
22762 comment and expected output to reflect that the multiple fix-it
22763 hints are now consolidated into one insertion.
22764 (selftest::test_one_liner_many_fixits_2): New test.
22765 (selftest::test_diagnostic_show_locus_one_liner): Update for
22767 (selftest::test_fixit_consolidation): Update for fix-it API
22769 * diagnostic.c (print_parseable_fixits): Likewise.
22770 * edit-context.c (edited_line::m_line_events): Convert from
22771 auto_vec <line_event *> to auto_vec <line_event>.
22772 (class line_event): Convert from abstract base class to a concrete
22773 class, taking over the role of replace_event.
22774 (class insert_event): Delete.
22775 (class replace_event): Rename to class line_event. Convert to
22777 (edit_context::add_fixits): Reimplement.
22778 (edit_context::apply_insert): Delete.
22779 (edit_context::apply_replace): Rename to...
22780 (edit_context::apply_fixit): ...this. Convert to half-open range.
22781 (edited_file::apply_insert): Delete.
22782 (edited_file::apply_replace): Rename to...
22783 (edited_file::apply_fixit): ...this.
22784 (edited_line::~edited_line): Drop deletion of events.
22785 (edited_line::apply_insert): Delete.
22786 (edited_line::apply_replace): Rename to...
22787 (edited_line::apply_fixit): ...this. Convert to half-open range.
22788 Update for change to type of m_line_events.
22789 * edit-context.h (edit_context::apply_insert): Delete.
22790 (edit_context::apply_replace): Rename to...
22791 (edit_context::apply_fixit): ...this.
22793 2017-05-01 Martin Sebor <msebor@redhat.com>
22795 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22798 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22801 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22802 __get_cpuid_max returns 0.
22803 (__get_cpuid_count): Ditto.
22805 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22807 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22808 replacement expression is another instance of one of its arguments.
22810 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22813 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22814 check for stack push/pop autoinc.
22815 * config/i386/i386.c (ix86_agi_dependent): Return false
22816 if the only reason why modified_in_p returned true is that
22817 addr is SP based and set_insn is a push or pop.
22819 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22821 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22824 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22827 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22828 (account_size_time): Use two predicates - exec_pred and
22830 (evaluate_conditions_for_known_args): Compute both clause and
22832 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22833 (inline_summary_t::duplicate): Update.
22834 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22836 (compute_inline_parameters): Likewise.
22837 (estimate_edge_size_and_time): Update caluclation of time.
22838 (estimate_node_size_and_time): Compute both time and nonspecialized
22840 (estimate_ipcp_clone_size_and_time): Update.
22841 (inline_merge_summary): Update.
22842 (do_estimate_edge_time): Update.
22843 (do_estimate_edge_size): Update.
22844 (do_estimate_edge_hints): Update.
22845 (inline_read_section, inline_write_summary): Stream both new predicates.
22846 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22848 (compute_inlined_call_time): Cleanup.
22849 (big_speedup_p): Update.
22850 (edge_badness): Update.
22851 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22852 (size_time_entry): Replace predicate by exec_predicate and
22853 nonconst_predicate.
22854 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22855 (estimate_edge_time): Return also nonspec_time.
22856 (reset_edge_growth_cache): Update.
22858 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22860 PR rtl-optimization/80491
22861 * ifcvt.c (noce_process_if_block): When looking for x setter
22862 with missing else_bb, don't check only the insn right before
22863 cond_earliest, but look for the last insn that x is modified in
22864 within the same bb.
22866 PR rtl-optimization/80491
22867 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22869 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22871 PR tree-optimization/80487
22872 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22874 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22876 PR tree-optimization/79697
22877 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22878 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22879 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22881 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22882 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22884 2017-04-28 Martin Sebor <msebor@redhat.com>
22886 PR tree-optimization/80523
22887 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22888 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22890 (maybe_warn, format_directive, parse_directive): Use new functions.
22891 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22893 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22895 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22897 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22899 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22900 target_header_dir): Set correctly.
22901 * configure: Regenerated.
22902 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22903 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22904 instead of SYSTEM_HEADER_DIR.
22906 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22908 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22909 (estimate_local_effects): Likewise.
22910 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22911 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22912 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22913 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22914 do_estimate_edge_time, estimate_edge_time): Likewise.
22915 * ipa-inline-analysis.c (estimate_node_size_and_time,
22916 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22917 (estimate_time_after_inlining): Remove.
22919 2017-04-28 Martin Liska <mliska@suse.cz>
22921 * doc/gcov.texi: Enhance documentation of gcov.
22923 2017-04-28 Martin Liska <mliska@suse.cz>
22925 * doc/gcov.texi: Sort options in alphabetic order.
22926 * doc/gcov-dump.texi: Likewise.
22927 * doc/gcov-tool.texi: Likewise.
22928 * gcov.c (print_usage): Likewise.
22929 * gcov-dump.c (print_usage): Likewise.
22930 * gcov-tool.c (print_merge_usage_message): Likewise.
22931 (print_rewrite_usage_message): Likewise.
22932 (print_overlap_usage_message): Likewise.
22934 2017-04-28 Martin Liska <mliska@suse.cz>
22936 PR gcov-profile/53915
22937 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22939 2017-04-28 Martin Liska <mliska@suse.cz>
22941 PR gcov-profile/79891
22942 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22943 is marked by compiler as living on a line.
22944 (get_cycles_count): Remove usage of the union.
22945 (output_intermediate_file): Likewise.
22946 (find_source): Fix GNU coding style.
22947 (accumulate_line_counts): Remove old non-all block mode.
22948 (output_lines): Remove usage of the union.
22949 * profile.c (output_location): Include all BBs, even if
22950 belonging to a same line (and file) as a previous BB.
22952 2017-04-28 Martin Liska <mliska@suse.cz>
22954 * gcov.c (process_args): Handle new argument 'w'.
22955 (read_graph_file): Assign ID to BBs.
22956 (output_branch_count): Display BB # if verbose flag is set.
22957 (output_lines): Likewise for arcs.
22958 (print_usage): Add '--verbose' option help.
22959 * doc/gcov.texi: Document --verbose (-w) option.
22961 2017-04-28 Martin Liska <mliska@suse.cz>
22963 * gcov.c (struct block_location_info): New struct.
22964 (process_file): Fill up the new structure.
22965 (read_graph_file): Replace usage of encoding by the newly added
22967 (add_line_counts): Likewise.
22968 (accumulate_line_counts): Remove usage of the union.
22969 (function_info::function_info): New function.
22970 (function_info::~function_info): Likewise.
22971 (process_file): Call delete instead of release_function.
22972 (release_function): Release the function.
22973 (release_structures): Call delete instead of release_function.
22974 (solve_flow_graph): Replace usage of num_blocks.
22975 (find_exception_blocks): Likewise.
22976 (output_lines): Fix GNU coding style.
22978 2017-04-28 Martin Liska <mliska@suse.cz>
22981 * coverage.c (coverage_remove_note_file): New function.
22982 * coverage.h: Declare the function.
22983 * toplev.c (finalize): Clean if an error has been seen.
22985 2017-04-28 Martin Liska <mliska@suse.cz>
22987 PR gcov-profile/80031
22988 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22989 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22990 * gcov.c (read_graph_file): Read just number of blocks.
22991 * profile.c (branch_prob): Do not stream 0 flags per a basic
22994 2017-04-28 Martin Liska <mliska@suse.cz>
22996 * gcov-dump.c (tag_*): Add new argument to declarations.
22997 (dump_gcov_file): Likewise.
22998 (tag_blocks): Add and use new argument depth.
22999 (tag_arcs): Likewise.
23000 (tag_lines): Likewise.
23001 (tag_counters): Likewise.
23002 (tag_summary): Likewise.
23003 (dump_working_sets): Use depth to do a proper indentation.
23005 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23008 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23009 * symtab.c (symtab_node::debug_symtab): Move definition here.
23011 2017-04-28 Richard Biener <rguenther@suse.de>
23013 * lto-streamer.h (LTO_major_version): Bump to 7.
23015 2017-04-28 Richard Biener <rguenther@suse.de>
23017 * tree-vrp.c (assert_info): New struct.
23018 (add_assert_info): New helper.
23019 (register_edge_assert_for_2): Refactor to add asserts to a vector
23021 (register_edge_assert_for_1): Likewise.
23022 (register_edge_assert_for): Likewise.
23023 (finish_register_edge_assert_for): New helper actually registering
23024 asserts where live on edge.
23025 (find_conditional_asserts): Adjust.
23026 (find_switch_asserts): Likewise.
23027 (evrp_dom_walker::try_find_new_range): Generalize.
23028 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23030 2017-04-27 Marek Polacek <polacek@redhat.com>
23033 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23034 arg10 and arg11 to itype.
23036 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23038 * doc/extend.texi (Object Size Checking): Improve grammar.
23040 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23043 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23044 that the logic for permitting reciprocal estimates matches that
23047 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23050 * tree.c (type_cache_hasher::equal): Only compare
23051 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23052 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23053 non-aggregate element types.
23054 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23055 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23057 2017-04-27 Richard Biener <rguenther@suse.de>
23059 PR middle-end/80533
23060 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23061 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23062 keeping a reference to a trailing array.
23064 2017-04-27 Richard Biener <rguenther@suse.de>
23066 PR middle-end/80539
23067 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23068 being in loop-closed SSA form conservatively.
23069 (chrec_fold_multiply_poly_poly): Likewise.
23071 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23073 PR middle-end/79665
23074 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23075 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23077 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23080 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23081 (aarch64_function_arg_alignment): Return unsigned int again, but still
23082 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23083 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23084 Don't emit -Wpsabi note.
23085 (aarch64_function_arg_boundary): Likewise.
23086 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23089 2017-04-26 Nathan Sidwell <nathan@acm.org>
23091 * tree.h (crc32_unsigned_n): Declare.
23092 (crc32_unsigned, crc32_unsigned): Make inline.
23093 * tree.c (crc32_unsigned_bits): Replace with ...
23094 (crc32_unsigned_n): ... this.
23095 (crc32_unsigned, crc32_byte): Remove.
23096 (crc32_string): Remove unnecessary braces.
23098 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23100 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23101 * ipa-inline-analysis.c (MAX_TIME): Remove.
23102 (account_size_time): Use sreal for time.
23103 (dump_inline_summary): Update.
23104 (estimate_function_body_sizes): Update.
23105 (estimate_edge_size_and_time): Update.
23106 (estimate_calls_size_and_time): Update.
23107 (estimate_node_size_and_time): Update.
23108 (inline_merge_summary): Update.
23109 (inline_update_overall_summary): Update.
23110 (estimate_time_after_inlining): Update.
23111 (inline_read_section): Update.
23112 (inline_write_summary): Update.
23113 * ipa-inline.c (compute_uninlined_call_time): Update.
23114 (compute_inlined_call_time): Update.
23115 (recursive_inlining): Update.
23116 (inline_small_functions): Update.
23117 (dump_overall_stats): Update.
23118 * ipa-inline.h: Include sreal.h.
23119 (size_time_entry): Turn time to sreal.
23120 (inline_summary): Turn self_time nad time to sreal.
23122 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23124 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23126 (sreal::stream_out, sreal::stream_in): New.
23127 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23129 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23131 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23134 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23137 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23139 (dimode_scalar_chain::compute_convert_gain): Ditto.
23140 (dimode_scalar_chain::make_vector_copies): Ditto.
23141 (dimode_scalar_chain::convert_reg): Ditto.
23142 (dimode_scalar_chain::convert_insn): Ditto.
23143 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23144 (VI248_AVX512BW_1): New mode iterator.
23145 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23146 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23149 2017-04-25 Martin Sebor <msebor@redhat.com>
23151 PR tree-optimization/80497
23152 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23153 constants are representable in HOST_WIDE_INT.
23154 (parse_directive): Ditto.
23156 2017-04-25 Martin Sebor <msebor@redhat.com>
23159 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23160 (new_zero_array): Adjust signature.
23161 (dom_info::dom_init): Used unsigned rather that size_t.
23162 (dom_info::dom_info): Same.
23164 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23165 Jakub Jelinek <jakub@redhat.com>
23168 * config/arm/arm.c: Include gimple.h.
23169 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23170 returns negative, increment ncrn only if it returned positive.
23171 (arm_needs_doubleword_align): Return int instead of bool,
23172 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23173 members, but if there is any such non-FIELD_DECL
23174 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23175 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23176 returns negative, increment nregs only if it returned positive.
23177 (arm_setup_incoming_varargs): Likewise.
23178 (arm_function_arg_boundary): Emit -Wpsabi note if
23179 arm_needs_doubleword_align returns negative, return
23180 DOUBLEWORD_ALIGNMENT only if it returned positive.
23182 2017-04-25 Marek Polacek <polacek@redhat.com>
23185 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23186 first argument to type.
23188 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23191 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23192 type checks to test for compatibility instead of equality.
23194 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23195 Jakub Jelinek <jakub@redhat.com>
23198 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23200 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23201 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23202 the alignment computation, but return their maximum in warn_alignment.
23203 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23204 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23206 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23207 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23210 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23212 * config/arc/simdext.md (dmpyh): Fix typo.
23214 2017-04-25 Richard Biener <rguenther@suse.de>
23216 PR tree-optimization/80492
23217 * alias.c (compare_base_decls): Handle registers with asm
23218 specification conservatively.
23219 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23220 compare_base_decls returning dont-know properly.
23222 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23224 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23225 (legitimate_offset_address_p): New function.
23226 (arc_legitimate_address_p): Use above function.
23228 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23230 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23232 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23234 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23235 ACCH registers whenever they are available.
23237 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23239 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23240 double regs fix when not used.
23242 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23244 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23246 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23247 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23249 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23251 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23252 when emitting short ADD instructions.
23254 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23256 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23258 (cmpsi_cc_c_insn): Likewise.
23259 (cbranchsi4_scratch): Compute proper instruction length using
23260 compact_hreg_operand.
23261 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23263 2017-04-25 Richard Biener <rguenther@suse.de>
23265 PR middle-end/80509
23266 * passes.c (pass_manager::pass_manager): Initialize
23267 m_name_to_pass_map.
23269 2017-04-25 Richard Biener <rguenther@suse.de>
23271 PR tree-optimization/79201
23272 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23274 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23277 * config/s390/vector.md: Split MEM->GPR vector moves for
23278 non-s_operand addresses.
23280 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23283 * config/s390/predicates.md (reload_const_wide_int_operand): New
23285 * config/s390/s390.md ("movti"): Remove d/P alternative.
23286 ("movti_bigconst"): New pattern definition.
23288 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23291 * s390-protos.h (s390_expand_cs_hqi): Removed.
23292 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23293 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23294 modes as well as CCZ1mode and CCZmode.
23295 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23296 signature of s390_emit_compare_and_swap.
23297 (s390_expand_cs_hqi): Likewise, make static.
23298 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23299 compare-and-swap, in some cases.
23300 (s390_expand_cs): Wrapper function.
23301 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23303 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23304 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23305 patterns for small and large integers. Forbid symref memory operands.
23306 Move expander to s390.c. Require cc register.
23307 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23308 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23309 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23310 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23311 symref memory operands. Remove CC mode and call s390_match_ccmode
23313 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23315 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23317 * config/s390/s390.md (define_peephole2): New peephole to help
23318 combining the load-and-test pattern with volatile memory.
23320 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23322 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23323 with CCZmode for TARGET_Z196.
23325 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23327 PR rtl-optimization/80501
23328 * combine.c (make_compound_operation_int): Set subreg_code to SET
23329 even for AND with mask of the sign bit of mode.
23331 PR rtl-optimization/80500
23332 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23333 sum's initial value.
23335 2017-04-25 Julian Brown <julian@codesourcery.com>
23336 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23338 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23340 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23342 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23344 2017-04-25 Julian Brown <julian@codesourcery.com>
23345 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23347 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23348 (thunderx2t99_sha): New Reservation.
23350 2017-04-25 Julian Brown <julian@codesourcery.com>
23351 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23353 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23354 type for 1-element load.
23356 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23358 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23360 2017-04-24 Martin Jambor <mjambor@suse.cz>
23362 PR tree-optimization/80293
23363 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23364 char arrays not totally scalarizable if it is false.
23365 (analyze_all_variable_accesses): Pass correct value in the new
23366 parameter. Add a statistics counter.
23368 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23370 PR middle-end/79931
23371 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23373 2017-04-24 Richard Biener <rguenther@suse.de>
23375 PR tree-optimization/80494
23376 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23377 out for complex types.
23379 2017-04-24 Richard Biener <rguenther@suse.de>
23381 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23382 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23383 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23384 (DFS): Adjust and never fail.
23385 (sccvn_dom_walker::fail): Remove.
23386 (sccvn_dom_walker::before_dom_children): Adjust.
23387 (run_scc_vn): Likewise and never fail.
23388 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23389 (pass_fre::execute): Likewise.
23391 2017-04-24 Richard Biener <rguenther@suse.de>
23393 PR tree-optimization/79725
23394 * tree-ssa-sink.c (statement_sink_location): Return whether
23395 failure reason was zero uses. Move that check later.
23396 (sink_code_in_bb): Deal with zero uses by removing the stmt
23399 2017-04-24 Richard Biener <rguenther@suse.de>
23402 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23403 pointer-based references.
23405 2017-04-24 Richard Biener <rguenther@suse.de>
23408 * pass_manager.h (pass_manager::operator new): Remove.
23409 (pass_manager::operator delete): Likewise.
23410 * passes.c (pass_manager::operator new): Remove.
23411 (pass_manager::operator delete): Likewise.
23412 (pass_manager::pass_manager): Zero individual pass members.
23414 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23417 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23418 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23419 Check "XEXP (src, 1)" operand here.
23420 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23421 Check "XEXP (src, 1)" operand here.
23422 (dimode_scalar_chain::make_vector_copies): Detect count register
23423 of a shift instruction. Zero extend count register from QImode
23424 to DImode to satisfy vector shift pattern count operand predicate.
23425 Substitute vector shift count operand with a DImode copy.
23426 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23429 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23431 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23432 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23433 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23434 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23435 (UNSPEC_NOREX_MEM): Remove definition.
23437 2017-04-21 Richard Biener <rguenther@suse.de>
23439 PR tree-optimization/79547
23440 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23441 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23442 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23443 without any constraints.
23445 2017-04-21 Richard Biener <rguenther@suse.de>
23447 PR tree-optimization/78847
23448 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23450 2017-04-21 Richard Biener <rguenther@suse.de>
23452 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23453 (build_distinct_type_copy): Likewise.
23454 (build_variant_type_copy): Likewise.
23455 * tree.c (build_qualified_type): Pass down mem-stat info.
23456 (build_distinct_type_copy): Likewise.
23457 (build_variant_type_copy): Likewise.
23459 2017-04-21 Richard Biener <rguenther@suse.de>
23461 PR tree-optimization/80237
23462 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23464 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23465 for a simplified result.
23467 2016-04-21 Richard Biener <rguenther@suse.de>
23469 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23470 sth as strict as a simple_iv but a chrec without symbols and an
23471 operand defined in the loop we are peeling (and not some subloop).
23472 (propagate_constants_for_unrolling): Propagate all constants.
23474 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23477 * config/i386/i386.c (print_reg): Remove assert for disalowed
23478 regno values, call output_operand_lossage instead.
23480 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23483 * config/i386/constraints.md (Yc): New register constraint.
23484 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23485 Use Yc constraint for alternative 2 of operand 0. Remove
23486 preferred_for_speed attribute.
23488 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23490 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23491 lastprivate clauses in SIMT case.
23493 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23495 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23497 2017-04-20 Richard Biener <rguenther@suse.de>
23499 PR tree-optimization/57796
23500 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23501 as N scalar stores.
23502 (vect_model_load_cost): Cost gathers as N scalar loads.
23504 2017-04-20 Richard Biener <rguenther@suse.de>
23506 * ggc-page.c (ggc_allocated_p): Rename to ...
23507 (safe_lookup_page_table_entry): ... this and return the lookup
23509 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23511 2017-04-20 Richard Biener <rguenther@suse.de>
23513 PR tree-optimization/80453
23514 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23515 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23516 from the conditions.
23517 (vn_phi_eq): Pass them down.
23518 (vn_phi_lookup): Record them.
23519 (vn_phi_insert): Likewise.
23521 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23523 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23524 uninitialized variable warning to avoid buffer overrun.
23526 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23529 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23530 is suppressed for '{ 0 }' in C.
23532 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23534 * BASE-VER: Set to 8.0.0.
23536 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23538 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23539 priority .init_array and .fini_array section with SECTION_NOTYPE
23542 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23544 PR middle-end/80423
23545 * tree.h (build_array_type): Add typeless_storage default argument.
23546 * tree.c (type_cache_hasher::equal): Also compare
23547 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23548 (build_array_type): Add typeless_storage argument, set
23549 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23551 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23552 (build_array_type): Likewise. Add typeless_storage argument.
23554 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23555 Jakub Jelinek <jakub@redhat.com>
23557 PR tree-optimization/80426
23558 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23559 operation on symbolic operands, also compute the overflow for the
23560 invariant part when the operation degenerates into a negation.
23562 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23565 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23566 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23569 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23571 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23574 * config/avr/avr.c (tree.h): Include it.
23575 (cgraph.h): Include it.
23576 (avr_encode_section_info): Don't warn for uninitialized progmem
23577 variable if it's just an alias.
23579 2017-04-19 Richard Biener <rguenther@suse.de>
23582 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23583 when needed by AutoPGO.
23585 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23588 * doc/lto.texi: Remove an extra 'that'.
23590 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23592 PR rtl-optimization/80429
23593 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23594 are only used in debug insns.
23596 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23597 Vladimir Makarov <vmakarov@redhat.com>
23599 * config/sparc/predicates.md (input_operand): Add comment. Return
23600 true for any memory operand when LRA is in progress.
23601 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23603 2017-04-18 Jeff Law <law@redhat.com>
23606 * mips.md ({return,simple_return}_internal): Do not overwrite
23609 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23611 PR tree-optimization/80443
23612 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23613 instead of adding 1, subtract -1 and similarly instead of subtracting
23616 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23618 PR rtl-optimization/80357
23619 * haifa-sched.c (tmp_bitmap): New variable.
23620 (model_recompute): Handle duplicate use records.
23621 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23622 (free_global_sched_pressure_data): Free it.
23624 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23627 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23628 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23629 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23630 instead of SYSTEM_HEADER_DIR.
23632 2017-04-18 Jeff Law <law@redhat.com>
23634 PR middle-end/80422
23635 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23636 predecessors after walking up the insn chain.
23638 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23641 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23642 sizetype type into debug info.
23644 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23647 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23648 unneeded test for TARGET_UPPER_REGS_SF.
23649 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23651 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23654 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23655 instead of gsi_after_labels.
23657 2017-04-18 Jeff Law <law@redhat.com>
23659 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23663 2017-04-13 Jeff Law <law@redhat.com>
23664 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23665 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23667 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23670 * config/avr/avr.c (intl.h): Include it.
23671 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23673 2017-04-18 Martin Liska <mliska@suse.cz>
23675 PR gcov-profile/78783
23676 * gcov-tool.c (gcov_output_files): Validate that destination
23677 file is either removed by the tool or by a user.
23679 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23680 Guy Benyei <guybe@mellanox.com>
23682 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23683 block, and do not negate it, the stored id is already negative.
23685 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23687 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23689 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23692 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23693 masks of options that should be turned off if the VSX vector
23694 options are turned off.
23695 (OTHER_P8_VECTOR_MASKS): Likewise.
23696 (OTHER_VSX_VECTOR_MASKS): Likewise.
23697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23698 rs6000_disable_incompatible_switches to validate no type switches
23700 (rs6000_incompatible_switch): New function to disallow turning on
23701 other vector options if -mno-vsx, -mno-power8-vector, or
23702 -mno-power9-vector are specified.
23704 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23706 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23708 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23710 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23711 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23712 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23713 (ARG_POINTER_CFA_OFFSET): Likewise.
23715 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23717 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23718 conditions to take advantage of various optimizations.
23720 2017-04-13 Jeff Law <law@redhat.com>
23722 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23723 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23724 (zero_extendsidi2_dext): Likewise.
23726 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23729 * fold-const.c (fold_ternary_loc): Revert
23730 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23733 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23735 PR rtl-optimization/80343
23736 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23737 new scratch pseudo.
23739 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23742 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23743 to ubsan_encode_value.
23745 2017-04-13 Jeff Law <law@redhat.com>
23747 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23748 appearing in DEBUG_INSNs.
23750 2017-04-13 Martin Liska <mliska@suse.cz>
23752 PR gcov-profile/80413
23753 * gcov-io.c (gcov_write_string): Copy to buffer just when
23754 allocated size is greater than zero.
23756 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23759 * dwarf2out.c (decls_for_scope): Ignore declarations of
23760 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23762 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23765 * ipa-visibility.c (non_local_p): Fix typos.
23766 (localize_node): When localizing symbol in same comdat group,
23767 dissolve the group only when we know external symbols are going
23769 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23771 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23773 PR tree-optimization/79390
23774 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23775 order does not result in usable sequence, retry with reversed operand
23781 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23782 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23783 op0 instead of fold_convert_loc (loc, type, arg0).
23785 2017-04-12 Jeff Law <law@redhat.com>
23787 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23788 has a delay slot in the generated code.
23790 * config/cris/cris.md (cris_preferred_reload_class): Return
23791 GENNONACR_REGS rather than GENERAL_REGS.
23793 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23796 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23797 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23798 signedness of the result type.
23800 2017-04-12 Richard Biener <rguenther@suse.de>
23801 Jeff Law <law@redhat.com>
23803 PR tree-optimization/80359
23804 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23805 trim stores to TARGET_MEM_REFs.
23807 2017-04-12 Richard Biener <rguenther@suse.de>
23809 PR tree-optimization/79390
23810 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23811 threading case even more.
23813 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23816 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23817 for quad_address_p for TImode, instead of just not indexed_address.
23819 2017-04-12 Richard Biener <rguenther@suse.de>
23820 Bernd Edlinger <bernd.edlinger@hotmail.de>
23822 PR middle-end/79671
23823 * alias.c (component_uses_parent_alias_set_from): Handle
23824 TYPE_TYPELESS_STORAGE.
23825 (get_alias_set): Likewise.
23826 * tree-core.h (tree_type_common): Add typeless_storage flag.
23827 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23828 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23829 for types containing members with TYPE_TYPELESS_STORAGE.
23830 (place_field): Likewise.
23831 (layout_type): Likewise for ARRAY_TYPE.
23832 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23833 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23834 TYPE_TYPELESS_STORAGE.
23835 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23837 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23840 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23841 first argument to type.
23843 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23847 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23848 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23849 (rs6000_expand_binop_builtin): Likewise.
23850 (rs6000_expand_ternop_builtin): Likewise; also add missing
23851 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23852 vshasigma built-ins.
23853 * doc/extend.texi: Document that vec_xxpermdi's third argument
23854 must be a constant.
23856 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23858 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23859 Use shift_const cost parameter when calculating gain of STV shifts.
23861 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23863 PR rtl-optimization/70478
23864 * lra-constraints.c (process_alt_operands): Check memory for
23865 disfavoring memory insn operand.
23867 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23869 PR middle-end/80100
23870 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23871 left shift in unsigned HOST_WIDE_INT type.
23873 PR rtl-optimization/80385
23874 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23875 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23878 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23879 if they have any depend clauses.
23881 2017-04-11 Martin Liska <mliska@suse.cz>
23884 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23885 * ipa-split.c (split_function): Create a local comdat symbol
23886 if caller is in a comdat group.
23888 2017-04-11 Martin Liska <mliska@suse.cz>
23891 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23894 2017-04-11 Martin Sebor <msebor@redhat.com>
23896 PR middle-end/80364
23897 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23898 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23900 (directive::set_width, directive::set_precision, format_character):
23902 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23905 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23908 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23909 conflict, set target->arch_name instead of target->cpu_name.
23911 2017-04-11 Richard Biener <rguenther@suse.de>
23913 PR tree-optimization/80374
23914 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23915 build_zero_cst, remove fold_convertible_p check again.
23917 2017-04-11 Martin Liska <mliska@suse.cz>
23920 * ubsan.c (instrument_object_size): Do not instrument register
23923 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23926 * config/i386/i386-builtin-types.def
23927 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23928 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23929 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23930 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23931 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23932 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23933 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23934 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23935 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23936 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23937 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23938 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23939 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23940 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23941 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23942 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23943 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23944 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23945 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23946 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23947 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23948 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23949 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23950 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23951 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23952 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23953 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23954 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23955 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23956 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23957 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23958 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23959 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23960 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23961 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23962 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23963 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23964 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23965 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23966 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23967 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23968 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23969 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23970 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23971 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23972 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23974 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23975 flag to second_arg_count, handle 4 argument function type _COUNT
23976 aliases, handle second_arg_count on second argument rather than last.
23978 2017-04-10 Jeff Law <law@redhat.com>
23980 PR tree-optimization/80374
23981 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23982 record anything if we can not convert integer_zero_node to the
23985 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23988 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23989 Enhance special handling given to the TARGET_P9_MINMAX option in
23990 relation to certain other options.
23992 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23994 PR tree-optimization/80153
23995 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23996 remove POINTER_PLUS_EXPR's base part directly, rather than through
23999 2017-04-10 Richard Biener <rguenther@suse.de>
24000 Bin Cheng <bin.cheng@arm.com>
24002 PR tree-optimization/80153
24003 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24004 the first element of pointer type aff_tree. Build result expr in
24006 (add_elt_to_tree): Convert to type unconditionally. Remove other
24007 fold_convert calls.
24008 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24009 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24011 2017-04-10 Richard Biener <rguenther@suse.de>
24013 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24016 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24018 PR rtl-optimization/70478
24019 * lra-constraints.c (curr_small_class_check): New.
24020 (update_and_check_small_class_inputs): New.
24021 (process_alt_operands): Update curr_small_class_check. Disfavor
24022 alternative insn memory operands. Check available regs for small
24025 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24028 * config/mips/mips.opt (-mvirt): Update description.
24029 * doc/invoke.texi (-mvirt): Likewise.
24031 2017-04-10 Richard Biener <rguenther@suse.de>
24033 PR middle-end/80362
24034 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24035 looking for NEGATE_EXPR in -A / -B to A / B folding.
24037 2017-04-10 Martin Liska <mliska@suse.cz>
24039 PR gcov-profile/80224
24040 * gcov.c (print_usage): Fix usage string.
24041 (get_gcov_intermediate_filename): Remove.
24042 (output_gcov_file): Use both for normal and intermediate format.
24043 (generate_results): Do not initialize special file for
24044 intermediate format.
24046 2017-04-10 Richard Biener <rguenther@suse.de>
24048 PR tree-optimization/80304
24049 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24052 2017-04-10 Nathan Sidwell <nathan@acm.org>
24055 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24056 (rs6000_init_builtins): Use it.
24058 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24060 * config/arm/arm.md (<mrc>): Add mode to SET source.
24061 (<mrrc>): Likewise.
24063 2017-04-10 Richard Biener <rguenther@suse.de>
24065 PR middle-end/80344
24066 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24068 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24071 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24072 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24073 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24074 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24075 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24076 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24077 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24078 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24079 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24080 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24081 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24082 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24083 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24084 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24085 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24086 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24087 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24088 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24089 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24090 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24091 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24092 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24093 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24095 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24097 PR rtl-optimization/70478
24098 * lra-constraints.c: Reverse the last patch.
24100 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24102 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24103 Add comment for WCHAR_T.
24105 2017-04-08 Martin Liska <mliska@suse.cz>
24108 2017-04-07 Martin Liska <mliska@suse.cz>
24111 * ipa-split.c (split_function): Add function part to a same comdat
24114 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24117 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24119 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24121 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24122 (vec_pairing): ...to this.
24123 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24124 (rs6000_sched_init): Adjust for name change.
24125 (struct rs6000_sched_context): Likewise.
24126 (rs6000_init_sched_context): Likewise.
24127 (rs6000_set_sched_context): Likewise.
24129 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24135 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24137 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24138 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24139 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24141 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24143 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24145 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24147 PR rtl-optimization/70703
24148 * ira-color.c (update_conflict_hard_regno_costs): Use
24149 int64_t instead of HOST_WIDE_INT.
24151 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24153 PR rtl-optimization/70478
24154 * lra-constraints.c (process_alt_operands): Disfavor alternative
24155 insn memory operands.
24157 2017-04-07 Jeff Law <law@redhat.com>
24159 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24160 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24162 2017-04-07 Martin Liska <mliska@suse.cz>
24165 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24166 Show error message instead of an ICE.
24168 2017-04-07 Martin Liska <mliska@suse.cz>
24171 * ipa-split.c (split_function): Add function part to a same comdat
24174 2017-04-07 Richard Biener <rguenther@suse.de>
24176 PR middle-end/80341
24177 * tree.c (get_unwidened): Also handle ! for_type case for
24179 * convert.c (do_narrow): Split out from ...
24180 (convert_to_integer_1): ... here. Do not pass final truncation
24181 type to get_unwidened for TRUNC_DIV_EXPR.
24183 2017-04-07 Richard Biener <rguenther@suse.de>
24185 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24187 (aff_combination_const): Adjust.
24188 (aff_combination_scale): Likewise.
24189 (aff_combination_add_elt): Likewise.
24190 (aff_combination_add_cst): Likewise.
24191 (aff_combination_convert): Likewise.
24192 (add_elt_to_tree): Likewise. Remove unused argument.
24193 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24195 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24197 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24199 * config/arm/arm.c (arm_default_short_enums): Use
24200 ARM_DEFAULT_SHORT_ENUMS.
24201 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24203 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24206 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24207 members with redundant out-of-class redeclaration.
24209 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24212 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24213 * config/i386/i386.md (*zero_extendsidi2):
24214 Add (?*x,*x) and (?*v,*v) alternatives.
24216 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24219 * config/i386/i386.c (ix86_expand_builtin)
24220 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24221 mode from insn data. Convert operands to insn operand mode.
24222 Copy operands that don't satisfy insn predicate to a register.
24224 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24226 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24229 2017-04-06 Richard Biener <rguenther@suse.de>
24231 PR tree-optimization/80334
24232 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24233 preserve alignment of accesses.
24235 2017-04-06 Richard Biener <rguenther@suse.de>
24237 PR tree-optimization/80262
24238 * tree-sra.c (build_ref_for_offset): Preserve address-space
24240 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24241 Drop useless address-space information on MEM_REF offsets.
24243 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24245 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24247 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24249 PR rtl-optimization/70703
24250 * ira-color.c (update_conflict_hard_regno_costs): Use
24251 HOST_WIDE_INT instead of long.
24253 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24256 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24257 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24259 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24260 for x86_64 target. Handle -m3dnowa option.
24262 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24264 PR rtl-optimization/70703
24265 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24266 (update_conflict_hard_regno_costs): Use long instead of unsigned
24267 arithmetic for cost calculation.
24269 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24270 Bernd Edlinger <bernd.edlinger@hotmail.de>
24273 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24276 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24279 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24280 ptr_mode with Pmode throughout.
24281 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24282 into probe_stack_range and use DImode.
24284 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24287 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24288 call_eh_return is true.
24290 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24292 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24293 Initialize last_match_fntype_index.
24295 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24298 * tree-nvr.c: Include internal-fn.h.
24299 (pass_return_slot::execute): Ignore internal calls without
24302 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24303 Richard Biener <rguenther@suse.de>
24306 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24307 captures used multiple times, except for the last use.
24308 * generic-match-head.c: Include gimplify.h.
24310 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24312 PR tree-optimization/79390
24313 * target.h (struct noce_if_info): Declare.
24314 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24315 * target.def (noce_conversion_profitable_p): New target hook.
24316 * ifcvt.h (struct noce_if_info): New type, moved from ...
24317 * ifcvt.c (struct noce_if_info): ... here.
24318 (noce_conversion_profitable_p): Renamed to ...
24319 (default_noce_conversion_profitable_p): ... this. No longer
24321 (noce_try_store_flag_constants, noce_try_addcc,
24322 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24323 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24324 instead of noce_conversion_profitable_p.
24325 * config/i386/i386.c: Include ifcvt.h.
24326 (ix86_option_override_internal): Don't override
24327 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24328 (ix86_noce_conversion_profitable_p): New function.
24329 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24330 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24331 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24332 * doc/tm.texi: Regenerated.
24334 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24336 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24339 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24342 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24343 instructions for small multiply cores.
24345 2017-04-04 Jeff Law <law@redhat.com>
24347 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24349 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24350 that are not set by the loop over the elements.
24352 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24355 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24356 int mode, convert_modes it to mode as unsigned, otherwise use
24357 lowpart_subreg to mode rather than SImode.
24358 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24359 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24360 Use DImode instead of SImode for the shift count operand.
24361 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24364 2017-04-04 Richard Biener <rguenther@suse.de>
24366 PR middle-end/80281
24367 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24368 arithmetic done for the negate or the plus. Simplify.
24369 (A - (-B) -> A + B): Likewise.
24370 * fold-const.c (split_tree): Make sure to not negate pointers.
24372 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24374 PR rtl-optimization/60818
24375 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24376 a compare of comparisons with the thing compared if this results
24377 in a different machine mode.
24379 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24381 * alias.c (base_alias_check): Fix typo in comment.
24382 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24383 * cgraphunit.c (symbol_table::compile): Likewise.
24384 * collect2.c (maybe_run_lto_and_relink): Likewise.
24385 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24386 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24387 * config/avr/avr.c (avr_map_op_t): Likewise.
24388 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24389 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24390 * config/epiphany/epiphany.md (movcc): Likewise.
24391 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24392 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24394 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24395 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24396 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24397 * config/sh/sh.c (sh_rtx_costs): Likewise.
24398 * fold-const.c (fold_truth_andor): Likewise.
24399 * genautomata.c (collapse_flag): Likewise.
24400 * gengtype.h (struct type::u::s): Likewise.
24401 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24402 * input.c (FORMAT_AMOUNT): Likewise.
24403 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24404 (known_aggs_to_agg_replacement_list): Likewise.
24405 * ipa-inline-analysis.c: Likewise.
24406 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24407 * ipa-polymorphic-call.c
24408 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24409 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24410 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24412 * modulo-sched.c (apply_reg_moves): Likewise.
24413 * omp-expand.c (build_omp_regions_1): Likewise.
24414 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24415 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24416 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24417 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24418 * value-prof.c: Likewise.
24419 * var-tracking.c (val_reset): Likewise.
24421 2017-04-03 Richard Biener <rguenther@suse.de>
24423 PR tree-optimization/80275
24424 * fold-const.c (split_address_to_core_and_offset): Handle
24427 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24429 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24430 descriptors is at least equal to that of functions.
24432 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24434 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24436 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24439 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24440 (mov<IMOD4:mode>): New expander.
24441 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24443 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24445 PR rtl-optimization/79405
24446 * fwprop.c (propagations_left): New variable.
24447 (forward_propagate_into): Decrement it.
24448 (fwprop_init): Initialize it.
24449 (fw_prop): If the variable has reached zero, stop propagating.
24450 (fwprop_addr): Ditto.
24452 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24455 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24456 a FUNCTION_DECL, pass it as decl instead of origin to
24459 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24461 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24464 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24467 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24468 TARGET_VSX_SMALL_INTEGER.
24470 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24472 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24473 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24475 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24477 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24478 extraction from odd-numbered MSA register.
24480 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24482 PR middle-end/80173
24483 * expmed.c (store_bit_field_1): Don't attempt to create
24484 a word subreg out of hard registers wider than word if they
24485 have HARD_REGNO_NREGS of 1 for their mode.
24487 PR middle-end/80163
24488 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24489 conversions to integer types wider than word and pointer.
24492 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24493 (rtx_equal_for_cselib_p): Pass 0 to it.
24494 * cselib.c (cselib_hasher::equal): Likewise.
24495 (rtx_equal_for_cselib_1): Add depth argument. If depth
24496 is 128, don't look up VALUE locs and punt. Increment
24497 depth in recursive calls when walking VALUE locs.
24499 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24501 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24502 (make_gcov_file_name): Use the canonical path name for generating
24504 (read_line): Fix handling of files with ascii null bytes.
24506 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24508 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24509 to initialise a vector register instead
24510 of using a const_int.
24512 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24514 PR translation/80189
24515 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24516 diagnostic messages.
24518 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24521 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24522 (dfp_diex_<mode>): Update mode of operand 1.
24523 * doc/extend.texi (dxex, dxexq): Document change to return type.
24524 (diex, diexq): Document change to argument type.
24526 2017-03-30 Martin Jambor <mjambor@suse.cz>
24529 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24530 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24531 it reflects the signature changes performed at the callee side.
24532 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24533 to cgraph_build_function_type_skip_args.
24534 (build_function_decl_skip_args): Adjust call to the above function.
24536 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24539 * config/i386/sse.md
24540 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24541 register as dest whenever it is a MEM not rtx_equal_p to the
24542 corresponding dup operand, and when forcing into reg move the
24543 reg into the memory afterwards.
24544 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24545 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24546 for the force_reg mode.
24547 (avx512vl_vextractf128<mode>): Use register as dest either
24548 always when a MEM, or when it is a MEM not rtx_equal_p to the
24549 corresponding dup operand, or even not when it is a CONST_VECTOR
24550 depending on the mode and lo vs. hi.
24551 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24553 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24554 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24555 Likewise. Require that operands[2] is even.
24556 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24557 Remove extraneous parens. Require that operands[2] is a multiple
24559 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24560 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24561 disallow memory then.
24563 2017-03-30 Richard Biener <rguenther@suse.de>
24565 PR tree-optimization/77498
24566 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24567 to non-constants over backedges.
24569 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24571 PR rtl-optimization/80233
24572 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24573 as last_combined_insn. Do not test for BARRIER_P separately.
24575 2017-03-29 Andreas Schwab <schwab@suse.de>
24578 * calls.c (prepare_call_address): Convert funexp to Pmode before
24579 copying to temp reg.
24581 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24583 PR tree-optimization/80158
24584 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24585 Handle possible future case of more than one alternate
24587 (replace_rhs_if_not_dup): Likewise.
24588 (replace_one_candidate): Likewise.
24590 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24592 PR rtl-optimization/80193
24593 * ira.c (ira): Do not check allocation for LRA.
24595 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24597 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24598 (nvptx_output_simt_exit): Declare.
24599 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24600 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24601 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24602 (nvptx_declare_function_name): ...here. Emit declaration of local
24603 memory space buffer for omp_simt_enter insn.
24604 (nvptx_output_unisimt_switch): New.
24605 (nvptx_output_softstack_switch): New.
24606 (nvptx_output_simt_enter): New.
24607 (nvptx_output_simt_exit): New.
24608 * config/nvptx/nvptx.h (struct machine_function): New fields
24609 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24610 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24611 (UNSPECV_SIMT_EXIT): Ditto.
24612 (omp_simt_enter_insn): New insn.
24613 (omp_simt_enter): New expansion.
24614 (omp_simt_exit): New insn.
24615 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24617 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24618 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24619 (expand_GOMP_SIMT_EXIT): New.
24620 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24621 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24622 (GOMP_SIMT_EXIT): Ditto.
24623 * target-insns.def (omp_simt_enter): New insn.
24624 (omp_simt_exit): Ditto.
24625 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24627 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24628 (lower_rec_input_clauses): Likewise.
24629 (lower_lastprivate_clauses): Handle SIMT privatization.
24631 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24632 (ompdevlow_adjust_simt_enter): New.
24633 (find_simtpriv_var_op): New.
24634 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24635 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24637 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24638 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24639 (copy_decl_for_dup_finish): Ditto.
24641 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24643 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24646 * config/i386/i386.c (ix86_option_override_internal): Always
24647 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24649 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24651 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24653 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24655 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24656 mark new edge's irreducible flag accordign to it.
24657 (vect_do_peeling): Check loop preheader edge's irreducible flag
24658 and pass it to function slpeel_add_loop_guard.
24660 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24662 PR tree-optimization/80218
24663 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24664 Update block frequencies and counts.
24666 2017-03-28 Richard Biener <rguenther@suse.de>
24668 PR tree-optimization/78644
24669 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24670 of a simplification result we may not use it at all.
24672 2017-03-28 Richard Biener <rguenther@suse.de>
24675 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24676 without arguments, generate default definition of a SSA name.
24678 2017-03-28 Richard Biener <rguenther@suse.de>
24680 PR middle-end/80222
24681 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24682 TYPE_REF_CAN_ALIAS_ALL references.
24683 * fold-const.c (fold_indirect_ref_1): Likewise.
24685 2017-03-28 Martin Liska <mliska@suse.cz>
24688 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24689 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24691 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24692 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24694 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24695 (EXTRA_SPECS): Define.
24696 (SUBTARGET_EXTRA_SPECS): Likewise.
24697 (SUBTARGET_CPP_SPEC): Likewise.
24698 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24699 SUBTARGET_EXTRA_SPECS.
24700 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24702 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24704 * config/arc/simdext.md (vst64_insn): Update pattern.
24705 (vld32wh_insn): Likewise.
24706 (vld32wl_insn): Likewise.
24707 (vld64_insn): Likewise.
24708 (vld32_insn): Likewise.
24710 2017-03-28 Marek Polacek <polacek@redhat.com>
24713 * fold-const.c (fold_comparison): Use protected_set_expr_location
24714 instead of SET_EXPR_LOCATION.
24716 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24718 * tree.c (add_expr): Avoid name lookup warning.
24720 2017-03-27 Jeff Law <law@redhat.com>
24722 PR tree-optimization/80216
24723 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24724 function name. Limit recursion depth.
24725 (record_temporary_equivalences): Corresponding changes.
24727 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24729 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24732 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24735 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24737 * cfgcleanup.c (reg_note_cfa_p): New array.
24738 (insns_have_identical_cfa_notes): New function.
24739 (old_insns_match_p): Don't cross-jump in between /f
24740 and non-/f instructions. If both i1 and i2 are frame related,
24741 verify all CFA notes, their order and content.
24743 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24746 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24747 HImode and SImode with zero extend to DImode to one insn.
24748 (bswap<mode>2_extenddi): Likewise.
24749 (bswapsi2_extenddi): Likewise.
24750 (bswaphi2_extendsi): Likewise.
24751 (bswaphi2): Combine bswap HImode and SImode into one insn.
24752 Separate memory insns from swapping register.
24753 (bswapsi2): Likewise.
24754 (bswap<mode>2): Likewise.
24755 (bswaphi2_internal): Delete, no longer used.
24756 (bswapsi2_internal): Likewise.
24757 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24758 store, and gpr<-gpr swap insns.
24759 (bswap<mode>2_store): Likewise.
24760 (bswaphi2_reg): Register only splitter, combine with the splitter.
24761 (bswaphi2 splitter): Likewise.
24762 (bswapsi2_reg): Likewise.
24763 (bswapsi2 splitter): Likewise.
24764 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24765 the insns into load, store, and register/register insns.
24766 (bswapdi2_ldbrx): Likewise.
24767 (bswapdi2_load): Likewise.
24768 (bswapdi2_store): Likewise.
24769 (bswapdi2_reg): Likewise.
24771 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24773 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24774 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24776 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24782 special handling for target option conflicts between dform
24783 options (-mpower9-dform, -mpower9-dform-vector,
24784 -mpower9-dform-scalar) and -mno-direct-move.
24786 2017-03-27 Richard Biener <rguenther@suse.de>
24788 PR tree-optimization/80181
24789 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24791 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24793 * config/arc/predicates.md (move_double_src_operand): Replace the
24794 call to move_double_src_operand with a call to address_operand.
24796 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24798 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24799 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24800 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24802 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24804 * config/arc/predicates.md (long_immediate_loadstore_operand):
24805 Consider scaled addresses cases.
24807 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24809 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24810 restored when in interrupt.
24811 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24812 doesn't have delay slot.
24814 2017-03-27 Richard Biener <rguenther@suse.de>
24817 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24818 inlined thunk clones.
24820 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24823 * asan.c (instrument_derefs): Copy over last operand from
24824 original COMPONENT_REF to the new COMPONENT_REF with
24825 DECL_BIT_FIELD_REPRESENTATIVE.
24826 * ubsan.c (instrument_object_size): Likewise.
24828 2017-03-27 Richard Biener <rguenther@suse.de>
24830 PR tree-optimization/80170
24831 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24832 sure DR/SCEV didnt fold in constants we do not see when looking
24833 at the reference base alignment.
24835 2017-03-27 Richard Biener <rguenther@suse.de>
24837 PR middle-end/80171
24838 * gimple-fold.c (fold_ctor_reference): Properly guard against
24839 NULL return value from canonicalize_constructor_val.
24841 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24844 * config/i386/i386.c (ix86_expand_builtin)
24845 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24846 flags reg setting and flags reg using instructions.
24847 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24848 clobbering instructions to zero extend op2.
24850 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24852 * doc/install.texi (Configuration) <--with-aix-soname>:
24853 Update link to AIX ld.
24855 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24857 PR rtl-optimization/80160
24858 PR rtl-optimization/80159
24859 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24860 reg_alternate_class into account.
24862 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24865 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24866 to consider in curr_insn_transform.
24868 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24870 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24871 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24872 and emit_mode_inner.
24874 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24876 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24877 argument to the overloaded builtin variants. Use the new flag to
24878 deprecate certain builtin variants.
24879 * config/s390/s390-builtin-types.def: Add new builtin types.
24880 * config/s390/s390-builtins.h: Support new flags field for
24881 overloaded builtins.
24882 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24883 (s390_macro_to_expand): Enable vector float data type.
24884 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24885 builtins by incrementing the __VEC__ version number.
24886 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24888 (s390_resolve_overloaded_builtin): Emit error messages depending
24889 on the builtin flags.
24890 * config/s390/s390.c (s390_expand_builtin): Support additional
24891 flags argument. Change error message to match the messages
24892 emitted in s390-c.c.
24893 * config/s390/s390.md: New UNSPEC_* constants.
24894 (op_type): Add new instruction types.
24895 * config/s390/vecintrin.h: Add new builtins and test data class
24897 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24898 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24899 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24900 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24901 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24902 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24904 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24905 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24906 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24907 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24909 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24910 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24911 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24912 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24914 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24915 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24916 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24917 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24918 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24919 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24920 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24922 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24923 ("vec_scatter_element<V_HW_4:mode>_DI")
24924 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24925 ("vec_fpint<mode>", "vflls")
24926 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24927 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24928 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24929 ("*vec_cmphe<mode>_cc"): ... these.
24931 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24932 mode constant instead of magic value.
24934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24936 * config/s390/s390.c (s390_expand_vec_compare): Support other
24937 vector floating point modes than just V2DF.
24938 (s390_expand_vcond): Likewise.
24939 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24940 (s390_cannot_change_mode_class): Prevent mode changes between TF
24941 and V1TF in vector registers.
24942 * config/s390/s390.md (DF, SF): New mode attributes.
24943 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24944 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24945 SFmode support for VRs.
24946 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24948 (VFT, VF_HW): New mode iterators.
24949 (vw, sdx): New mode attributes.
24950 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24951 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24952 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24953 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24954 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24955 also the new vector floating point modes. Renaming to ...
24957 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24958 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24959 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24960 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24961 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24962 ("vec_unordered<mode>"): ... these.
24964 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24965 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24966 ("*vec_extendv2df"): New insn definitions.
24968 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24970 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24971 ("mulditi3_2", "*muldi3_sign"): New patterns.
24972 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24973 rename the pattern definition.
24975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24977 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24979 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24981 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24983 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24984 instruction if possible.
24985 * config/s390/vector.md (vec_halfnumelts): New mode
24987 ("*vec_vllezlf<mode>"): New pattern.
24989 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24991 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24992 ("popcountv4si2", "popcountv2di2"): Rename to ...
24993 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24994 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24996 ("popcount<mode>2_vxe"): New pattern.
24998 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25000 * common/config/s390/s390-common.c (processor_flags_table): Add
25002 * config.gcc: Add arch12.
25003 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25004 Default to arch12 for unknown CPU model numbers.
25005 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25006 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25007 PROCESSOR_max sanity check.
25008 * config/s390/s390-opts.h (enum processor_type): Add
25010 * config/s390/s390.c (processor_table): Add arch12.
25011 (s390_expand_builtin): Add check for B_VXE flag.
25012 (s390_issue_rate): Add PROCESSOR_ARCH12.
25013 (s390_get_sched_attrmask): Likewise.
25014 (s390_get_unit_mask): Likewise.
25015 (s390_sched_score): Enable z13 scheduling for arch12.
25016 (s390_sched_reorder): Likewise.
25017 (s390_sched_variable_issue): Likewise.
25018 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25020 (s390_tune_attr): Use z13 scheduling also for arch12.
25021 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25022 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25023 (TARGET_VXE_P): New macros.
25024 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25025 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25026 * config/s390/s390.opt: Add arch12 as processor_type.
25028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25030 * config/s390/s390.md
25031 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25032 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25033 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25035 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25036 Rename expanders to ...
25038 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25039 ("fixuns_truncdddi2_emu"): ... these.
25041 ("fixuns_trunc<mode>si2_emu"): New expander.
25043 ("*fixuns_truncdfdi2_z13"): Rename to ...
25044 ("*fixuns_truncdfdi2_vx"): ... this.
25046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25048 * config/s390/2964.md: Remove the single element vector compare
25049 instructions which are no longer used.
25050 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25052 (s390_canonicalize_comparison): Remove handling of DFmode
25054 (s390_expand_vec_compare_scalar): Remove function.
25055 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25056 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25058 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25062 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25064 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25065 will anyway by matched by mov<mode>_64dfp.
25067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25069 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25070 vlef/vstef. Add missing operand to vleif.
25072 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25074 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25075 pair for all vector types with 64 bit elements.
25076 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25077 * config/s390/vector.md (V_HW_64): ... here.
25078 (V_128_NOSINGLE): New mode iterator.
25079 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25080 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25081 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25082 ("*vec_load_pairv2di"): Change to ...
25083 ("*vec_load_pair<mode>"): ... this one.
25085 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25087 * config/s390/constraints.md: Add comments.
25088 (jKK): Reject element sizes > 8 bytes.
25089 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25091 * config/s390/s390.md: Add the s_operand checks formerly in
25092 s390_split_ok_p to various splitters where they are still
25094 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25095 for 128 bit vectors. Plus two splitters.
25097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25099 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25102 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25105 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25106 error if the boundary argument is not constant.
25108 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25110 PR rtl-optimization/80112
25111 * loop-doloop.c (doloop_condition_get): Don't check condition
25112 if cmp isn't SET with IF_THEN_ELSE src.
25114 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25116 PR tree-optimization/80158
25117 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25118 replacing a candidate statement, also replace it for the
25119 candidate's alternate interpretation.
25120 (replace_rhs_if_not_dup): Likewise.
25121 (replace_one_candidate): Likewise.
25123 2017-03-24 Richard Biener <rguenther@suse.de>
25125 PR tree-optimization/80167
25126 * graphite-isl-ast-to-gimple.c
25127 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25129 (translate_isl_ast_to_gimple::get_rename): Likewise.
25131 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25133 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25134 handling of certain combinations of target options, including the
25135 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25136 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25138 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25141 * config/arm/arm.md (*load_multiple): Add reload_completed to
25142 matching condition.
25144 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25145 Richard Biener <rguenth@suse.de>
25147 PR tree-optimization/79908
25148 PR tree-optimization/80136
25149 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25150 been cast away, gimplify_and_add suffices.
25152 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25154 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25156 2017-03-23 Richard Biener <rguenther@suse.de>
25158 PR tree-optimization/80032
25159 * gimplify.c (gimple_push_cleanup): Forced unconditional
25160 cleanups still have to go to the conditional_cleanups
25163 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25165 PR tree-optimization/80072
25166 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25168 (next_operand_entry_id): Change type to unsigned int.
25169 (sort_by_operand_rank): Make sure to return the right return value
25170 even if unsigned fields are bigger than INT_MAX.
25171 (struct oecount): Change cnt and id type to unsigned int.
25172 (oecount_hasher::equal): Formatting fix.
25173 (oecount_cmp): Make sure to return the right return value
25174 even if unsigned fields are bigger than INT_MAX.
25175 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25178 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25179 TREE_READONLY on result if writing it more than once.
25182 * doc/invoke.texi (-fsanitize=thread): Document that with
25183 -fnon-call-exceptions atomics are not able to throw
25187 * tsan.c: Include tree-eh.h.
25188 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25189 maybe_clean_or_replace_eh_stmt where needed.
25190 (instrument_memory_accesses): Add cfg_changed argument.
25191 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25192 if it returned true.
25193 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25195 PR rtl-optimization/63191
25196 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25197 wrapper function, moved the whole old content into ...
25198 (ix86_delegitimize_address_1): ... this. New inline function.
25199 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25200 true as last argument instead of ix86_delegitimize_address.
25202 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25204 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25205 cortexa57_branch_cost.
25207 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25209 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25211 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25214 * doc/md.texi (Constraints): Document wA constraint.
25215 * config/rs6000/constraints.md (wA): New.
25216 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25217 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25218 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25219 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25221 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25224 * gimplify.c (is_oacc_declared): New function.
25225 (oacc_default_clause): Use it to set default flags for acc declared
25226 variables inside parallel regions.
25227 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25228 declared variables.
25229 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25230 declare attribute to any decl as necessary.
25232 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25235 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25236 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25237 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25238 (arm_arch_lpae): This.
25239 * config/arm/arm.c (arm_arch7ve): Rename into ...
25240 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25241 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25244 2017-03-22 Martin Liska <mliska@suse.cz>
25247 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25248 error message instead of an ICE.
25250 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25252 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25254 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25256 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25258 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25261 2017-03-21 Martin Sebor <msebor@redhat.com>
25263 * doc/extend.texi: Use "cannot" instead of "can't."
25264 * doc/hostconfig.texi: Same.
25265 * doc/install.texi: Same.
25266 * doc/invoke.texi: Same.
25267 * doc/loop.texi: Same.
25268 * doc/md.texi: Same.
25269 * doc/objc.texi: Same.
25270 * doc/rtl.texi: Same.
25271 * doc/tm.texi: Same.
25272 * doc/tm.texi.in: Same.
25273 * doc/trouble.texi: Same.
25275 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25278 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25279 (collect_checksum_attributes): Set it.
25280 (die_checksum_ordered): Use it.
25282 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25284 PR tree-optimization/79908
25285 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25286 change: For a VA_ARG whose LHS has been cast away, use
25287 force_gimple_operand to construct the side effects.
25289 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25291 PR translation/80001
25292 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25293 more amenable to translation.
25294 (oacc_loop_auto_partitions): Likewise.
25296 2017-03-21 Marek Polacek <polacek@redhat.com>
25297 Martin Sebor <msebor@redhat.com>
25299 PR tree-optimization/80109
25300 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25301 on INTEGRAL_TYPE_P.
25303 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25304 Segher Boessenkool <segher@kernel.crashing.org>
25307 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25308 check reg_used_between_p between insn and one of succ or succ2
25309 depending on if succ is artificial insn not inserted into insn
25312 2017-03-21 Martin Liska <mliska@suse.cz>
25314 PR gcov-profile/80081
25315 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25316 * doc/gcc.texi: Include gcov-dump stuff.
25317 * doc/gcov-dump.texi: New file.
25319 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25321 PR rtl-optimization/79150
25322 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25323 conditional jump, if the jump is the last insn of the loop.
25325 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25326 Richard Biener <rguenth@suse.de>
25328 PR tree-optimization/79908
25329 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25330 been cast away, use force_gimple_operand to construct the side
25333 2017-03-21 Martin Liska <mliska@suse.cz>
25335 PR libfortran/79956
25336 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25339 2017-03-21 Brad Spengler <spender@grsecurity.net>
25342 * plugin.c (htab_hash_plugin): New function.
25343 (add_new_plugin): Use it and adjust.
25344 (parse_plugin_arg_opt): Adjust.
25345 (init_one_plugin): Likewise.
25347 2017-03-21 Richard Biener <rguenther@suse.de>
25349 PR tree-optimization/80032
25350 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25351 if set force the cleanup to happen unconditionally.
25352 (gimplify_target_expr): Push inserted clobbers with force_uncond
25353 to avoid them being removed by control-dependent DCE.
25355 2017-03-21 Richard Biener <rguenther@suse.de>
25357 PR tree-optimization/80122
25358 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25359 va_arg_pack_len when the inlined call stmt requires pack
25361 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25363 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25366 * tsan.c (instrument_builtin_call): If the memory model argument
25367 is not a constant, assume it is valid.
25370 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25373 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25375 PR rtl-optimization/79910
25376 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25377 if its dest is used by an insn before I2 (other than the combined
25378 insns themselves, which are properly handled already).
25380 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25383 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25385 * combine.c (record_used_regs): New static function.
25386 (try_combine): Handle situations where there is an additional
25387 instruction between I2 and I3 which needs to have a LOG_LINK
25391 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25393 * combine.c (try_combine): Delete redundant i1 test. Call
25394 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25396 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25399 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25400 alternatives 13/14.
25402 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25404 PR tree-optimization/80054
25405 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25406 the optimization if a PHI or any of its arguments is not dominated
25407 by the candidate's basis. Use gphi* rather than gimple* as
25409 (replace_profitable_candidates): Clean up a gimple* variable that
25410 should be a gphi* variable.
25412 2017-03-20 Martin Sebor <msebor@redhat.com>
25415 * doc/extend.texi (attribute constructor): Document present limitation.
25417 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25420 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25421 __POWER9_VECTOR__ #ifdef control, change template definition to
25422 use Power9-specific built-in function.
25423 (vec_any_eq): Likewise.
25424 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25425 to control outcomes from this test.
25426 (vector_ae_<mode>p): For VEC_F modes, likewise.
25428 2017-03-20 Ian Lance Taylor <iant@google.com>
25430 * config/i386/i386.c (ix86_function_regparm): Save an extra
25431 register for -fsplit-stack with DECL_STATIC_CHAIN.
25433 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25436 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25437 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25439 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25441 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25443 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25446 2017-03-20 Marek Polacek <polacek@redhat.com>
25449 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25451 2017-03-20 Richard Biener <rguenther@suse.de>
25453 PR tree-optimization/80113
25454 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25455 allocate extra SSA name for PHI def.
25456 (add_close_phis_to_outer_loops): Likewise.
25457 (add_close_phis_to_merge_points): Likewise.
25458 (copy_loop_close_phi_args): Likewise.
25459 (copy_cond_phi_nodes): Likewise.
25461 2017-03-20 Martin Liska <mliska@suse.cz>
25463 PR middle-end/79753
25464 * tree-chkp.c (chkp_build_returned_bound): Do not build
25465 returned bounds for a LHS that's not a BOUNDED_P type.
25467 2017-03-20 Martin Liska <mliska@suse.cz>
25471 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25472 COMPLEX_CST and VECTOR_CST.
25474 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25477 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25478 target operand. A new splitter adds the clobber statement in case
25479 the target operand is dead anyway.
25481 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25483 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25484 to age-old versions of binutils and glibc.
25486 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25488 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25490 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25492 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25494 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25496 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25497 requirement for binutils 2.13.
25499 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25501 * combine.c (try_combine): Delete redundant i1 test. Call
25502 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25504 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25506 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25507 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25509 <riscv64-*-elf>: Re-arrange section
25510 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25511 <riscv32-*-linux>: Likewise.
25512 <riscv64-*-elf>: Likewise
25513 <riscv64-*-linux>: Likewise.
25515 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25518 * aarch64.opt(verbose-cost-dump): Fix typo.
25520 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25523 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25524 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25526 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25528 * reload.c (find_reloads): When reloading a nonoffsettable address,
25529 use RELOAD_OTHER for it and its address reloads.
25531 PR rtl-optimization/79910
25532 * combine.c (record_used_regs): New static function.
25533 (try_combine): Handle situations where there is an additional
25534 instruction between I2 and I3 which needs to have a LOG_LINK
25537 2017-03-17 Jeff Law <law@redhat.com>
25539 PR tree-optimization/71437
25540 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25541 conditional in the hash table first.
25542 (vrp_dom_walker::before_dom_children): Extract condition from
25543 ASSERT_EXPR. Record condition, its inverion and any implied
25544 conditions as well.
25546 2017-03-17 Marek Polacek <polacek@redhat.com>
25547 Markus Trippelsdorf <markus@trippelsdorf.de>
25549 PR tree-optimization/80079
25550 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25553 2017-03-17 Richard Biener <rguenther@suse.de>
25555 PR middle-end/80075
25556 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25557 Properly verify the LHS before the RHS possibly claims to be
25559 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25562 2017-03-17 Martin Jambor <mjambor@suse.cz>
25564 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25565 (List of -O2 options): Likewise.
25566 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25569 2017-03-17 Tom de Vries <tom@codesourcery.com>
25571 * gcov-dump.c (print_usage): Print bug_report_url.
25573 2017-03-17 Richard Biener <rguenther@suse.de>
25575 PR middle-end/80050
25576 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25577 (parser::peek): Likewise.
25579 2017-03-17 Richard Biener <rguenther@suse.de>
25581 PR tree-optimization/80048
25582 * sese.c (free_sese_info): Properly release rename_map and
25583 copied_bb_map elements.
25585 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25587 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25588 Add linked-list forward and backlinks. Insert on
25589 construction, remove on destruction.
25590 (class pass_store_merging): Add m_stores_head field.
25591 (pass_store_merging::terminate_and_process_all_chains):
25592 Iterate over m_stores_head list.
25593 (pass_store_merging::terminate_all_aliasing_chains):
25595 (pass_store_merging::execute): Check for debug stmts first.
25596 Push new chains onto the m_stores_head stack.
25598 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25601 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25602 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25603 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25605 2017-03-16 Jeff Law <law@redhat.com>
25607 PR tree-optimization/71437
25608 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25609 member function. Implementation moved into after_dom_children
25610 member function and into the threader's thread_outgoing_edges
25612 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25613 some code into new thread_outgoing_edges.
25614 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25615 definition. Simplify marker handling (do it here). Assume we always
25616 have the available expression and the const/copies tables.
25617 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25619 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25620 * tree-vrp.c (equiv_stack): No longer file scoped.
25621 (vrp_dom_walker): New class.
25622 (vrp_dom_walker::before_dom_children): New member function.
25623 (vrp_dom_walker::after_dom_children): Likewise.
25624 (identify_jump_threads): Setup domwalker. Use it rather than
25625 walking edges in a random order by hand. Simplify setup/finalization.
25626 (finalize_jump_threads): Remove.
25627 (vrp_finalize): Do not call identify_jump_threads here.
25628 (execute_vrp): Do it here instead and call thread_through_all_blocks
25631 PR tree-optimization/71437
25632 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25634 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25636 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25637 (dom_opt_dom_walker::thread_across_edge): Remove
25638 handle_dominating_asserts argument. All callers changed.
25639 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25640 changes. Remove calls to lhs_of_dominating_assert. Other
25641 uses of handle_dominating_asserts turn into unconditional code
25642 (simplify_control_stmt_condition_1): Likewise.
25643 (simplify_control_stmt_condition): Likewise.
25644 (thread_through_normal_block, thread_across_edge): Likewise.
25645 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25646 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25647 object if it is not an SSA_NAME.
25648 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25649 before calling into the VRP specific simplifiers.
25650 (identify_jump_threads): Remove handle_dominating_asserts
25653 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25656 * tree-diagnostic.c (default_tree_printer): No longer static.
25657 * tree-diagnostic.h (default_tree_printer): New prototype.
25659 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25661 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25662 Change ins into fmov.
25664 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25666 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25667 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25668 Use h_con constraint for operand 1.
25669 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25670 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25672 2017-03-15 Jeff Law <law@redhat.com>
25674 PR tree-optimization/71437
25675 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25676 (record_temporary_equivalences): Use it.
25678 PR tree-optimization/71437
25679 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25680 tree-ssa-scopedtables.
25681 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25682 (record_conditions, record_cond, vuse_eq): Likewise.
25683 (record_edge_info): Adjust to API tweak of record_conditions.
25684 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25685 (record_temporary_equivalences, optimize_stmt): Likewise.
25686 (eliminate_redundant_computations): Likewise.
25687 (record_equivalences_from_stmt): Likewise.
25688 * tree-ssa-scopedtables.c: Include options.h and params.h.
25689 (vuse_eq): New function, moved from tree-ssa-dom.c
25690 (build_and_record_new_cond): Likewise.
25691 (record_conditions): Likewise. Accept vector of conditions rather
25692 than edge_equivalence structure for first argument.
25693 for the first argument.
25694 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25695 from tree-ssa-dom.c.
25696 (avail_exprs_stack::record_cond): Likewise.
25697 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25698 from tree-ssa-dom.c.
25699 (avail_exprs_stack): Add new member functions lookup_avail_expr
25701 (record_conditions): Declare.
25703 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25706 * lra-constraints.c (process_alt_operands): Increase reject for
25707 reloading an input/output operand.
25709 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25712 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25713 insns to convert from signed/unsigned char/short to IEEE 128-bit
25715 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25717 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25720 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25721 subreg of inner mode for values already in registers.
25723 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25725 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25726 iteration reg is used after the loop.
25728 2017-03-14 Martin Sebor <msebor@redhat.com>
25730 PR tree-optimization/79800
25731 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25732 precision in negative-positive range.
25733 (format_floating): Call non-const overload with adjusted precision.
25735 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25738 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25741 2017-03-14 Martin Sebor <msebor@redhat.com>
25743 PR middle-end/80020
25744 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25745 * builtins.def (aligned_alloc): Use it.
25748 * Makefile.in (GTFILES): Add calls.c.
25749 * calls.c: Include "gt-calls.h".
25751 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25753 PR rtl-optimization/79728
25754 * regs.h (struct target_regs): New field
25755 x_contains_allocatable_regs_of_mode.
25756 (contains_allocatable_regs_of_mode): New macro.
25757 * reginfo.c (init_reg_sets_1): Initialize it, and change
25758 contains_reg_of_mode so it includes global regs as well.
25759 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25760 rather than contains_regs_of_mode.
25762 2017-03-14 Martin Liska <mliska@suse.cz>
25764 * doc/invoke.texi: Document options that can't be combined with
25765 -fcheck-pointer-bounds.
25767 2017-03-14 Martin Liska <mliska@suse.cz>
25769 PR middle-end/79831
25770 * doc/invoke.texi (-Wchkp): Document the option.
25772 2017-03-14 Martin Liska <mliska@suse.cz>
25774 * Makefile.in: Install gcov-dump.
25776 2017-03-14 Martin Liska <mliska@suse.cz>
25778 * multiple_target.c (expand_target_clones): Bail out for
25779 an invalid attribute.
25781 2017-03-14 Richard Biener <rguenther@suse.de>
25783 * alias.c (struct alias_set_entry): Pack properly.
25784 * cfgloop.h (struct loop): Likewise.
25785 * cse.c (struct set): Likewise.
25786 * ipa-utils.c (struct searchc_env): Likewise.
25787 * loop-invariant.c (struct invariant): Likewise.
25788 * lra-remat.c (struct cand): Likewise.
25789 * recog.c (struct change_t): Likewise.
25790 * rtl.h (struct address_info): Likewise.
25791 * symbol-summary.h (function_summary): Likewise.
25792 * tree-loop-distribution.c (struct partition): Likewise.
25793 * tree-object-size.c (struct object_size_info): Likewise.
25794 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25795 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25796 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25797 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25798 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25799 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25800 (struct _stmt_vec_info): Likewise.
25802 2017-03-14 Martin Liska <mliska@suse.cz>
25805 * multiple_target.c (create_dispatcher_calls): Check that
25806 a target can create a function dispatcher.
25808 2017-03-14 Martin Liska <mliska@suse.cz>
25811 * multiple_target.c (expand_target_clones): Drop local.local
25812 flag for default implementation.
25814 2017-03-14 Richard Biener <rguenther@suse.de>
25816 PR tree-optimization/80030
25817 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25819 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25821 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25822 gcc_fallthrough() instead of __attribute__((fallthrough));
25824 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25826 * doc/gcc.texi: Remove "up" link to (DIR).
25827 * doc/gccint.texi: Ditto.
25829 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25831 * doc/install.texi (Specific) <avr>: Remove reference to
25834 2017-03-13 Jeff Law <law@redhat.com>
25836 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25837 attribute rather than comments.
25839 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25840 match_scratch operand is highest.
25842 2017-03-13 Martin Liska <mliska@suse.cz>
25844 PR middle-end/78339
25845 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25846 is a CHKP clone, use original declaration.
25848 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25850 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25851 (arc_conditional_register_usage): Use a different allocation order
25852 when optimizing for size.
25853 * common/config/arc/arc-common.c (arc_option_optimization_table):
25854 Section anchors default on when optimizing for size.
25856 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25858 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25860 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25862 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25863 * config/arc/arc.md (cpu_facility): Add cd variant.
25864 (*movqi_insn): Add code density variant.
25865 (*movhi_insn): Likewise.
25866 (*movqi_insn): Likewise.
25867 (*addsi3_mixed): Likewise.
25868 (subsi3_insn): Likewise.
25870 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25872 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25874 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25876 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25877 expressions with MINUS and UNARY ops.
25879 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25882 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25884 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25885 between vec_select and vector argument.
25886 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25887 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25888 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25889 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25890 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25891 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25893 2017-03-13 Richard Biener <rguenther@suse.de>
25896 * params.def (vect-max-peeling-for-alignment): Fix typo.
25898 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25900 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25901 issue that only occurred with binutils below 2.18.
25903 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25905 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25906 refer to binutils 2.11/2.12 minimum.
25908 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25910 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25911 ftp.kernel.org and simplify binutils requirement.
25913 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25915 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25917 (Optimize Options): Ditto. Also remove redundancy.
25919 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25921 PR translation/79848
25922 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25924 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25925 to G_ to avoid double translation.
25927 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25929 PR translation/79923
25930 * auto-profile.c (get_combined_location): Convert leading
25931 character of diagnostics to lower case and remove trailing period.
25932 (read_profile): Likewise for various diagnostics.
25933 * config/arm/arm.c (arm_option_override): Remove trailing period
25934 from various diagnostics.
25935 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25936 (msp430_expand_delay_cycles): Likewise.
25938 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25941 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25942 full command-line argument, rather than just "str".
25943 (aarch64_validate_march): Likewise.
25944 (aarch64_validate_mtune): Likewise.
25946 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25948 PR rtl-optimization/78911
25949 * lra-assigns.c (must_not_spill_p): New function.
25950 (spill_for): Use it.
25952 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25954 PR tree-optimization/79981
25955 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25956 ATOMIC_COMPARE_EXCHANGE ifn result.
25957 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25958 IFN_ATOMIC_COMPARE_EXCHANGE.
25960 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25963 * opts.c (parse_sanitizer_options): Add missing question mark to
25964 "did you mean" message.
25966 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25968 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25970 (VMULEUH_UNS): Likewise.
25971 (VMULOUB_UNS): Likewise.
25972 (VMULOUH_UNS): Likewise.
25973 * config/rs6000/rs6000.c (builtin_function_type): Remove
25974 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25976 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25979 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25980 x with result of extra_parsing_for_operand_code_0.
25981 (function_reader::extra_parsing_for_operand_code_0): Convert
25982 return type from void to rtx, returning x. When reading
25983 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25984 larger size containing struct block_symbol.
25986 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25989 -mfloat128-hardware without -m64.
25991 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25994 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25995 entries to the case statement that marks unsigned arguments to
25996 overloaded functions.
25998 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26001 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26003 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26006 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26007 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26009 2017-03-10 Martin Liska <mliska@suse.cz>
26013 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26014 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26017 2017-03-10 Olivier Hainque <hainque@adacore.com>
26019 * tree-switch-conversion (array_value_type): Start by resetting
26020 candidate type to it's main variant.
26022 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26024 PR rtl-optimization/79909
26025 * combine.c (try_combine): Use simplify_replace_rtx on individual
26026 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26027 of the whole CALL_INSN_FUNCTION_USAGE.
26029 PR tree-optimization/79972
26030 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26031 get_range_info on SSA_NAMEs. Formatting fixes.
26033 2017-03-10 Richard Biener <rguenther@suse.de>
26034 Jakub Jelinek <jakub@redhat.com>
26036 PR tree-optimization/77975
26037 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26038 edge to be constant.
26039 (get_val_for): For constant x return it. Formatting fix.
26040 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26041 would use the same bases as the current one.
26043 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26045 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26046 instead of vec_select for V1TImode.
26047 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26049 (VSX_LE_128): Add V1TI to this mode iterator.
26050 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26051 (*vsx_le_perm_store_<mode>): Likewise.
26052 (pre-reload splitter for VSX stores): Likewise.
26053 (post-reload splitter for VSX stores): Likewise.
26054 (*vsx_xxpermdi2_le_<mode>): Likewise.
26055 (*vsx_lxvd2x2_le_<mode>): Likewise.
26056 (*vsx_stxvd2x2_le_<mode>): Likewise.
26058 2017-03-09 Michael Eager <eager@eagercon.com>
26060 Correct failures with --enable-checking=yes,rtl.
26062 * config/microblaze/microblaze.c (microblaze_expand_shift):
26063 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26064 test for const0_rtx.
26065 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26066 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26068 2017-03-09 Richard Biener <rguenther@suse.de>
26070 PR tree-optimization/79977
26071 * graphite-scop-detection.c (scop_detection::merge_sese):
26072 Handle the case of extra exits to blocks dominating the entry.
26074 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26076 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26079 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26081 PR rtl-optimization/79949
26082 * lra-constraints.c (process_alt_operands): Check memory when
26083 trying to predict a cycle. Print about the overall increase.
26085 2017-03-09 Richard Biener <rguenther@suse.de>
26087 PR middle-end/79971
26088 * gimple-expr.c (useless_type_conversion_p): Preserve
26089 TYPE_SATURATING for fixed-point types.
26091 2017-03-09 Richard Biener <rguenther@suse.de>
26094 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26095 alignment of BLKmode params.
26097 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26100 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26101 (VALL_NO_V2Q): Likewise.
26103 * config/aarch64/aarch64-simd.md
26104 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26106 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26107 VALL_NO_V2Q mode iterator.
26108 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26110 2017-03-09 Martin Liska <mliska@suse.cz>
26112 PR tree-optimization/79631
26113 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26114 tree_int_cst_sign_bit just for INTEGER constants.
26116 2017-03-09 Martin Liska <mliska@suse.cz>
26120 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26123 2017-03-09 Marek Polacek <polacek@redhat.com>
26126 * tree.c (inchash::add_expr): Handle TREE_VEC.
26128 2017-03-09 Martin Liska <mliska@suse.cz>
26131 (chkp_narrow_size_and_offset): New function.
26132 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26133 (void chkp_parse_bit_field_ref): New function.
26134 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26135 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26137 2017-03-09 Martin Liska <mliska@suse.cz>
26140 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26141 (chkp_find_bounds_1): Remove gcc_unreachable.
26143 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26146 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26147 BUILT_IN_SYNC*, determine the access type from the size suffix and
26148 always build a MEM_REF with that type. Handle forgotten
26149 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26152 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26153 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26154 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26155 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26156 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26157 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26158 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26159 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26160 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26161 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26162 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26163 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26164 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26165 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26166 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26167 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26168 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26169 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26170 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26171 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26172 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26173 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26174 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26175 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26176 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26177 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26178 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26179 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26180 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26181 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26182 definitions outside of __OPTIMIZE__ guarded section.
26185 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26186 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26187 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26188 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26191 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26193 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26194 ("vfenez<mode>"): Add missing constraints.
26196 2017-03-08 Martin Sebor <msebor@redhat.com>
26199 * config/nds32/nds32.c (nds32_option_override):
26200 Fix misspelled diagnostic.
26202 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26205 * gimplify.c (gimplify_omp_for): Replace index var in outer
26206 taskloop statement with an artificial variable and add
26207 OMP_CLAUSE_PRIVATE clause for it.
26209 2017-03-08 Richard Biener <rguenther@suse.de>
26211 PR tree-optimization/79955
26212 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26213 for accesses that are completely outside of the variable.
26215 2017-03-08 Andrew Haley <aph@redhat.com>
26217 PR tree-optimization/79943
26218 * tree-ssa-loop-split.c (compute_new_first_bound): When
26219 calculating the new upper bound, (END-BEG) should be added, not
26222 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26224 * config/avr/avr.md (setmemhi): Make sure match_dup
26225 operand number comes before match_scratch.
26227 2017-03-08 Richard Biener <rguenther@suse.de>
26229 PR tree-optimization/79920
26230 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26231 with ncopies == 1 to ...
26232 (vect_transform_slp_perm_load): ... here. Properly compute
26233 all element loads by iterating VF times over the group. Do
26234 not handle ncopies (computed in a broken way) in
26235 vect_create_mask_and_perm.
26237 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26240 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26241 is a uniform vector, use uniform_vector_p return value instead of
26242 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26244 2017-03-07 Marek Polacek <polacek@redhat.com>
26246 PR middle-end/79809
26247 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26248 (alloca_call_type): Likewise.
26250 2017-03-07 Martin Liska <mliska@suse.cz>
26252 * gcov.c (process_args): Put comment to correct location.
26254 2017-03-07 Martin Liska <mliska@suse.cz>
26256 PR middle-end/68270
26257 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26258 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26259 (chkp_narrow_bounds_for_field): Likewise.
26260 (chkp_parse_array_and_component_ref): Pass one more argument to
26263 2017-03-07 Richard Biener <rguenther@suse.de>
26265 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26268 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26270 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26271 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26273 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26276 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26277 to end of description.
26278 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26280 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26282 PR rtl-optimization/79901
26283 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26285 (*avx512f_<code><mode>3<mask_name>): ... this.
26286 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26287 iterator instead of VI8_AVX2_AVX512BW.
26289 PR rtl-optimization/79901
26290 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26291 min/max expander, expand it using expand_vec_cond_expr.
26294 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26297 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26300 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26301 to void * for PCH reasons.
26302 * dwarf2out.c (output_loc_operands, output_die): Cast
26303 v.val_vec.array to unsigned char *.
26305 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26308 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26311 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26313 PR rtl-optimization/79571
26314 * lra-constraints.c (process_alt_operands): Calculate static
26315 reject and subtract it from overall when only addresses will be
26318 2017-03-06 Julia Koval <julia.koval@intel.com>
26321 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26322 incoming stack boundary to 128 for 64-bit targets.
26324 2017-03-06 Richard Biener <rguenther@suse.de>
26326 PR tree-optimization/79894
26327 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26328 to NULL after folding it.
26330 2017-03-06 Richard Biener <rguenther@suse.de>
26332 PR tree-optimization/79824
26333 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26334 check disabling peeling for gaps.
26336 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26338 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26339 attributes): Document gettimeofday.
26341 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26343 * config/s390/s390.c (s390_option_override_internal): Set
26344 PARAM_MIN_VECT_LOOP_BOUND
26346 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26348 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26349 * config/s390/s390.md: Likewise.
26351 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26354 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26355 (<avx2_avx512>_perm<mode>): Rename to ...
26356 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26358 (<avx512>_perm<mode>_mask): Rename to ...
26359 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26361 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26362 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26363 instead of VI8F_256_512.
26364 (avx512f_perm<mode>): New define_expand.
26365 (avx512f_perm<mode>_mask): Likewise.
26366 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26367 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26369 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26371 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26372 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26374 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26376 2017-03-06 Martin Liska <mliska@suse.cz>
26379 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26380 when having a SSA NAME w/o VAR_DECL assigned to it.
26382 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26384 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26385 msa_dpsub_<su>_d): Fix MODE for vec_select.
26387 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26389 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26391 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26393 2017-03-06 Richard Biener <rguenther@suse.de>
26395 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26396 * plugin.c (register_plugin_info): Likewise.
26397 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26399 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26401 * config/i386/sse.md (sse_storehps, sse_storelps,
26402 avx_<castmode><avxsizesuffix>_<castmode>,
26403 avx512f_<castmode><avxsizesuffix>_<castmode>,
26404 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26405 in condition that at least one operand is not a MEM.
26407 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26409 PR middle-end/79805
26410 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26411 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26413 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26414 gimple_call_nothrow_p flag based on whether original builtin can throw.
26415 If it can, emit following stmts on the fallthrough edge.
26416 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26417 don't create new bb if inserting just debug stmts on the edge, try to
26418 insert them on the fallthru bb or just reset debug stmts.
26420 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26423 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26424 restore recog_data (including the operand rtxes inside it) around
26425 the call to get_insn_template.
26427 2017-03-03 Martin Sebor <msebor@redhat.com>
26429 PR tree-optimization/79699
26430 * context.c (context::~context): Free MPFR caches to avoid
26431 a memory leak on program exit.
26433 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26435 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26436 Use wide_int::ulow () instead of .elt (0).
26438 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26440 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26441 (*pushxf): Limit oF constraint to 32bit targets and add oC
26442 constraint for 64bit targets.
26443 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26444 (*pushdf): Change rmF constraint to rmC.
26446 2017-03-03 Martin Liska <mliska@suse.cz>
26448 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26449 Remove unused variable.
26451 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26454 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26455 is a memory operand, increase num_memory.
26456 (ix86_expand_args_builtin): Likewise.
26458 2017-03-03 Jan Hubicka <jh@suse.cz>
26461 * ipa-devirt.c (maybe_record_node): Properly handle
26462 __cxa_pure_virtual visibility.
26464 2017-03-03 Martin Liska <mliska@suse.cz>
26466 PR tree-optimization/79803
26467 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26469 (pass_loop_prefetch::execute): Disabled optimization if an
26470 assumption about L1 cache size is not met.
26472 2017-03-03 Martin Liska <mliska@suse.cz>
26474 PR rtl-optimization/79574
26475 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26476 (hash_scan_set): Likewise.
26477 (dump_hash_table): Likewise.
26478 (hoist_code): Likewise.
26480 2017-03-03 Richard Biener <rguenther@suse.de>
26482 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26483 in place of uhigh ().
26484 (fixed_convert_from_real): Likewise.
26486 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26489 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26491 2017-03-03 Richard Biener <rguenther@suse.de>
26493 PR middle-end/79818
26494 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26495 TYPE_OVERFLOW_UNDEFINED check.
26497 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26499 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26501 (vector_ae_<mode>_p): Likewise.
26502 (vector_nez_<mode>_p): Likewise.
26503 (vector_ne_v2di_p): Likewise.
26504 (vector_ae_v2di_p): Likewise.
26505 (vector_ne_<mode>_p): Likewise.
26506 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26508 (vsx_tsqrt<mode>2_fe): Likewise.
26510 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26513 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26515 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26517 PR rtl-optimization/79780
26518 * cprop.c (one_cprop_pass): When second and further conditional trap
26519 in a single basic block is turned into an unconditional trap, turn it
26520 into a deleted note to avoid RTL verification failures.
26522 2017-03-02 Richard Biener <rguenther@suse.de>
26524 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26526 2017-03-02 Richard Biener <rguenther@suse.de>
26528 PR tree-optimization/79345
26530 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26531 param and abort the walk, returning -1 if it is hit.
26532 (walk_aliased_vdefs): Take a limit param and pass it on.
26533 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26534 defaulting to 0 and return a signed int.
26535 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26536 (check_defs): New helper.
26537 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26538 about uninitialized memory.
26539 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26540 bogus uninitialized warning.
26541 (fixed_convert_from_real): Likewise.
26543 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26545 PR tree-optimization/66768
26546 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26547 iv_use if base object can't be determined.
26549 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26551 PR tree-optimization/79345
26552 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26553 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26554 (get_pattern_stats): Initialize it.
26555 * genemit.c (gen_expand): Verify match_scratch numbers come after
26556 match_operand/match_dup numbers.
26557 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26558 match_scratch numbers.
26559 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26561 * config/s390/s390.md (trunctdsd2): Likewise.
26563 2017-03-02 Richard Biener <rguenther@suse.de>
26565 * wide-int.h (wide_int_storage::operator=): Implement in terms
26568 2017-03-02 Richard Biener <rguenther@suse.de>
26570 PR tree-optimization/79777
26571 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26572 the to insert expression to sth existing.
26574 2017-03-01 Martin Sebor <msebor@redhat.com>
26576 PR middle-end/79692
26577 * gimple-ssa-sprintf.c
26578 (directive::known_width_and_precision): New function.
26579 (format_integer): Use it.
26580 (get_mpfr_format_length): Consider the full range of precision
26581 when computing %g output with the # flag. Set the likely byte
26582 count to 3 rather than 1 when precision is indeterminate.
26583 (format_floating): Correct the lower bound of precision.
26585 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26587 * doc/invoke.texi: Document default code model for 64-bit Linux.
26589 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26592 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26593 udiv rather than div since input pattern is unsigned.
26595 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26597 * config/i386/i386.c (print_reg): Warn for values of
26598 unsupported size in integer register.
26600 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26603 * config/rs6000/predicates.md (current_file_function_operand): Do
26604 not allow self calls to be local if the function is replaceable.
26606 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26609 * config/rs6000/altivec.h (vec_ctz and others): Change the
26610 preprocessor macro that controls conditional compilation from
26611 _ARCH_PWR9 to __POWER9_VECTOR__.
26612 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26613 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26614 control (instead of _ARCH_PWR9 control) so that template
26615 definition uses power9-specific function.
26616 (vec_any_eq): Likewise.
26617 (vec_all_ne): Change macro definition to use a power9-specific
26618 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26619 _ARCH_PWR9 control).
26620 (vec_any_eq) Likewise.
26621 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26622 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26623 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26624 support for xvcmpnedp instruction.
26625 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26626 macro expansion so that Power9 implementation of vec_all_ne does
26627 not use the AltiVec predicate framework.
26628 (VCMPNEH_P): Likewise.
26629 (VCMPNEW_P): Likewise.
26630 (VCMPNED_P): Likewise.
26631 (VCMPNEFP_P): Likewise.
26632 (VCMPNEDP_P): Likewise.
26633 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26634 implementation of vec_any_eq to not use AltiVec predicate
26636 (VCMPAEH_P): Likewise.
26637 (VCMPAEW_P): Likewise.
26638 (VCMPAED_P): Likewise.
26639 (VCMPAEFP_P): Likewise.
26640 (VCMPAEDP_P): Likewise.
26641 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26642 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26643 not use the AltiVec predicate framework.
26644 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26645 of vec_any_eq to not use AltiVec predicate framework.
26646 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26647 support for predefined __POWER9_VECTOR__ macro to indicate that
26648 Power9 instruction selection is enabled.
26649 (altivec_overloaded_builtins): Remove extraneous
26650 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26651 function argument types RS6000_BTI_bool_V16QI and
26652 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26653 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26654 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26655 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26656 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26657 Power9 for implementations of vec_cmpne. Change the signature for
26658 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26659 (representing vec_all_ne) to remove the previously described first
26660 argument of type RS6000_BTI_INTSI, as this was an artifact of
26661 reliance on the AltiVec predicate framework, which is no longer
26662 used in the implementation of these functions. Add
26663 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26664 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26665 since, unlike the AltiVec predicate framework implementation, we
26666 do not share function descriptors between vec_alle and vec_anyeq.
26667 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26668 set of modes that receive special treatment even when
26669 TARGET_P9_VECTOR is true. The special treatment emits code that
26670 does not depend on Power9 instructions.
26671 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26672 define_expand to not rely on AltiVec predicate framework.
26673 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26675 (vector_ne_v2di_p): Change this define_expand to not rely on
26676 AltiVec predicate framework.
26677 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26679 (vector_ne_<mode>_p): Change this define_expand to not rely on
26680 AltiVec predicate framework.
26681 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26683 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26684 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26685 define_insn pattern.
26686 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26687 define_insn pattern because the xvcmpne<VSs>. instruction is not
26689 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26690 instruction is not supported.
26692 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26694 * config/nvptx/nvptx.c: Include intl.h.
26696 2017-03-01 Martin Jambor <mjambor@suse.cz>
26699 * ipa-prop.h (ipa_bits): Removed field known.
26700 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26701 to pointers. Adjusted their comments to warn about their sharing.
26702 (ipcp_transformation_summary): Change bits to a vector of pointers.
26703 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26704 (ipa_get_ipa_bits_for_value): Declare.
26705 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26706 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26707 (ipa_bits_hash_table): Likewise.
26708 (ipa_vr_ggc_hash_traits): Likewise.
26709 (ipa_vr_hash_table): Likewise.
26710 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26711 being pointers and vr_known being removed.
26712 (ipa_set_jf_unknown): Likewise.
26713 (ipa_get_ipa_bits_for_value): New function.
26714 (ipa_set_jfunc_bits): Likewise.
26715 (ipa_get_value_range): New overloaded functions.
26716 (ipa_set_jfunc_vr): Likewise.
26717 (ipa_compute_jump_functions_for_edge): Use the above functions to
26718 construct bits and vr parts of jump functions.
26719 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26720 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26722 (ipcp_grow_transformations_if_necessary): Also allocate
26723 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26725 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26726 them. Fix too long lines.
26727 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26728 vr_known being removed.
26729 (ipa_read_jump_function): Use new setter functions to construct bits
26730 and vr parts of jump functions or set them to NULL.
26731 (write_ipcp_transformation_info): Adjust for bits being pointers.
26732 (read_ipcp_transformation_info): Likewise.
26733 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26735 Include gt-ipa-prop.h.
26736 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26738 (ipcp_store_bits_results): Likewise.
26739 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26740 Do not write to existing jump functions but use a temporary instead.
26742 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26745 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26746 attempt to use its first operand as BIT_FIELD_REF base.
26748 2017-03-01 Richard Biener <rguenther@suse.de>
26750 PR middle-end/79721
26751 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26752 interpolating formula in wrapping arithmetic.
26753 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26755 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26757 PR tree-optimization/79734
26758 * tree-vect-generic.c (expand_vector_condition): Optimize
26759 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26760 Handle VEC_COND_EXPR where comparison has different inner width from
26761 type's inner width.
26763 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26765 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26766 markup, and similar issues. Remove @opindex entries for things
26767 that aren't options. Add missing -mmpy-option entries.
26769 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26771 PR tree-optimization/79737
26772 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26773 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26774 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26775 instead of byte_size. Formatting fix.
26776 (shift_bytes_in_array_right): Formatting fix.
26778 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26781 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26782 condition on optimize for the leaf function test.
26784 2017-02-28 Martin Liska <mliska@suse.cz>
26787 * read-rtl-function.c (function_reader::handle_unknown_directive):
26788 Bail out when one uses -flto.
26790 2017-02-28 Martin Liska <mliska@suse.cz>
26792 * common.opt: Replace space with tabular for options of <number>
26794 * config/i386/i386.opt: Show <number> value for
26795 -mlarge-data-threshold.
26796 * opts.c (print_filtered_help): Do not display number in hexadecimal
26799 2017-02-28 Martin Liska <mliska@suse.cz>
26801 * common.opt: Fix --help=option -Q for options which are of
26804 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26806 * config/i386/i386.c (print_reg): Error out for values
26807 of 8-bit size in invalid integer register.
26809 2017-02-28 Martin Sebor <msebor@redhat.com>
26811 PR tree-optimization/79691
26812 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26814 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26817 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26818 gcc_unreachable with output_operand_lossage.
26820 2017-02-28 Richard Biener <rguenther@suse.de>
26822 PR tree-optimization/79740
26823 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26825 (visit_nary_op): Insert the nary into the hashtable if we
26826 pattern-matched sth.
26827 * tree-ssa-pre.c (eliminate_insert): Robustify.
26829 2017-02-28 Richard Biener <rguenther@suse.de>
26831 PR middle-end/79731
26832 * fold-const.c (decode_field_reference): Reject out-of-bound
26835 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26837 * config/i386/i386.c: Include intl.h.
26838 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26839 instead of just cond ? "..." : "...".
26840 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26841 * coverage.c (read_counts_file): Likewise.
26842 * omp-offload.c: Include intl.h.
26843 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26844 of just cond ? "..." : "...".
26845 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26846 of just cond ? "..." : "...".
26848 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26851 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26853 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26854 'tune for' CPU name.
26855 * config/arm/arm-cpu-data.h: Regenerated.
26857 2017-02-28 Richard Biener <rguenther@suse.de>
26859 PR tree-optimization/79732
26860 * tree-inline.c (expand_call_inline): Do not shadow var.
26862 2017-02-28 Richard Biener <rguenther@suse.de>
26864 PR tree-optimization/79723
26865 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26866 address-space properly.
26868 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26870 * doc/optinfo.texi (Optimization groups): Fix option used for
26872 * doc/invoke.texi (-fopt-info): Document "omp".
26873 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26874 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26875 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26876 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26877 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26879 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26881 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26882 OPTGROUP_OMP with "omp".
26884 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26887 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26888 for arithmetic shift of unsigned V2DI.
26890 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26892 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26893 arc/linux.h headers.
26894 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26895 (LINK_SPEC): Likewise.
26896 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26897 (EXTRA_SPECS): Likewise.
26898 (STARTFILE_SPEC): Likewise.
26899 (ENDFILE_SPEC): Likewise.
26900 (LIB_SPEC): Likewise.
26901 (TARGET_SDATA_DEFAULT): Likewise.
26902 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26903 (MULTILIB_DEFAULTS): Likewise.
26904 (DWARF2_UNWIND_INFO): Likewise.
26905 * config/arc/big.h: New file.
26906 * config/arc/elf.h: Likewise.
26907 * config/arc/linux.h: Likewise.
26908 * config/arc/t-uClibc: Remove.
26910 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26912 PR tree-optimization/77536
26913 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26914 (tree_transform_and_unroll_loop): Use above function to compute the
26915 estimated niter of unrolled loop and use it when scaling profile.
26916 Also use count info rather than frequency if it's non-zero.
26917 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26918 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26919 (vect_transform_loop): Call above function.
26921 2017-02-27 Richard Biener <rguenther@suse.de>
26923 PR tree-optimization/45397
26924 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26925 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26926 (visit_nary_op): Add pattern matching for CSEing sign-changed
26927 or truncated operations with wider ones.
26929 2017-02-27 Richard Biener <rguenther@suse.de>
26931 PR tree-optimization/79690
26932 * tree-vect-stmts.c (vectorizable_store): Use vector type
26933 built from the DR with address-space.
26935 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26937 * doc/invoke.texi (Optimize Options): Refine the description
26938 of asan-use-after-return.
26940 2017-02-25 Alan Modra <amodra@gmail.com>
26942 PR rtl-optimization/79584
26943 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26944 base, not ad->base_term, the reg within base. Remove assertion
26945 that ad->base == ad->base_term. Replace gen_int_mode using
26946 bogus mode with const0_rtx.
26948 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26950 PR middle-end/79396
26951 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26952 FMA_EXPR like tcc_binary or tcc_unary.
26954 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26957 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26959 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26960 (output_loc_operands): Handle DW_OP_call_ref and
26961 DW_OP_GNU_variable_value.
26962 (struct variable_value_struct): New type.
26963 (struct variable_value_hasher): Likewise.
26964 (variable_value_hash): New variable.
26965 (string_types): Remove.
26966 (copy_loc_descr): New function.
26967 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26968 (prepend_loc_descr_to_each): New function.
26969 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26970 instead of add_loc_descr_to_each if the first argument is single
26971 location list and the second has multiple.
26972 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26973 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26974 when looking for variable value which doesn't have other location info.
26975 (loc_list_from_tree): Formatting fix.
26976 (gen_array_type_die): Simplify DW_AT_string_length handling.
26977 (adjust_string_types): Remove.
26978 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26979 string_types. Call resolve_variable_values.
26980 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26981 (resolve_addr_in_expr): Likewise. Add A argument.
26982 (copy_deref_exprloc): Remove deref argument. Adjust for the
26983 original expression being DW_OP_GNU_variable_value with optionally
26984 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26985 optionally after it.
26986 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26987 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26988 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26989 (variable_value_hasher::hash, variable_value_hasher::equal): New
26991 (resolve_variable_value_in_expr, resolve_variable_value,
26992 resolve_variable_values, note_variable_value_in_expr,
26993 note_variable_value): New functions.
26994 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26997 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27000 * opts.h (handle_generated_option): Add GENERATED_P argument.
27001 * opts-common.c (handle_option): Adjust function comment.
27002 (handle_generated_option): Add GENERATED_P argument, pass it to
27004 (control_warning_option): Pass false to handle_generated_option
27006 * opts.c (maybe_default_option): Pass true to handle_generated_option
27008 * optc-gen.awk: Likewise.
27010 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27012 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27013 a REG, look at the REG it is a SUBREG of.
27014 (splitter for cmpeqsi_t): Ditto.
27016 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27018 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27019 the special USEs with the pattern of the insn, not the insn itself.
27021 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27024 * doc/invoke.texi: Document -mload-store-pairs.
27026 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27027 Sandra Loosemore <sandra@codesourcery.com>
27029 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27030 argument isn't a CONST_INT.
27031 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27032 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27033 (nios2_validate_compare): Bypass alternate compare logic if *op2
27034 is not a CONST_INT.
27035 (ldstwm_operation_p): Return false if first_base is not a REG or
27036 if first_offset is not a CONST_INT.
27038 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27040 * config/cris/cris.md: Use correct operand in a define_peephole2.
27042 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27044 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27046 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27048 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27049 this_insn if it is an INSN or JUMP_INSN.
27050 (force_offsettable): Look at base, not at addr.
27051 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27052 on things that aren't necessarily CONST_INTs.
27054 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27056 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27057 -mfpmath=sse is the default also for x86-32 targets with SSE2
27058 instruction set when @option{-ffast-math} is enabled
27060 2017-02-24 Jeff Law <law@redhat.com>
27062 PR rtl-optimizatoin/79286
27063 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27066 2017-02-24 Richard Biener <rguenther@suse.de>
27068 PR tree-optimization/79389
27069 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27072 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27074 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27075 function comment to reflect reality.
27076 (loop_exits_before_overflow): Fix typo in function description.
27078 2017-02-24 Richard Biener <rguenther@suse.de>
27080 PR tree-optimization/79389
27081 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27082 properly that a threading opportunity exists. Detect conditional
27083 copy/constant propagation opportunities.
27085 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27087 * config/visium/visium.md (type): Add trap.
27088 (b): New mode attribute.
27089 (*btst): Rename into...
27090 (*btst<mode>): ...this and adjust.
27091 (*cbranchsi4_btst_insn): Rename into...
27092 (*cbranch<mode>4_btst_insn): ...this and adjust.
27093 (trap): New define_insn.
27095 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27097 PR tree-optimization/79389
27098 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27099 (noce_reversed_cond_code): New function.
27100 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27101 reversed_comparison_code. Formatting fix.
27102 (noce_try_store_flag): Test rev_cond != NULL in addition to
27103 reversed_comparison_code.
27104 (noce_try_store_flag_constants): Likewise.
27105 (noce_try_store_flag_mask): Likewise.
27106 (noce_try_addcc): Use rev_cond if non-NULL instead of
27107 reversed_comparison_code.
27108 (noce_try_cmove_arith): Likewise. Formatting fixes.
27109 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27110 (noce_find_if_block): Initialize rev_cond.
27111 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27112 instead of false as last argument never attempt to reverse it
27115 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27117 PR tree-optimization/79663
27118 * tree-predcom.c (combine_chains): Process refs in reverse order
27119 only for ZERO length chains, and add explaining comment.
27121 2017-02-23 Jeff Law <law@redhat.com>
27123 PR tree-optimization/79578
27124 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27125 in call to operand_equal_p.
27127 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27130 * config/i386/cpuid.h: Fix another undefined behavior.
27132 2017-02-23 Richard Biener <rguenther@suse.de>
27134 PR tree-optimization/79683
27135 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27136 vector types for data-refs.
27138 2017-02-23 Martin Liska <mliska@suse.cz>
27140 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27142 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27144 PR middle-end/79665
27145 * internal-fn.c (get_range_pos_neg): Moved to ...
27146 * tree.c (get_range_pos_neg): ... here. No longer static.
27147 * tree.h (get_range_pos_neg): New prototype.
27148 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27149 are known to be in between 0 and signed maximum inclusive, try to
27150 expand both unsigned and signed divmod and use the cheaper one from
27153 2017-02-22 Jeff Law <law@redhat.com>
27155 PR tree-optimization/79578
27156 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27157 to compare base operands.
27159 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27162 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27163 gpc_reg_operand instead of fpr_reg_operand.
27165 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27167 * config/mips/mips.c (mips_return_in_memory): Force FP
27168 vector types to be returned in memory for o32 ABI.
27170 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27172 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27173 instead of DW_TAG_member for static data member declarations and don't
27174 set no_linkage_name for static inline data members.
27175 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27178 2017-02-22 Martin Liska <mliska@suse.cz>
27180 * doc/invoke.texi: Replace inequality signs with square brackets
27183 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27185 PR tree-optimization/68644
27186 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27188 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27191 * lra-constraints.c (simplify_operand_subreg): Handle
27192 WORD_REGISTER_OPERATIONS targets.
27194 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27197 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27198 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27199 elimination by swapping fld*.
27201 2017-02-22 Richard Biener <rguenther@suse.de>
27203 PR tree-optimization/79673
27204 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27205 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27206 irrelevant address-space qualifiers and avoiding a
27207 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27209 2017-02-22 Richard Biener <rguenther@suse.de>
27211 PR tree-optimization/79666
27212 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27213 to not symbolically negate if that may introduce undefined
27216 2017-02-22 Martin Liska <mliska@suse.cz>
27219 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27220 * data-streamer-out.c (streamer_write_gcov_count_stream):
27222 * value-prof.c (stream_out_histogram_value): Make assert more
27223 precise based on type of counter.
27225 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27228 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27229 Use nonimmediate_operand instead of memory_operand for operand 1.
27230 (float-extend standard_x87sse_constant_load splitter): Ditto.
27232 2017-02-21 Jeff Law <law@redhat.com>
27234 PR tree-optimization/79621
27235 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27236 blocks with edges to themselves.
27238 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27241 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27242 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27243 Use gimple_call_builtin_p.
27246 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27247 on temporarily removed DEBUG_INSNs.
27249 PR tree-optimization/79649
27250 * tree-loop-distribution.c (classify_partition): Give up on
27251 non-generic address space loads/stores.
27253 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27255 * doc/loop.texi (Loop manipulation): Remove nonexistent
27256 tree_ssa_loop_version from the documentation.
27257 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27259 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27262 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27263 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27264 * config/rs6000/rs6000.c: Include except.h.
27265 (rs6000_expand_split_stack_prologue): Call
27266 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27268 2017-02-21 Martin Jambor <mjambor@suse.cz>
27271 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27272 have been analyzed.
27274 2017-02-21 Martin Jambor <mjambor@suse.cz>
27276 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27277 for backward compatibility only.
27278 * doc/invoke.texi (Option Summary): Remove all references to
27279 -fipa-cp-alignment.
27281 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27285 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27287 * lra-constraints.c (curr_insn_transform): Handle
27288 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27290 2017-02-21 Martin Liska <mliska@suse.cz>
27292 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27294 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27296 PR translation/79638
27297 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27299 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27302 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27303 (arm_function_ok_for_sibcall): Return false for an indirect call by
27304 descriptor if all the argument registers are used.
27305 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27306 alignment of the function.
27308 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27310 PR tree-optimization/61441
27311 * simplify-rtx.c (simplify_const_unary_operation): For
27312 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27313 the sNaN unmodified.
27315 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27317 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27318 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27319 instead of SYSTEM_HEADER_DIR.
27321 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27322 Martin Liška <mliska@suse.cz>
27324 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27325 Fix typos and grammar, use active voice, and clarify.
27327 2017-02-20 Marek Polacek <polacek@redhat.com>
27329 PR middle-end/79537
27330 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27333 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27335 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27338 * config/i386/i386.c (ix86_expand_builtin): Handle
27339 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27340 ix86_builtins_isa[fcode].isa as a requirement of those
27341 flags and any other flag in the bitmask.
27342 (ix86_init_mmx_sse_builtins): Use 0 instead of
27343 ~OPTION_MASK_ISA_64BIT as mask.
27344 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27345 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27346 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27347 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27349 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27352 * lra-constraints.c (split_reg): Check requested split mode
27353 is supported by the register.
27355 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27357 * lra-constraints.c (simplify_operand_subreg): Remove early
27360 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27363 * lra-constraints.c (curr_insn_transform): Tighten condition
27364 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27366 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27369 * lra-constraints.c (curr_insn_transform): Handle
27370 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27372 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27375 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27377 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27379 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27382 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27385 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27387 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27390 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27392 * final.c (last_columnnum, override_columnnum): New variables.
27393 (final_start_function): Set last_columnnum, pass it to begin_prologue
27394 hook and pass 0 to dwarf2out_begin_prologue.
27395 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27396 to source_line debug hook.
27397 (notice_source_line): Compute last_columnnum and for debug_column_info
27398 return true on column changes.
27399 * debug.h (struct gcc_debug_hooks): Add column argument to
27400 source_line and begin_prologue hooks.
27401 (debug_nothing_int_charstar_int_bool): Remove prototype.
27402 (debug_nothing_int_int_charstar,
27403 debug_nothing_int_int_charstar_int_bool): New prototypes.
27404 (dwarf2out_begin_prologue): Add column argument.
27405 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27406 begin_prologue hooks.
27407 (debug_nothing_int_charstar_int_bool): Remove.
27408 (debug_nothing_int_int_charstar,
27409 debug_nothing_int_int_charstar_int_bool): New functions.
27410 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27411 through to dwarf2out_source_line.
27412 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27413 (dwarf2out_source_line): Add column argument, emit it if requested.
27414 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27416 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27417 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27418 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27419 through to dwarf2out_begin_prologue.
27420 (vmsdbgout_source_line): Add column argument, pass it through to
27421 dwarf2out_source_line.
27422 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27423 dbxout_source_line caller.
27424 (dbxout_source_line): Add column argument.
27426 * common.opt (gno-column-info, gcolumn-info): New options.
27427 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27428 (check_die): Also test for multiple DW_AT_decl_column attributes.
27429 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27430 DW_AT_decl_column if requested.
27431 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27433 (gen_variable_die): Likewise.
27434 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27435 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27438 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27439 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27440 (ix86_handle_option): Handle OPT_m3dnowa.
27441 * doc/invoke.texi (-m3dnowa): Document.
27442 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27443 -m3dnowa instead of -m3dnow -march=athlon.
27446 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27447 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27449 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27452 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27453 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27454 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27455 generator for vsx_xxpermdi_<mode>_be.
27456 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27457 force big-endian semantics.
27458 (vsx_xxpermdi_<mode>_be): New define_expand with same
27459 implementation as previous version of vsx_xxpermdi_<mode>.
27461 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27463 PR tree-optimization/79327
27464 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27465 variable, its initialization and use.
27467 2017-02-17 Julia Koval <julia.koval@intel.com>
27469 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27470 (OPTION_MASK_ISA_PKU_UNSET): New.
27471 (ix86_handle_option): Handle -mrdpid.
27472 * config/i386/cpuid.h (bit_RDPID): New.
27473 * config/i386/driver-i386.c (host_detect_local_cpu):
27474 Detect RDPID feature.
27475 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27476 * config/i386/i386-c.c (ix86_target_macros_internal):
27478 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27479 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27480 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27481 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27482 * config/i386/i386.md (define_insn "rdpid"): New.
27483 * config/i386/i386.opt Add -mrdpid.
27484 * config/i386/immintrin.h (_rdpid_u32): New.
27486 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27488 PR rtl-optimization/79541
27489 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27490 instead of transforming it into USE.
27492 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27494 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27495 If HONOR_SNANS (SFmode) force the input to a register.
27496 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27497 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27498 an frsp or similar insn.
27500 2017-02-17 Martin Liska <mliska@suse.cz>
27502 PR rtl-optimization/79577
27503 * params.def (selsched-max-sched-times): Increase minimum to 1.
27505 2017-02-17 Martin Liska <mliska@suse.cz>
27507 PR rtl-optimization/79574
27508 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27510 2017-02-17 Martin Liska <mliska@suse.cz>
27512 PR tree-optimization/79529
27513 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27514 ssa_defined_default_def_p to handle cases which are implicitly
27516 * tree-ssa.c (ssa_defined_default_def_p): New function.
27517 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27518 which are implicitly defined.
27519 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27521 2017-02-17 Richard Biener <rguenther@suse.de>
27523 PR middle-end/79576
27524 * params.def (max-ssa-name-query-depth): Limit to 10.
27526 2017-02-17 Richard Biener <rguenther@suse.de>
27528 PR tree-optimization/79552
27529 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27532 2017-02-17 Richard Biener <rguenther@suse.de>
27535 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27537 2017-02-17 Marek Polacek <polacek@redhat.com>
27539 PR middle-end/79536
27540 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27541 (fold_negate_expr): New wrapper.
27543 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27545 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27546 Correct terminology and de-emphasize pre-standard behavior.
27548 2017-02-16 Alan Modra <amodra@gmail.com>
27550 PR rtl-optimization/79286
27551 * ira.c (def_dominates_uses): New function.
27552 (update_equiv_regs): Don't create an equivalence for insns that
27553 may trap where the register def does not dominate the use.
27555 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27557 PR rtl-optimization/78127
27558 * lra.c (lra): Call lra_eliminate before finish the loop after
27561 2017-02-16 Richard Biener <rguenther@suse.de>
27563 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27565 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27566 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27567 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27568 (isl_val_int_from_wi): New function.
27569 (extract_affine_gmp): Rename to ...
27570 (extract_affine_wi): ... this, take a widest_int.
27571 (extract_affine_int): Just wrap extract_affine_wi.
27572 (add_param_constraints): Use isl_val_int_from_wi.
27573 (add_loop_constraints): Likewise, and extract_affine_wi.
27575 2017-02-15 Jeff Law <law@redhat.com>
27577 PR middle-end/79521
27578 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27579 ira_init_register_move_cost_if_necessary.
27581 2017-02-15 Martin Sebor <msebor@redhat.com>
27583 PR middle-end/32003
27584 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27585 removed in a prior commit.
27587 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27589 PR tree-optimization/79347
27590 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27591 counters during peeling.
27593 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27595 * Makefile.in (site.exp): Remove "set ISLVER".
27597 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27600 * real.c (real_from_integer): Call real_convert even for decimal.
27602 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27605 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27607 2017-02-14 Andrew Pinski <apinski@cavium.com>
27609 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27610 cores and change the partno/implementer to be correct.
27611 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27612 the 'B" as the implementer.
27613 * config/aarch64/aarch64-tune.md: Regenerate.
27615 2017-02-14 Carl Love <cel@us.ibm.com>
27617 * config/rs6000/rs6000.c: Add case statement entry to make the
27618 xvcvuxdsp built-in argument unsigned.
27619 * config/rs6000/vsx.md: Fix the source and return operand types so they
27620 match the instruction definitions from the ISA document. Fix typo
27621 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27624 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27627 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27628 member early_clobber_alts.
27629 * lra-lives.c (reg_early_clobber_p): New.
27630 (process_bb_lives): Use it.
27631 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27632 (debug_operand_data): Initialize early_clobber_alts.
27633 (setup_operand_alternative): Set up early_clobber_alts.
27634 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27635 alternatives to new_insn_reg.
27636 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27638 (lra_update_insn_regno_info): Pass the new arg.
27640 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27642 PR middle-end/79505
27643 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27644 (new_oacc_loop_raw): Don't clear already cleared fields.
27647 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27648 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27649 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27651 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27654 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27656 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27659 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27660 the extra instruction to the right place to store 128-bit constant
27663 2017-02-14 Martin Sebor <msebor@redhat.com>
27665 PR middle-end/79448
27666 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27667 warning for strings of unknown length.
27669 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27671 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27673 2017-02-14 Jeff Law <law@redhat.com>
27676 * ira-costs.c (scan_one_insn): Initialize register move costs
27677 for pseudos seen in USE/CLOBBER insns.
27679 PR tree-optimization/79095
27680 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27681 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27682 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27683 if the operands are known to be not equal, then the resulting range
27685 (intersect_ranges): If the new range is ~[0,0] and the old range is
27686 wide, then prefer ~[0,0].
27687 * tree-vrp.c (overflow_comparison_p_1): New function.
27688 (overflow_comparison_p): New function.
27689 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27690 if NAME is used in an overflow test.
27691 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27692 overflow check that can be expressed as an equality test, then adjust
27693 ops to be that equality test.
27695 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27697 * config/s390/s390-builtin-types.def: Remove flags argument.
27698 * config/s390/s390.c (s390_init_builtins): Likewise.
27700 2017-02-14 Martin Liska <mliska@suse.cz>
27702 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27703 vector. Fix trailing white spaces.
27705 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27707 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27710 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27712 PR rtl-optimization/68664
27713 * config/arm/arm.c (arm_sched_can_speculate_insn):
27714 New function. Declare prototype.
27715 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27717 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27719 PR rtl-optimization/68664
27720 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27722 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27724 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27726 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27727 max skip bytes for function, loop and jump.
27729 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27731 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27732 ABS_EXPR for gimple dump.
27734 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27737 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27739 PR tree-optimization/79408
27740 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27741 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27742 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27743 also if rhs1 is INTEGER_CST.
27745 2017-02-14 Richard Biener <rguenther@suse.de>
27747 PR middle-end/79432
27748 * tree-into-ssa.c (insert_phi_nodes): When the function can
27749 have abnormal edges rewrite SSA names with broken use-def
27750 dominance out of SSA and register them for PHI insertion.
27752 2017-02-13 Martin Sebor <msebor@redhat.com>
27754 PR middle-end/79496
27755 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27756 clearing info.nowrite flag when snprintf size argument is a range.
27758 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27760 * cprop.c (cprop_jump): Add missing space in string literal.
27761 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27762 (get_constraint_for_component_ref): Likewise.
27763 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27764 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27765 * lra-constraints.c (process_alt_operands): Likewise.
27766 * ipa-inline.c (inline_small_functions): Likewise.
27767 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27768 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27769 * trans-mem.c (diagnose_tm_1_op): Likewise.
27770 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27771 (grid_parallel_clauses_gridifiable): Likewise.
27773 * config/nvptx/mkoffload.c (process): Add space in between
27776 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27777 "MOD4_SSE_REGS" and "ALL_REGS".
27779 * spellcheck.c (test_data): Add , in between "foo" and "food".
27781 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27784 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27785 boundary crossing check and subsequent code generation agree.
27787 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27789 * config/aarch64/aarch64.c (has_memory_op): Delete.
27790 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27793 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27795 PR rtl-optimization/79388
27796 PR rtl-optimization/79450
27797 * combine.c (distribute_notes): When removing TEM_INSN for which
27798 corresponding dest has last value recorded, invalidate that last
27801 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27803 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27804 of explicit '@'. Add missing assembly comment marker on branch costs
27807 2017-02-13 Nathan Sidwell <nathan@acm.org>
27809 * gengtype-lex.l (<in_struct>): Add '/'.
27811 2017-02-13 Martin Liska <mliska@suse.cz>
27814 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27816 2017-02-13 Richard Biener <rguenther@suse.de>
27818 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27820 * configure: Re-generate.
27821 * config.in: Likewise.
27822 * graphite-dependences.c: Simplify as if
27823 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27824 * graphite-isl-ast-to-gimple.c: Likewise.
27825 * graphite-optimize-isl.c: Likewise.
27826 * graphite-poly.c: Likewise.
27827 * graphite-sese-to-poly.c: Likewise.
27828 * graphite.h: Likewise.
27829 * toplev.c: Include isl/version.h and use isl_version () for
27830 printing the ISL version.
27831 * doc/install.texi: Update ISL requirement.
27833 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27835 * doc/standards.texi (Standards): Update reference to
27838 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27840 * doc/extend.texi (Named Address Spaces): sourceware.org now
27842 * doc/install.texi (Binaries): Ditto.
27845 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27847 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27848 standard terminology "stringize"/"stringizing" throughout.
27849 * doc/cppinternals.texi: Likewise.
27851 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27853 * doc/extend.texi: Fix some spelling mistakes and typos.
27854 * doc/invoke.texi: Likewise.
27856 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27859 * params.def (inline-min-speedup) Change from 10 to 8.
27861 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27863 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27866 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27869 * ipa-inline-analysis.c (get_minimal_bb): New function.
27870 (record_modified): Use it.
27871 (remap_edge_change_prob): Handle also ancestor functions.
27873 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27875 * doc/contrib.texi (Contributors): Remove broken link into
27876 the Mauve CVS repository.
27878 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27880 PR middle-end/79454
27881 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27882 result computation whenever lhs doesn't have vector mode, not
27883 just when it has BLKmode.
27885 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27887 * doc/makefile.texi (profiledbootstrap): Refer to the
27888 installation instructions only in textual form.
27890 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27893 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27895 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27897 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27898 (Specific): Update mingw-w64 reference.
27900 (Specific): Remove broken link to Renesas RX processor.
27902 2017-02-10 Richard Biener <rguenther@suse.de>
27904 * toplev.c (process_options): Do not mention obsolete graphite
27905 options when printing sorry message about missing graphite support.
27906 Mention -floop-nest-optimize.
27908 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27910 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27911 (vtst_p16): Likewise.
27912 (vtstq_p8): Likewise.
27913 (vtstq_p16): Likewise.
27915 (vtstq_p64): Likewise.
27916 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27917 (vset_lane_p64): New.
27918 (vsetq_lane_p64): New.
27920 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27922 PR tree-optimization/79411
27923 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27924 stmt operands are SSA_NAMEs used in abnormal phis.
27925 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27928 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27931 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27934 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27936 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27938 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27940 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27944 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27945 not arbitrary TREE_CONSTANT.
27948 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27949 "omp declare target link" attribute unless is_global_var.
27950 * omp-offload.c (find_link_var_op): Likewise.
27952 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27953 Chung-Lin Tang <cltang@codesourcery.com>
27955 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27957 (gimplify_adjust_omp_clauses): Don't delete TILE.
27958 (gimplify_omp_for): Deal with TILE.
27959 * internal-fn.c (expand_GOACC_TILE): New function.
27960 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27962 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27963 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27965 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27966 avoid DIV for outermost collapse var.
27967 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27968 Remove out of date comments, fix whitespace.
27969 * omp-general.c (omp_extract_for_data): Deal with tiling.
27970 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27971 adjust OLF_DIM_BASE value.
27972 (struct omp_for_data): Add tiling field.
27973 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27974 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27975 for auto loops. Remove default auto determining, moved to
27976 oacc_loop_fixed_partitions.
27977 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27978 stmts, add e_mask field.
27979 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27980 (oacc_thread_numbers): Use oacc_dim_call.
27981 (oacc_xform_tile): New.
27982 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27983 (finish_oacc_loop): Adjust for ifns vector.
27984 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27985 add case for GOACC_TILE fns.
27986 (oacc_loop_xform_loop): Delete.
27987 (oacc_loop_process): Iterate over call list directly, and add
27988 handling for GOACC_TILE fns.
27989 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27991 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27992 vector partitioning to outer loops. Assign 2 partitions to loops
27993 when available. Add TILE handling.
27994 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27995 (execite_oacc_device_lower): Process GOACC_TILE fns,
27996 ignore unknown specs.
27997 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27998 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27999 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28001 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28003 * configure.ac (ACX_BUGURL): Update.
28004 * configure: Regenerate.
28006 2017-02-09 Richard Biener <rguenther@suse.de>
28008 PR tree-optimization/69823
28009 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28010 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28012 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28014 * config/arc/arc-c.def: Add __NPS400__ definition.
28015 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28016 (TARGET_NPS400): Define.
28018 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28020 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28022 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28023 pointer, arch_info.
28024 (arc_cpu_types): Fill the arch_info field with a pointer into the
28025 arc_arch_types table.
28026 (arc_selected_cpu): Declare.
28027 * config/arc/arc.c (arc_selected_cpu): Make global.
28028 (arc_selected_arch): Delete.
28029 (arc_base_cpu): Delete.
28030 (arc_override_options): Remove references to deleted variables,
28031 update access to arch information.
28032 (ARC_OPT): Update access to arch information.
28033 (ARC_OPTX): Likewise.
28034 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28035 (TARGET_ARC600): Update access to arch information.
28036 (TARGET_ARC601): Likewise.
28037 (TARGET_ARC700): Likewise.
28038 (TARGET_EM): Likewise.
28039 (TARGET_HS): Likewise.
28040 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28043 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28046 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28047 condition/operands for integer GE/LE/GEU/LEU operations.
28049 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28051 PR translation/79397
28052 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28055 2017-02-08 Martin Jambor <mjambor@suse.cz>
28058 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28059 whether allocation happened.
28060 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28061 nothing was allocated.
28063 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28065 PR tree-optimization/79408
28066 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28067 constant, but SSA_NAME with a known integer range, use the minimum
28068 of that range instead of op1 to determine if modulo can be replaced
28069 with its first operand.
28071 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28073 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28075 2017-02-08 Richard Biener <rguenther@suse.de>
28077 PR tree-optimization/71824
28078 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28079 Check all loops contained in the merged region.
28081 2017-02-07 Andrew Pinski <apinski@cavium.com>
28083 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28085 2017-02-07 Andrew Pinski <apinski@cavium.com>
28087 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28088 (thunderxt88): Likewise.
28089 (thunderxt81): Disable LSE and change v8.1 to v8.
28090 (thunderxt83): Likewise.
28092 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28093 Richard Biener <rguenther@suse.de>
28095 PR middle-end/79399
28096 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28097 type from int to size_t.
28098 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28100 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28102 PR rtl-optimization/79386
28103 * cprop.c (bypass_conditional_jumps): Initialize
28104 bypass_last_basic_block already before splitting bbs after
28105 unconditional traps...
28106 (bypass_conditional_jumps): ... rather than here.
28109 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28110 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28111 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28112 fix -masm=intel patterns.
28114 2017-02-07 Richard Biener <rguenther@suse.de>
28116 PR tree-optimization/79256
28117 PR middle-end/79278
28118 * builtins.c (get_object_alignment_2): Use min_align_of_type
28119 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28120 and ADJUST_FIELD_ALIGN.
28122 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28124 * doc/tm.texi: Regenerate.
28125 * stor-layout.c (layout_decl): Adjust.
28126 (update_alignment_for_field): Likewise.
28127 (place_field): Likewise.
28128 (min_align_of_type): Likewise.
28129 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28130 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28131 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28132 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28133 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28134 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28135 * config/i386/i386.c (x86_field_alignment): Likewise.
28136 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28137 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28138 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28139 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28140 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28141 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28145 2017-01-30 Richard Biener <rguenther@suse.de>
28147 PR tree-optimization/79256
28148 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28149 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28152 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28154 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28155 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28156 builtins to SImode and emit a zero-extend, if necessary.
28158 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28160 * docs/invoke.texi (RISC-V Options): Alphabetize.
28162 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28164 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28167 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28169 * config/riscv/riscv.c: New file.
28170 * common/config/riscv/riscv-common.c: Likewise.
28171 * config.gcc: Likewise.
28172 * config/riscv/constraints.md: Likewise.
28173 * config/riscv/elf.h: Likewise.
28174 * config/riscv/generic.md: Likewise.
28175 * config/riscv/linux.h: Likewise.
28176 * config/riscv/multilib-generator: Likewise.
28177 * config/riscv/peephole.md: Likewise.
28178 * config/riscv/pic.md: Likewise.
28179 * config/riscv/predicates.md: Likewise.
28180 * config/riscv/riscv-builtins.c: Likewise.
28181 * config/riscv/riscv-c.c: Likewise.
28182 * config/riscv/riscv-ftypes.def: Likewise.
28183 * config/riscv/riscv-modes.def: Likewise.
28184 * config/riscv/riscv-opts.h: Likewise.
28185 * config/riscv/riscv-protos.h: Likewise.
28186 * config/riscv/riscv.h: Likewise.
28187 * config/riscv/riscv.md: Likewise.
28188 * config/riscv/riscv.opt: Likewise.
28189 * config/riscv/sync.md: Likewise.
28190 * config/riscv/t-elf-multilib: Likewise.
28191 * config/riscv/t-linux: Likewise.
28192 * config/riscv/t-linux-multilib: Likewise.
28193 * config/riscv/t-riscv: Likewise.
28194 * configure.ac: Likewise.
28195 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28196 Waterman as RISC-V maintainers.
28197 * doc/install.texi: Add RISC-V entries.
28198 * doc/invoke.texi: Add RISC-V options section.
28199 * doc/md.texi: Add RISC-V constraints section.
28200 * configure: Regenerated.
28202 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28205 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28206 false values to be constant vectors with all 0 or all 1 bits set.
28207 (vcondu<mode><mode>): Likewise.
28208 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28210 (fpmask_comparison_operator): Update comment.
28211 (vecint_comparison_operator): New predicate.
28212 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28213 vector conditionals when the true and false values are constant
28214 vectors with all 0 bits or all 1 bits set.
28216 2017-02-06 Martin Sebor <msebor@redhat.com>
28218 PR tree-optimization/79376
28219 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28221 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28223 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28224 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28225 to simplify split condition.
28227 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28229 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28230 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28233 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28235 PR rtl-optimization/68664
28236 * target.def (can_speculate_insn): New hook.
28237 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28238 * doc/tm.texi: Regenerate.
28239 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28240 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28241 (rs6000_sched_can_speculate_insn): New function.
28243 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28245 PR tree-optimization/79284
28246 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28247 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28248 vectorizable_mask_load_store, vectorizable_operation,
28249 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28250 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28251 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28252 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28253 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28254 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28255 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28256 is_gimple_assign (stmt). Replace another such test with
28257 is_gimple_assign (stmt).
28259 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28262 * config/avr/avr.c (rtl-iter.h): Include it.
28263 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28264 (avr_legitimate_combined_insn): ...and implementation.
28266 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28268 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28269 * config/s390/s390.c (s390_const_operand_ok)
28270 (s390_canonicalize_comparison, s390_extract_part)
28271 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28272 (s390_contiguous_bitmask_p, s390_rtx_costs)
28273 (legitimize_pic_address): Likewise.
28274 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28275 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28276 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28277 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28278 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28280 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28282 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28283 REGNO($0) == REGNO($1).
28285 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28287 * config/s390/linux.h(SIZE_TYPE): Add comment.
28289 2017-02-06 Julian Brown <julian@codesourcery.com>
28290 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28291 Virendra Pathak <virendra.pathak@broadcom.com>
28293 * config/aarch64/aarch64-cores.def: Change the scheduler
28295 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28296 * config/aarch64/thunderx2t99.md: New file.
28298 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28300 * doc/standards.texi (Go Language): Update link to language
28303 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28305 * tree-eh.c (lower_resx): Sanitize profile.
28306 (cleanup_empty_eh_move_lp): Likewise.
28308 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28311 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28313 * cfgloopmanip.h (loop_version): Update prototype.
28314 * modulo-sched.c (sms_schedule): Update call of loop_version.
28315 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28316 * tree-parloops.c (gen_parallel_loop): Likewise.
28317 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28318 * tree-ssa-loop-split.c (split_loop): Likewise.
28319 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28320 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28322 2017-02-05 Martin Liska <mliska@suse.cz>
28325 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28327 (print_operand_address): Initialize a struct to zero.
28329 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28331 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28332 garbage collector only in textual form.
28334 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28336 * doc/extend.texi (x86 specific memory model extensions for
28337 transactional memory): Simplify a phrase.
28339 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28342 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28343 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28344 (atomic_storedi_1): Likewise.
28346 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28348 PR tree-optimization/79338
28349 * tree-parloops.c (gather_scalar_reductions): Don't call
28350 vect_analyze_loop_form for loop->inner before destroying loop's
28353 2017-02-03 Martin Sebor <msebor@redhat.com>
28355 PR tree-optimization/79327
28356 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28357 when precision has resulted in leading zeros.
28358 (format_integer): Adjust the likely counter to assume an unknown
28359 argument that may be zero is non-zero.
28361 2017-02-03 Jason Merrill <jason@redhat.com>
28364 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28365 avoid copying non-taken branch.
28367 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28369 PR tree-optimization/79340
28370 * tree-vect-loop.c (vectorizable_reduction): Release
28371 vec_defs elements after safe_splicing them into other vectors.
28374 PR tree-optimization/79327
28375 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28376 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28378 (format_integer): Use wide_int_to_tree instead of build_int_cst
28379 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28380 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28381 of shortest and longest sequence.
28383 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28385 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28386 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28388 2017-02-03 Walter Lee <walt@tilera.com>
28391 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28392 after initial stackframe link reg save.
28393 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28395 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28398 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28399 wu for stxssp alternative.
28401 2017-02-03 Martin Sebor <msebor@redhat.com>
28403 PR tree-optimization/79352
28404 * gimple-fold.c (get_range_strlen): Add argument.
28405 (get_range_strlen): Change return type to bool.
28406 (get_maxval_strlen): Pass in a dummy argument.
28407 * gimple-fold.h (get_range_strlen): Change return type to bool.
28408 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28409 * tree.h (array_at_struct_end_p): Add argument.
28410 * tree.c (array_at_struct_end_p): Handle it.
28412 2017-02-03 Martin Liska <mliska@suse.cz>
28415 * multiple_target.c (create_dispatcher_calls): Redirect edge
28416 from a caller of a dispatcher.
28417 (expand_target_clones): Make the clones local.
28418 (ipa_target_clone): Do both target clones and resolvers.
28419 (ipa_dispatcher_calls): Remove the pass.
28420 (pass_dispatcher_calls::gate): Likewise.
28421 (make_pass_dispatcher_calls): Likewise.
28422 * passes.def (pass_target_clone): Put as very first IPA early
28425 2017-02-03 Martin Liska <mliska@suse.cz>
28427 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28428 in case of a function with ifunc attribute.
28430 2017-02-03 Martin Liska <mliska@suse.cz>
28432 * cgraph.c (cgraph_node::dump): Dump function version info.
28433 * symtab.c (symtab_node::dump_base): Add missing new line.
28435 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28437 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28438 (ifcombine_ifandif): Use it.
28440 2017-02-03 Martin Liska <mliska@suse.cz>
28442 * doc/invoke.texi: Document default value for
28443 use-after-scope-direct-emission-threshold.
28445 2017-02-03 Martin Liska <mliska@suse.cz>
28447 PR tree-optimization/79339
28448 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28449 (format_floating): Likewise.
28451 2017-02-03 Martin Liska <mliska@suse.cz>
28454 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28456 (ipa_node_params_t::remove): Likewise.
28457 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28458 initialization from removed ipa_node_params_t::insert.
28459 (ipa_node_params::~ipa_node_params): Move from removed
28460 ipa_node_params_t::release.
28461 * symbol-summary.h (symbol_summary::m_released): New member.
28462 Do not release a summary twice. Do not allow to call finalizer
28463 for types of a summary that live in GGC memory.
28465 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28467 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28470 2017-02-02 Martin Sebor <msebor@redhat.com>
28472 PR middle-end/79275
28473 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28474 (format_string): Tighten up the range of output for non-constant
28475 strings and correct the expected range for wide non-constant strings.
28477 2017-02-02 Martin Sebor <msebor@redhat.com>
28479 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28481 PR middle-end/32003
28482 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28484 (-fdump-tree-@var): Add to index and document how to come up
28485 with pass-specific option and dump file names.
28486 (-fdump-passes): Clarify where to look for output.
28488 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28490 PR middle-end/77445
28491 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28492 statistics of the analyzed path; allow threading for speed when
28493 any of BBs along the path are optimized for speed.
28495 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28497 PR middle-end/78468
28498 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28499 settings of the virtual registers.
28502 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28504 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28505 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28508 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28510 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28511 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28513 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28515 * config/s390/s390.md: Add missing comments with the expanded
28517 * config/s390/vector.md: Likewise.
28518 * config/s390/vx-builtins.md: Likewise.
28520 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28523 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28524 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28525 conditions on a single line.
28527 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28529 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28530 __S390_VX__ to __VX__.
28532 2017-02-01 Andrew Pinski <apinski@cavium.com>
28534 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28535 stmt_info to record_stmt_cost.
28536 (vect_get_known_peeling_cost): Pass stmt_info if known to
28538 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28539 cpu_vector_cost field into
28540 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28541 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28542 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28543 splitting of scalar_stmt_cost and vec_stmt_cost.
28544 (thunderx_vector_cost): Likewise.
28545 (cortexa57_vector_cost): LIkewise.
28546 (exynosm1_vector_cost): Likewise.
28547 (xgene1_vector_cost): Likewise.
28548 (thunderx2t99_vector_cost): Improve after the splitting of the two
28550 (aarch64_builtin_vectorization_cost): Update for the splitting of
28551 scalar_stmt_cost and vec_stmt_cost.
28553 2017-02-01 Torvald Riegel <triegel@redhat.com>
28554 Richard Henderson <rth@redhat.com>
28556 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28557 conditional on existance of a fast atomic load.
28558 * optabs-query.c (can_atomic_load_p): New function.
28559 * optabs-query.h (can_atomic_load_p): Declare it.
28560 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28561 no fast atomic load is available for the particular size of access.
28562 (expand_atomic_compare_and_swap): Likewise.
28563 (expand_atomic_load): Likewise.
28564 (expand_atomic_store): Likewise.
28565 (expand_atomic_fetch_op): Likewise.
28566 * testsuite/lib/target-supports.exp
28567 (check_effective_target_sync_int_128): Remove x86 because it provides
28568 no fast atomic load.
28569 (check_effective_target_sync_int_128_runtime): Likewise.
28571 2017-02-01 Richard Biener <rguenther@suse.de>
28573 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28574 (graphite_transform_loops): Provide opt-info for optimized nests.
28575 * tree-parloop.c (parallelize_loops): Provide opt-info for
28576 parallelized loops.
28578 2017-02-01 Richard Biener <rguenther@suse.de>
28580 PR middle-end/79315
28581 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28582 was not set before.
28584 2017-02-01 Richard Biener <rguenther@suse.de>
28586 PR tree-optimization/71824
28587 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28588 Verify the loops are valid in the merged SESE region.
28589 (scop_detection::can_represent_loop_1): Check analyzing the
28590 evolution of the number of iterations in the region succeeds.
28592 2017-01-31 Ian Lance Taylor <iant@golang.org>
28594 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28595 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28597 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28599 PR preprocessor/79210
28600 * input.c (get_substring_ranges_for_loc): Replace line_width
28601 assertion with error-handling.
28603 2017-01-31 Richard Biener <rguenther@suse.de>
28605 PR tree-optimization/77318
28606 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28607 (create_pw_aff_from_tree): Take loop parameter.
28608 (add_condition_to_pbb): Pass loop of the condition to
28609 create_pw_aff_from_tree.
28611 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28613 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28614 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28616 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28620 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28622 (convert_int_to_float128): Likewise.
28623 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28624 (convert_int_to_float128): Likewise.
28625 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28626 (UNSPEC_IEEE128_CONVERT): Likewise.
28627 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28628 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28629 Use local variables for IBM extended format.
28630 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28631 (fix_trunc<mode>si2_fprs): Likewise.
28632 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28633 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28634 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28635 to know that we can now have integers of all sizes in vector
28637 (fix<uns>_<mode>di2_hw): Likewise.
28638 (float<uns>_<mode>si2_hw): Likewise.
28639 (fix_<mode>si2_hw): Likewise.
28640 (fixuns_<mode>si2_hw): Likewise.
28641 (float<uns>_<mode>di2_hw): Likewise.
28642 (float_<mode>di2_hw): Likewise.
28643 (float_<mode>si2_hw): Likewise.
28644 (floatuns_<mode>di2_hw): Likewise.
28645 (floatuns_<mode>si2_hw): Likewise.
28646 (xscvqp<su>wz_<mode>): Delete, no longer used.
28647 (xscvqp<su>dz_<mode>): Likewise.
28648 (xscv<su>dqp_<mode>): Likewise.
28649 (ieee128_mfvsrd_64bit): Likewise.
28650 (ieee128_mfvsrd_32bit): Likewise.
28651 (ieee128_mfvsrwz): Likewise.
28652 (ieee128_mtvsrw): Likewise.
28653 (ieee128_mtvsrd_64bit): Likewise.
28654 (ieee128_mtvsrd_32bit): Likewise.
28656 2017-01-31 Martin Liska <mliska@suse.cz>
28659 * ipa-prop.c (ipa_free_all_node_params): Call release method
28660 instead of ~sumbol_summary to not to trigger double times
28663 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28665 PR tree-optimization/71691
28666 * bitmap.h (class auto_bitmap): New.
28667 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28668 is_maybe_undefined instead of ssa_undefined_value_p.
28670 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28672 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28673 __S390_ARCH_LEVEL__ to __ARCH__.
28675 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28677 PR tree-optimization/79267
28678 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28679 if should_remove_lhs_p is true.
28681 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28684 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28685 (add_alignment_attribute): New.
28686 (base_type_die): Add alignment attribute.
28687 (subrange_type_die): Likewise.
28688 (modified_type_die): Likewise.
28689 (gen_array_type_die): Likewise.
28690 (gen_descr_array_type_die: Likewise.
28691 (gen_enumeration_type_die): Likewise.
28692 (gen_subprogram_die): Likewise.
28693 (gen_variable_die): Likewise.
28694 (gen_field_die): Likewise.
28695 (gen_ptr_to_mbr_type_die): Likewise.
28696 (gen_struct_or_union_type_die): Likewise.
28697 (gen_subroutine_type_die): Likewise.
28698 (gen_typedef_die): Likewise.
28699 (base_type_cmp): Compare alignment attribute.
28701 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28704 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28705 (setb_unsigned) New pattern for setb with CCUNS.
28706 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28707 subfc./subfe sequence to avoid overflow problems. Generate a
28708 shorter sequence with cmpld/setb for power9.
28709 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28710 for generating subfc. instruction.
28711 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28712 now uses this instruction.
28714 2017-01-30 Ian Lance Taylor <iant@google.com>
28717 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28718 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28720 2017-01-30 Martin Sebor <msebor@redhat.com>
28722 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28723 Move constant to the right of a relational operator.
28724 (get_mpfr_format_length, format_character, format_string): Ditto.
28725 (should_warn_p, maybe_warn): Same.
28727 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28729 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28732 * asan.c (get_translation_unit_decl): Remove function.
28733 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28735 2017-01-30 Martin Liska <mliska@suse.cz>
28737 PR gcov-profile/79259
28738 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28739 -fprofile-generate.
28741 2017-01-30 Martin Liska <mliska@suse.cz>
28744 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28745 Initialize variables with NULL value.
28747 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28750 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28752 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28754 2017-01-30 Richard Biener <rguenther@suse.de>
28756 PR tree-optimization/79276
28757 * tree-vrp.c (process_assert_insertions): Properly adjust common
28758 when removing a duplicate.
28760 * gcc.dg/torture/pr79276.c: New testcase.
28762 2017-01-30 Richard Biener <rguenther@suse.de>
28764 PR tree-optimization/79256
28765 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28766 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28768 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28770 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28773 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28774 ("*r<noxa>sbg_<mode>_sll_bitmask")
28775 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28776 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28777 Use contiguous_bitmask_nowrap_operand.
28779 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28782 * config/rs6000/altivec.h (vec_xl): Revise #define.
28783 (vec_xst): Likewise.
28785 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28787 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28789 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28791 PR rtl-optimization/79194
28792 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28793 traps before call to bypass_conditional_jumps.
28795 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28797 PR tree-optimization/71374
28798 * lra-constraints.c (check_conflict_input_operands): New.
28799 (match_reload): Use it.
28801 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28804 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28805 account to calculate conflict_set.
28807 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28809 PR rtl-optimization/78559
28810 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28811 other_insn in combine.
28813 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28815 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28816 uint16_type_node for BT_UINT16.
28818 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28820 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28821 "RTL Tests" to menu.
28822 (GIMPLE Tests): New node.
28823 (RTL Tests): New node.
28825 2017-01-27 Richard Biener <rguenther@suse.de>
28827 PR tree-optimization/79245
28828 * tree-loop-distribution.c (distribute_loop): Apply cost
28829 modeling also to detected patterns.
28831 2017-01-27 Richard Biener <rguenther@suse.de>
28833 PR tree-optimization/71433
28834 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28835 (compare_assert_loc): New function.
28836 (process_assert_insertions): Sort and optimize assert locations
28837 to remove duplicates and push down identical assertions on
28838 edges to their destination block.
28840 2017-01-27 Richard Biener <rguenther@suse.de>
28842 PR tree-optimization/79244
28843 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28844 out SSA names even if abnormal.
28846 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28848 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28849 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28850 instead of MPFR_RNDN.
28852 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28855 * arm.c (arm_option_override): Don't call build_target_option_node
28856 until after doing all option overrides.
28857 (arm_valid_target_attribute_tree): Likewise.
28859 2017-01-27 Martin Liska <mliska@suse.cz>
28861 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28862 for {cd}tors and C++ {cd}tors.
28864 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28866 * config/s390/s390.md ("*setmem_long_and")
28867 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28869 2017-01-26 Martin Sebor <msebor@redhat.com>
28871 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28874 2017-01-26 Martin Sebor <msebor@redhat.com>
28876 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28877 HAVE_DFmode before using XFmode or DFmode.
28878 (parse_directive): Avoid using the z length modifier to avoid
28879 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28881 PR middle-end/78703
28882 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28883 to accept adjustment as an array.
28884 (get_int_range): New function.
28885 (struct directive): Make width and prec arrays.
28886 (directive::set_width, directive::set_precision): Call get_int_range.
28887 (format_integer, format_floating): Handle width and precision ranges.
28888 (format_string, parse_directive): Same.
28890 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28893 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28894 comdat_type_p set, just clone them, but keep the children in the
28898 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28899 which have direct callers with -fvar-tracking-assignments enabled
28901 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28902 inside of type units.
28904 2017-01-26 Martin Sebor <msebor@redhat.com>
28906 PR middle-end/78703
28907 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28909 (struct format_result): Replace number_chars, number_chars_min,
28910 and number_chars_max with a single member of struct result_range.
28912 (format_result::operator+=): Adjust.
28913 (struct fmtresult): Remove bounded. Handle likely and unlikely
28915 (fmtresult::adjust_for_width_or_precision): New function.
28916 (fmtresult:type_max_digits): New function.
28917 (bytes_remaining): Handle likely and unlikely counters.
28918 (min_bytes_remaining): Remove.
28919 (format_percent): Simplify.
28920 (format_integer, format_floating): Set likely and unlikely counters.
28921 (get_string_length, format_character, format_string): Same.
28922 (format_plain, should_warn_p): New function.
28923 (maybe_warn): Call should_warn_p. Update diagnostic messages
28924 and handle those for all directives, including plain strings.
28925 (format_directive): Handle likely and unlikely counters.
28926 Remove unnecessary quoting from diagnostics. Add an informational
28928 (add_bytes): Remove.
28929 (pass_sprintf_length::compute_format_length): Simplify.
28930 (try_substitute_return_value): Handle likely and unlikely counters.
28932 2017-01-26 Carl Love <cel@us.ibm.com>
28934 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28935 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28937 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28940 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28941 endianess for subregs into account.
28942 * lra-constraints.c (lra_constraints): Do risky transformations
28943 always on the first iteration.
28944 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28945 last_call_used_reg_set.
28946 (process_bb_lives): Define and use last_call_used_reg_set.
28947 * lra.c (lra): Always continue after lra_constraints on the first
28950 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28952 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28954 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28956 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28958 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28959 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28960 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28961 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28962 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28963 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28964 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28965 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28966 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28968 2017-01-26 Marek Polacek <polacek@redhat.com>
28971 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28972 for the third operand.
28974 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28976 PR middle-end/79236
28977 * omp-low.c (struct omp_context): Add simt_stmt field.
28978 (scan_omp_for): Return omp_context *.
28979 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28980 context to the _simt_ SIMD stmt.
28981 (lower_omp_for): For combined SIMD with sibling _simt_
28982 SIMD, make sure to use the same decls in _looptemp_
28983 clauses as in the sibling.
28985 2017-01-26 David Sherwood <david.sherwood@arm.com>
28987 PR middle-end/79212
28988 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28991 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28994 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28995 emit fld b; fld a; if possible.
28997 * brig-builtins.def: Update copyright years.
28998 * config/arm/arm_acle_builtins.def: Update copyright years.
29000 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29003 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29004 constraint instead of o for the stxsd instruction.
29006 2017-01-25 Carl Love <cel@us.ibm.com>
29008 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29009 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29011 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29013 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29015 2017-01-25 Richard Biener <rguenther@suse.de>
29017 PR tree-optimization/69264
29018 * target.def (vector_alignment_reachable): Improve documentation.
29019 * doc/tm.texi: Regenerate.
29020 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29022 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29023 earlier changes with respect to TYPE_USER_ALIGN.
29024 (vector_alignment_reachable_p): Likewise. Improve dumping.
29026 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29029 * config/arm/arm.md (xordi3): Force constant operand into a register
29032 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29034 * doc/invoke.texi (-fstore-merging): Correct default optimization
29035 levels at which it is enabled.
29036 (-O): Move -fstore-merging from list to...
29039 2017-01-25 Richard Biener <rguenther@suse.de>
29042 * omp-expand.c: Include debug.h.
29043 (expand_omp_taskreg): Make sure to generate early debug before
29044 outlining anything from a function.
29045 (expand_omp_target): Likewise.
29046 (grid_expand_target_grid_body): Likewise.
29048 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29051 * asan.c (get_translation_unit_decl): New function.
29052 (asan_add_global): Extract modules file name from globals
29053 TRANSLATION_UNIT_DECL name.
29055 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29058 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29059 for long calls with APCS frame and VFP.
29061 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29063 * cfg.c (original_copy_tables_initialized_p): New function.
29064 * cfg.h (original_copy_tables_initialized_p): New decl.
29065 * cfgrtl.c (relink_block_chain): Guard the call to
29066 free_original_copy_tables with a call to
29067 original_copy_tables_initialized_p.
29068 * cgraph.h (symtab_node::native_rtl_p): New decl.
29069 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29070 (symtab_node::needed_p): Don't assert for early assembly output
29071 for __RTL functions.
29072 (cgraph_node::finalize_function): Set "force_output" for __RTL
29074 (cgraph_node::analyze): Bail out early for __RTL functions.
29075 (analyze_functions): Update assertion to support __RTL functions.
29076 (cgraph_node::expand): Bail out early for __RTL functions.
29077 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29079 * function.h (struct function): Update comment for field
29081 * gimple-expr.c: Include "tree-pass.h".
29082 (gimple_has_body_p): Return false for __RTL functions.
29083 * Makefile.in (OBJS): Add run-rtl-passes.o.
29084 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29086 (gcc::pass_manager::get_clean_slate): New accessor.
29087 * passes.c: Include "insn-addr.h".
29088 (should_skip_pass_p): Add logging. Update logic for running
29089 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29090 property-provider override so it is only done for gimple passes.
29092 (skip_pass): New function.
29093 (execute_one_pass): Call skip_pass when skipping passes.
29094 * read-md.c (md_reader::read_char): Support filtering
29095 the input to a subset of line numbers.
29096 (md_reader::md_reader): Initialize fields
29097 m_first_line and m_last_line.
29098 (md_reader::read_file_fragment): New function.
29099 * read-md.h (md_reader::read_file_fragment): New decl.
29100 (md_reader::m_first_line): New field.
29101 (md_reader::m_last_line): New field.
29102 * read-rtl-function.c (function_reader::create_function): Only
29103 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29104 curr_properties. Set DECL_INITIAL to a dummy block.
29105 (read_rtl_function_body_from_file_range): New function.
29106 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29108 * run-rtl-passes.c: New file.
29109 * run-rtl-passes.h: New file.
29111 2017-01-24 Jeff Law <law@redhat.com>
29113 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29116 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29118 PR tree-optimization/79159
29119 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29120 (record_nonwrapping_iv): Improve boundary using above function if no
29121 value range information.
29123 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29124 Martin Jambor <mjambor@suse.cz>
29126 * brig-builtins.def: New file.
29127 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29128 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29129 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29130 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29131 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29132 * builtin-types.def (BT_INT8): New.
29133 (BT_INT16): Likewise.
29134 (BT_UINT8): Likewise.
29135 (BT_UINT16): Likewise.
29136 (BT_FN_ULONG): Likewise.
29137 (BT_FN_UINT_INT): Likewise.
29138 (BT_FN_UINT_ULONG): Likewise.
29139 (BT_FN_UINT_LONG): Likewise.
29140 (BT_FN_UINT_PTR): Likewise.
29141 (BT_FN_ULONG_PTR): Likewise.
29142 (BT_FN_INT8_FLOAT): Likewise.
29143 (BT_FN_INT16_FLOAT): Likewise.
29144 (BT_FN_UINT32_FLOAT): Likewise.
29145 (BT_FN_UINT16_FLOAT): Likewise.
29146 (BT_FN_UINT8_FLOAT): Likewise.
29147 (BT_FN_UINT64_FLOAT): Likewise.
29148 (BT_FN_UINT16_UINT32): Likewise.
29149 (BT_FN_UINT32_UINT16): Likewise.
29150 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29151 (BT_FN_INT_PTR_INT): Likewise.
29152 (BT_FN_UINT_PTR_UINT): Likewise.
29153 (BT_FN_LONG_PTR_LONG): Likewise.
29154 (BT_FN_ULONG_PTR_ULONG): Likewise.
29155 (BT_FN_VOID_UINT64_UINT64): Likewise.
29156 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29157 (BT_FN_INT8_INT8_INT8): Likewise.
29158 (BT_FN_INT16_INT16_INT16): Likewise.
29159 (BT_FN_INT_INT_INT): Likewise.
29160 (BT_FN_UINT_FLOAT_UINT): Likewise.
29161 (BT_FN_FLOAT_UINT_UINT): Likewise.
29162 (BT_FN_ULONG_UINT_UINT): Likewise.
29163 (BT_FN_ULONG_UINT_PTR): Likewise.
29164 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29165 (BT_FN_UINT_UINT_UINT): Likewise.
29166 (BT_FN_VOID_UINT_PTR): Likewise.
29167 (BT_FN_UINT_UINT_PTR: Likewise.
29168 (BT_FN_UINT32_UINT64_PTR): Likewise.
29169 (BT_FN_INT_INT_UINT_UINT): Likewise.
29170 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29171 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29172 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29173 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29174 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29175 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29176 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29177 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29178 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29179 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29180 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29181 * doc/frontends.texi: List BRIG FE.
29182 * doc/install.texi (Testing): Add BRIG tesring requirements.
29183 * doc/invoke.texi (Overall Options): Mention BRIG.
29184 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29186 2017-01-24 Richard Biener <rguenther@suse.de>
29188 PR translation/79208
29189 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29191 2017-01-24 Martin Jambor <mjambor@suse.cz>
29194 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29195 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29196 and known_contexts.
29198 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29200 PR middle-end/79123
29201 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29202 casts from signed to unsigned really don't have a range.
29204 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29206 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29207 GMP_RNDx for compatiblity.
29209 2017-01-24 Martin Liska <mliska@suse.cz>
29212 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29213 that would prevent us to call alloca with -1 as argument.
29215 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29217 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29218 Avoid -Wformat-security warning.
29220 2017-01-23 Andrew Pinski <apinski@cavium.com>
29222 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29225 2017-01-23 Martin Sebor <msebor@redhat.com>
29227 PR middle-end/78703
29228 * gimple-ssa-sprintf.c (warn_level): New global.
29229 (format_integer): Use it here and throughout the rest of the file.
29230 Use the same switch to compute sign as base.
29231 (maybe_warn): New function.
29232 (format_directive): Factor out warnings into maybe_warn.
29233 Add debugging output. Use warn_level.
29234 (add_bytes): Use warn_level.
29235 (pass_sprintf_length::compute_format_length): Add debugging output.
29236 (try_substitute_return_value): Same.
29237 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29239 PR middle-end/78703
29240 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29241 (struct fmtresult, format_integer, format_floating): Adjust.
29242 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29243 (get_string_length, format_string,format_directive): Same.
29244 (pass_sprintf_length::compute_format_length): Same.
29245 (try_substitute_return_value): Simplify slightly.
29247 PR middle-end/78703
29248 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29249 (fmtresult::operator+=): Outlined.
29250 (struct fmtresult): Add ctors.
29251 (struct conversion_spec): Rename...
29252 (struct directive): ...to this. Add and remove data members.
29253 (directive::set_width, directive::set_precision): New functions.
29254 (format_percent): Use fmtresult ctor.
29255 (get_width_and_precision): Remove.
29256 (format_integer): Make naming changes. Avoid computing width and
29258 (format_floating): Same. Adjust indentation.
29259 (format_character, format_none): New functions.
29260 (format_string): Moved character handling to format_character.
29261 (format_directive): Remove arguments, change return type.
29262 (parse_directive): New function.
29263 (pass_sprintf_length::compute_format_length): Move directive
29264 parsing to parse_directive.
29266 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29268 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29269 (assign_assembler_name_if_needed): ... this.
29270 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29271 (assign_assembler_name_if_needed): ... this.
29272 (free_lang_data_in_cgraph): Adjust callers.
29273 * cgraphunit.c (cgraph_node::analyze): Likewise.
29274 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29277 2017-01-23 Richard Biener <rguenther@suse.de>
29279 PR tree-optimization/79088
29280 PR tree-optimization/79188
29281 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29282 resetting loop bounds after last path deletion. Reset loop
29283 bounds of the target loop, make code match the comments.
29284 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29285 Make sure loops need no fixups.
29287 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29289 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29290 exponent support with double type for first argument.
29291 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29292 type returned by __builtin_vec_extract_sig,
29293 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29294 functions from "vector int" to "vector unsigned int" or from
29295 "vector long long int" to "vector unsigned long long int".
29296 Changed type returned by __builtin_vec_extract_exp,
29297 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29298 functions from "vector int" to "vector unsigned int" or from
29299 "vector long long int" to "vector unsigned long long int".
29300 Changed return type of __builtin_vec_test_data_class,
29301 __builtin_vec_test_data_class_sp, and
29302 __builtin_vec_test_data_class_dp from "vector int" to
29303 "vector bool int" or from "vector long long int" to "vector bool
29304 long long int" and changed second argument type from "unsigned
29305 int" to "int". Added new overloaded function forms "vector float
29306 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29307 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29308 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29309 double, vector unsigned long long int)" and "vector double
29310 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29311 long int)". Changed return type of
29312 __builtin_scalar_test_data_class and
29313 __builtin_scalar_test_data_class_sp and
29314 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29315 int" and changed second argument from "unsigned int" to "int".
29316 Changed type returned by __builtin_scalar_test_neg,
29317 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29318 from "int" to "bool int". Added new overloaded function form
29319 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29320 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29321 exponent double-precision with floating point first argument.
29322 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29323 documentation of scalar_test_data_class, scalar_test_neg,
29324 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29325 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29326 vec_test_data_class built-in functions to reflect refinements in
29327 their type signatures.
29329 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29331 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29333 (aarch64_elf_asm_destructor): Likewise.
29335 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29337 PR rtl-optimization/78634
29338 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29339 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29340 * ifcvt.c (noce_try_cmove): Add missing cost check.
29342 PR rtl-optimization/71724
29343 * combine.c (if_then_else_cond): Look for situations where it is
29344 beneficial to undo the work of one of the recursive calls.
29346 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29348 PR tree-optimization/70754
29349 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29350 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29351 combined stmt before it if not NULL.
29352 (combine_chains): Process refs reversely and compute dominance point
29355 2017-01-23 Martin Liska <mliska@suse.cz>
29357 PR tree-optimization/79196
29358 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29359 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29361 (strlen_optimize_stmt): Call the renamed function.
29363 2017-01-23 Michael Matz <matz@suse.de>
29365 PR tree-optimization/78384
29366 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29368 2017-01-23 Richard Biener <rguenther@suse.de>
29370 PR tree-optimization/79186
29371 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29372 both incoming edges before moving an assert.
29374 2017-01-23 Martin Jambor <mjambor@suse.cz>
29376 * ipa-prop.c (load_from_param_1): Removed.
29377 (load_from_unmodified_param): Bits from load_from_param_1 put back
29379 (load_from_param): Removed.
29380 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29381 with stmt. Reverted back to use of load_from_unmodified_param.
29383 2017-01-23 Martin Jambor <mjambor@suse.cz>
29386 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29387 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29388 field a pointer to garbage collected vector, mark lattices and
29389 ipcp_orig_node with GTY((skip)).
29390 (ipa_get_param_count): Adjust to descriptors being a pointer.
29391 (ipa_get_param): Likewise.
29392 (ipa_get_type): Likewise.
29393 (ipa_get_param_move_cost): Likewise.
29394 (ipa_set_param_used): Likewise.
29395 (ipa_get_controlled_uses): Likewise.
29396 (ipa_set_controlled_uses): Likewise.
29397 (ipa_is_param_used): Likewise.
29398 (ipa_node_params_t): Move into garbage collector. New methods insert
29400 (ipa_node_params_sum): Annotate wth GTY(()).
29401 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29403 (ipa_load_from_parm_agg): Adjust declaration.
29404 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29405 * ipa-profile.c (ipa_profile): Likewise.
29406 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29407 (ipa_populate_param_decls): Make descriptors parameter garbage
29409 (ipa_dump_param): Adjust to descriptors being a pointer.
29410 (ipa_alloc_node_params): Likewise.
29411 (ipa_initialize_node_params): Likewise.
29412 (load_from_param_1): Make descriptors parameter garbage collected.
29413 (load_from_unmodified_param): Likewise.
29414 (load_from_param): Likewise.
29415 (ipa_load_from_parm_agg): Likewise.
29416 (ipa_node_params::~ipa_node_params): Removed.
29417 (ipa_free_all_node_params): Remove call to delete operator.
29418 (ipa_node_params_t::insert): New.
29419 (ipa_node_params_t::remove): Likewise.
29420 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29421 copy known_csts and known_contexts vectors.
29422 (ipa_read_node_info): Adjust to descriptors being a pointer.
29423 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29425 (ipcp_transform_function): Make descriptors variable garbage
29428 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29430 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29431 * config/i386/avx512dqintrin.h: Ditto.
29432 * config/i386/avx512fintrin.h: Ditto.
29433 * config/i386/i386.c: Handle new builtins.
29434 * config/i386/i386-builtin.def: Add new builtins.
29435 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29436 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29438 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29439 Martin Liska <mliska@suse.cz>
29441 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29442 * asan.c (asan_expand_poison_ifn): Support stores and use
29443 appropriate ASAN report function.
29444 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29445 * internal-fn.def (ASAN_POISON_USE): Declare.
29446 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29447 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29448 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29449 ASAN_POISON calls w/o LHS.
29450 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29451 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29452 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29453 * gimplify.c (asan_poison_variables): Add attribute
29454 use_after_scope_memory to variables that really needs to live
29456 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29457 having the attribute.
29459 2017-01-23 Martin Liska <mliska@suse.cz>
29461 * asan.c (create_asan_shadow_var): New function.
29462 (asan_expand_poison_ifn): Likewise.
29463 * asan.h (asan_expand_poison_ifn): New declaration.
29464 * internal-fn.c (expand_ASAN_POISON): Likewise.
29465 * internal-fn.def (ASAN_POISON): New builtin.
29466 * sanopt.c (pass_sanopt::execute): Expand
29467 asan_expand_poison_ifn.
29468 * tree-inline.c (copy_decl_for_dup_finish): Make function
29470 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29471 * tree-ssa.c (is_asan_mark_p): New function.
29472 (execute_update_addresses_taken): Rewrite local variables
29473 (identified just by use-after-scope as addressable) into SSA.
29475 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29477 * doc/install.texi (Specific): opensource.apple.com uses https
29478 now. Remove trailing slash.
29480 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29482 * README.Portability: Remove note on an Irix compatibility issue.
29484 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29486 * gcov.c (INCLUDE_ALGORITHM): Define.
29487 (INCLUDE_VECTOR): Define.
29488 No longer include <vector> and <algorithm> directly.
29490 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29492 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29494 * doc/invoke.texi (Code Gen Options): Ditto.
29496 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29499 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29501 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29503 rtl-optimization/79125
29504 * cprop.c (local_cprop_pass): Handle cases where we make an
29505 unconditional trap.
29507 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29511 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29512 read from, for big endian.
29514 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29516 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29517 register pauth builtins for LP64 only.
29519 2017-01-20 Marek Polacek <polacek@redhat.com>
29522 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29525 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29527 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29529 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29530 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29531 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29533 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29536 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29537 in big-endian mode when they are not a single duplicated value.
29539 2017-01-20 Richard Biener <rguenther@suse.de>
29541 * BASE-VER: Bump to 7.0.1.
29543 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29545 * omp-low.c (omplow_simd_context): New struct. Use it...
29546 (lower_rec_simd_input_clauses): ...here and...
29547 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29548 references to idx, lane, max_vf, is_simt.
29550 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29552 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29555 2017-01-20 Martin Jambor <mjambor@suse.cz>
29557 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29558 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29560 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29561 (GTFILES): Rename hsa.c to hsa-common.c.
29562 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29563 * hsa-dump.c: Likewise.
29564 * hsa-gen.c: Likewise.
29565 * hsa-regalloc.c: Likewise.
29566 * ipa-hsa.c: Likewise.
29567 * omp-expand.c: Likewise.
29568 * omp-low.c: Likewise.
29569 * toplev.c: Likewise.
29571 2017-01-20 Marek Polacek <polacek@redhat.com>
29574 * doc/invoke.texi: Document -Wduplicated-branches.
29575 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29576 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29577 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29578 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29579 return 0 only when not OEP_LEXICOGRAPHIC.
29580 (fold_build_cleanup_point_expr): Use the expression
29581 location when building CLEANUP_POINT_EXPR.
29582 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29583 * tree.c (add_expr): Handle error_mark_node.
29585 2017-01-20 Martin Liska <mliska@suse.cz>
29588 * tree-profile.c (init_ic_make_global_vars): Do not call
29590 (gimple_init_gcov_profiler): Likewise.
29592 2017-01-20 Martin Liska <mliska@suse.cz>
29595 * cgraph.h (maybe_create_reference): Remove argument and
29597 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29599 * ipa-cp.c (create_specialized_node): Likewise.
29600 * symtab.c (symtab_node::maybe_create_reference): Handle
29601 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29603 2017-01-20 Martin Liska <mliska@suse.cz>
29605 * read-rtl-function.c (function_reader::create_function): Use
29606 build_decl instread of build_decl_stat.
29608 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29610 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29611 * config/i386/avx512dqintrin.h: Ditto.
29612 * config/i386/avx512fintrin.h: Ditto.
29613 * config/i386/i386-builtin-types.def: Add new types.
29614 * config/i386/i386.c: Handle new types.
29615 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29616 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29617 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29618 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29619 (__builtin_ia32_kshiftridi): New.
29620 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29622 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29626 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29627 define to rs6000_init_stack_protect_guard.
29628 (rs6000_init_stack_protect_guard): New function.
29630 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29631 Yunqiang Su <yunqiang.su@imgtec.com>
29633 * config.gcc (supported_defaults): Add madd4.
29634 (with_madd4): Add validation.
29635 (all_defaults): Add madd4.
29636 * config/mips/mips.opt (mmadd4): New option.
29637 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29639 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29641 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29642 (ISA_HAS_FUSED_MADD4): Likewise.
29643 * doc/invoke.texi (-mmadd4): Document the new option.
29644 * doc/install.texi (--with-madd4): Document the new option.
29646 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29648 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29649 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29650 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29651 (aarch64_init_pauth_hint_builtins): New.
29652 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29653 (aarch64_expand_builtin): Expand new builtins.
29655 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29657 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29658 * combine-stack-adj.c (no_unhandled_cfa): Handle
29659 REG_CFA_TOGGLE_RA_MANGLE.
29660 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29661 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29662 info for return address signing.
29663 (aarch64_expand_epilogue): Likewise.
29665 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29667 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29668 * config/aarch64/aarch64-protos.h
29669 (aarch64_return_address_signing_enabled): New declaration.
29670 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29672 (aarch64_expand_prologue): Sign return address before it's pushed onto
29674 (aarch64_expand_epilogue): Authenticate return address fetched from
29676 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29677 (aarch64_attributes): New function attributes for "sign-return-address".
29678 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29679 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29680 ("*do_return"): Generate combined instructions according to key index.
29681 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29682 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29684 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29685 * config/aarch64/aarch64.opt (msign-return-address=): New.
29686 * doc/extend.texi (AArch64 Function Attributes): Documents
29687 "sign-return-address=".
29688 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29690 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29692 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29693 overall option summary.
29695 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29697 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29698 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29699 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29700 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29702 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29704 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29705 -mpower9-minmax by default for -mcpu=power9.
29706 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29707 128-bit floating point.
29709 2017-01-20 Alan Modra <amodra@gmail.com>
29711 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29712 optimizing for size.
29714 2017-01-20 Alan Modra <amodra@gmail.com>
29717 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29718 for strcmp and strncmp from corresponding builtin decl.
29720 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29722 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29723 instead of i386/rtems-64.h.
29724 * config/i386/rtems-64.h: Remove.
29726 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29730 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29732 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29734 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29736 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29737 Change int to HOST_WIDE_INT.
29738 * config/aarch64/aarch64-protos.h
29739 (aarch64_simd_gen_const_vector_dup): Likewise.
29740 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29742 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29744 * langhooks-def.h (lhd_type_for_size): New decl.
29745 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29746 * langhooks.c (lhd_type_for_size): New function, taken from
29749 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29751 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29752 define_bypass for CR latency.
29753 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29754 (power9-alu2): Add define_bypass for CR latency.
29756 (power9-mul): Update insn latency.
29757 (power9-mul-compare): Update insn latency, bypass latency and remove
29760 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29762 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29764 * config/aarch64/aarch64.md
29765 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29766 aarch64_nopcrelative_literal_loads.
29767 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29769 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29771 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29772 TARGET_LOONGSON_3A.
29773 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29775 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29778 * config.gcc (supported_defaults): Add lxc1-sxc1.
29779 (with_lxc1_sxc1): Add validation.
29780 (all_defaults): Add lxc1-sxc1.
29781 * config/mips/mips.opt (mlxc1-sxc1): New option.
29782 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29784 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29785 __mips_no_lxc1_sxc1.
29786 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29787 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29788 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29790 2017-01-19 Richard Biener <rguenther@suse.de>
29792 PR tree-optimization/72488
29793 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29794 sure to restore SSA info.
29795 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29797 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29799 PR rtl-optimization/79121
29800 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29801 of the inner type when shifting an extended value.
29803 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29806 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29807 interposable aliases.
29809 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29812 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29813 Use the evmergelohi instruction.
29814 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29815 (mov_sitf_e500_subreg8_2_be): Likewise.
29816 (mov_sitf_e500_subreg12_2_le): Likewise.
29817 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29818 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29819 (mov_sitf_e500_subreg8_2_le): Likewise.
29820 (mov_sitf_e500_subreg12_2_be): Likewise.
29822 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29824 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29825 attribute from vecsimple to vecperm.
29826 (altivec_vbpermq2): Likewise.
29828 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29831 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29833 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29834 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29835 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29836 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29837 case where N arg is SIZE_MAX.
29838 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29839 (cmpstrsi): Add pattern.
29841 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29843 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29844 __builtin_vec_revb builtins.
29845 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29846 built-in functions to support generation of the ISA 3.0 XXBR<x>
29847 vector byte reverse instructions.
29848 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29849 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29850 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29851 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29852 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29853 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29854 (P9V_BUILTIN_VEC_REVB): Likewise.
29855 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29856 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29857 (p9_xxbrq_v16qi): Likewise.
29858 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29859 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29860 (p9_xxbrh_v8hi): Likewise.
29861 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29862 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29863 vec_revb built-in functions.
29865 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29867 PR rtl-optimization/78952
29868 * config/i386/i386.md (any_extract): New code iterator.
29869 (*insvqi_2): Use any_extract for source operand.
29870 (*insvqi_3): Use any_shiftrt for source operand.
29872 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29874 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29876 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29878 2017-01-18 Matthias Klose <doko@ubuntu.com>
29880 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29882 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29884 * config/rs6000/altivec.h (vec_bperm): Change #define.
29885 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29886 (altivec_vbpermq2): New define_insn.
29887 (altivec_vbpermd): Likewise.
29888 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29889 function interface.
29890 (VBPERMD): Likewise.
29891 (VBPERM): New polymorphic function interface.
29892 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29893 Add entries for P9V_BUILTIN_VEC_VBPERM.
29894 * doc/extend.texi: Add interfaces for vec_bperm.
29896 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29898 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29899 first letter of error messages.
29900 (s390_resolve_overloaded_builtin): Likewise.
29901 * config/s390/s390.c (s390_expand_builtin): Likewise.
29902 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29903 (s390_valid_target_attribute_inner_p): Likewise.
29904 * config/s390/s390.md ("tabort"): Likewise.
29906 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29908 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29909 (ISA_AVOID_DIV_HILO): New macro.
29910 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29911 (ISA_HAS_DDIV): Likewise.
29913 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29915 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29917 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29919 * doc/invoke.texi (fabi-version): Spelling fix.
29921 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29924 * doc/invoke.texi (fabi-version): Mention mangling fix for
29927 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29930 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29932 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29935 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29936 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29938 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29940 (stack_protect_test): Ditto.
29941 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29942 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29944 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29945 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29946 -mstack-protector-guard-offset=.
29947 (RS/6000 and PowerPC Options): Ditto.
29949 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29951 * config/i386/i386.h (MASK_CLASS_P): New define.
29952 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29953 there are no registers from different register sets also when
29954 mask registers are used. Update function comment.
29955 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29956 to (*k/*r) and (*k/*km) alternatives.
29958 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29960 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29961 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29962 (EH_RETURN_HANDLER_RTX): New define.
29963 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29964 Force frame pointer in EH return functions.
29965 (aarch64_expand_epilogue): Add barrier for eh_return.
29966 (aarch64_final_eh_return_addr): Remove.
29967 (aarch64_eh_return_handler_rtx): New function.
29968 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29970 (aarch64_eh_return_handler_rtx): New prototype.
29972 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29974 * config/rs6000/altivec.h (vec_rlmi): New #define.
29975 (vec_vrlnm): Likewise.
29976 (vec_rlnm): Likewise.
29977 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29978 (UNSPEC_VRLNM): Likewise.
29979 (VIlong): New mode iterator.
29980 (altivec_vrl<VI_char>mi): New define_insn.
29981 (altivec_vrl<VI_char>nm): Likewise.
29982 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29984 (VRLDNM): Likewise.
29985 (RLNM): New polymorphic function entry.
29986 (VRLWMI): New monomorphic function entry.
29987 (VRLDMI): Likewise.
29988 (RLMI): New polymorphic function entry.
29989 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29990 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29991 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29994 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29997 * dwarf2out.c (field_byte_offset): Restore the
29998 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29999 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30000 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30003 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30006 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30008 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30011 * dwarf2out.c (add_data_member_location_attribute): For constant
30012 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30013 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30014 DW_AT_byte_size attributes.
30016 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30018 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30019 after forcing to constant memory when the code model is medium.
30021 2017-01-17 Julia Koval <julia.koval@intel.com>
30024 * config/i386/avx512fintrin.h
30025 (_mm512_i32gather_ps): Change __addr type to void const*.
30026 (_mm512_mask_i32gather_ps): Ditto.
30027 (_mm512_i32gather_pd): Ditto.
30028 (_mm512_mask_i32gather_pd): Ditto.
30029 (_mm512_i64gather_ps): Ditto.
30030 (_mm512_mask_i64gather_ps): Ditto.
30031 (_mm512_i64gather_pd): Ditto.
30032 (_mm512_mask_i64gather_pd): Ditto.
30033 (_mm512_i32gather_epi32): Ditto.
30034 (_mm512_mask_i32gather_epi32): Ditto.
30035 (_mm512_i32gather_epi64): Ditto.
30036 (_mm512_mask_i32gather_epi64): Ditto.
30037 (_mm512_i64gather_epi32): Ditto.
30038 (_mm512_mask_i64gather_epi32): Ditto.
30039 (_mm512_i64gather_epi64): Ditto.
30040 (_mm512_mask_i64gather_epi64): Ditto.
30041 (_mm512_i32scatter_ps): Change __addr type to void*.
30042 (_mm512_mask_i32scatter_ps): Ditto.
30043 (_mm512_i32scatter_pd): Ditto.
30044 (_mm512_mask_i32scatter_pd): Ditto.
30045 (_mm512_i64scatter_ps): Ditto.
30046 (_mm512_mask_i64scatter_ps): Ditto.
30047 (_mm512_i64scatter_pd): Ditto.
30048 (_mm512_mask_i64scatter_pd): Ditto.
30049 (_mm512_i32scatter_epi32): Ditto.
30050 (_mm512_mask_i32scatter_epi32): Ditto.
30051 (_mm512_i32scatter_epi64): Ditto.
30052 (_mm512_mask_i32scatter_epi64): Ditto.
30053 (_mm512_i64scatter_epi32): Ditto.
30054 (_mm512_mask_i64scatter_epi32): Ditto.
30055 (_mm512_i64scatter_epi64): Ditto.
30056 (_mm512_mask_i64scatter_epi64): Ditto.
30057 * config/i386/avx512pfintrin.h
30058 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30059 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30060 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30061 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30062 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30063 (_mm512_prefetch_i32scatter_ps): Ditto.
30064 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30065 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30066 (_mm512_prefetch_i64scatter_pd): Ditto.
30067 (_mm512_prefetch_i64scatter_ps): Ditto.
30068 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30069 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30070 * config/i386/avx512vlintrin.h
30071 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30072 (_mm_mmask_i32gather_ps): Ditto.
30073 (_mm256_mmask_i32gather_pd): Ditto.
30074 (_mm_mmask_i32gather_pd): Ditto.
30075 (_mm256_mmask_i64gather_ps): Ditto.
30076 (_mm_mmask_i64gather_ps): Ditto.
30077 (_mm256_mmask_i64gather_pd): Ditto.
30078 (_mm_mmask_i64gather_pd): Ditto.
30079 (_mm256_mmask_i32gather_epi32): Ditto.
30080 (_mm_mmask_i32gather_epi32): Ditto.
30081 (_mm256_mmask_i32gather_epi64): Ditto.
30082 (_mm_mmask_i32gather_epi64): Ditto.
30083 (_mm256_mmask_i64gather_epi32): Ditto.
30084 (_mm_mmask_i64gather_epi32): Ditto.
30085 (_mm256_mmask_i64gather_epi64): Ditto.
30086 (_mm_mmask_i64gather_epi64): Ditto.
30087 (_mm256_i32scatter_ps): Change __addr type to void*.
30088 (_mm256_mask_i32scatter_ps): Ditto.
30089 (_mm_i32scatter_ps): Ditto.
30090 (_mm_mask_i32scatter_ps): Ditto.
30091 (_mm256_i32scatter_pd): Ditto.
30092 (_mm256_mask_i32scatter_pd): Ditto.
30093 (_mm_i32scatter_pd): Ditto.
30094 (_mm_mask_i32scatter_pd): Ditto.
30095 (_mm256_i64scatter_ps): Ditto.
30096 (_mm256_mask_i64scatter_ps): Ditto.
30097 (_mm_i64scatter_ps): Ditto.
30098 (_mm_mask_i64scatter_ps): Ditto.
30099 (_mm256_i64scatter_pd): Ditto.
30100 (_mm256_mask_i64scatter_pd): Ditto.
30101 (_mm_i64scatter_pd): Ditto.
30102 (_mm_mask_i64scatter_pd): Ditto.
30103 (_mm256_i32scatter_epi32): Ditto.
30104 (_mm256_mask_i32scatter_epi32): Ditto.
30105 (_mm_i32scatter_epi32): Ditto.
30106 (_mm_mask_i32scatter_epi32): Ditto.
30107 (_mm256_i32scatter_epi64): Ditto.
30108 (_mm256_mask_i32scatter_epi64): Ditto.
30109 (_mm_i32scatter_epi64): Ditto.
30110 (_mm_mask_i32scatter_epi64): Ditto.
30111 (_mm256_i64scatter_epi32): Ditto.
30112 (_mm256_mask_i64scatter_epi32): Ditto.
30113 (_mm_i64scatter_epi32): Ditto.
30114 (_mm_mask_i64scatter_epi32): Ditto.
30115 (_mm256_i64scatter_epi64): Ditto.
30116 (_mm256_mask_i64scatter_epi64): Ditto.
30117 (_mm_i64scatter_epi64): Ditto.
30118 (_mm_mask_i64scatter_epi64): Ditto.
30119 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30120 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30121 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30122 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30123 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30124 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30125 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30126 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30127 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30128 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30129 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30130 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30131 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30132 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30133 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30134 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30135 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30136 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30137 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30138 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30139 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30140 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30141 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30142 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30143 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30144 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30145 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30146 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30147 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30148 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30149 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30150 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30151 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30152 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30153 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30154 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30155 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30156 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30157 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30158 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30159 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30160 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30161 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30162 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30163 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30164 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30165 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30166 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30167 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30168 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30169 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30170 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30171 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30172 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30173 definitions accordingly.
30175 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30176 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30179 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30182 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30185 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30186 hard regs for inner regno.
30188 2017-01-17 Martin Liska <mliska@suse.cz>
30191 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30192 assumption and add comment.
30194 2017-01-17 Nathan Sidwell <nathan@acm.org>
30196 * ipa-visibility.c (localize_node): New function, broken out of ...
30197 (function_and_variable_visibility): ... here. Call it.
30199 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30201 PR middle-end/77445
30202 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30203 correctly set frequency of oudgoing edge.
30204 (duplicate_thread_path): Fix profile updating.
30206 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30209 * configure.ac: Add GCC_BASE_VER.
30210 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30211 version from BASE-VER file.
30212 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30213 (gcc.o): Depend on $(BASEVER).
30214 * common.opt (dumpfullversion): New option.
30215 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30216 * doc/invoke.texi: Document -dumpfullversion.
30217 * doc/install.texi: Document --with-gcc-major-version-only.
30218 * configure: Regenerated.
30220 2017-01-17 Richard Biener <rguenther@suse.de>
30222 PR tree-optimization/71433
30223 * tree-vrp.c (register_new_assert_for): Merge same asserts
30224 on all incoming edges.
30225 (process_assert_insertions_for): Handle insertions at the
30228 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30230 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30231 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30233 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30236 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30239 2017-01-17 Alan Modra <amodra@gmail.com>
30242 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30243 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30244 symbolic stack limit when pic.
30246 2017-01-16 Martin Sebor <msebor@redhat.com>
30248 PR tree-optimization/78608
30249 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30251 2017-01-16 Jeff Law <law@redhat.com>
30254 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30255 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30256 for several include directories that may be relative to sysroot.
30257 * config/i386/x-mingw32 (gplus_includedir): Define.
30258 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30259 (native_system_includedir): Likewise.
30260 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30261 override if TARGET_SYSTEM_ROOT is defined.
30262 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30264 PR tree-optimization/79090
30265 PR tree-optimization/33562
30266 PR tree-optimization/61912
30267 PR tree-optimization/77485
30268 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30269 and computed trims into the dump file.
30271 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30273 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30275 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30278 * gimplify.c (gimplify_init_constructor): If want_value and
30279 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30283 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30284 sequence. Formatting fixes.
30285 (doloop_optimize): Formatting fixes.
30288 * gcc.c (debug_level_greater_than_spec_func): New function.
30289 (static_spec_functions): Add debug-level-gt spec function.
30290 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30292 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30293 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30294 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30295 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30296 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30297 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30299 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30301 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30302 QImode fixups to general and mask registers only.
30304 2017-01-16 Carl Love <cel@us.ibm.com>
30306 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30307 for built-in functions
30308 vector signed char vec_nabs (vector signed char)
30309 vector signed short vec_nabs (vector signed short)
30310 vector signed int vec_nabs (vector signed int)
30311 vector signed long long vec_nabs (vector signed long long)
30312 vector float vec_nabs (vector float)
30313 vector double vec_nabs (vector double)
30314 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30316 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30317 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30318 * doc/extend.texi: Update the documentation file for the new built-in
30321 2017-01-16 Martin Sebor <msebor@redhat.com>
30323 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30326 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30328 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30329 UNSPEC_VSX__XXSPLTD to require special splat handling.
30331 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30334 * system.h: Poison strndup.
30336 2017-01-16 Alan Modra <amodra@gmail.com>
30339 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30342 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30344 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30346 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30348 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30349 call recog here. Assert that INSN_CODE (insn) is non-negative.
30351 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30354 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30356 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30357 in the currently scheduled RTL fragment.
30359 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30361 PR rtl-optimization/78751
30362 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30365 2017-01-14 Jeff Law <law@redhat.com>
30367 PR tree-optimization/79090
30368 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30369 variable length stores.
30370 (compute_trims): Delete dead assignment to *trim_tail.
30371 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30374 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30376 PR rtl-optimization/78626
30377 PR rtl-optimization/78727
30378 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30379 of a block, and split such blocks after everything else is finished.
30381 2017-01-14 Alan Modra <amodra@gmail.com>
30384 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30385 target legitimate_combined_insn.
30386 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30387 (rs6000_legitimate_combined_insn): New function.
30388 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30390 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30391 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30392 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30394 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30396 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30398 2017-01-13 Jeff Law <law@redhat.com>
30400 PR tree-optimization/33562
30401 PR tree-optimization/61912
30402 PR tree-optimization/77485
30403 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30405 (delete_dead_assignment): Likewise.
30406 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30407 statement to delete_dead_call and delete_dead_assignment.
30409 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30412 * substring-locations.c (format_warning_va): Strengthen case 1 so
30413 that both endpoints of the substring must be within the format
30414 range for just the substring to be printed.
30416 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30418 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30419 * config/i386/i386.c (ix86_target_string): Add missing options
30420 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30421 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30422 flags_other and ix86_target_other to flags2_other. Display unknown
30424 (ix86_valid_target_attribute_inner_p): Add missing options and
30425 reorder options by implied ISAs, as in ix86_target_string.
30427 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30429 * hash-table.h (hash_table::too_empty_p): New function.
30430 (hash_table::expand): Use it.
30431 (hash_table::traverse): Likewise.
30432 (hash_table::empty_slot): Use sizeof (value_type) instead of
30433 sizeof (PTR) to convert bytes to elements. Shrink the table
30434 if the current size is excessive for the current number of
30437 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30439 * ira-costs.c (record_reg_classes): Break from the inner loop
30440 early once alt_fail is known to be true. Update outer loop
30441 handling accordingly.
30443 2017-01-13 Jeff Law <law@redhat.com>
30445 * tree-ssa-dse.c (decrement_count): New function.
30446 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30447 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30448 when we know the partially dead statement is a mem* function.
30450 PR tree-optimization/61912
30451 PR tree-optimization/77485
30452 * tree-ssa-dse.c: Include expr.h.
30453 (maybe_trim_constructor_store): New function.
30454 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30456 PR tree-optimization/33562
30457 PR tree-optimization/61912
30458 PR tree-optimization/77485
30459 * doc/invoke.texi: Document new dse-max-object-size param.
30460 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30461 * tree-ssa-dse.c: Include params.h.
30462 (dse_store_status): New enum.
30463 (initialize_ao_ref_for_dse): New, partially extracted from
30465 (valid_ao_ref_for_dse, normalize_ref): New.
30466 (setup_live_bytes_from_ref, compute_trims): Likewise.
30467 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30468 (maybe_trim_partially_dead_store): Likewise.
30469 (maybe_trim_complex_store): Likewise.
30470 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30471 Track what bytes live from the original store. Return tri-state
30472 for dead, partially dead or live.
30473 (dse_dom_walker): Add constructor, destructor and new private members.
30474 (delete_dead_call, delete_dead_assignment): New extracted from
30476 (dse_optimize_stmt): Make a member of dse_dom_walker.
30477 Use initialize_ao_ref_for_dse.
30479 PR tree-optimization/33562
30480 PR tree-optimization/61912
30481 PR tree-optimization/77485
30482 * sbitmap.h (bitmap_count_bits): Prototype.
30483 (bitmap_clear_range, bitmap_set_range): Likewise.
30484 * sbitmap.c (bitmap_clear_range): New function.
30485 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30487 2017-01-13 Martin Liska <mliska@suse.cz>
30490 * function.c (set_cfun): Add new argument force.
30491 * function.h (set_cfun): Likewise.
30492 * ipa-inline-transform.c (inline_call): Use the function when
30493 strict alising from is dropped for function we inline to.
30495 2017-01-13 Richard Biener <rguenther@suse.de>
30497 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30498 for dumping GIMPLE INTEGER_CSTs.
30500 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30502 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30503 to 201112L since C++17.
30505 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30508 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30509 if -fsanitize=kernel-address is present.
30511 2017-01-13 Richard Biener <rguenther@suse.de>
30513 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30514 as _Literal ( type ) number in case usual suffixes do not
30515 preserve all information.
30517 2017-01-13 Richard Biener <rguenther@suse.de>
30519 PR tree-optimization/77283
30520 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30521 and ssa-iterators.h.
30522 (is_feasible_trace): Implement a cost model based on joiner
30525 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30528 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30529 char or short to __float128/_Float128 directly.
30531 2017-01-12 Martin Sebor <msebor@redhat.com>
30533 to -Wformat-overflow.
30534 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30535 (min_bytes_remaining): Same.
30536 (get_string_length): Same.
30537 (format_string): Same.
30538 (format_directive): Same.
30540 (pass_sprintf_length::handle_gimple_call): Same.
30542 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30544 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30545 info.nowrite calls with no lhs that can't throw. Return bool
30546 whether gsi_remove has been called or not.
30547 (pass_sprintf_length::handle_gimple_call): Return bool whether
30548 try_substitute_return_value called gsi_remove. Formatting fix.
30549 (pass_sprintf_length::execute): Don't use gsi_remove if
30550 handle_gimple_call returned true.
30553 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30554 be removed due to side-effects, don't remove following barrier nor
30555 turn the successor edge into fallthru edge.
30557 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30560 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30561 element-reversing loads and stores as not swappable.
30563 2017-01-12 Nathan Sidwell <nathan@acm.org>
30564 Nicolai Stange <nicstange@gmail.com>
30566 * combine.c (try_combine): Don't ignore result of overlap checking
30567 loop. Combine overlap & asm check into single loop.
30569 2017-01-12 Richard Biener <rguenther@suse.de>
30571 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30572 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30574 2017-01-12 Richard Biener <rguenther@suse.de>
30576 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30577 and TS_TARGET_OPTION directly derive from TS_BASE.
30578 * tree-core.h (tree_optimization_option): Derive from tree_base.
30579 (tree_target_option): Likewise.
30581 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30583 * config/i386/i386.c (memory_address_length): Increase len
30584 only when rip_relative_addr_p returns false.
30586 2017-01-11 Julia Koval <julia.koval@intel.com>
30588 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30589 (OPTION_MASK_ISA_SGX_SET): New.
30590 (ix86_handle_option): Handle OPT_msgx.
30591 * config.gcc: Added sgxintrin.h.
30592 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30593 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30594 * config/i386/i386.c (ix86_target_string): Add -msgx.
30596 (ix86_option_override_internal): Handle new options.
30597 (ix86_valid_target_attribute_inner_p): Add sgx.
30598 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30599 * config/i386/i386.opt: Add msgx.
30600 * config/i386/sgxintrin.h: New file.
30601 * config/i386/x86intrin.h: Add sgxintrin.h.
30603 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30606 * fold-const.c (maybe_nonzero_address): Return 1 for function
30608 (tree_single_nonzero_warnv_p): Don't handle function local objects
30612 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30615 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30618 * opts.c: Include "spellcheck.h"
30619 (struct string_fragment): New struct.
30620 (struct edit_distance_traits<const string_fragment &>): New
30622 (get_closest_sanitizer_option): New function.
30623 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30625 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30627 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30629 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30630 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30631 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30632 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30633 for initial die_offset if dwarf_split_debug_info.
30634 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30635 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30637 (output_skeleton_debug_sections): Formatting fix. Use
30638 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30639 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30641 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30643 * config/arm/cortex-a53.md: Add bypasses for
30644 cortex_a53_r2f_cvt.
30645 (cortex_a53_r2f): Only use for transfers.
30646 (cortex_a53_f2r): Likewise.
30647 (cortex_a53_r2f_cvt): Add reservation for conversions.
30648 (cortex_a53_f2r_cvt): Likewise.
30650 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30652 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30653 to all inlined functions, change static to extern.
30655 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30658 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30660 (arm_assemble_integer): Likewise.
30662 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30664 * config.gcc: Use new awk script to check CPU, FPU and architecture
30665 parameters for --with-... options.
30666 * config/arm/parsecpu.awk: New file
30667 * config/arm/arm-cpus.in: New file.
30668 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30670 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30672 * config/arm/t-arm: Update dependency rules.
30673 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30674 of processing .def files.
30675 * config/arm/genopt.sh: Deleted.
30676 * config/arm/gentune.sh: Deleted.
30677 * config/arm/arm-cores.def: Deleted.
30678 * config/arm/arm-arches.def: Deleted.
30679 * config/arm/arm-fpus.def: Deleted.
30680 * config/arm/arm-tune.md: Regenerated.
30681 * config/arm/arm-tables.opt: Regenerated.
30682 * config/arm/arm-cpu.h: New generated file.
30683 * config/arm/arm-cpu-data.h: New generated file.
30684 * config/arm/arm-cpu-cdata.h: New generated file.
30686 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30689 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30691 (input_varpool_node): Unpack dynamically_initialized bit.
30693 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30695 PR rtl-optimization/79032
30696 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30697 the alignment of the adjusted memory reference against that of MODE,
30698 instead of the alignment of the original memory reference.
30700 2017-01-11 Martin Jambor <mjambor@suse.cz>
30702 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30704 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30705 decorated functions.
30707 2017-01-11 Richard Biener <rguenther@suse.de>
30709 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30710 set range/nonnull info for PHI results. Do not set it on
30711 stmts marked for removal.
30713 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30715 * expr.c (store_field): In the bitfield case, fetch the return value
30716 from the registers before applying a single big-endian adjustment.
30717 Always do a final load for a BLKmode value not larger than a word.
30719 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30722 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30723 that we correctly handle column numbers greater than
30724 LINE_MAP_MAX_COLUMN_NUMBER.
30726 2017-01-10 Martin Sebor <msebor@redhat.com>
30728 PR middle-end/78245
30729 * gimple-ssa-sprintf.c (get_destination_size): Call
30730 {init,fini}object_sizes.
30731 * tree-object-size.c (addr_object_size): Adjust.
30732 (pass_through_call): Adjust.
30733 (pass_object_sizes::execute): Adjust.
30734 * tree-object-size.h (fini_object_sizes): Declare.
30736 2017-01-10 Martin Sebor <msebor@redhat.com>
30738 PR tree-optimization/78775
30739 * builtins.c (get_size_range): Move...
30740 * calls.c: ...to here.
30741 (alloc_max_size): Accept zero argument.
30742 (operand_signed_p): Remove.
30743 (maybe_warn_alloc_args_overflow): Call get_size_range.
30744 * calls.h (get_size_range): Declare.
30746 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30748 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30749 from TI's devices.csv file as of September 2016.
30750 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30752 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30754 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30755 * doc/invoke.texi: Likewise.
30756 * doc/md.texi: Likewise.
30757 * doc/objc.texi: Likewise.
30759 2017-01-10 Joshua Conner <joshconner@google.com>
30761 * config/arm/fuchsia-elf.h: New file.
30762 * config/fuchsia.h: New file.
30763 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30764 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30766 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30768 2016-01-10 Richard Biener <rguenther@suse.de>
30770 PR tree-optimization/79034
30771 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30772 Propagate out degenerate PHIs in the joiner.
30774 2017-01-10 Martin Liska <mliska@suse.cz>
30776 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30777 (sort_congruence_classes_by_decl_uid): Likewise.
30778 (sort_congruence_class_groups_by_decl_uid): Likewise.
30779 (sem_item_optimizer::merge_classes): Sort class, groups in these
30780 classes and members in the groups by DECL_UID of declarations.
30781 This would make merge operations stable.
30783 2017-01-10 Martin Liska <mliska@suse.cz>
30785 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30786 usage of m_classes_vec.
30787 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30788 (sem_item_optimizer::get_group_by_hash): Likewise.
30789 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30790 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30791 (sem_item_optimizer::verify_classes): Likewise.
30792 (sem_item_optimizer::process_cong_reduction): Likewise.
30793 (sem_item_optimizer::dump_cong_classes): Likewise.
30794 (sem_item_optimizer::merge_classes): Likewise.
30795 * ipa-icf.h (congruence_class_hash): Rename from
30796 congruence_class_group_hash. Remove declaration of m_classes_vec.
30798 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30800 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30801 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30802 * config.gcc: Add avx512vpopcntdqintrin.h.
30803 * config/i386/avx512vpopcntdqintrin.h: New.
30804 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30805 * config/i386/i386-builtin-types.def: Add new types.
30806 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30807 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30808 __builtin_ia32_vpopcountq_v8di_mask): New.
30809 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30810 __AVX512VPOPCNTDQ__.
30811 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30812 (PTA_AVX512VPOPCNTDQ): Define.
30813 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30814 TARGET_AVX512VPOPCNTDQ_P): Define.
30815 * config/i386/i386.opt: Add mavx512vpopcntdq.
30816 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30817 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30819 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30821 PR middle-end/77484
30822 * predict.def (PRED_CALL): Set to 67.
30824 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30826 * expr.c (store_field): In the bitfield case, if the value comes from
30827 a function call and is of an aggregate type returned in registers, do
30828 not modify the field mode; extract the value in all cases if the mode
30829 is BLKmode and the size is not larger than a word.
30831 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30834 * config/i386/cpuid.h: Fix undefined behavior.
30836 2017-01-04 Jeff Law <law@redhat.com>
30838 PR tree-optimization/79007
30839 PR tree-optimization/67955
30840 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30841 conservative for pt.null when flag_non_call_exceptions is on.
30843 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30845 PR translation/79019
30846 PR translation/79020
30847 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30848 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30849 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30851 * config/avr/avr.opt (maccumulate-args): Likewise.
30852 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30853 * common.opt (freport-bug): Likewise.
30854 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30855 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30856 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30857 translatable string.
30858 * config/i386/i386.c (function_value_32): Likewise.
30859 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30860 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30862 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30863 * common/config/msp430/msp430-common.c (msp430_handle_option):
30865 * symtab.c (symtab_node::verify_base): Likewise.
30866 * opts.c (set_debug_level): Likewise.
30867 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30868 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30869 missing whitespace to translatable strings.
30870 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30871 * config/sh/superh.h: Likewise.
30872 * config/i386/xopintrin.h: Likewise.
30873 * config/i386/znver1.md: Likewise.
30874 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30875 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30876 * double-int.h (struct double_int): Likewise.
30877 * double-int.c (div_and_round_double): Likewise.
30878 * wide-int.cc: Likewise.
30879 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30880 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30881 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30882 (crossjumps_occurred): ... this.
30883 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30886 PR tree-optimization/78899
30887 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30888 returning bool return struct loop *, NULL for failure and the new
30890 (versionable_outer_loop_p): Don't version outer loop if it has
30891 dont_vectorized bit set.
30892 (tree_if_conversion): When versioning outer loop, ensure
30893 tree_if_conversion is performed also on the inner loop of the
30894 non-vectorizable outer loop copy.
30895 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30896 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30897 prevent vectorization of it.
30898 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30899 the outer loop vectorization of the non-scalar version is attempted
30900 before vectorization of the inner loop in scalar version. If
30901 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30902 vectorization of its inner loop.
30903 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30904 has 2 inner loops, rename also on edges from bb whose single pred
30905 is outer_loop->header. Fix typo in function comment.
30907 2017-01-09 Martin Sebor <msebor@redhat.com>
30910 * asan.c (asan_emit_stack_protection): Increase local buffer size
30911 to avoid snprintf truncation warning.
30913 2017-01-09 Andrew Pinski <apinski@cavium.com>
30915 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30916 to reference thunderx2t99 for the tuning structure
30917 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30919 (thunderx2t99_extra_costs): This.
30920 * config/aarch64/aarch64-tune.md: Regenerate.
30921 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30922 (vulcan_addrcost_table): This.
30923 (vulcan_regmove_cost): Rename to ...
30924 (thunderx2t99_regmove_cost): This.
30925 (vulcan_vector_cost): Rename to ...
30926 (thunderx2t99_vector_cost): this.
30927 (vulcan_branch_cost): Rename to ...
30928 (thunderx2t99_branch_cost): This.
30929 (vulcan_tunings): Rename to ...
30930 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30931 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30933 2017-01-09 Martin Jambor <mjambor@suse.cz>
30937 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30938 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30939 (propagate_vr_accross_jump_function): Use the above function for all
30940 value range computations for pass-through jump functions and type
30941 converasion from explicit value range values.
30942 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30943 parameters from TYPE_ARG_TYPES.
30944 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30945 (ipa_write_node_info): Stream type of the actual argument.
30946 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30948 2017-01-09 Martin Liska <mliska@suse.cz>
30951 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30952 (lookup_compiler): Do not show error message with have_E.
30954 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30956 PR tree-optimization/78938
30957 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30958 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30959 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30960 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30963 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30965 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30968 2017-01-09 Richard Biener <rguenther@suse.de>
30970 PR tree-optimization/78997
30971 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30972 name condition properly.
30974 2017-01-09 Richard Biener <rguenther@suse.de>
30977 * dwarf2out.c (is_cxx): New overload with context.
30978 (is_naming_typedef_decl): Use it.
30980 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30982 * invoke.texi (Option Summary): Correct spacing in option lists
30983 and add line breaks to fix over-long lines.
30985 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30987 PR middle-end/17660
30989 * extend.texi (Common Variable Attributes): Add xref to GCC
30990 Internals manual to explain mode attribute keywords.
30992 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30995 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30996 and Preprocessor Options.
30997 (Options for Linking): Document -pthread here....
30998 (RS/6000 and PowerPC Options): ...not here.
30999 (Solaris 2 Options): ...or here.
31000 * doc/cppopts.texi: Document -pthread.
31002 2017-01-08 Martin Sebor <msebor@redhat.com>
31004 PR middle-end/77708
31005 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31006 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31007 New member functions.
31008 (format_directive): Used them.
31010 (pass_sprintf_length::handle_gimple_call): Same.
31011 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31012 to avoid truncation for any argument.
31013 (extract_affine_mul): Same.
31014 * tree.c (get_file_function_name): Same.
31016 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31018 PR middle-end/77484
31019 * predict.def (PRED_INDIR_CALL): Set to 86.
31021 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31023 PR preprocessor/54124
31024 * doc/cppopts.texi: Reformat -d subtable to list the full name
31025 of the options. Add cross-reference to the docs for the general
31026 compiler -d options.
31027 * doc/invoke.texi (Developer Options): Add cross-reference to the
31028 preprocessor-specific -d option documentation.
31030 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31032 PR preprocessor/13498
31033 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31034 redudant material, and reflect new command-line options.
31035 (System Headers): Likewise.
31037 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31039 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31040 -isystem, and -idirafter. Copy-edit.
31041 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31042 default for -ftrack-macro-expansion. Delete obsolete and
31043 badly-formatted implementation details about -fdebug-cpp output.
31044 * doc/cppwarnopts.texi: Copy-edit.
31046 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31049 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31050 that the transition from a max line width >= 1<<10 to narrower
31051 lines works correctly.
31053 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31055 * doc/options.texi (PerFunction): New.
31056 * opt-functions.awk (switch_flags): Map both Optimization and
31057 PerFunction to CL_OPTIMIZATION.
31058 * opth-gen.awk: Test for PerFunction flag along with
31060 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31061 it only when the latter is present. Skip those that don't in
31062 the hash function generator.
31063 * common.opt (fvar-tracking): Mark as PerFunction instead of
31065 (fvar-tracking-assignments): Likewise.
31066 (fvar-tracking-assignments-toggle): Likewise.
31067 (fvar-tracking-uninit): Likewise.
31069 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31071 PR translation/79018
31072 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31075 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31078 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31079 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31080 TARGET_LONG_JUMP_TABLE_OFFSETS.
31081 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31082 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31083 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31084 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31085 * config/m68k/m68k.md (tablejump expander): Likewise.
31086 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31087 TARGET_LONG_JUMP_TABLE_OFFSETS.
31088 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31089 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31091 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31092 David Holsgrove <david.holsgrove@xilinx.com>
31094 * common/config/microblaze/microblaze-common.c
31095 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31096 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31098 * config/microblaze/microblaze.c (microblaze_must_save_register)
31099 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31101 (microblaze_eh_return): New function.
31102 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31103 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31104 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31105 * config/microblaze/microblaze.md (eh_return): New pattern.
31107 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31109 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31110 GCC_DIAGNOSTIC_STRINGIFY): Define.
31112 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31114 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31116 * config/arm/arm.md (<mcrr>): New.
31118 * config/arm/arm.c (arm_arch5te): New.
31119 (arm_option_override): Set arm_arch5te.
31120 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31122 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31123 (arm_mcrr_qualifiers): ... this. New.
31124 (MRRC_QUALIFIERS): Define to...
31125 (arm_mrrc_qualifiers): ... this. New.
31126 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31128 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31129 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31130 (MRRCI, mrrc, MRRC): New.
31131 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31132 VUNSPEC_MRRC2): New.
31134 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31136 * config/arm/arm.md (<mcr>): New.
31138 * config/arm/arm.c (arm_coproc_builtin_available): Add
31139 support for mcr, mrc, mcr2 and mrc2.
31140 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31141 (arm_mcr_qualifiers): ... this. New.
31142 (MRC_QUALIFIERS): Define to ...
31143 (arm_mrc_qualifiers): ... this. New.
31144 (MCR_QUALIFIERS): Define to ...
31145 (arm_mcr_qualifiers): ... this. New.
31146 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31148 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31149 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31150 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31151 VUNSPEC_MRC2): New.
31153 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31155 * config/arm/arm.md (*ldc): New.
31159 * config/arm/arm.c (arm_coproc_builtin_available): Add
31160 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31161 (arm_coproc_ldc_stc_legitimate_address): New.
31162 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31163 'qualifier_const_pointer'.
31164 (LDC_QUALIFIERS): Define to...
31165 (arm_ldc_qualifiers): ... this. New.
31166 (STC_QUALIFIERS): Define to...
31167 (arm_stc_qualifiers): ... this. New.
31168 * config/arm/arm-protos.h
31169 (arm_coproc_ldc_stc_legitimate_address): New.
31170 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31171 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31172 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31173 stc2, stcl, stc2l): New.
31174 * config/arm/constraints.md (Uz): New.
31175 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31176 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31177 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31178 VUNSPEC_STC2L): New.
31180 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31182 * config/arm/arm.md (<cdp>): New.
31183 * config/arm/arm.c (neon_const_bounds): Rename this ...
31184 (arm_const_bounds): ... this.
31185 (arm_coproc_builtin_available): New.
31186 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31187 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31188 (CDP_QUALIFIERS): Define to...
31189 (arm_cdp_qualifiers): ... this. New.
31191 (arm_expand_builtin_args): Add case for 6 arguments.
31192 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31193 (arm_const_bounds): ... this.
31194 (arm_coproc_builtin_available): New.
31195 * config/arm/arm_acle.h (__arm_cdp): New.
31197 * config/arm/arm_acle_builtins.def (cdp): New.
31199 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31200 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31201 'arm_const_bounds'.
31202 * config/arm/types.md (coproc): New.
31203 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31204 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31205 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31206 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31208 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31210 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31211 (UBINOP_QUALIFIERS): New.
31213 (acle_builtin_data): New. Change comment.
31214 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31215 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31216 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31217 arm_acle_builtins.def.
31218 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31219 (arm_init_acle_builtins): New.
31220 (CRC32_BUILTIN): Remove.
31221 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31222 crc32cb, crc32ch and crc32cw.
31223 (arm_init_crc32_builtins): Remove.
31224 (arm_init_builtins): Use arm_init_acle_builtins rather
31225 than arm_init_crc32_builtins.
31226 (arm_expand_acle_builtin): New.
31227 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31228 * config/arm/arm_acle_builtins.def: New.
31230 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31232 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31233 (arm_builtin_datum): ... this.
31234 (arm_init_neon_builtin): Rename to ...
31235 (arm_init_builtin): ... this. Add a new parameters PREFIX
31236 and USE_SIG_IN_NAME.
31237 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31238 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31239 'arm_builtin_datum'.
31240 (arm_init_vfp_builtins): Likewise.
31241 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31242 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31243 (arm_expand_neon_args): Rename to ...
31244 (arm_expand_builtin_args): ... this. Rename builtin_arg
31245 enum values and differentiate between ARG_BUILTIN_MEMORY
31246 and ARG_BUILTIN_NEON_MEMORY.
31247 (arm_expand_neon_builtin_1): Rename to ...
31248 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31249 values, arm_expand_builtin_args and add bool parameter NEON.
31250 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31251 (arm_expand_vfp_builtin): Likewise.
31252 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31254 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31256 PR middle-end/77484
31257 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31258 * predict.c (tree_estimate_probability_bb): Reverse direction of
31259 polymorphic call predictor.
31261 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31263 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31264 (determine_pass_name_match): ...this new function and...
31265 (should_skip_pass_p): ...this new function.
31267 2017-01-06 Nathan Sidwell <nathan@acm.org>
31269 * ipa-visibility.c (function_and_variable_visibility): Reformat
31270 comments and long lines. Remove extrneous if.
31271 * symtab.c (symtab_node::make_decl_local): Fix code format.
31272 (symtab_node::set_section_for_node): Fix comment typo.
31274 2017-01-06 Martin Liska <mliska@suse.cz>
31277 * lra-constraints.c: Rename invariant to lra_invariant.
31278 * predict.c (set_even_probabilities): Initialize e to NULL.
31280 2017-01-05 Martin Sebor <msebor@redhat.com>
31282 PR tree-optimization/78910
31283 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31284 (format_integer): Correct off-by-one error in the handling
31285 of precision with negative numbers in signed conversions..
31287 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31289 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31291 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31293 PR tree-optimization/71016
31294 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31295 factor_out_conditional_conversion. Formatting fix.
31296 (factor_out_conditional_conversion): Add cond_stmt argument.
31297 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31298 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31301 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31303 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31304 read-rtl-function.o, and selftest-rtl.o.
31305 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31306 (selftest::aarch64_test_loading_full_dump): New function.
31307 (selftest::aarch64_run_selftests): New function.
31308 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31309 selftest::aarch64_run_selftests.
31310 * config/i386/i386.c
31311 (selftest::ix86_test_loading_dump_fragment_1): New function.
31312 (selftest::ix86_test_loading_call_insn): New function.
31313 (selftest::ix86_test_loading_full_dump): New function.
31314 (selftest::ix86_test_loading_unspec): New function.
31315 (selftest::ix86_run_selftests): Call the new functions.
31316 * emit-rtl.c (maybe_set_max_label_num): New function.
31317 * emit-rtl.h (maybe_set_max_label_num): New decl.
31318 * function.c (instantiate_decls): Guard call to
31319 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31320 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31322 * gensupport.c (gen_reader::gen_reader): Pass "false"
31323 for new "compact" param of rtx_reader.
31324 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31325 rather than an empty string for NULL strings.
31326 * read-md.c: Potentially include config.h rather than bconfig.h.
31327 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31328 (have_error): New global, copied from errors.c.
31329 (md_reader::read_name): Rename to...
31330 (md_reader::read_name_1): ...this, adding "out_loc" param,
31331 and converting "missing name or number" to returning false, rather
31333 (md_reader::read_name): Reimplement in terms of read_name_1.
31334 (md_reader::read_name_or_nil): New function.
31335 (md_reader::read_string): Handle "(nil)" by returning NULL.
31336 (md_reader::md_reader): Add new param "compact".
31337 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31338 (md_reader::read_file): New method.
31339 * read-md.h (md_reader::md_reader): Add new param "compact".
31340 (md_reader::read_file): New method.
31341 (md_reader::is_compact): New accessor.
31342 (md_reader::read_name): Convert return type from void to file_location.
31343 (md_reader::read_name_or_nil): New decl.
31344 (md_reader::read_name_1): New decl.
31345 (md_reader::m_compact): New field.
31346 (noop_reader::noop_reader): Pass "false" for new "compact" param
31348 (rtx_reader::rtx_reader): Add new "compact" param.
31349 (rtx_reader::read_rtx_operand): Make virtual and convert return
31350 type from void to rtx.
31351 (rtx_reader::read_until): New decl.
31352 (rtx_reader::handle_any_trailing_information): New virtual function.
31353 (rtx_reader::postprocess): New virtual function.
31354 (rtx_reader::finalize_string): New virtual function.
31355 (rtx_reader::m_in_call_function_usage): New field.
31356 (rtx_reader::m_reuse_rtx_by_id): New field.
31357 * read-rtl-function.c: New file.
31358 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31359 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31360 (selftest::verify_three_block_rtl_cfg): New decl.
31361 * read-rtl-function.h: New file.
31362 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31363 For host, include function.h, memmodel.h, and emit-rtl.h.
31364 (one_time_initialization): New function.
31365 (struct compact_insn_name): New struct.
31366 (compact_insn_names): New array.
31367 (find_code): Handle insn codes in compact dumps.
31368 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31369 (bind_subst_iter_and_attr): Likewise.
31370 (add_condition_to_string): Likewise.
31371 (add_condition_to_rtx): Likewise.
31372 (apply_attribute_uses): Likewise.
31373 (add_current_iterators): Likewise.
31374 (apply_iterators): Likewise.
31375 (initialize_iterators): Guard usage of apply_subst_iterator with
31376 #ifdef GENERATOR_FILE.
31377 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31378 (md_reader::read_mapping): Likewise.
31379 (add_define_attr_for_define_subst): Likewise.
31380 (add_define_subst_attr): Likewise.
31381 (read_subst_mapping): Likewise.
31382 (check_code_iterator): Likewise.
31383 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31385 (one_time_initialization): New function.
31386 (rtx_reader::read_until): New method.
31387 (read_flags): New function.
31388 (parse_reg_note_name): New function.
31389 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31390 Handle reuse_rtx ids.
31391 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31392 Add parsing support for RTL dumps, mirroring the special-cases in
31393 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31394 values, and calling handle_any_trailing_information.
31395 (rtx_reader::read_rtx_operand): Convert return type from void
31396 to rtx, returning return_rtx. Handle case 'e'. Call
31397 finalize_string on XSTR and XTMPL fields.
31398 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31399 "(nil)" values were omitted. Call the postprocess vfunc on the
31401 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31402 class ctor. Initialize m_in_call_function_usage. Call
31403 one_time_initialization.
31404 * rtl-tests.c (selftest::test_uncond_jump): Call
31405 set_new_first_and_last_insn.
31406 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31407 * selftest-rtl.c: New file.
31408 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31409 (selftest::get_insn_by_uid): New decl.
31410 * selftest-run-tests.c (selftest::run_tests): Call
31411 read_rtl_function_c_tests.
31412 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31413 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31416 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31418 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31419 operands in a special way. Assert that pos+len <= mode precision.
31421 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31423 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31424 3 argument Alias with unlimited for the negative form.
31425 (fno-vect-cost-model): Removed.
31427 2017-01-05 Martin Liska <mliska@suse.cz>
31429 * hsa-gen.c (gen_hsa_divmod): New function.
31430 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31432 2017-01-05 Martin Liska <mliska@suse.cz>
31435 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31438 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31440 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31441 small constant length operands.
31443 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31445 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31446 between loop iterations.
31448 2017-01-05 Martin Liska <mliska@suse.cz>
31451 * gimplify.c (gimplify_decl_expr): Compare to
31452 asan_poisoned_variables instread of checking flags.
31453 (gimplify_target_expr): Likewise.
31454 (gimplify_expr): Likewise.
31455 (gimplify_function_tree): Conditionally initialize
31456 asan_poisoned_variables.
31458 2017-01-04 Jeff Law <law@redhat.com>
31460 PR tree-optimizatin/78812
31461 * rtl.h (contains_mem_rtx_p): Prototype.
31462 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31463 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31464 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31465 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31466 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31468 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31470 * input.c (assert_char_at_range): Default-initialize actual_range.
31472 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31474 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31475 to match the caller.
31477 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31479 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31480 insns after final jump in test to emit dummy move.
31482 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31484 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31485 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31487 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31489 * multiple_target.c (create_dispatcher_calls): Init e_next.
31490 * tree-ssa-loop-split.c (split_loop): Init border.
31491 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31494 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31499 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31500 (altivec_register_operand): Do not return true if the operand
31501 contains a SUBREG mixing SImode and SFmode.
31502 (vsx_register_operand): Likewise.
31503 (vsx_reg_sfsubreg_ok): New predicate.
31504 (vfloat_operand): Do not return true if the operand contains a
31505 SUBREG mixing SImode and SFmode.
31506 (vint_operand): Likewise.
31507 (vlogical_operand): Likewise.
31508 (gpc_reg_operand): Likewise.
31509 (int_reg_operand): Likewise.
31510 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31511 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31512 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31514 (rs6000_emit_move_si_sf_subreg): New helper function.
31515 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31516 fixup SUBREGs involving SImode and SFmode.
31517 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31518 numbers for the new peephole2 optimization.
31519 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31520 the GLIBC math library that do AND/IOR/XOR operations on single
31521 precision floating point.
31522 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31523 target macros to say whether we need to avoid SUBREGs mixing
31525 (TARGET_ALLOW_SF_SUBREG): Likewise.
31526 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31527 (UNSPEC_SI_FROM_SF): Likewise.
31528 (iorxor): Change spacing.
31529 (and_ior_xor): New iterator for AND, IOR, and XOR.
31530 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31531 (movdi_from_sf_zero_ext): Likewise.
31532 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31533 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31534 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31535 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31536 (fms<mode>4): Likewise.
31537 (fnma<mode>4): Likewise.
31538 (fnms<mode>4): Likewise.
31539 (nfma<mode>4): Likewise.
31540 (nfms<mode>4): Likewise.
31542 2017-01-04 Marek Polacek <polacek@redhat.com>
31545 * doc/invoke.texi: Document -Wpointer-compare.
31547 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31549 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31552 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31553 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31556 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31559 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31560 documentation of the powerpc_popcntb_ok attribute.
31561 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31562 code to issue warning messages if a requested CPU configuration is
31563 not supported by the binary (assembler and loader) toolchain.
31564 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31565 made to define a built-in function that has been disabled.
31566 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31567 made to define a built-in function that has been disabled.
31568 (altivec_init_builtins): Add comment explaining why definition
31569 of the DST built-in functions is not preceded by an assertion
31570 check. Add assertions to prevent ICE if attempts are made to
31571 define an altivec predicate or an abs* built-in function that has
31573 (htm_init_builtins): Add comment explaining why definition of the
31574 htm built-in functions is not preceded by an assertion check.
31576 2017-01-04 Jeff Law <law@redhat.com>
31578 PR tree-optimizatin/67955
31579 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31580 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31581 the points-to solution does not include pt_null. Use DECL_PT_UID
31584 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31586 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31587 Use gen_int_mode instead of gen_lopwart for const_int operands.
31589 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31591 PR tree-optimization/71563
31592 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31593 out of range value - has low bits known to be zero.
31595 2017-01-04 Alan Modra <amodra@gmail.com>
31597 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31598 * configure: Regenerate.
31599 * config.in: Regenerate.
31601 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31604 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31605 a substring of the message, but strcmp with the whole message. Ifdef
31606 ENABLE_NLS, translate the message first using dgettext.
31608 2017-01-03 Jeff Law <law@redhat.com>
31610 PR tree-optimizatin/78856
31611 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31612 (mark_threaded_blocks): Remove code to truncate thread paths that
31613 cross multiple loop headers. Instead invalidate the cached loop
31614 iteration information and handle case of a thread path walking
31615 into an irreducible region.
31617 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31620 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31621 assertions. Add support for doing the signbit if the IEEE 128-bit
31622 floating point value is in a GPR.
31623 * config/rs6000/rs6000.md (Fsignbit): Delete.
31624 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31625 Update the length attribute if the value is in a GPR.
31626 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31627 the sign or zero extension instruction, since the value is always 0/1.
31628 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31631 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31632 extracting SImode to a GPR register so that we can generate a
31633 store, limit the vector to be in a traditional Altivec register
31634 for the vextuwrx instruction.
31636 2017-01-03 Ian Lance Taylor <iant@google.com>
31638 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31640 2017-01-03 Martin Sebor <msebor@redhat.com>
31642 PR tree-optimization/78696
31643 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31644 precision. Use MPFR for %f for greater fidelity. Correct handling
31646 (pass_sprintf_length::compute_format_length): Set width and precision
31647 specified by asrerisk to void_node for vararg functions.
31648 (try_substitute_return_value): Adjust dump output.
31650 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31652 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31654 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31656 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31657 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31658 -mlra/-mno-lra was passed to the compiler.
31660 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31662 PR rtl-optimization/65618
31663 * emit-rtl.c (try_split): Move initialization of "before" and
31664 "after" to just before the call to emit_insn_after_setloc.
31666 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31668 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31670 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31672 * dwarf2out.c (gen_enumeration_type_die): When
31673 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31675 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31677 PR tree-optimization/78965
31678 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31679 Change first argument from const call_info & to call_info &. For %n
31680 set info.nowrite to false.
31682 PR middle-end/78901
31683 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31684 possibly throwing calls.
31686 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31687 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31688 and fns handling, rather than in a separate case SSA_NAME.
31690 2017-01-02 Jeff Law <law@redhat.com>
31692 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31693 fixes for first_period and second_period variables.
31695 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31698 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31699 (*insvqi_1): New insn pattern.
31700 (*insvqi_1_mem_rex64): Ditto.
31701 (*insvqi_2): Ditto.
31702 (*insvqi_3): Rename from *insvqi.
31704 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31706 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31708 * doc/cfg.texi (Edges): Remove reference to Java.
31709 (Maintaining the CFG): Ditto.
31711 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31713 PR middle-end/77674
31714 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31715 transparent aliases.
31717 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31719 PR middle-end/77484
31720 * predict.def (PRED_CALL): Update hitrate.
31721 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31722 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31723 into direct/indirect/polymorphic variants.
31725 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31727 Update copyright years.
31729 * gcc.c (process_command): Update copyright notice dates.
31730 * gcov-dump.c (print_version): Ditto.
31731 * gcov.c (print_version): Ditto.
31732 * gcov-tool.c (print_version): Ditto.
31733 * gengtype.c (create_file): Ditto.
31734 * doc/cpp.texi: Bump @copying's copyright year.
31735 * doc/cppinternals.texi: Ditto.
31736 * doc/gcc.texi: Ditto.
31737 * doc/gccint.texi: Ditto.
31738 * doc/gcov.texi: Ditto.
31739 * doc/install.texi: Ditto.
31740 * doc/invoke.texi: Ditto.
31742 Copyright (C) 2017 Free Software Foundation, Inc.
31744 Copying and distribution of this file, with or without modification,
31745 are permitted in any medium without royalty provided the copyright
31746 notice and this notice are preserved.