1 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5 Check special value of flag_omit_frame_pointer.
6 (aarch64_can_eliminate): Likewise.
7 (aarch64_override_options_after_change_1): Simplify handling of
8 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10 2017-10-24 Richard Biener <rguenther@suse.de>
12 PR tree-optimization/82697
13 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
14 zero for conditional load and unconditional store.
16 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
18 * doc/install.texi: Document bootstrap-cet.
20 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
23 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
24 ENDBR instruction at function entrance if function is only
27 2017-10-24 Jakub Jelinek <jakub@redhat.com>
30 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
31 patterns to better describe from which operation the CF is computed.
32 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
33 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
34 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
35 is 0, use _0 suffixed expanders instead of emitting a comparison
38 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
40 * config/i386/i386.md(*movsf_internal, *movdf_internal):
41 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
43 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
46 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
47 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
48 * loop-iv.c (iv_get_reaching_def): Likewise.
49 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
50 variable is promoted and the partition contains undefined values.
52 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
54 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
56 (nios2_address_cost): Define.
57 (nios2_legitimize_address): Recognize (exp + constant) directly.
58 (TARGET_ADDRESS_COST): Define.
60 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
62 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
63 (nios2_symbolic_memory_operand_p): Declare.
64 (nios2_split_large_constant): Declare.
65 (nios2_split_symbolic_memory_operand): Declare.
66 * config/nios2/nios2.c: Adjust includes.
67 (nios2_symbolic_constant_allowed): New.
68 (nios2_symbolic_constant_p): New.
69 (nios2_plus_symbolic_constant_p): New.
70 (nios2_valid_addr_expr_p): Recognize addresses involving
72 (nios2_legitimate_address_p): Likewise, also LO_SUM.
73 (nios2_symbolic_memory_operand_p): New.
74 (nios2_large_constant_p): New.
75 (nios2_split_large_constant): New.
76 (nios2_split_plus_large_constant): New.
77 (nios2_split_symbolic_memory_operand): New.
78 (nios2_legitimize_address): Code refactoring. Handle addresses
79 involving symbolic constants.
80 (nios2_emit_move_sequence): Likewise.
81 (nios2_print_operand): Improve error output.
82 (nios2_print_operand_address): Handle LO_SUM.
83 (nios2_cdx_narrow_form_p): Likewise.
84 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
85 operands involving symbolic constants.
86 (movhi_internal, movsi_internal): Likewise.
87 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
88 (extendhisi2, extendqi<mode>2): Likewise.
90 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
92 * tree-pass.h (PROP_rtl_split_insns): Define.
93 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
95 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
97 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
99 2017-10-23 Jakub Jelinek <jakub@redhat.com>
102 * target.def (const_not_ok_for_debug_p): Default to
103 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
104 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
105 * targhooks.c (default_const_not_ok_for_debug_p): New function.
106 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
107 which targetm.const_not_ok_for_debug_p returned true.
108 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
110 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
112 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
113 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
114 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
116 (ix86_const_not_ok_for_debug_p): New function.
117 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
118 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
120 2017-10-23 David Malcolm <dmalcolm@redhat.com>
123 * system.h: Conditionally include "unique-ptr.h" if
124 INCLUDE_UNIQUE_PTR is defined.
125 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
126 of defining INCLUDE_UNIQUE_PTR before including "system.h".
128 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
130 * config/rl78/rl78.md: New define_expand "subdi3".
132 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
135 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
136 DF_REF_INSN if DF_REF_INSN_INFO is false.
138 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
140 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
141 xmm_move instead of sse_move.
142 (sse_store_index): New function.
143 (ix86_register_move_cost): Be more sensible about mismatch stall;
144 model AVX moves correctly; make difference between sse->integer and
146 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
147 moves; make difference between SSE and AVX.
148 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
149 and zmm_move. Increase size of sse load and store tables;
150 add unaligned load and store tables; add ssemmx_to_integer.
151 * x86-tune-costs.h: Update all entries according to real
152 move latencies from Agner Fog's manual and chip documentation.
154 2017-10-23 Jakub Jelinek <jakub@redhat.com>
157 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
158 * config/i386/constraints.md (Wf): New constraint.
159 * config/i386/i386.md (UNSPEC_SBB): New unspec.
160 (cmp<dwi>_doubleword): Removed.
161 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
162 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
163 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
164 expand with cmp<dwi>_doubleword. For LTU and GEU use
165 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
167 * common.opt (gcolumn-info): Enable by default.
168 * doc/invoke.texi (gcolumn-info): Document new default.
170 2017-10-23 Richard Biener <rguenther@suse.de>
172 PR tree-optimization/82672
173 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
174 Fold the stmt if we propagated into it.
176 2017-10-23 Richard Biener <rguenther@suse.de>
178 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
179 (bitmap_remove_expr_from_set): ... this. All callers call this
180 for non-constant values.
181 (bitmap_set_subtract): Rename to...
182 (bitmap_set_subtract_expressions): ... this. Adjust and
184 (bitmap_set_contains_value): Remove superfluous check.
185 (bitmap_set_replace_value): Inline into single caller ...
186 (bitmap_value_replace_in_set): ... here and simplify.
187 (dependent_clean): Merge into ...
188 (clean): ... this using an overload. Adjust.
189 (prune_clobbered_mems): Adjust.
190 (compute_antic_aux): Likewise.
191 (compute_partial_antic_aux): Likewise.
193 2017-10-23 Richard Biener <rguenther@suse.de>
195 PR tree-optimization/82129
197 2017-08-01 Richard Biener <rguenther@suse.de>
199 PR tree-optimization/81181
200 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
201 (compute_antic): ... end of iteration here.
203 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
205 * target.def (starting_frame_offset): New hook.
206 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
207 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
208 * doc/tm.texi.in: Regenerate.
209 * hooks.h (hook_hwi_void_0): Declare.
210 * hooks.c (hook_hwi_void_0): New function.
211 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
212 STARTING_FRAME_OFFSET.
213 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
214 * reload1.c (reload): Likewise.
215 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
216 instead of STARTING_FRAME_OFFSET.
217 * function.c (try_fit_stack_local): Likewise.
218 (assign_stack_local_1): Likewise
219 (instantiate_virtual_regs): Likewise.
220 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
221 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
222 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
223 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
224 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
225 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
226 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
227 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
228 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
229 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
230 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
231 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
232 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
233 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
234 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
235 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
236 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
237 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
238 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
239 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
240 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
241 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
242 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
243 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
244 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
245 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
246 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
247 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
248 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
249 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
250 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
251 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
252 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
253 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
254 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
255 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
256 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
257 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
258 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
259 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
260 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
261 * config/avr/avr.c (avr_starting_frame_offset): Make static and
262 return a HOST_WIDE_INT.
263 (avr_builtin_setjmp_frame_value): Use it instead of
264 STARTING_FRAME_OFFSET.
265 (TARGET_STARTING_FRAME_OFFSET): Redefine.
266 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
267 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
269 (TARGET_STARTING_FRAME_OFFSET): Redefine.
270 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
271 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
272 (TARGET_CONSTANT_ALIGNMENT): Redefine.
273 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
274 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
275 (TARGET_STARTING_FRAME_OFFSET): Redefine.
276 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
277 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
278 (TARGET_STARTING_FRAME_OFFSET): Redefine.
279 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
280 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
282 (TARGET_STARTING_FRAME_OFFSET): Redefine.
283 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
284 * config/mips/mips.c (mips_compute_frame_info): Refer to
285 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
286 (mips_starting_frame_offset): New function.
287 (TARGET_STARTING_FRAME_OFFSET): Redefine.
288 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
289 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
290 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
291 and return a HOST_WIDE_INT.
292 (TARGET_STARTING_FRAME_OFFSET): Redefine.
293 (mmix_initial_elimination_offset): Refer to
294 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
295 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
296 * config/pa/pa.c (pa_starting_frame_offset): New function.
297 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
298 (pa_expand_prologue): Likewise.
299 (TARGET_STARTING_FRAME_OFFSET): Redefine.
300 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
301 !FRAME_GROWS_DOWNWARD handling to...
302 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
303 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
304 !FRAME_GROWS_DOWNWARD handling to...
305 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
306 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
307 !FRAME_GROWS_DOWNWARD handling to...
308 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
309 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
311 (rs6000_starting_frame_offset): New function.
312 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
313 !FRAME_GROWS_DOWNWARD handling to...
314 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
315 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
316 !FRAME_GROWS_DOWNWARD handling to...
317 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
318 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
319 !FRAME_GROWS_DOWNWARD handling to...
320 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
321 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
322 (rs6000_starting_frame_offset): New function.
323 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
324 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
325 * config/vax/vax.c (vax_starting_frame_offset): New function.
326 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
327 (TARGET_STARTING_FRAME_OFFSET): Redefine.
328 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
329 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
330 (TARGET_STARTING_FRAME_OFFSET): Redefine.
331 * system.h (STARTING_FRAME_OFFSET): Poison.
333 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
335 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
336 SCALAR_TYPE_MODE instead of TYPE_MODE.
338 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
344 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
345 Alan Hayward <alan.hayward@arm.com>
346 David Sherwood <david.sherwood@arm.com>
348 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
350 2017-10-23 Richard Biener <rguenther@suse.de>
352 PR tree-optimization/82129
353 * tree-ssa-pre.c (bitmap_set_and): Remove.
354 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
355 canonicalizing expressions in the set to those with lowest
356 ID rather than taking that from the first edge.
358 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
360 * combine.c (rtx_equal_for_field_assignment_p): Use
363 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
364 Alan Hayward <alan.hayward@arm.com>
365 David Sherwood <david.sherwood@arm.com>
367 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
368 to a SUBREG_PROMOTED_VAR.
370 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
371 Alan Hayward <alan.hayward@arm.com>
372 David Sherwood <david.sherwood@arm.com>
374 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
375 (expand_debug_source_expr): Likewise.
376 * combine.c (combine_simplify_rtx): Likewise.
377 * cse.c (fold_rtx): Likewise.
378 * optabs.c (expand_float): Likewise.
379 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
380 (simplify_binary_operation_1): Likewise.
382 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
383 Alan Hayward <alan.hayward@arm.com>
384 David Sherwood <david.sherwood@arm.com>
386 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
387 (record_promoted_value): Likewise.
388 * expr.c (expand_expr_real_2): Likewise.
389 * ree.c (update_reg_equal_equiv_notes): Likewise.
390 (combine_set_extension): Likewise.
391 * rtlanal.c (low_bitmask_len): Likewise.
392 * simplify-rtx.c (neg_const_int): Likewise.
393 (simplify_binary_operation_1): Likewise.
395 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
396 Alan Hayward <alan.hayward@arm.com>
397 David Sherwood <david.sherwood@arm.com>
399 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
400 * regcprop.c (maybe_mode_change): Likewise.
401 * reload1.c (alter_reg): Likewise.
403 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
405 * inchash.h (inchash::hash::add_wide_int): New function.
406 * lto-streamer-out.c (hash_tree): Use it.
408 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
410 * inchash.h (inchash::hash::add_wide_int): Rename to...
411 (inchash::hash::add_hwi): ...this.
412 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
413 (polymorphic_call_target_hasher::hash): Likewise.
414 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
415 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
416 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
417 * lto-streamer-out.c (hash_tree): Likewise.
418 * optc-save-gen.awk: Likewise.
419 * tree.c (add_expr): Likewise.
421 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
424 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
425 for ordered inequality comparisons even with TARGET_IEEE_FP.
427 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
430 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
431 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
432 Expand with cmp<dwi>_doubleword.
434 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
436 * extend.texi: Add x86 specific to 'nocf_check' attribute.
438 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
439 specific to -fcf-protection option.
441 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
443 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
444 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
445 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
446 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
447 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
448 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
449 (extra_objs): Add cet.o for Linux/x86 targets.
450 (tmake_file): Add i386/t-cet for Linux/x86 targets.
451 * config/i386/cet.c: New file.
452 * config/i386/cetintrin.h: Likewise.
453 * config/i386/t-cet: Likewise.
454 * config/i386/cpuid.h (bit_SHSTK): New.
456 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
457 pass IBT and SHSTK bits.
458 * config/i386/i386-builtin-types.def
459 (VOID_FTYPE_UNSIGNED_PVOID): New.
460 (VOID_FTYPE_UINT64_PVOID): Likewise.
461 * config/i386/i386-builtin.def: Add CET intrinsics.
462 * config/i386/i386-c.c (ix86_target_macros_internal): Add
463 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
464 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
465 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
467 * config/i386/i386.c (rest_of_insert_endbranch): New.
468 (pass_data_insert_endbranch): Likewise.
469 (pass_insert_endbranch): Likewise.
470 (make_pass_insert_endbranch): Likewise.
471 (ix86_notrack_prefixed_insn_p): Likewise.
472 (ix86_target_string): Add -mibt, -mshstk flags.
473 (ix86_option_override_internal): Add flag_cf_protection
475 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
476 (ix86_print_operand): Add 'notrack' prefix output.
477 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
478 (ix86_expand_builtin): Expand CET intrinsics.
479 (x86_output_mi_thunk): Add 'endbranch' instruction.
480 * config/i386/i386.h (TARGET_IBT): New.
481 (TARGET_IBT_P): Likewise.
482 (TARGET_SHSTK): Likewise.
483 (TARGET_SHSTK_P): Likewise.
484 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
485 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
486 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
487 (builtin_setjmp_setup): New pattern.
488 (builtin_longjmp): Likewise.
489 (rdssp<mode>): Likewise.
490 (incssp<mode>): Likewise.
491 (saveprevssp): Likewise.
492 (rstorssp): Likewise.
493 (wrss<mode>): Likewise.
494 (wruss<mode>): Likewise.
495 (setssbsy): Likewise.
496 (clrssbsy): Likewise.
497 (nop_endbr): Likewise.
498 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
500 * config/i386/immintrin.h: Include <cetintrin.h>.
501 * config/i386/linux-common.h
502 (file_end_indicate_exec_stack_and_cet): New prototype.
503 (TARGET_ASM_FILE_END): New.
505 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
507 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
508 latencies instead of having separate table; make difference between
509 integer and float costs.
510 * i386.h (processor_costs): Remove scalar_stmt_cost,
511 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
512 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
514 * x86-tune-costs.h: Remove entries which has been removed in
515 procesor_costs from all tables; make cond_taken_branch_cost
516 and cond_not_taken_branch_cost COST_N_INSNS based.
518 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
520 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
522 2017-10-20 Jakub Jelinek <jakub@redhat.com>
524 * config/i386/i386.md (isa): Remove fma_avx512f.
525 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
526 <avx512>_fmadd_<mode>_mask3<round_name>,
527 <avx512>_fmsub_<mode>_mask<round_name>,
528 <avx512>_fmsub_<mode>_mask3<round_name>,
529 <avx512>_fnmadd_<mode>_mask<round_name>,
530 <avx512>_fnmadd_<mode>_mask3<round_name>,
531 <avx512>_fnmsub_<mode>_mask<round_name>,
532 <avx512>_fnmsub_<mode>_mask3<round_name>,
533 <avx512>_fmaddsub_<mode>_mask<round_name>,
534 <avx512>_fmaddsub_<mode>_mask3<round_name>,
535 <avx512>_fmsubadd_<mode>_mask<round_name>,
536 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
537 (*vec_widen_umult_even_v16si<mask_name>,
538 *vec_widen_smult_even_v16si<mask_name>): Likewise.
539 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
541 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
543 * extend.texi: Add 'nocf_check' documentation.
544 * gimple.texi: Add second parameter to
545 gimple_build_call_from_tree.
546 * invoke.texi: Add -fcf-protection documentation.
547 * rtl.texi: Add REG_CALL_NOTRACK documenation.
549 2017-10-20 Richard Biener <rguenther@suse.de>
551 PR tree-optimization/82473
552 * tree-vect-loop.c (vectorizable_reduction): Properly get at
553 the largest input type.
555 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
557 * c-attribs.c (handle_nocf_check_attribute): New function.
558 (c_common_attribute_table): Add 'nocf_check' handling.
559 * gimple-parser.c: Add second argument NULL to
560 gimple_build_call_from_tree.
561 * attrib.c (comp_type_attributes): Check nocf_check attribute.
562 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
564 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
565 * common.opt: Add fcf-protection flag.
566 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
567 * flag-types.h: Add enum cf_protection_level.
568 * gimple.c (gimple_build_call_from_tree): Add second parameter.
569 Add 'nocf_check' attribute propagation to gimple call.
570 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
571 (gimple_build_call_from_tree): Update prototype.
572 (gimple_call_nocf_check_p): New function.
573 (gimple_call_set_nocf_check): Likewise.
574 * gimplify.c: Add second argument to gimple_build_call_from_tree.
575 * ipa-icf.c: Add nocf_check attribute in statement hash.
576 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
577 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
578 * toplev.c (process_options): Add flag_cf_protection handling.
580 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
582 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
584 2017-10-20 Richard Biener <rguenther@suse.de>
586 PR tree-optimization/82603
587 * tree-if-conv.c (predicate_mem_writes): Make sure to only
588 remove false predicated stores.
590 2017-10-20 Richard Biener <rguenther@suse.de>
592 * graphite-isl-ast-to-gimple.c
593 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
594 Remove return value and simplify, dump copied stmt after lhs
596 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
597 Reduce dump verbosity.
598 (gsi_insert_earliest): Likewise.
599 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
600 * graphite.c (print_global_statistics): Adjust dumping.
601 (print_graphite_scop_statistics): Likewise.
602 (print_graphite_statistics): Do not dump loops here.
603 (graphite_transform_loops): But here.
605 2017-10-20 Nicolas Roche <roche@adacore.com>
607 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
608 * configure: Regenerate.
610 2017-10-20 Jakub Jelinek <jakub@redhat.com>
613 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
614 functions when optimizing replace GIMPLE_RETURN stmts with
615 calls to __builtin_unreachable ().
618 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
619 for -fsanitize=thread link of executables.
620 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
624 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
626 (<shift_insn><mode>3<mask_name>): Change the last of the 3
627 define_insns for logical vector shifts to use VI248_AVX512BW
628 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
629 condition, useless isa and prefix attributes. Change the first
630 2 of these define_insns to ...
631 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
632 define_insn for avx512vl.
633 (<shift_insn><mode>3): ... and this, new define_insn without
634 masking for non-avx512vl.
637 * config/i386/sse.md (*andnot<mode>3,
638 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
639 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
640 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
641 not applied use empty suffix even for TARGET_AVX512VL.
642 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
643 is applied, supply evex,evex or evex,evex,evex instead of just
646 2017-10-20 Julia Koval <julia.koval@intel.com>
648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
649 (OPTION_MASK_ISA_GFNI_UNSET): New.
650 (ix86_handle_option): Handle OPT_mgfni.
651 * config/i386/cpuid.h (bit_GFNI): New.
652 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
653 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
654 * config/i386/i386.c (ix86_target_string): Add -mgfni.
655 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
656 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
657 * config/i386/i386.opt: Add mgfni.
659 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
661 * config/msp430/msp430.c (msp430_option_override): Disable
662 -fdelete-null-pointer-checks.
663 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
665 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
667 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
668 of x87 and SSE instructions.
670 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
672 * asan.c (create_cond_insert_point): Do not update edge count.
673 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
674 (afdo_propagate_circuit): Likewise.
675 (afdo_calculate_branch_prob): Likewise.
676 (afdo_annotate_cfg): Likewise.
677 * basic-block.h (struct edge_def): Remove count.
678 (edge_def::count): New accessor.
679 * bb-reorder.c (rotate_loop): Update.
680 (find_traces_1_round): Update.
681 (connect_traces): Update.
682 (sanitize_hot_paths): Update.
683 * cfg.c (unchecked_make_edge): Update.
684 (make_single_succ_edge): Update.
685 (check_bb_profile): Update.
686 (dump_edge_info): Update.
687 (update_bb_profile_for_threading): Update.
688 (scale_bbs_frequencies_int): Update.
689 (scale_bbs_frequencies_gcov_type): Update.
690 (scale_bbs_frequencies_profile_count): Update.
691 (scale_bbs_frequencies): Update.
692 * cfganal.c (connect_infinite_loops_to_exit): Update.
693 * cfgbuild.c (compute_outgoing_frequencies): Update.
694 (find_many_sub_basic_blocks): Update.
695 * cfgcleanup.c (try_forward_edges): Update.
696 (try_crossjump_to_edge): Update
697 * cfgexpand.c (expand_gimple_cond): Update
698 (expand_gimple_tailcall): Update
699 (construct_exit_block): Update
700 * cfghooks.c (verify_flow_info): Update
701 (redirect_edge_succ_nodup): Update
703 (make_forwarder_block): Update
704 (duplicate_block): Update
705 (account_profile_record): Update
706 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
707 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
708 * cfgloopmanip.c (scale_loop_profile): Update.
710 (lv_adjust_loop_entry_edge): Update.
711 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
712 (force_nonfallthru_and_redirect): Update.
713 (purge_dead_edges): Update.
714 (rtl_flow_call_edges_add): Update.
715 * cgraphunit.c (init_lowered_empty_function): Update.
716 (cgraph_node::expand_thunk): Update.
717 * gimple-pretty-print.c (dump_probability): Update.
718 (dump_edge_probability): Update.
719 * gimple-ssa-isolate-paths.c (isolate_path): Update.
720 * haifa-sched.c (sched_create_recovery_edges): Update.
721 * hsa-gen.c (convert_switch_statements): Update.
722 * ifcvt.c (dead_or_predicable): Update.
723 * ipa-inline-transform.c (inline_transform): Update.
724 * ipa-split.c (split_function): Update.
725 * ipa-utils.c (ipa_merge_profiles): Update.
726 * loop-doloop.c (add_test): Update.
727 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
728 * lto-streamer-in.c (input_cfg): Update.
729 (input_function): Update.
730 * lto-streamer-out.c (output_cfg): Update.
731 * modulo-sched.c (sms_schedule): Update.
732 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
733 * predict.c (maybe_hot_edge_p): Update.
734 (unlikely_executed_edge_p): Update.
735 (probably_never_executed_edge_p): Update.
736 (dump_prediction): Update.
737 (drop_profile): Update.
738 (propagate_unlikely_bbs_forward): Update.
739 (determine_unlikely_bbs): Update.
740 (force_edge_cold): Update.
741 * profile.c (compute_branch_probabilities): Update.
742 * reg-stack.c (better_edge): Update.
743 * shrink-wrap.c (handle_simple_exit): Update.
744 * tracer.c (better_p): Update.
745 * trans-mem.c (expand_transaction): Update.
746 (split_bb_make_tm_edge): Update.
747 * tree-call-cdce.c: Update.
748 * tree-cfg.c (gimple_find_sub_bbs): Update.
749 (gimple_split_edge): Update.
750 (gimple_duplicate_sese_region): Update.
751 (gimple_duplicate_sese_tail): Update.
752 (gimple_flow_call_edges_add): Update.
753 (insert_cond_bb): Update.
754 (execute_fixup_cfg): Update.
755 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
756 * tree-complex.c (expand_complex_div_wide): Update.
757 * tree-eh.c (lower_resx): Update.
758 (unsplit_eh): Update.
759 (cleanup_empty_eh_move_lp): Update.
760 * tree-inline.c (copy_edges_for_bb): Update.
761 (freqs_to_counts): Update.
762 (copy_cfg_body): Update.
763 * tree-ssa-dce.c (remove_dead_stmt): Update.
764 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
765 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
766 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
767 (unloop_loops): Update.
768 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
769 * tree-ssa-loop-split.c (connect_loops): Update.
770 (split_loop): Update.
771 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
772 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
773 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
774 * tree-ssa-reassoc.c (branch_fixup): Update.
775 * tree-ssa-tail-merge.c (replace_block_by): Update.
776 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
777 (compute_path_counts): Update.
778 (update_profile): Update.
779 (recompute_probabilities): Update.
780 (update_joiner_offpath_counts): Update.
781 (estimated_freqs_path): Update.
782 (freqs_to_counts_path): Update.
783 (clear_counts_path): Update.
784 (ssa_fix_duplicate_block_edges): Update.
785 (duplicate_thread_path): Update.
786 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
787 (case_bit_test_cmp): Update.
788 (collect_switch_conv_info): Update.
789 (gen_inbound_check): Update.
790 (do_jump_if_equal): Update.
791 (emit_cmp_and_jump_insns): Update.
792 * tree-tailcall.c (decrease_profile): Update.
793 (eliminate_tail_call): Update.
794 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
795 (vect_do_peeling): Update.
796 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
797 * ubsan.c (ubsan_expand_null_ifn): Update.
798 (ubsan_expand_ptr_ifn): Update.
799 * value-prof.c (gimple_divmod_fixed_value): Update.
800 (gimple_mod_pow2): Update.
801 (gimple_mod_subtract): Update.
803 (gimple_stringop_fixed_value): Update.
805 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
808 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
810 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
812 PR rtl-optimization/82395
813 * ira-color.c (allocno_priority_compare_func): Fix comparison step
814 based on non_spilled_static_chain_regno_p.
816 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
818 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
819 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
820 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
822 2017-10-19 Martin Sebor <msebor@redhat.com>
824 PR tree-optimization/82596
825 * tree.c (array_at_struct_end_p): Handle STRING_CST.
827 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
829 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
830 (get_mem_refs_of_builtin_call): Likewise.
831 * builtins.c (expand_builtin_apply): Adjust call to
832 allocate_dynamic_stack_space.
833 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
834 the third argument to allocate_dynamic_stack_space, otherwise -1.
835 (expand_builtin): Deal with all alloca variants.
836 (is_inexpensive_builtin): Likewise.
837 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
838 * calls.c (special_function_p): Deal with all alloca variants.
839 (initialize_argument_information): Adjust call to
840 allocate_dynamic_stack_space.
841 (expand_call): Likewise.
842 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
843 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
844 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
845 use it for the stack usage computation.
846 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
847 * function.c (gimplify_parameters): Call build_alloca_call_expr.
848 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
849 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
850 (in_loop_p): Remove first argument and useless check.
851 (pass_walloca::execute): Remove useless test and adjust call to above.
852 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
853 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
854 (gimplify_call_expr): Deal with all alloca variants.
855 * hsa-gen.c (gen_hsa_alloca): Likewise.
856 (gen_hsa_insns_for_call): Likewise.
857 * ipa-pure-const.c (special_builtin_state): Likewise.
858 * tree-chkp.c (chkp_build_returned_bound): Likewise.
859 * tree-object-size.c (alloc_object_size): Likewise.
860 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
861 (call_may_clobber_ref_p_1): Likewise.
862 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
863 (ccp_fold_stmt): Likewise.
864 (optimize_stack_restore): Likewise.
865 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
866 (mark_all_reaching_defs_necessary_1): Likewise.
867 (propagate_necessity): Likewise.
868 (eliminate_unnecessary_stmts): Likewise.
869 * tree.c (build_common_builtin_nodes): Build
870 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
871 (build_alloca_call_expr): New function.
872 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
873 (CASE_BUILT_IN_ALLOCA): Likewise.
874 (build_alloca_call_expr): Declare.
875 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
877 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
880 * dwarf2out.c (new_die_raw): New static inline function.
881 (new_die): Use it to create the DIE.
882 (add_AT_external_die_ref): Likewise.
883 (clone_die): Likewise.
884 (clone_as_declaration): Likewise.
885 (dwarf2out_vms_debug_main_pointer): Likewise.
886 (base_type_die): Likewise. Remove early return for corner cases.
887 Do not call add_pubtype on the DIE here.
888 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
889 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
890 typedefs for base types with DW_AT_endianity. Make sure a DIE with
891 native order exists for base types, attach the DIE manually and call
892 add_pubtype on it. Do not equate a reverse order DIE to the type.
894 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
896 * config/arm/arm.c (align_ok_ldrd_strd): New function.
897 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
899 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
901 2017-10-19 Jakub Jelinek <jakub@redhat.com>
903 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
904 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
905 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
906 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
907 * opts.c (sanitizer_opts): Add builtin.
908 * ubsan.c (instrument_builtin): New function.
909 (pass_ubsan::execute): Call it.
910 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
911 * doc/invoke.texi: Document -fsanitize=builtin.
913 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
914 builtins, store max (log2 (align), 0) into uchar field instead of
915 align into uptr field.
916 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
917 store uchar 0 field instead of uptr 0 field.
918 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
919 instead of passing one address of struct with 2 locations pass
920 two addresses of structs with 1 location each.
921 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
922 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
923 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
924 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
925 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
926 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
927 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
928 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
930 2017-10-19 Martin Liska <mliska@suse.cz>
933 * file-find.c (remove_prefix): Remove.
934 * file-find.h (remove_prefix): Likewise.
935 * gcc-ar.c: Remove smartness of lookup.
937 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
939 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
940 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
941 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
943 2017-10-19 Jakub Jelinek <jakub@redhat.com>
946 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
947 (setcc + and to xor + setcc): New peephole2.
949 2017-10-19 Tom de Vries <tom@codesourcery.com>
951 * doc/sourcebuild.texi (Test Directives, Variants of
952 dg-require-support): Add dg-require-stack-size.
954 2017-10-19 Martin Liska <mliska@suse.cz>
957 * gimplify.c (gimplify_decl_expr): Do not instrument variables
958 that have a large alignment.
959 (gimplify_target_expr): Likewise.
961 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
963 PR rtl-optimization/82602
964 * ira.c (rtx_moveable_p): Return false for volatile asm.
966 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
969 * config/i386/i386-modes.def (CCGZ): New CC mode.
970 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
971 * config/i386/predicates.md (ix86_comparison_operator):
973 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
974 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
975 with double-word subtraction.
976 (put_condition_code): Handle CCGZmode.
978 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
980 * wide-int.cc (debug (const wide_int &)): New.
981 (debug (const wide_int *)): New.
982 (debug (const widest_int &)): New.
983 (debug (const widest_int *)): New.
985 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
988 * lra-constraints.c (curr_insn_transform): Use non-input operand
989 instead of output one for matched reload.
991 2017-10-18 Bin Cheng <bin.cheng@arm.com>
993 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
994 (tree-ssa-loop-ivopts.h): New header file.
995 (struct builtin_info): New fields.
996 (classify_builtin_1): Compute and record base and offset parts for
997 memset builtin partition by calling strip_offset.
998 (offset_cmp, fuse_memset_builtins): New functions.
999 (finalize_partitions): Fuse adjacent memset partitions by calling
1001 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1002 Expose the interface.
1003 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1005 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1007 PR tree-optimization/82574
1008 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1009 that data reference must be executed exactly once per iteration
1010 against the outermost loop in nest.
1011 (classify_partition): Update call to above function.
1013 2017-10-18 Richard Biener <rguenther@suse.de>
1015 PR tree-optimization/82591
1016 * graphite.c (graphite_transform_loops): Move code gen message
1018 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1019 Here. Handle scop_to_isl_ast failing.
1020 (scop_to_isl_ast): Limit the number of ISL operations.
1022 2017-10-18 Richard Biener <rguenther@suse.de>
1024 * graphite-isl-ast-to-gimple.c
1025 (translate_isl_ast_to_gimple::set_rename): Simplify.
1026 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1027 (graphite_copy_stmts_from_block): ... here.
1028 (copy_bb_and_scalar_dependences): Simplify.
1029 (add_parameters_to_ivs_params): Canonicalize.
1030 (generate_entry_out_of_ssa_copies): Simplify.
1031 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1032 by passing in ISL dimension.
1033 (parameter_index_in_region_1): Rename to ...
1034 (parameter_index_in_region): ... this.
1035 (extract_affine): Adjust assert, pass down parameter index.
1036 (add_param_constraints): Use range-info when available.
1037 (build_scop_context): Adjust.
1038 * sese.c (new_sese_info): Adjust.
1039 (free_sese_info): Likewise.
1040 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1041 Remove unused typedefs.
1042 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1044 2017-10-18 Martin Liska <mliska@suse.cz>
1046 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1048 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1050 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1051 (s390_sched_init): Do not reset s390_sched_state if we entered the
1052 current basic block via a fallthru edge and all others are unlikely.
1054 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1056 * config/s390/s390.c (NUM_SIDES): New variable.
1057 (LONGRUNNING_THRESHOLD): New variable.
1058 (LATENCY_FACTOR): New variable.
1059 (s390_sched_score): Decrease score for long-running instructions on
1061 (s390_sched_variable_issue): Perform bookkeeping for long-running
1064 2017-10-18 Richard Biener <rguenther@suse.de>
1066 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1067 Simplify with removal of the parameter rename map.
1068 (set_rename): Likewise.
1069 (should_copy_to_new_region): Likewise.
1070 (graphite_copy_stmts_from_block): Likewise.
1071 (copy_bb_and_scalar_dependences): Remove initialization of
1072 unused copied_bb_map.
1074 (copy_internal_parameters): Likewise.
1075 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1076 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1077 Use INTEGRAL_TYPE_P.
1078 (parameter_index_in_region_1): Rename to ...
1079 (assign_parameter_index_in_region): ... this. Assert we have
1080 a parameter we handle.
1081 (scan_tree_for_params): Adjust.
1082 * sese.h (parameter_rename_map_t): Remove.
1083 (struct sese_info_t): Remove unused parameter_rename_map and
1084 copied_bb_map members.
1085 * sese.c (new_sese_info): Adjust.
1086 (free_sese_info): Likewise.
1088 2017-10-18 Martin Liska <mliska@suse.cz>
1091 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1092 on an abnormal edge.
1094 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1096 * doc/invoke.texi (ffunction-sections and fdata-sections):
1099 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1101 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1102 the use statement can throw internally.
1104 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1106 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1107 any RTX present on the RHS of a SET.
1108 * compare-elim.c (try_eliminate_compare): Restore comment.
1110 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1112 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1113 may be also called on tcc_type nodes.
1114 * langhooks.c (lhd_tree_size): Likewise.
1116 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1118 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1119 format_warning_at_substring.
1120 (maybe_warn): Convert source_range * param to a location_t. Pass
1121 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1122 (format_directive): Remove code to extract source_ranges and
1123 source_range * in favor of just a location_t.
1124 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1126 * substring-locations.c (format_warning_va): Convert
1127 source_range * param to a location_t.
1128 (format_warning_at_substring): Likewise.
1129 * substring-locations.h (format_warning_va): Likewise.
1130 (format_warning_at_substring): Likewise.
1132 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1134 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1136 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1137 and scatter/gather ops.
1139 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1140 vec_gather_load and vec_scatter_store.
1141 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1142 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1143 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1144 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1145 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1146 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1148 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1150 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1151 Detect FTST instruction and handle its register pops. Only pop
1152 second operand if can_pop_second_op is true.
1153 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1154 set can_pop_second_op to false in the compare_for_stack_reg call.
1156 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1157 output_fp_compare for stack register operands.
1158 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1159 instructions here. Do not emit stack register pops here. Assert
1160 that FCOMPP pops next to top stack register. Rewrite function.
1162 2017-10-17 Nathan Sidwell <nathan@acm.org>
1165 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1166 use DECL_ASSEMBLER_NAME_RAW.
1169 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1172 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1173 Wilco Dijkstra <wilco.dijkstra@arm.com>
1175 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1176 converstion to Pmode from the buf_addr.
1178 2017-10-17 Richard Biener <rguenther@suse.de>
1180 * graphite-dependences.c (scop_get_reads_and_writes): Change
1181 output parameters to references.
1183 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1185 PR 71026/tree-optimization
1186 * fold-const.c (distribute_real_division): Removed.
1187 (fold_binary_loc): Remove calls to distribute_real_divison.
1189 2017-10-17 Richard Biener <rguenther@suse.de>
1191 * graphite-scop-detection.c
1192 (scop_detection::stmt_has_simple_data_refs_p): Always use
1193 the full nest as region.
1194 (try_generate_gimple_bb): Likewise.
1195 * sese.c (scalar_evolution_in_region): Simplify now that
1196 SCEV can handle instantiation in regions.
1197 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1198 in the non-loop part of a function if requested.
1200 2017-10-17 Richard Biener <rguenther@suse.de>
1202 PR tree-optimization/82563
1203 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1205 (graphite_regenerate_ast_isl): Call it.
1206 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1208 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1210 PR tree-optimization/82549
1211 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1212 Formatting fixes. Instead of calling make_bit_field_ref with negative
1215 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1217 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1218 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1219 _mm_maskz_reduce_ss): New.
1220 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1221 __builtin_ia32_reducess_mask): Ditto..
1222 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1223 * config/i386/sse.md (reduces<mode>): Renamed to ...
1224 (reduces<mode><mask_scalar_name>): ... this.
1225 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1227 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1228 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1230 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1232 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1233 * selftest-run-tests.c (selftest::run_tests): Call
1234 selftest::unique_ptr_tests_cc_tests.
1235 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1236 * unique-ptr-tests.cc: New file.
1238 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1241 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1243 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1244 (make_hard_regno_born, make_hard_regno_dead): Update
1245 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1247 2017-10-16 Jeff Law <law@redhat.com>
1249 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1251 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1253 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1255 2017-10-16 Olivier Hainque <hainque@adacore.com>
1257 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1258 with_cpu if we were configured for an e500v2 target cpu name.
1260 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1262 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1263 * doc/invoke.texi: Document +nodsp as a valid extension for
1266 2017-10-16 Martin Liska <mliska@suse.cz>
1268 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1269 (test_set_range): Likewise.
1270 (test_range_functions): Rename to ...
1271 (test_bit_in_range): ... this.
1272 (sbitmap_c_tests): Add new test.
1274 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1276 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1277 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1278 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1280 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1282 * config/aarch64/aarch64-builtins.c
1283 (aarch64_types_quadopu_lane_qualifiers): New.
1284 (TYPES_QUADOPU_LANE): New.
1285 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1286 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1287 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1288 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1289 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1290 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1291 (Vdottype, DOTPROD): New.
1292 (sur): Add SDOT and UDOT.
1294 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1296 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1297 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1298 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1299 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1300 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1301 (cortex-a75.cortex-a55): Likewise.
1302 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1304 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1306 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1307 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1308 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1309 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1310 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1311 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1312 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1313 * config/arm/types.md (neon_dot, neon_dot_q): New.
1314 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1316 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1318 * config/arm/arm.h (TARGET_DOTPROD): New.
1319 * config/arm/arm.c (arm_arch_dotprod): New.
1320 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1321 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1322 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1323 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1324 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1325 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1326 * doc/invoke.texi (armv8.2-a): Document dotprod
1328 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1330 * i386.c (ix86_vec_cost): New function.
1331 (ix86_rtx_costs): Handle vector operations better.
1332 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1333 * x86-tune-costs.h: Add new costs to all tables.
1335 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1337 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1339 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1340 divsd, sqrtss and sqrtsd
1341 * x86-tune-costs.h: Add new entries to all costs.
1342 (znver1_cost): Fix to match real instruction latencies.
1344 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1345 Michael Collison <michael.collison@arm.com>
1347 * compare-elim.c: Include emit-rtl.h.
1348 (can_merge_compare_into_arith): New function.
1349 (try_validate_parallel): Likewise.
1350 (try_merge_compare): Likewise.
1351 (try_eliminate_compare): Call the above when no previous clobber
1353 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1356 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1360 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1361 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1365 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1366 to be any operand_equal_p operands. For & (B - 1) require
1367 B to be power of 2. Recognize
1368 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1370 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1373 * optabs.c (expand_memory_blockage): Fix call of
1374 targetm.have_memory_blockage.
1376 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1379 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1380 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1381 objects to extra_objs instead of overwriting it.
1383 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1385 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1386 Use any_fp_register_operand as operand[3] predicate. Simplify
1387 equality test for operands[2] and operands[4] memory location.
1388 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1389 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1390 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1391 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1392 any_fp_register_operand as operand[1] predicate. Simplify
1393 equality test for operands[0] and operands[3] memory location.
1394 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1395 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1396 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1398 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1400 * target-insns.def: Add memory_blockage.
1401 * optabs.c (expand_memory_blockage): New function.
1402 (expand_asm_memory_barrier): Rename ...
1403 (expand_asm_memory_blockage): ... to this.
1404 (expand_mem_thread_fence): Call expand_memory_blockage
1405 instead of expand_asm_memory_barrier.
1406 (expand_mem_singnal_fence): Ditto.
1407 (expand_atomic_load): Ditto.
1408 (expand_atomic_store): Ditto.
1409 * doc/md.texi (Standard Pattern Names For Generation):
1410 Document memory_blockage instruction pattern.
1412 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1414 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1415 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1416 * config/rl78/rl78.md: New define_expand "adddi3".
1418 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1420 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1423 2017-10-13 Jeff Law <law@redhat.com>
1425 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1427 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1430 * internal-fn.c (expand_mul_overflow): If both operands have
1431 the same highpart of -1 or 0 and the topmost bit of lowpart
1432 is different, overflow is if res <= 0 rather than res < 0.
1434 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1436 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1437 TARGET_P9_VECTOR code for unaligned_load case.
1439 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1441 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1443 2017-10-13 Nathan Sidwell <nathan@acm.org>
1445 * tree-core.h (tree_contains_struct): Make bool.
1446 * tree.c (tree_contains_struct): Likewise.
1447 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1448 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1449 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1450 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1451 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1453 2017-10-13 Richard Biener <rguenther@suse.de>
1455 * graphite-isl-ast-to-gimple.c
1456 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1457 parameters and dominance check.
1458 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1459 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1460 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1461 Do not update SSA form here or do intermediate IL verification.
1462 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1463 (graphite_initialize): Remove check on the number of loops in
1464 the function and inline into graphite_transform_loops.
1465 (graphite_finalize): Inline into graphite_transform_loops.
1466 (graphite_transform_loops): Perform SSA update and IL verification
1468 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1470 2017-10-13 Richard Biener <rguenther@suse.de>
1472 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1473 graphite_expression_type_precision): Avoid global constructor
1475 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1476 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1477 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1478 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1479 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1480 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1482 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1485 * config/i386/i386.h (ix86_red_zone_size): New.
1486 * config/i386/i386.md (push peephole2s): Replace
1487 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1489 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * combine.c (can_change_dest_mode): Reject changes in
1494 REGMODE_NATURAL_SIZE.
1496 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1497 Alan Hayward <alan.hayward@arm.com>
1498 David Sherwood <david.sherwood@arm.com>
1500 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1501 (expand_debug_source_expr): Likewise.
1502 * combine.c (combine_simplify_rtx): Likewise.
1503 * cse.c (fold_rtx): Likewise.
1504 * fwprop.c (canonicalize_address): Likewise.
1505 * targhooks.c (default_shift_truncation_mask): Likewise.
1507 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1508 Alan Hayward <alan.hayward@arm.com>
1509 David Sherwood <david.sherwood@arm.com>
1511 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1512 (widened_mode): Likewise.
1513 (expand_unop): Likewise.
1514 * ree.c (transform_ifelse): Likewise.
1515 (merge_def_and_ext): Likewise.
1516 (combine_reaching_defs): Likewise.
1517 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1519 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1520 Alan Hayward <alan.hayward@arm.com>
1521 David Sherwood <david.sherwood@arm.com>
1523 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1524 * combine.c (gen_lowpart_for_combine): Likewise.
1525 * dwarf2out.c (rtl_for_decl_location): Likewise.
1526 * final.c (alter_subreg): Likewise.
1527 * rtlhooks.c (gen_lowpart_general): Likewise.
1528 (gen_lowpart_if_possible): Likewise.
1530 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1531 Alan Hayward <alan.hayward@arm.com>
1532 David Sherwood <david.sherwood@arm.com>
1534 * calls.c (expand_call): Use subreg_lowpart_offset.
1535 * cse.c (cse_insn): Likewise.
1536 * regcprop.c (copy_value): Likewise.
1537 (copyprop_hardreg_forward_1): Likewise.
1539 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1542 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1543 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1544 =Q constraints to +Q and into insn condition add check
1545 that operands[0] and operands[1] are equal.
1546 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1547 =Q constraints to +Q and into insn condition add check
1548 that operands[0] is equal to either operands[1] or operands[2].
1551 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1552 instead of handling MINUS_EXPR twice (once for each argument),
1553 canonicalize operand order and handle just once, use rtype where
1554 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1557 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1558 any values of __C while still being pattern recognizable as a simple
1561 2017-10-13 Richard Biener <rguenther@suse.de>
1563 PR tree-optimization/82451
1565 2017-10-02 Richard Biener <rguenther@suse.de>
1567 PR tree-optimization/82355
1568 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1569 a mapping for the enclosing loop but avoid generating one for
1571 (copy_bb_and_scalar_dependences): Remove premature codegen
1572 error on PHIs in blocks duplicated into multiple places.
1573 * graphite-scop-detection.c
1574 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1575 in the region use it as loop and nest to analyze the DR in.
1576 (try_generate_gimple_bb): Likewise.
1577 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1578 (add_loop_constraints): For blocks in a loop not in the region
1579 create a dimension with a single iteration.
1580 * sese.h (gbb_loop_at_index): Remove assert.
1582 * cfgloop.c (loop_preheader_edge): For the loop tree root
1583 return the single successor of the entry block.
1584 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1585 Reset the SCEV hashtable and niters.
1586 * graphite-scop-detection.c
1587 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1588 assert that we only have POLYNOMIAL_CHREC that vary in loops
1589 contained in the region.
1590 (scop_detection::graphite_can_represent_expr): Adjust.
1591 (scop_detection::stmt_has_simple_data_refs_p): For loops
1592 not in the region set loop to NULL. The nest is now the
1593 entry edge to the region.
1594 (try_generate_gimple_bb): Likewise.
1595 * sese.c (scalar_evolution_in_region): Adjust for
1596 instantiate_scev change.
1597 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1598 Make nest parameter the edge into the region.
1599 (create_data_ref): Likewise.
1600 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1601 entry edge into a region and adjust instantiate_scev calls.
1602 (create_data_ref): Likewise.
1603 (graphite_find_data_references_in_stmt): Likewise.
1604 (find_data_references_in_stmt): Pass the loop preheader edge
1605 from the nest argument.
1606 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1607 parameter the edge into the region.
1608 (instantiate_parameters): Use the loop preheader edge as entry.
1609 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1611 (get_instantiated_value_entry): Make instantiate_below parameter
1612 the edge into the region.
1613 (instantiate_scev_name): Likewise. Adjust dominance checks,
1614 when we cannot use loop-based instantiation instantiate by
1615 walking use-def chains.
1616 (instantiate_scev_poly): Adjust.
1617 (instantiate_scev_binary): Likewise.
1618 (instantiate_scev_convert): Likewise.
1619 (instantiate_scev_not): Likewise.
1620 (instantiate_array_ref): Remove.
1621 (instantiate_scev_3): Likewise.
1622 (instantiate_scev_2): Likewise.
1623 (instantiate_scev_1): Likewise.
1624 (instantiate_scev_r): Do not blindly handle N-operand trees.
1625 Do not instantiate array-refs. Handle all constants and invariants.
1626 (instantiate_scev): Make instantiate_below parameter
1627 the edge into the region.
1628 (resolve_mixers): Use the loop preheader edge for the region
1629 parameter to instantiate_scev_r.
1630 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1632 2017-10-13 Richard Biener <rguenther@suse.de>
1634 PR tree-optimization/82525
1635 * graphite-isl-ast-to-gimple.c
1636 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1638 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1639 Fail code generation when we cannot represent the isl integer.
1640 (binary_op_to_tree): Elide modulo operations that are no-ops
1641 in the type we code generate. Remove now superfluous code
1644 2017-10-13 Richard Biener <rguenther@suse.de>
1646 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1647 (scop_detection::harmful_loop_in_region): Remove premature
1648 IV type restriction.
1649 (scop_detection::graphite_can_represent_scev): We can handle
1650 pointer IVs just fine.
1652 2017-10-13 Alan Modra <amodra@gmail.com>
1654 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1655 "Clobbers and Scratch Registers". Add paragraph on
1656 alternative to clobbers for scratch registers and OpenBLAS
1659 2017-10-13 Alan Modra <amodra@gmail.com>
1661 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1662 example of a memory input for a string of known length. Move
1663 commentary out of table. Add a number of new examples
1664 covering array memory inputs.
1666 2017-10-12 Martin Liska <mliska@suse.cz>
1668 PR tree-optimization/82493
1669 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1670 (test_range_functions): New function.
1671 (sbitmap_c_tests): Likewise.
1672 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1673 * selftest.h (sbitmap_c_tests): New function.
1675 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1677 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1679 * config/rs6000/amo.h: Fix spacing issue.
1681 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1684 * config/i386/i386.md (*ashl<mode>3_mask_1,
1685 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1686 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1689 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1691 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1692 (profile_probability): Set max_probability
1693 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1695 * profile-count.c (profile_probability::differs_from_p): Do not
1696 rely on max_probaiblity == 10000
1698 2017-10-12 Jeff Law <law@redhat.com>
1700 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1703 2017-10-12 Martin Sebor <msebor@redhat.com>
1707 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1708 (handle_alias_pairs): Call it.
1709 * common.opt (-Wattribute-alias): New option.
1710 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1711 * doc/invoke.texi (-Wattribute-alias): Document.
1713 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1716 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1718 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1720 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1721 (make_hard_regno_born, make_hard_regno_dead): Update
1722 bb_killed_pseudos and bb_gen_pseudos.
1724 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1726 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1728 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1730 * config/alpha/alpha.c (alpha_split_conditional_move):
1731 Use std::swap instead of manually swapping.
1732 (alpha_stdarg_optimize_hook): Ditto.
1733 (alpha_canonicalize_comparison): Ditto.
1735 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1737 * tree-loop-distribution.c (struct builtin_info): New struct.
1738 (struct partition): Refactor fields into struct builtin_info.
1739 (partition_free): Free struct builtin_info.
1740 (build_size_arg_loc, build_addr_arg_loc): Delete.
1741 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1742 information from struct builtin_info.
1743 (find_single_drs): New function refactored from classify_partition.
1744 Also moved builtin validity checks to this function.
1745 (compute_access_range, alloc_builtin): New functions.
1746 (classify_builtin_st, classify_builtin_ldst): New functions.
1747 (classify_partition): Refactor code into functions find_single_drs,
1748 classify_builtin_st and classify_builtin_ldst.
1749 (distribute_loop): Don't do runtime alias check when distributing
1751 (find_seed_stmts_for_distribution): New function.
1752 (pass_loop_distribution::execute): Refactor code finding seed
1753 stmts into above function. Support distribution for the innermost
1754 two-level loop nest. Adjust dump information.
1756 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1758 * tree-loop-distribution.c: Adjust the general comment.
1759 (NUM_PARTITION_THRESHOLD): New macro.
1760 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1761 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1762 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1763 in call to build_partition_graph.
1764 (finalize_partitions): New parameter. Make loop distribution more
1765 conservative by fusing more partitions.
1766 (distribute_loop): Don't do runtime alias check in case of loop nest
1768 (find_seed_stmts_for_distribution): New function.
1769 (prepare_perfect_loop_nest): New function.
1770 (pass_loop_distribution::execute): Refactor code finding seed stmts
1771 and loop nest into above functions. Support loop nest distribution.
1772 Adjust dump information accordingly.
1774 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1776 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1777 and set PTYPE_SEQUENTIAL for merged partition.
1779 2017-10-12 Richard Biener <rguenther@suse.de>
1781 PR tree-optimization/69728
1783 2017-09-19 Richard Biener <rguenther@suse.de>
1785 PR tree-optimization/69728
1786 * graphite-sese-to-poly.c (schedule_error): New global.
1787 (add_loop_schedule): Handle empty domain by failing the
1789 (build_original_schedule): Handle schedule_error.
1791 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1792 domain by returning an unchanged schedule.
1794 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1796 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1797 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1799 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1801 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1804 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1807 * expr.c (store_field): Don't optimize away bitsize == 0 store
1808 from CALL_EXPR with addressable return type.
1810 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1812 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1813 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1814 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1815 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1816 TARGET_ISEL instead of TARGET_ISEL<sel>.
1818 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1820 * config/rs6000/rs6000.c
1821 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1823 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1825 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1827 (reg_or_cint_operand): Fix comment.
1828 (reg_or_zero_operand): New predicate.
1829 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1830 * config/rs6000/rs6000.c (output_isel): Delete.
1831 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1832 instead of reg_or_cint_operand. Output instruction directly (not via
1834 (isel_unsigned_<mode>): Ditto.
1835 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1836 gpc_reg_operand. Add an instruction alternative for this. Output
1837 instruction directly.
1838 (*isel_reversed_unsigned_<mode>): Ditto.
1840 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1842 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1843 (TARGET_CANONICALIZE_COMPARISON): Define.
1845 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1848 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1849 Check whether the dest is REG before adding REG_EQUIV note.
1851 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1854 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1856 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1857 (make_hard_regno_born, make_hard_regno_dead): Update
1858 bb_killed_pseudos and bb_gen_pseudos.
1860 2017-10-11 Nathan Sidwell <nathan@acm.org>
1862 * incpath.h (enum incpath_kind): Name enum, prefix values.
1863 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1864 * incpath.c (heads, tails): Use INC_MAX.
1865 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1866 (merge_include_chains, split_quote_chain,
1867 register_include_chains): Update incpath_kind names.
1868 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1869 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1871 (add_framework_path, darwin_register_objc_includes): Likewise.
1872 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1874 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1876 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1877 Do not use float_operator operator predicate.
1878 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1879 * config/i386/predicates.md (float_operator): Remove predicate.
1881 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1883 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1884 (*jccxf_i387): Ditto.
1885 (*jcc<mode>_i387): Ditto.
1886 (*jccu<mode>_i387): Ditto.
1887 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1888 (*jcc_*_i387 splitters): Remove.
1889 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1890 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1891 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1894 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1896 * profile-count.h (slow_safe_scale_64bit): New function.
1897 (safe_scale_64bit): New inline.
1898 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1899 * profile-count.c: Include wide-int.h
1900 (slow_safe_scale_64bit): New.
1902 2017-10-11 Nathan Sidwell <nathan@acm.org>
1904 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1905 HAS_DECL_ASSEMBLER_NAME_P.
1906 * gimple-expr.c (gimple_decl_printable_name: Check
1907 HAS_DECL_ASSEMBLER_NAME_P too.
1908 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1909 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1910 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1911 * passes.c (rest_of_decl_compilation): Check
1912 HAS_DECL_ASSEMBLER_NAME_P too.
1913 * recog.c (verify_changes): Likewise.
1914 * tree-pretty-print.c (dump_decl_name): Likewise.
1915 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1917 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1918 (SET_DECL_ASSEMBLER_NAME): Use it.
1919 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1920 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1921 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1923 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1925 * config.gcc (i386, x86_64): Add extra objects.
1926 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1927 (ix86_min_insn_size): Declare.
1928 (ix86_issue_rate): Declare.
1929 (ix86_adjust_cost): Declare.
1930 (ia32_multipass_dfa_lookahead): Declare.
1931 (ix86_macro_fusion_p): Declare.
1932 (ix86_macro_fusion_pair_p): Declare.
1933 (ix86_bd_has_dispatch): Declare.
1934 (ix86_bd_do_dispatch): Declare.
1935 (ix86_core2i7_init_hooks): Declare.
1936 (ix86_atom_sched_reorder): Declare.
1937 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1938 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1939 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1940 (rip_relative_addr_p): Rename to ...
1941 (ix86_rip_relative_addr_p): ... this one; export.
1942 (memory_address_length): Update.
1943 (ix86_issue_rate): Move to x86-tune-sched.c.
1944 (ix86_flags_dependent): Move to x86-tune-sched.c.
1945 (ix86_agi_dependent): Move to x86-tune-sched.c.
1946 (exact_dependency_1): Move to x86-tune-sched.c.
1947 (exact_store_load_dependency): Move to x86-tune-sched.c.
1948 (ix86_adjust_cost): Move to x86-tune-sched.c.
1949 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1950 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1951 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1952 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1953 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1954 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1955 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1956 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1957 (min_insn_size): Rename to ...
1958 (ix86_min_insn_size): ... this one; export.
1959 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1960 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1961 (core2i7_first_cycle_multipass_backtrack): Move to
1962 x86-tune-sched-core.c.
1963 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1964 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1965 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1966 (ix86_avoid_jump_mispredicts): Update.
1967 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1968 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1969 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1970 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1971 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1972 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1973 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1974 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1975 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1976 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1977 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1978 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1979 (BIG): Move to ix86-tune-sched-bd.c.
1980 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1981 (enum insn_path): Move to ix86-tune-sched-bd.c.
1982 (get_mem_group): Move to ix86-tune-sched-bd.c.
1983 (is_cmp): Move to ix86-tune-sched-bd.c.
1984 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1985 (is_branch): Move to ix86-tune-sched-bd.c.
1986 (is_prefetch): Move to ix86-tune-sched-bd.c.
1987 (init_window): Move to ix86-tune-sched-bd.c.
1988 (allocate_window): Move to ix86-tune-sched-bd.c.
1989 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1990 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1991 (process_end_window): Move to ix86-tune-sched-bd.c.
1992 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1993 (find_constant): Move to ix86-tune-sched-bd.c.
1994 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1995 (has_immediate): Move to ix86-tune-sched-bd.c.
1996 (get_insn_path): Move to ix86-tune-sched-bd.c.
1997 (get_insn_group): Move to ix86-tune-sched-bd.c.
1998 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1999 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2000 (add_insn_window): Move to ix86-tune-sched-bd.c.
2001 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2002 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2003 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2004 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2005 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2006 (do_dispatch): Move to ix86-tune-sched-bd.c.
2007 (has_dispatch): Move to ix86-tune-sched-bd.c.
2008 * config/i386/t-i386: Add new object files.
2009 * config/i386/x86-tune-costs.h: New file.
2010 * config/i386/x86-tune-sched-atom.c: New file.
2011 * config/i386/x86-tune-sched-bd.c: New file.
2012 * config/i386/x86-tune-sched-core.c: New file.
2013 * config/i386/x86-tune-sched.c: New file.
2015 2017-10-11 Liu Hao <lh_mouse@126.com>
2017 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2018 the generic version below instead.
2019 (should_colorize): Recognize Windows consoles as terminals
2021 * pretty-print.c [__MINGW32__] (write_all): New function.
2022 [__MINGW32__] (find_esc_head): Likewise.
2023 [__MINGW32__] (find_esc_terminator): Likewise.
2024 [__MINGW32__] (eat_esc_sequence): Likewise.
2025 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2027 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2030 2017-10-11 Richard Biener <rguenther@suse.de>
2032 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2033 Properly call analyze_scalar_evolution with the loop of the stmt.
2035 2017-10-11 Richard Biener <rguenther@suse.de>
2037 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2038 * tree-core.h (tree_base): Add chrec_var union member.
2039 * tree.h (CHREC_VAR): Remove.
2040 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2041 * tree-chrec.h (build_polynomial_chrec): Adjust.
2042 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2043 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2045 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2047 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2048 * match.pd: ... here.
2049 ((T) X == (T) Y): Relax condition.
2051 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2053 PR tree-optimization/82472
2054 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2056 (break_alias_scc_partitions): Update postorder number.
2058 2017-10-11 Martin Liska <mliska@suse.cz>
2061 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2063 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2065 2017-10-11 Martin Liska <mliska@suse.cz>
2067 * print-rtl.c (print_insn): Move declaration of idbuf
2068 to same scope as name.
2070 2017-10-11 Martin Liska <mliska@suse.cz>
2075 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2078 2017-10-11 Martin Liska <mliska@suse.cz>
2081 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2084 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2086 * config.gcc (default_gnu_indirect_function): Default to yes for
2087 arm*-*-linux* with glibc.
2089 2017-10-11 Richard Biener <rguenther@suse.de>
2091 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2092 default-defs and types we do not want to analyze.
2093 (interpret_loop_phi): Replace unreachable code with an assert.
2094 (compute_scalar_evolution_in_loop): Remove and inline ...
2095 (analyze_scalar_evolution_1): ... here, replacing condition with
2096 what makes the intent clearer. Remove handling of cases
2097 get_scalar_evolution now handles.
2099 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2101 PR rtl-optimization/81434
2102 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2103 comment for main loop. In sched_group_found if, also add checks for
2104 pass and min_cost_group.
2106 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2108 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2109 (rs6000_insn_cost): New function.
2110 * config/rs6000/rs6000.md (cost): New attribute.
2112 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2113 H.J. Lu <hongjiu.lu@intel.com>
2117 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2118 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2119 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2120 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2121 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2122 OPTION_MASK_ISA_MMX and some other bit set require both
2123 mmx and the ISAs without the mmx bit.
2124 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2125 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2126 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2127 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2128 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2129 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2130 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2131 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2132 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2133 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2134 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2135 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2136 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2137 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2138 Add OPTION_MASK_ISA_MMX.
2140 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2142 * config.gcc (armv7*-*-freebsd*): New target.
2143 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2145 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2147 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2148 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2151 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2153 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2154 a template parameter.
2155 (WIDE_INT_REF_FOR): Update accordingly.
2156 * tree.h (wi::int_traits <const_tree>): Delete.
2157 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2158 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2159 (wi::tree_to_wide_ref): New typedef.
2160 (wi::to_wide): New function.
2161 * calls.c (get_size_range): Use wi::to_wide when operating on
2163 * cgraph.c (cgraph_node::create_thunk): Likewise.
2164 * config/i386/i386.c (ix86_data_alignment): Likewise.
2165 (ix86_local_alignment): Likewise.
2166 * dbxout.c (stabstr_O): Likewise.
2167 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2168 * expr.c (const_vector_from_tree): Likewise.
2169 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2170 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2171 (fold_negate_expr_1, int_const_binop_1, const_binop)
2172 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2173 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2174 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2175 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2176 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2177 (fold_not_const, round_up_loc): Likewise.
2178 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2179 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2180 (alloca_call_type): Likewise.
2181 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2182 * godump.c (go_output_typedef): Likewise.
2183 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2184 * internal-fn.c (get_min_precision): Likewise.
2185 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2186 * ipa-polymorphic-call.c
2187 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2188 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2189 (ipa_modify_call_arguments): Likewise.
2190 * match.pd: Likewise.
2191 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2192 * print-tree.c (print_node_brief, print_node): Likewise.
2193 * stmt.c (expand_case): Likewise.
2194 * stor-layout.c (layout_type): Likewise.
2195 * tree-affine.c (tree_to_aff_combination): Likewise.
2196 * tree-cfg.c (group_case_labels_stmt): Likewise.
2197 * tree-data-ref.c (dr_analyze_indices): Likewise.
2198 (prune_runtime_alias_test_list): Likewise.
2199 * tree-dump.c (dequeue_and_dump): Likewise.
2200 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2201 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2202 * tree-pretty-print.c (dump_generic_node): Likewise.
2203 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2204 (simple_iv_with_niters): Likewise.
2205 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2206 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2207 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2208 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2209 (refine_value_range_using_guard, number_of_iterations_ne_max)
2210 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2211 (get_cst_init_from_scev, record_nonwrapping_iv)
2212 (scev_var_range_cant_overflow): Likewise.
2213 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2214 * tree-ssa-pre.c (compute_avail): Likewise.
2215 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2216 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2217 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2218 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2219 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2220 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2221 (dump_case_nodes, try_switch_expansion): Likewise.
2222 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2223 (vect_do_peeling): Likewise.
2224 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2225 * tree-vect-stmts.c (vectorizable_load): Likewise.
2226 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2227 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2228 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2229 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2230 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2231 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2232 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2233 (evrp_dom_walker::before_dom_children): Likewise.
2234 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2235 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2236 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2237 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2238 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2239 (get_range_pos_neg): Likewise.
2240 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2241 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2242 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2243 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2244 * config/avr/avr.c (avr_fold_builtin): Likewise.
2245 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2246 * config/msp430/msp430.c (msp430_attr): Likewise.
2247 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2248 * config/powerpcspe/powerpcspe-c.c
2249 (altivec_resolve_overloaded_builtin): Likewise.
2250 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2251 (rs6000_expand_ternop_builtin): Likewise.
2252 * config/rs6000/rs6000-c.c
2253 (altivec_resolve_overloaded_builtin): Likewise.
2254 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2255 (rs6000_expand_ternop_builtin): Likewise.
2256 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2258 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2260 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2261 when copying loop nest with only one inner loop.
2263 2017-10-10 Richard Biener <rguenther@suse.de>
2265 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2266 blocks if SCEV is active.
2267 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2269 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2270 (scev_initialize): Assert we are not yet initialized.
2272 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2274 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2275 inner loop's exit stmt by making it always exit the loop, otherwise
2276 we would generate an infinite empty loop.
2278 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2280 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2281 renaming variables in new preheader if it's deleted.
2283 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2285 * tree-loop-distribution.c (struct partition): Remove unused field
2286 loops of the structure.
2287 (partition_alloc, partition_free): Ditto.
2288 (build_rdg_partition_for_vertex): Ditto.
2290 2017-10-09 Jeff Law <law@redhat.com>
2292 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2293 return type to match prototype and documentation.
2295 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2297 * config/rs6000/rs6000.c (processor_costs): Move to ...
2298 * config/rs6000/rs6000.h: ... here.
2299 (rs6000_cost): Declare.
2301 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2303 * except.c (setjmp_fn): New global variable.
2304 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2305 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2306 if DONT_USE_BUILTIN_SETJMP is defined.
2308 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2310 * target.def (insn_cost): New hook.
2311 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2312 * doc/tm.texi: Regenerate.
2313 * rtlanal.c (insn_cost): Use the new hook.
2315 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2317 * combine.c (combine_validate_cost): Compute the new insn_cost,
2318 not just pattern_cost.
2319 (try_combine): Adjust comment.
2321 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2323 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2325 * combine.c (uid_insn_cost): Adjust comment.
2326 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2328 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2329 * dse.c (find_shift_sequence): Ditto.
2330 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2331 (bb_valid_for_noce_process_p): Use pattern_cost.
2332 * rtl.h (insn_rtx_cost): Delete.
2333 (pattern_cost): New prototype.
2334 (insn_cost): New prototype.
2335 * rtlanal.c (insn_rtx_cost): Rename to...
2336 (pattern_cost): ... this.
2339 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2341 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2342 (*jcc<mode>_0_r_i387): Ditto.
2343 (*jccxf_r_i387): Ditto.
2344 (*jcc<mode>_r_i387): Ditto.
2345 (*jccu<mode>_r_i387): Ditto.
2346 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2347 (*jcc): Rename from *jcc_1.
2349 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2351 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2352 deferred rescans after the lvx/stvx recombination pre-pass.
2354 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2356 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2357 memory operation instruction support.
2358 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2359 (rs6000-ibm-aix[789]*): Likewise.
2360 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2361 Document new functions.
2363 2017-10-09 Richard Biener <rguenther@suse.de>
2365 PR tree-optimization/82397
2366 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2367 equality only for semantically equal trees.
2369 2017-10-09 Richard Biener <rguenther@suse.de>
2371 PR tree-optimization/82449
2372 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2373 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2374 allow constant addresses.
2375 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2378 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2380 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2383 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2386 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2389 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2392 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2394 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2397 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2398 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2400 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2402 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2403 (WI_BINARY_PREDICATE_RESULT): Likewise.
2404 (wi::binary_traits::operator_result): New type.
2405 (wi::binary_traits::predicate_result): Likewise.
2406 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2407 (generic_wide_int::operator==, generic_wide_int::operator!=)
2408 (generic_wide_int::operator&, generic_wide_int::and_not)
2409 (generic_wide_int::operator|, generic_wide_int::or_not)
2410 (generic_wide_int::operator^, generic_wide_int::operator+
2411 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2413 (operator~, unary operator-, operator==, operator!=, operator&)
2414 (operator|, operator^, operator+, binary operator-, operator*): New
2416 * expr.c (get_inner_reference): Use wi::bit_and_not.
2417 * fold-const.c (fold_binary_loc): Likewise.
2418 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2419 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2420 (bit_value_binop): Likewise.
2421 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2422 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2423 (extract_range_from_binary_expr_1): Likewise.
2424 (masked_increment): Likewise.
2425 (simplify_bit_ops_using_ranges): Likewise.
2427 2017-10-09 Martin Jambor <mjambor@suse.cz>
2430 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2431 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2432 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2434 (hsa_fixup_mov_insn_type): New function.
2435 (hsa_op_with_type::get_in_type): Use it.
2436 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2437 immediates in an assert.
2438 (hsa_op_with_type::extend_int_to_32bit): New method.
2439 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2440 types. Convert to dest type if necessary.
2441 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2442 (reg_for_gimple_ssa): Pass false as min32int to
2443 hsa_type_for_scalar_tree_type.
2444 (gen_hsa_addr): Fixup type when creating addresable temporary.
2445 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2446 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2447 necessary. Call hsa_fixup_mov_insn_type.
2448 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2449 extend operands and convert to dest type if necessary.
2450 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2451 to dest type if necessary.
2452 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2453 if conversion nt necessary and size matches.
2454 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2455 to dest type if necessary.
2456 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2457 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2459 (gen_hsa_clrsb): Likewise.
2460 (gen_hsa_ffs): Likewise.
2461 (gen_hsa_divmod): Extend operands and convert to dest type if
2463 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2465 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2467 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2468 Remove empty default arguments. Use a brace block as output
2470 (conditional return): Ditto.
2472 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2473 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2475 (group_ending_nop): Ditto.
2476 (doloop_end): Ditto.
2477 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2478 (splitters for those): Ditto.
2480 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2482 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2483 a conditional jump (and the compare for it) so that pc_rtx is the
2485 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2486 for the deleted and renamed ctr<mode>_internal[234] patterns.
2487 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2488 Delete second conditional return pattern.
2489 (ctr<mode>_internal2): Delete this second bdnz pattern.
2490 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2491 (ctr<mode>_internal4): Delete this second bdz pattern.
2493 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2495 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2496 (always_initialized_rtx_for_ssa_name_p): New predicate.
2497 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2498 (finish_out_of_ssa): Free new field of SA.
2499 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2500 * tree-ssa-coalesce.c: Include tree-ssa.h.
2501 (get_parm_default_def_partitions): Remove extern keyword.
2502 (get_undefined_value_partitions): New function.
2503 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2504 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2506 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2508 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2510 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2512 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2514 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2515 for targets that preffer 128bit.
2517 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2519 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2521 2017-10-08 Olivier Hainque <hainque@adacore.com>
2523 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2524 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2526 (thumb_set_return_address): Likewise.
2528 2017-10-08 Olivier Hainque <hainque@adacore.com>
2530 * common/config/arm/arm-common.c (arm_except_unwind_info):
2531 Handle DWARF2_UNWIND_INFO.
2533 2017-10-07 Michael Collison <michael.collison@arm.com>
2535 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2538 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2540 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2541 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2542 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2543 defined, force the creation of a new block for a dispatch label.
2545 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2547 * invoke.texi (Wsuggest-attribute=cold): Document.
2548 * common.opt (Wsuggest-attribute=cold): New
2549 * ipa-pure-const.c (warn_function_cold): New function.
2550 * predict.c (compute_function_frequency): Use it.
2551 * predict.h (warn_function_cold): Declare.
2553 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2555 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2558 2017-10-06 Martin Liska <mliska@suse.cz>
2560 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2561 keyword for member functions.
2562 (struct sanopt_tree_couple): New struct.
2563 (struct sanopt_tree_couple_hash): New function.
2564 (struct sanopt_ctx): Add new hash_map.
2565 (has_dominating_ubsan_ptr_check): New function.
2566 (record_ubsan_ptr_check_stmt): Likewise.
2567 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2568 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2569 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2571 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2574 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2575 aarch64_simd_valid_immediate on CONST_VECTORs.
2576 (aarch64_reg_or_bic_imm): Likewise.
2578 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2580 PR rtl-optimization/82396
2581 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2583 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2585 * graphite-dependences.c (scop_get_reads): Move code to...
2586 (scop_get_must_writes): Move code to...
2587 (scop_get_may_writes): Move code to...
2588 (scop_get_reads_and_writes): ... here.
2589 (scop_get_dependences): Call scop_get_reads_and_writes.
2591 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2593 PR tree-optimization/82434
2594 * fold-const.h (can_native_encode_type_p,
2595 can_native_encode_string_p): Remove.
2596 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2597 don't encode anything, just return what would be otherwise returned.
2598 (native_encode_fixed, native_encode_complex, native_encode_vector):
2600 (native_encode_string): Likewise. Inline by hand
2601 can_native_encode_string_p.
2602 (can_native_encode_type_p): Remove.
2603 (can_native_encode_string_p): Remove.
2604 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2605 STRING_CSTs using can_native_encode_string_p, test all
2606 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2607 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2608 argument from native_encode_expr.
2609 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2610 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2613 2017-10-06 Richard Biener <rguenther@suse.de>
2615 PR tree-optimization/82397
2616 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2617 operand_equal_p but rely on data_ref_compare_tree for detecting
2619 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2620 to match up with dr_group_sort_cmp.
2622 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2625 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2627 * config/s390/s390-builtin-types.def: Regenerate.
2629 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2632 * config/s390/s390-builtin-types.def: Regenerate.
2633 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2634 Change flag from B_VXE to B_VX.
2635 (s390_vec_min_dbl): Remove B_VXE flag.
2637 2017-10-06 Richard Biener <rguenther@suse.de>
2639 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2640 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2641 translate_isl_ast_to_gimple::is_valid_rename,
2642 translate_isl_ast_to_gimple::get_rename,
2643 translate_isl_ast_to_gimple::get_def_bb_for_const,
2644 translate_isl_ast_to_gimple::get_new_name,
2645 translate_isl_ast_to_gimple::collect_all_ssa_names,
2646 translate_isl_ast_to_gimple::copy_loop_phi_args,
2647 translate_isl_ast_to_gimple::collect_all_ssa_names,
2648 translate_isl_ast_to_gimple::copy_loop_phi_args,
2649 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2650 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2651 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2652 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2653 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2654 translate_isl_ast_to_gimple::copy_cond_phi_args,
2655 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2656 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2657 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2658 translate_isl_ast_to_gimple::rename_uses,
2659 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2660 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2661 (set_rename_for_each_def): Likewise.
2662 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2663 here. Handle rewriting SCEV analyzable uses here.
2664 (copy_bb_and_scalar_dependences): Generate code for PHI
2666 (graphite_regenerate_ast_isl): Adjust.
2667 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2668 (add_write, add_read): New functions.
2669 (build_cross_bb_scalars_def): Use it and simplify.
2670 (build_cross_bb_scalars_use): Likewise.
2671 (graphite_find_cross_bb_scalar_vars): Inline into...
2672 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2673 simulating out-of-SSA. Compute liveout and add dependencies.
2674 (build_scops): Force an empty entry block.
2675 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2677 (sese_build_liveouts): Declare.
2678 (sese_trivially_empty_bb_p): Likewise.
2679 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2680 compute liveout and debug_liveout.
2681 (sese_bad_liveouts_use): Remove.
2682 (sese_reset_debug_liveouts_bb): Likewise.
2683 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2684 (sese_build_liveouts): Build liveout and debug_liveout and store
2686 (new_sese_info): Adjust.
2687 (free_sese_info): Likewise.
2688 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2689 do not build liveout here.
2690 (move_sese_in_condition): Adjust region entry.
2691 (scev_analyzable_p): Match up with chrec_apply requirements.
2692 (sese_trivially_empty_bb_p): New.
2693 * tree-into-ssa.c (get_reaching_def): Properly support generating
2694 default-defs for incremental rewrite of anonymous names.
2696 2017-10-06 Richard Biener <rguenther@suse.de>
2698 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2699 precision do not perform modulo reduction.
2701 2017-10-06 Richard Biener <rguenther@suse.de>
2703 PR tree-optimization/82436
2704 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2705 conservatively choose the vectorization factor when checking
2706 whether we can perform the required load permutation.
2707 (vect_transform_slp_perm_load): Assert when we may not fail.
2709 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2711 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2712 message for incompatible -msdata=* and -mcall-* options.
2714 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2716 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2717 rate for post-reload scheduling.
2719 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2721 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2723 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2725 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2726 to improve monte carlo in scimark.
2728 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2730 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2731 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2732 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2733 pentium4_cost, nocona_cost): Set reassociation width to 1.
2734 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2735 width to 2 for fp operations and 1 otherwise.
2736 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2738 (atom_cost): Set reassociation width to 2.
2739 (slm_cost, generic_cost): Set fp reassociation width
2740 to 2 and 1 otherwise.
2741 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2742 (core_cost): Set fp reassociation width to 4 and vector to 2.
2743 (ix86_reassociation_width): Rewrite using cost table; special case
2744 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2745 and TARGET_AVX128_OPTIMAL.
2746 * config/i386/i386.h (processor_costs): Add
2747 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2748 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2749 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2750 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2751 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2752 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2754 2017-10-05 Nathan Sidwell <nathan@acm.org>
2756 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2758 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2760 * config/arm/arm.c (arm_test_fpu_data): New.
2761 (arm_run_selftests): Call arm_test_fpu_data.
2763 2017-10-04 Nathan Sidwell <nathan@acm.org>
2765 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2767 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2769 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2771 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2772 check type for aarch64_simd_valid_immediate.
2773 (aarch64_output_simd_mov_immediate): Update prototype.
2774 (aarch64_simd_valid_immediate): Update prototype.
2775 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2776 support for ORR-immediate.
2777 (and<mode>3): modified pattern to add support for BIC-immediate.
2778 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2779 now checks for valid immediate for BIC and ORR based on new enum
2781 (aarch64_output_simd_mov_immediate): Function now used to output
2782 BIC/ORR imm as well based on new enum argument.
2783 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2785 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2786 (aarch64_reg_or_bic_imm): Likewise.
2788 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2790 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2791 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2793 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2797 PR rtl-optimization/82396
2798 * haifa-sched.c (autopref_multipass_init): Simplify
2800 (autopref_rank_data): Simplify sort order.
2801 * sched-int.h (autopref_multipass_data_): Remove
2802 multi_mem_insn_p, min_offset and max_offset.
2804 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2806 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2808 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2810 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2813 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2815 * doc/sourcebuild.texi: Document vect_long_mult.
2817 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2819 PR tree-optimization/82413
2820 * fold-const.c (build_range_check): Use widest_int when comparing
2821 the maximum ETYPE value with HIGH.
2823 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2825 PR rtl-optimization/82396
2826 * haifa-sched.c (autopref_multipass_init): Simplify
2828 (autopref_rank_data): Simplify sort order.
2829 * sched-int.h (autopref_multipass_data_): Remove
2830 multi_mem_insn_p, min_offset and max_offset.
2832 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2834 PR tree-optimization/82381
2835 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2836 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2839 PR tree-optimization/82374
2840 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2841 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2842 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2843 current_function_decl to the new decl.
2845 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2847 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2848 helper macro for IEEE float128 hardware built-in functions.
2849 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2851 (TRUNCF128_ODD): Likewise.
2852 (ADDF128_ODD): Likewise.
2853 (SUBF128_ODD): Likewise.
2854 (MULF128_ODD): Likewise.
2855 (DIVF128_ODD): Likewise.
2856 (FMAF128_ODD): Likewise.
2857 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2858 UNSPEC_TRUNC_ROUND_TO_ODD.
2859 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2860 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2861 floating point round to odd instructions.
2862 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2863 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2864 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2865 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2866 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2867 (trunc<mode>sf2_hw): Change the truncate with round to odd
2868 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2869 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2870 to odd hardware instructions.
2871 (sub<mode>3_odd): Likewise.
2872 (mul<mode>3_odd): Likewise.
2873 (div<mode>3_odd): Likewise.
2874 (sqrt<mode>2_odd): Likewise.
2875 (fma<mode>4_odd): Likewise.
2876 (fms<mode>4_odd): Likewise.
2877 (nfma<mode>4_odd): Likewise.
2878 (nfms<mode>4_odd): Likewise.
2879 (trunc<mode>df2_odd): Change the truncate with round to odd
2880 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2882 * doc/extend.texi (PowerPC built-in functions): Update documentation
2883 for existing IEEE float128-bit built-in functions. Add built-in
2884 functions that generate the IEEE 128-bit floating point round to
2887 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2889 PR rtl-optimization/77729
2890 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2891 to (X&(C1&~C2))|C2 transformations.
2893 2017-10-03 Martin Jambor <mjambor@suse.cz>
2895 PR tree-optimization/82363
2896 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2897 mismatch, mark lacc written regardless of racc.
2899 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2901 PR tree-optimization/82381
2902 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2903 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2904 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2905 fallthrough into reassoc_stmt_dominates_stmt_p.
2908 * combine.c (combine_instructions): Don't combine in unreachable
2911 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2914 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2915 function to not use the have_cpu variable. Do not set cpu_index,
2916 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2918 (rs6000_valid_attribute_p): Remove duplicate initializations of
2919 old_optimize and func_optimize.
2920 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2921 (rs6000_activate_target_options): Make global.
2922 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2925 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2927 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2928 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2929 Set *poffset to 0 and *psize and *pmax_size to -1 if
2930 *poffset + *psize overflows in HOST_WIDE_INT.
2932 PR tree-optimization/82387
2933 PR tree-optimization/82388
2934 PR tree-optimization/82389
2935 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2936 instead of live_bytes non-NULL.
2938 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2941 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2944 2017-10-02 Richard Biener <rguenther@suse.de>
2946 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2947 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2948 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2950 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2952 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2953 requested precision matches the type's.
2954 * calls.c (alloc_max_size): Calculate the new candidate size as
2955 a widest_int and use wi::to_widest when comparing it with the
2956 current candidate size.
2957 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2958 zero rather than integer_zero_node.
2959 * match.pd: Check for a no-op conversion before using wi::add
2960 rather than after. Use tree_to_uhwi when summing small shift
2961 counts into an unsigned int.
2963 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2964 Alan Hayward <alan.hayward@arm.com>
2965 David Sherwood <david.sherwood@arm.com>
2968 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2969 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2970 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2971 POINTER_AND_FP_REGS.
2973 2017-10-02 Richard Biener <rguenther@suse.de>
2975 PR tree-optimization/82355
2976 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2977 a mapping for the enclosing loop but avoid generating one for
2979 (copy_bb_and_scalar_dependences): Remove premature codegen
2980 error on PHIs in blocks duplicated into multiple places.
2981 * graphite-scop-detection.c
2982 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2983 in the region use it as loop and nest to analyze the DR in.
2984 (try_generate_gimple_bb): Likewise.
2985 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2986 (add_loop_constraints): For blocks in a loop not in the region
2987 create a dimension with a single iteration.
2988 * sese.h (gbb_loop_at_index): Remove assert.
2990 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2992 * omp-expand.c (adjust_context_scope): New function.
2993 (expand_parallel_call): Call adjust_context_scope.
2995 2017-10-01 Jeff Law <law@redhat.com>
2997 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2998 dom_opt_dom_walker class with direct access to private members.
2999 Add comments. Call test_for_singularity.
3000 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3001 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3002 m_dummy_cond anymore.
3003 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3005 (pass_dominator:execute): Build the dummy_cond here and pass it
3006 to the dom_opt_dom_walker ctor.
3007 (test_for_singularity): New function.
3009 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3010 Maya Rashish <coypu@sdf.org>
3012 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3013 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3014 (alpha*-*-netbsd*) Use nbsd_tm_file.
3015 (arm*-*-netbsdelf*) Likewise.
3016 (i[34567]86-*-netbsdelf*) Likewise.
3017 (x86_64-*-netbsd*) Likewise.
3018 (mips*-*-netbsd*) Likewise.
3019 (powerpc-*-netbsd*) Likewise.
3020 (sh*-*-netbsd*) Likewise.
3021 (sparc-*-netbsdelf*) Likewise.
3022 (sparc64-*-netbsd*) Likewise.
3023 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3025 (vax-*-netbsdelf*) Likewise.
3026 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3027 (UINT_FAST8_TYPE) Likewise.
3028 (INT_FAST16_TYPE) Check CHAR_FAST16.
3029 (UINT_FAST16_TYPE) Likewise.
3031 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3034 * config/i386/i386.md
3035 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3036 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3037 *divmodsi4_zext_2): New define_insn_and_split.
3038 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3039 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3040 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3041 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3042 New define_insn_and_split.
3043 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3044 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3045 operands[1] having DImode when mode is SImode.
3047 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3048 always SImode for DIV and MOD in REG_EQUAL notes.
3050 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3053 * match.pd: Fix handling of NaNs in pattern.
3055 2017-09-29 Jeff Law <law@redhat.com>
3057 * sbitmap.c (bitmap_bit_in_range_p): New function.
3058 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3059 * tree-ssa-dse.c (live_bytes_read): New function.
3060 (dse_classify_store): Ignore reads of dead bytes.
3062 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3063 typos and whitespace errors.
3064 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3065 * config/s390/s390.c (s390_emit_prologue): Likewise.
3067 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3070 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3071 with a symbol for LRA.
3073 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3075 PR rtl-optimization/82338
3076 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3078 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3080 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3081 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3082 (qsort_chk): Declare.
3083 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3084 (qsort_chk): New function.
3086 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3088 PR tree-optimization/82337
3089 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3090 phi definition if the PHI result appears in an abnormal PHI.
3091 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3092 basis appears in an abnormal PHI.
3094 2017-09-29 Richard Biener <rguenther@suse.de>
3096 * graphite-isl-ast-to-gimple.c
3097 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3098 (binary_op_to_tree): Use it.
3099 (get_rename_from_scev): Likewise.
3100 (copy_loop_phi_nodes): Likewise.
3101 (copy_bb_and_scalar_dependences): Likewise.
3102 (translate_pending_phi_nodes): Likewise.
3104 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3107 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3108 for movabsq $(i32 << shift), r64.
3110 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3112 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3113 index when encoding %esp as %rsp to avoid 0x67 prefix.
3115 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3117 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3118 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3120 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3122 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3123 Extensions with more than 16 double VFP registers.
3124 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3125 to_clear_mask and all code related to it. Replace the remaining
3126 entry by a sbitmap and adapt code accordingly.
3128 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3130 * brig-builtins.def: Change pure attributes to const.
3132 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3134 * config.gcc (default_gnu_indirect_function): Default to yes for
3135 sparc*-*-linux* with glibc.
3137 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3139 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3140 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3141 when creating .init_array and .fini_array sections with priority
3144 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3147 * config/aarch64/aarch64.c
3148 (aarch64_builtin_support_vector_misalignment): Always return false
3149 when misalignment is unknown.
3151 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3153 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3154 this function to return false if the definition used by the swap
3155 instruction is artificial, or if the memory address from which the
3156 constant value is loaded is not represented by a base address held
3157 in a register or if the base address register is a frame or stack
3158 pointer. Additionally, return false if the base address of the
3159 loaded constant is a SYMBOL_REF but is not considered to be a
3161 (replace_swapped_load_constant): New function.
3162 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3163 loaded constant vector with a load of a swapped constant vector.
3165 2017-09-27 Carl Love <cel@us.ibm.com>
3167 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3168 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3169 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3172 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3174 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3175 first, always call autopref_rank_data otherwise.
3177 2017-09-27 Richard Biener <rguenther@suse.de>
3179 * graphite-scop-detection.c (find_scop_parameters): Move
3180 loop bound handling ...
3181 (gather_bbs::before_dom_children): ... here, avoiding the need
3182 to build scop_info->loop_nest.
3183 (record_loop_in_sese): Remove.
3184 * sese.h (sese_info_t::loop_nest): Remove.
3185 * sese.c (new_sese_info): Do not allocate loop_nest.
3186 (free_sese_info): Do not free loop_nest.
3188 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3191 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3192 lhs from calls if the lhs has addressable type.
3194 2017-09-27 Richard Biener <rguenther@suse.de>
3196 * graphite.h (scop::max_alias_set): New member.
3197 * graphite-scop-detection.c: Remove references to non-existing
3198 --param in comments.
3199 (build_alias_sets): Record the maximum alias set used for drs.
3200 (build_scops): Support zero as unlimited for
3201 --param graphite-max-arrays-per-scop.
3202 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3204 (build_poly_sr_1): ... here. Compute alias set based on the
3205 maximum alias set used for drs rather than
3206 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3208 2017-09-27 Richard Biener <rguenther@suse.de>
3210 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3211 --param loop-block-tile-size=0 to disable tiling.
3213 2017-09-27 Richard Biener <rguenther@suse.de>
3215 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3216 (graphite-max-nb-scop-params): Document special value zero.
3217 * domwalk.h (dom_walker::STOP): New symbolical constant.
3218 (dom_walker::dom_walker): Add optional parameter for bb to
3220 (dom_walker::~dom_walker): Declare.
3221 (dom_walker::before_dom_children): Document STOP return value.
3222 (dom_walker::m_user_bb_to_rpo): New member.
3223 (dom_walker::m_bb_to_rpo): Likewise.
3224 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3225 mapping here if not provided by the user.
3226 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3227 provided by the user.
3228 (dom_walker::STOP): Define.
3229 (dom_walker::walk): Do not compute bb to RPO mapping here.
3230 Support STOP return value from before_dom_children to stop
3232 * graphite-optimize-isl.c (optimize_isl): If the schedule
3233 is the same still generate code if -fgraphite-identity
3234 or -floop-parallelize-all are given.
3235 * graphite-scop-detection.c: Include cfganal.h.
3236 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3238 (gather_bbs::before_dom_children): Return STOP for BBs
3240 (build_scops): Compute bb to RPO mapping and pass it to
3241 the domwalk. Treat --param graphite-max-nb-scop-params=0
3242 as not limiting the number of params.
3243 * graphite.c (graphite_initialize): Remove limit on the
3244 number of basic-blocks in a function.
3245 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3246 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3247 default value of 10.
3249 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3251 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3252 Adjust code to eliminate needing to do the shift right 32-bits
3253 operation after XSCVDPSPN.
3255 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3257 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3258 ((X / Y) != 0 -> X >= Y): Likewise.
3260 2017-09-26 Carl Love <cel@us.ibm.com>
3262 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3263 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3264 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3265 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3266 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3267 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3268 definitions and overloading.
3269 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3270 statement for P9V_BUILTIN_XST_LEN_R.
3271 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3272 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3273 define_expand and define_insn for the instructions and builtins.
3274 * doc/extend.texi: Update the built-in documentation file for the new
3276 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3277 define_insn for the instructions
3279 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3282 * gcc/config/netbsd-protos.h: New file.
3283 * gcc/config/netbsd.c: New file.
3284 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3285 * gcc/config/t-netbsd: New file.
3286 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3287 (tmake_file) Add t-netbsd.
3288 (extra_objs) Add netbsd.o.
3290 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3294 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3296 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3298 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3299 sign extension from a vector register to a GPR by doing a 32-bit
3300 direct move and then an EXTSW.
3301 (extendsi<mode>2 splitter): Likewise.
3302 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3303 right or vector extract after doing XSCVDPSPN. Use
3304 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3306 (movdi_from_sf_zero_ext): Likewise.
3307 (reload_gpr_from_vsxsf): Likewise.
3308 (p8_mfvsrd_4_disf): Delete, no longer used.
3309 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3310 then needing to move the SFmode to a GPR to use the XSCVDPSP
3311 instruction instead of FRSP and XSCVDPSPN.
3312 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3313 it is adjacent to the other XSCVSPDP insns.
3314 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3315 SFmode to be in traditional Altivec registers.
3316 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3317 (vsx_xscvspdpn): Likewise.
3318 (vsx_xscvspdpn_scalar): Likewise.
3320 2017-09-26 Martin Jambor <mjambor@suse.cz>
3322 * tree-sra.c (compare_access_positions): Put integral types first,
3323 stabilize sorting of integral types, remove conditions putting
3324 non-full-precision integers last.
3325 (sort_and_splice_var_accesses): Disable scalarization if a
3326 non-integert would be represented by a non-full-precision integer.
3328 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3330 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3331 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3332 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3333 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3334 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3335 conditionals inside the function instead of around it. Call
3336 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3337 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3339 2017-09-26 Richard Biener <rguenther@suse.de>
3341 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3343 (scop_detection::build_scop_breadth): ... this. Removed.
3344 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3345 (scop_detection::harmful_stmt_in_bb): Likewise.
3346 (scop_detection::graphite_can_represent_stmt): Likewise.
3347 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3348 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3349 (scop_detection::can_represent_loop_1): ... this. Removed.
3350 (scop_detection::harmful_loop_in_region): Simplify after inlining
3351 the above and remove more quadraticness.
3352 (build_scops): Adjust.
3353 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3356 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3359 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3360 REGNO (base) == SP_REG if base is a REG.
3363 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3364 if it is different SSA_NAME.
3365 (optimize_range_tests_cmp_bitwise): New function.
3366 (optimize_range_tests): Call it.
3368 2017-09-26 Richard Biener <rguenther@suse.de>
3370 PR tree-optimization/82321
3371 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3372 for the def being inside the loop.
3374 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3376 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3378 * config/s390/s390-builtins.def: Fix constraint on op4.
3380 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3382 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3383 independent expanders.
3384 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3385 ("vec_ordered", "vec_unordered"): New expanders.
3387 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3389 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3392 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3394 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3395 vec_unpacks_lo_v16qi.
3396 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3398 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3400 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3401 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3402 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3404 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3406 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3408 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3410 ("*vec_slb<mode>"): New insn pattern.
3411 ("vec_shr_<mode>"): New expander.
3412 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3413 and force the shift count operand to V16QImode.
3414 ("vec_srb<mode>"): Set shift count mode to V16QI.
3416 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3418 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3419 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3420 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3422 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3425 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3427 2017-09-26 Richard Biener <rguenther@suse.de>
3429 PR tree-optimization/82320
3430 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3433 2017-09-25 Jeff Law <law@redhat.com>
3435 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3436 prototype for new argument.
3437 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3438 mostly extracted from rs6000_emit_allocate_stack.
3439 (rs6000_emit_probe_stack_range_stack_clash): New function.
3440 (rs6000_emit_allocate_stack): Call
3441 rs6000_emit_probe_stack_range_stack_clash as needed.
3442 (rs6000_emit_probe_stack_range): Add additional argument
3443 to call to gen_probe_stack_range{si,di}.
3444 (output_probe_stack_range): New.
3445 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3446 (output_probe_stack_range_stack_clash): New.
3447 (rs6000_emit_prologue): Emit notes into dump file as requested.
3448 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3449 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3450 Add additional operand and pass it to output_probe_stack_range.
3452 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3454 PR tree-optimization/82163
3455 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3456 (checking_verify_loop_closed_ssa): New parameter.
3457 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3458 (check_loop_closed_ssa_stmt): Delete.
3459 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3460 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3461 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3464 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3466 * brig-builtins.def: Treat HSAIL barrier builtins as
3467 setjmp/longjump style functions.
3469 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3471 * target.def (constant_alignment): New hook.
3472 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3473 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3474 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3475 * doc/tm.texi: Regenerate.
3476 * targhooks.h (default_constant_alignment): Declare.
3477 (constant_alignment_word_strings): Likewise.
3478 * targhooks.c (default_constant_alignment): New function.
3479 (constant_alignment_word_strings): Likewise.
3480 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3481 instead of CONSTANT_ALIGNMENT.
3482 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3483 (force_const_mem): Likewise.
3484 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3485 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3486 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3487 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3488 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3490 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3491 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3492 constant_alignment_word_strings.
3493 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3494 (CONSTANT_ALIGNMENT): Likewise.
3495 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3496 (arm_constant_alignment): New function.
3497 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3498 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3499 constant_alignment_word_strings.
3500 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3501 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3502 constant_alignment_word_strings.
3503 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3504 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3505 (cris_constant_alignment): New function.
3506 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3507 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3508 (epiphany_constant_alignment): New function.
3509 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3510 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3511 constant_alignment_word_strings.
3512 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3513 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3514 constant_alignment_word_strings.
3515 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3516 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3517 constant_alignment_word_strings.
3518 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3519 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3520 * config/i386/i386.c (ix86_constant_alignment): Make static.
3521 Use the same interface as the target hook.
3522 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3523 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3524 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3525 constant_alignment_word_strings.
3526 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3527 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3528 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3529 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3530 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3531 constant_alignment_word_strings.
3532 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3533 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3534 constant_alignment_word_strings.
3535 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3536 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3537 constant_alignment_word_strings.
3538 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3539 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3541 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3542 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3543 * config/mips/mips.c (mips_constant_alignment): New function.
3544 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3545 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3546 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3547 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3548 (mmix_constant_alignment): Make static. Use the same interface
3550 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3551 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3552 constant_alignment_word_strings.
3553 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3554 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3555 constant_alignment_word_strings.
3556 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3557 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3558 constant_alignment_word_strings.
3559 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3560 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3561 (rs6000_constant_alignment): New function.
3562 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3563 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3564 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3565 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3566 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3567 (rs6000_constant_alignment): New function.
3568 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3569 * config/s390/s390.c (s390_constant_alignment): New function.
3570 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3571 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3572 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3573 constant_alignment_word_strings.
3574 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3575 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3576 (sparc_constant_alignment): New function.
3577 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3578 * config/spu/spu.c (spu_constant_alignment): New function.
3579 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3580 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3581 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3582 constant_alignment_word_strings.
3583 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3584 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3585 constant_alignment_word_strings.
3586 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3587 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3588 constant_alignment_word_strings.
3589 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3590 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3591 (visium_constant_alignment): New function.
3592 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3593 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3594 (xtensa_constant_alignment): New function.
3595 * system.h (CONSTANT_ALIGNMENT): Poison.
3597 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3600 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3601 (rs6000_builtin_valid_without_lhs): New helper function.
3602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3603 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3605 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3607 * target.h (vec_perm_indices): Use unsigned short rather than
3609 (auto_vec_perm_indices): Likewise.
3610 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3611 Use unsigned int rather than unsigned char.
3612 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3614 2017-09-25 Richard Biener <rguenther@suse.de>
3616 * cfgloop.h (sort_sibling_loops): Declare.
3617 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3618 (sort_sibling_loops): New function sorting the sibling loop list
3620 * graphite.c (graphite_transform_loops): Sort sibling loops.
3622 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3624 * target.def (vec_perm_const_ok): Change sel parameter to
3626 * optabs-query.c (can_vec_perm_p): Update accordingly.
3627 * doc/tm.texi: Regenerate.
3628 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3629 auto_vec_perm_indices and remove separate nelt field.
3630 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3631 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3632 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3633 (aarch64_expand_vec_perm_const): Update accordingly.
3634 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3635 to vec_perm_indices.
3636 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3637 auto_vec_perm_indices and remove separate nelt field.
3638 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3639 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3640 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3642 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3643 to vec_perm_indices.
3644 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3645 sel to vec_perm_indices.
3646 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3647 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3648 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3650 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3653 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3656 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3657 on the FUNCTION_DECL function context if it has a DIE that is a
3660 2017-09-25 Richard Biener <rguenther@suse.de>
3662 PR tree-optimization/82285
3663 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3666 2017-09-25 Tom de Vries <tom@codesourcery.com>
3670 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3673 2017-09-25 Richard Biener <rguenther@suse.de>
3675 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3676 ISL errors other than isl_error_quota happen. Dump if the
3677 schedule is the same.
3678 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3679 errors instead of aborting inside ISL.
3681 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3684 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3685 of libgcc_eh for m64.
3686 * config/i386/darwin64.h: Likewise.
3688 2017-09-25 Richard Biener <rguenther@suse.de>
3691 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3692 attribute for incomplete types nor twice for complete ones.
3694 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3697 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3698 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3701 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3704 * config/i386/i386.opt (mprefer-avx256): Use
3705 ix86_target_flags variable.
3706 * config/i386/i386.c (ix86_target_string): Move
3707 -mprefer-avx256 to flag2_opts.
3709 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3712 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3713 and x != -1 | y != -1 into (x & y) != -1.
3715 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3717 * config.gcc: Add new case statement to set
3718 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3719 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3720 s390x-*-linux* case statements. Added aarch64 to the list of
3721 supported architectures.
3723 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3725 PR tree-optimization/82289
3726 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3727 STMT_VINFO_RELEVANT_P.
3729 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3730 Alan Hayward <alan.hayward@arm.com>
3731 David Sherwood <david.sherwood@arm.com>
3733 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3734 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3735 (extract_range_from_binary_expr_1): Don't call
3736 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3738 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3739 Alan Hayward <alan.hayward@arm.com>
3740 David Sherwood <david.sherwood@arm.com>
3742 * target.def (preferred_vector_alignment): New hook.
3743 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3745 * doc/tm.texi: Regenerate.
3746 * targhooks.h (default_preferred_vector_alignment): Declare.
3747 * targhooks.c (default_preferred_vector_alignment): New function.
3748 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3750 (DR_TARGET_ALIGNMENT): New macro.
3751 (aligned_access_p): Update commentary.
3752 (vect_known_alignment_in_bytes): New function.
3753 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3755 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3756 Calculate the misalignment based on the target alignment rather than
3758 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3759 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3760 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3761 the target alignment, rather than masking the element misalignment
3762 with the number of elements in a vector. Also use the target
3763 alignment when calculating the maximum number of peels.
3764 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3765 instead of TYPE_ALIGN_UNIT.
3766 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3767 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3768 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3769 (vect_create_data_ref_ptr): Likewise.
3770 (vect_setup_realignment): Realign by ANDing with
3771 -DR_TARGET_MISALIGNMENT.
3772 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3773 the number of peels based on DR_TARGET_ALIGNMENT.
3774 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3775 with the guaranteed alignment boundary when deciding whether
3777 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3778 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3779 (ensure_base_align): Remove stmt_info parameter. Get the
3780 target base alignment from DR_TARGET_ALIGNMENT.
3781 (vectorizable_store): Update call accordingly. Interpret
3782 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3784 (vectorizable_load): Likewise.
3786 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3787 Alan Hayward <alan.hayward@arm.com>
3788 David Sherwood <david.sherwood@arm.com>
3790 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3791 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3792 (vect_enhance_data_refs_alignment): Likewise.
3794 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3796 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3797 error. Only quit immediately if parsing is complete.
3798 (BEGIN): Initialize fatal_err and parse_done.
3799 (begin fpu, end fpu): Check number of arguments.
3800 (begin arch, end arch): Likewise.
3801 (begin cpu, end cpu): Likewise.
3802 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3803 (optalias): Likewise.
3805 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3807 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3808 * config/arm/arm-isa.h: Delete. Move definitions to ...
3809 * arm-cpus.in: ... here. Use new feature and fgroup values.
3810 * config/arm/arm.c (arm_option_override): Use lower case for feature
3812 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3813 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3814 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3816 (print_isa_bits_for): New function.
3817 (gen_isa): New function.
3818 (gen_comm_data): Use print_isa_bits_for.
3819 (define feature): New keyword.
3820 (define fgroup): New keyword.
3821 * config/arm/t-arm (TM_H): Remove.
3822 (GTM_H): Add arm-isa.h.
3823 (arm-isa.h): Add rule to generate file.
3824 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3825 case for feature bit names.
3827 2017-09-22 Richard Biener <rguenther@suse.de>
3829 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3831 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3832 print of no dependency loops ...
3833 * graphite.c (graphite_transform_loops): ... here.
3834 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3836 (same_close_phi_node, remove_duplicate_close_phi,
3837 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3838 (canonicalize_loop_closed_ssa): ... here and simplify.
3839 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3840 (optimize_isl): Use dump_printf_loc to tell when we stopped
3841 optimizing because of an ISL timeout.
3843 2017-09-22 Richard Biener <rguenther@suse.de>
3845 PR tree-optimization/82291
3846 * tree-if-conv.c (predicate_mem_writes): Make sure to
3847 remove writes in blocks predicated with false.
3849 2017-09-22 Richard Biener <rguenther@suse.de>
3851 * sese.c: Include cfganal.h.
3852 (if_region_set_false_region): Remove.
3853 (create_if_region_on_edge): Likewise.
3854 (move_sese_in_condition): Re-implement without destroying
3857 2017-09-22 Richard Biener <rguenther@suse.de>
3859 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3860 Verify both BBs contain loop PHI nodes before dispatching to
3862 (graphite_regenerate_ast_isl): Do not recompute dominators,
3863 do not verify three times. Restructure for clarity.
3864 * graphite-scop-detection.c (same_close_phi_node,
3865 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3866 defined_in_loop_p, canonicalize_loop_closed_ssa,
3867 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3868 checking and SSA rewrite, move to ...
3869 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3870 (graphite_initialize): Do not pass in ctx, do not reset the
3871 SCEV cache, compute only dominators.
3872 (graphite_transform_loops): Allocate ISL ctx after
3873 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3874 Maintain post-dominators only around build_scops.
3875 * sese.c (if_region_set_false_region): Make static. Free
3876 and recompute dominators.
3877 (move_sese_in_condition): Assert we don't get called with
3878 post-dominators computed.
3879 * sese.h (if_region_set_false_region): Remove.
3881 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3883 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3884 mode attribute for TARGET_AVX512VL.
3886 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3888 * config/i386/i386.opt (mprefer-avx256): New option.
3889 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3891 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3892 for TARGET_PREFER_AVX256.
3893 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3895 2017-09-21 Jeff Law <law@redhat.com>
3897 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3898 Fix dump output if the only stack space is for pushed registers.
3900 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3902 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3905 2017-09-21 Martin Sebor <msebor@redhat.com>
3908 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3909 code (in C++) or code that triggers warnings.
3911 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3913 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3915 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3917 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3918 * sched-rgn.c: Ditto.
3919 * sel-sched-ir.c: Ditto.
3921 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3923 * toplev.h (set_random_seed): Adjust return type.
3924 * toplev.c (init_local_tick): Move eager initialization of random_seed
3925 to get_random_seed. Adjust comment.
3926 (init_random_seed): Inline to get_random_seed, delete.
3927 (get_random_seed): Initialize random_seed lazily.
3928 (set_random_seed): Do not return previous value.
3929 (print_switch_value): Do not call get_random_seed.
3931 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3933 * cgraph.c (delete_function_version): New, broken out from...
3934 (cgraph_node::delete_function_version): ...here. Rename to
3935 cgraph_node::delete_function_version_by_decl. Update all uses.
3936 (cgraph_node::remove): Call delete_function_version.
3938 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3941 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3942 VAR_DECLs to which addressable non-volatile parameters are mapped
3943 and for id->retvar after the return value assignment. Clear
3944 id->retval and id->retbnd after inlining.
3946 2017-09-21 Richard Biener <rguenther@suse.de>
3948 PR tree-optimization/82276
3949 PR tree-optimization/82244
3950 * tree-vrp.c (build_assert_expr_for): Set
3951 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3953 (remove_range_assertions): Revert earlier change.
3955 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3958 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3960 2017-09-21 Richard Biener <rguenther@suse.de>
3962 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3963 Restore valid IL after code generation errors.
3964 * graphite.c (graphite_transform_loops): Diagnose code
3965 generation issues as MSG_MISSED_OPTIMIZATION and continue
3966 with processing SCOPs.
3968 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3969 Alan Hayward <alan.hayward@arm.com>
3970 David Sherwood <david.sherwood@arm.com>
3972 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3973 rather than choosing between plus_constant and gen_rtx_<CODE>.
3974 * expr.c (emit_push_insn): Likewise.
3975 (expand_expr_real_2): Likewise.
3977 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3978 Alan Hayward <alan.hayward@arm.com>
3979 David Sherwood <david.sherwood@arm.com>
3981 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3983 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3984 Alan Hayward <alan.hayward@arm.com>
3985 David Sherwood <david.sherwood@arm.com>
3987 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3988 calling tree_to_uhwi.
3990 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3991 Alan Hayward <alan.hayward@arm.com>
3992 David Sherwood <david.sherwood@arm.com>
3994 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3995 INTEGER_CST rather than a negative test for ADDR_EXPR.
3997 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3998 Alan Hayward <alan.hayward@arm.com>
3999 David Sherwood <david.sherwood@arm.com>
4001 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4002 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4004 2017-09-21 Richard Biener <rguenther@suse.de>
4006 PR tree-optimization/71351
4007 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4008 graphite_create_new_loop_guard): Remove, fold remaining parts
4010 (translate_isl_ast_node_for): ... here and simplify.
4012 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4015 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4016 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4017 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4018 alternative always use QI mode, for -Os imov (=R,R) alternative
4019 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4022 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4023 Jeff Law <law@redhat.com>
4025 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4026 (allocate_stack_space): New function, partially extracted from
4028 (s390_emit_prologue): Track offset to most recent stack probe.
4029 Code to allocate space moved into allocate_stack_space.
4030 Dump actions when no stack is allocated.
4031 (s390_prologue_plus_offset): New function.
4032 (s390_emit_stack_probe): Likewise.
4034 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4036 * common.opt (Wa, Wl, Wp, g, gz=): Add
4038 (gno-column-info): Remove.
4039 (gcolumn-info): Drop RejectNegative.
4041 (gno-record-gcc-switches): Remove.
4042 (grecord-gcc-switches): Drop RejectNegative.
4043 (gno-split-dwarf): Remove.
4044 (gsplit-dwarf): Drop RejectNegative.
4045 (gno-strict-dwarf): Remove.
4046 (gstrict-dwarf): Drop RejectNegative.
4047 * config/darwin.opt (gfull, gused): Add RejectNegative.
4048 * dwarf2out.c (gen_producer_string): Drop
4049 gno-record-gcc-switches handler.
4050 * optc-gen.awk: Add g to prefixes with negative forms.
4051 * opts-common.c (remapping_prefix_p): New.
4052 (find_opt): Check it.
4053 (generate_canonical_option): Test g prefix.
4054 (option_map): Add -gno- mapping.
4055 (add_misspelling_candidates): Check remapping_prefix_p.
4057 2017-09-20 Jeff Law <law@redhat.com>
4059 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4060 thinko in stack clash protection support.
4062 * explow.c (compute_stack_clash_protection_loop_data): Use
4063 CONST_INT_P instead of explicit test. Verify object is a
4064 CONST_INT_P before looking at INTVAL.
4065 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4066 instead of explicit test.
4068 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4071 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4072 address instead of to r1 and r11.
4074 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4076 * config.gcc: Support "knm".
4077 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4078 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4080 * config/i386/i386.c (m_KNM): Define.
4081 (processor_target_table): Add "knm".
4083 (ix86_option_override_internal): Add "knm".
4084 (ix86_issue_rate): Add PROCESSOR_KNM.
4085 (ix86_adjust_cost): Ditto.
4086 (ia32_multipass_dfa_lookahead): Ditto.
4087 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4088 (fold_builtin_cpu): Add M_INTEL_KNM.
4089 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4090 (processor_type): Add PROCESSOR_KNM.
4091 * config/i386/x86-tune.def: Add m_KNM.
4092 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4094 2017-09-20 Richard Biener <rguenther@suse.de>
4096 PR tree-optimization/80213
4097 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4098 are allowed in empty BBs as well.
4099 (canonicalize_loop_closed_ssa): Also look for other complex
4101 (scop_detection::get_sese): Include the loop-closed PHI block
4103 (scop_detection::merge_sese): Remove code adding extra blocks.
4104 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4105 (build_scops): Assert the final returned scop is invalid.
4107 2017-09-20 Richard Biener <rguenther@suse.de>
4109 PR tree-optimization/82264
4110 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4112 (vn_phi_lookup): Likewise.
4113 (vn_phi_insert): Likewise.
4115 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4117 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4118 that fits into uhwi or shwi, add DW_AT_const_value regardless
4119 of early_dwarf without going through RTL, using add_AT_unsigned
4122 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4123 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4124 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4125 (DEBUG_MACRO_SECTION): Likewise.
4126 (DEBUG_LTO_MACRO_SECTION): Likewise.
4127 (DEBUG_STR_DWO_SECTION): Likewise.
4128 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4129 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4130 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4131 (DEBUG_LTO_LINE_STR_SECTION): Define.
4132 (init_sections_and_labels): Initialize debug_line_str_section
4133 variable. Initialize debug_loc_section for -gdwarf-5 to
4134 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4136 2017-09-20 Richard Biener <rguenther@suse.de>
4138 * graphite-sese-to-poly.c (extract_affine): Properly handle
4139 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4141 2017-09-20 Richard Biener <rguenther@suse.de>
4143 PR tree-optimization/81373
4144 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4145 Force SESE live-out defs to be handled even if they are
4148 2017-09-19 Jeff Law <law@redhat.com>
4150 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4151 nothing for stack adjustments with REG_STACK_CHECK.
4152 * sched-deps.c (parse_add_or_inc): Reject insns with
4153 REG_STACK_CHECK from dependency breaking.
4154 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4155 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4156 * reg-notes.def (STACK_CHECK): New note.
4158 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4159 (ix86_expand_prologue): Dump stack clash info as needed.
4160 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4162 * function.c (dump_stack_clash_frame_info): New function.
4163 * function.h (dump_stack_clash_frame_info): Prototype.
4164 (enum stack_clash_probes): New enum.
4166 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4167 flag_stack_clash_protection.
4168 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4169 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4170 (arm_frame_pointer_required): Likewise.
4171 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4172 (ia64_expand_prologue): Likewise.
4173 * config/mips/mips.c (mips_expand_prologue): Likewise.
4174 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4175 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4176 (sparc_flat_expand_prologue): Likewise.
4177 * config/spu/spu.c (spu_expand_prologue): Likewise.
4179 * explow.c: Include "params.h".
4180 (anti_adjust_stack_and_probe_stack_clash): New function.
4181 (get_stack_check_protect): Likewise.
4182 (compute_stack_clash_protection_loop_data): Likewise.
4183 (emit_stack_clash_protection_loop_start): Likewise.
4184 (emit_stack_clash_protection_loop_end): Likewise.
4185 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4186 Use anti_adjust_stack_and_probe_stack_clash.
4187 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4188 (emit_stack_clash_protection_loop_start): Likewise.
4189 (emit_stack_clash_protection_loop_end): Likewise.
4190 * rtl.h (get_stack_check_protect): Prototype.
4191 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4192 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4193 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4195 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4197 * doc/tm.texi: Rebuilt.
4198 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4199 get_stack_check_protect.
4200 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4201 * config/arm/arm.c (arm_expand_prologue): Likewise.
4202 (arm_frame_pointer_required): Likewise.
4203 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4204 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4205 * config/mips/mips.c (mips_expand_prologue): Likewise.
4206 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4207 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4208 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4209 (sparc_flat_expand_prologue): Likewise.
4211 * common.opt (-fstack-clash-protection): New option.
4212 * flag-types.h (enum stack_check_type): Note difference between
4213 -fstack-check= and -fstack-clash-protection.
4214 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4215 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4216 * toplev.c (process_options): Issue warnings/errors for cases
4217 not handled with -fstack-clash-protection.
4218 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4219 (-fstack-check): Note additional problem with -fstack-check=generic.
4220 Note that -fstack-check is primarily for Ada and refer users
4221 to -fstack-clash-protection for stack-clash-protection.
4222 Document new params for stack clash protection.
4224 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4226 * config/i386/i386.c (ix86_split_long_move): Do not handle
4227 address used for LEA in a special way.
4229 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4231 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4233 2017-09-19 Martin Sebor <msebor@redhat.com>
4236 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4237 of incompatible types.
4239 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4242 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4243 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4244 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4246 2017-09-19 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/82244
4249 * tree-vrp.c (remove_range_assertions): Do not propagate
4250 a constant to abnormals but replace the assert with a copy.
4252 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4254 PR rtl-optimization/57878
4255 PR rtl-optimization/68988
4256 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4257 avoidance test involving non_reload_pseudos. Move frequency test
4258 below the general fragmentation avoidance test.
4260 2017-09-19 Richard Biener <rguenther@suse.de>
4262 PR tree-optimization/69728
4263 * graphite-sese-to-poly.c (schedule_error): New global.
4264 (add_loop_schedule): Handle empty domain by failing the
4266 (build_original_schedule): Handle schedule_error.
4268 2017-09-19 Richard Biener <rguenther@suse.de>
4270 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4271 Do not iterate to sibling loops but only to siblings of inner
4274 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4277 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4278 that the registers are different.
4280 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4282 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4283 to processor_model and "amdfam17h" to arch_names_table.
4284 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4286 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4289 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4291 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4292 Alan Hayward <alan.hayward@arm.com>
4293 David Sherwood <david.sherwood@arm.com>
4295 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4297 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4299 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4300 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4301 vect_schedule_slp_instance.
4302 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4303 Update call to vect_slp_analyze_node_operations. Simplify return
4305 (vect_slp_analyze_bb_1): Update call accordingly.
4306 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4307 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4308 (vect_schedule_slp): Update call accordingly.
4310 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4311 Alan Hayward <alan.hayward@arm.com>
4312 David Sherwood <david.sherwood@arm.com>
4314 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4315 with types that aren't in fact scalar.
4317 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4319 * tree-vect-slp.c (vect_record_max_nunits): New function,
4321 (vect_build_slp_tree_1): ...here.
4322 (vect_build_slp_tree_2): Call it for phis too.
4324 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4326 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4327 to vect_get_vec_def_for_operand when getting the mask operand.
4329 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4330 Alan Hayward <alan.hayward@arm.com>
4331 David Sherwood <david.sherwood@arm.com>
4333 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4336 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4337 Alan Hayward <alan.hayward@arm.com>
4338 David Sherwood <david.sherwood@arm.com>
4340 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4341 calculation for vector booleans.
4343 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4344 Alan Hayward <alan.hayward@arm.com>
4345 David Sherwood <david.sherwood@arm.com>
4347 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4349 (vect_transform_stmt): ...here.
4350 (vect_analyze_stmt): Use it instead of calling
4351 vectorizable_live_operation directly.
4353 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4355 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4356 non-SIMT targets in acc vector loops.
4358 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4360 * configure.ac: Add arc and check if assembler supports gdwarf2.
4361 * configure: Regenerate.
4363 2017-09-18 Richard Biener <rguenther@suse.de>
4365 PR tree-optimization/82220
4366 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4367 epilogue niters from the min_profitable_iters compute.
4369 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4372 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4373 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4374 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4376 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4379 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4380 switching to a new text section.
4382 2017-09-18 Richard Biener <rguenther@suse.de>
4384 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4386 (build_alias_set): Reject aliases with no access function.
4388 2017-09-18 Richard Biener <rguenther@suse.de>
4390 PR tree-optimization/79622
4391 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4393 (build_cross_bb_scalars_use): Likewise.
4395 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4397 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4399 2017-09-18 Alan Modra <amodra@gmail.com>
4402 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4403 stack_pointer_rtx for count 0. Update comments. Break up
4404 large rtl expression.
4406 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4408 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4409 Increase to 20 bytes.
4410 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4411 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4412 or avx version of the stub.
4414 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4417 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4418 compute the minimum stack alignment. Also update preferred stack
4419 boundary for leaf functions.
4421 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4423 PR tree-optimization/82228
4424 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4427 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4429 * common/config/nds32/nds32-common.c
4430 (nds32_option_optimization_table): Refine formatting.
4431 (nds32_option_optimization_table): Use -fsched-pressure and
4432 -fomit-frame-pointer for specific optimization level.
4434 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4436 * config/nds32/nds32.c: Refine formatting and comments.
4437 * config/nds32/nds32.h: Likewise.
4438 * config/nds32/nds32.md: Likewise.
4439 * config/nds32/nds32-cost.c: Likewise.
4440 * config/nds32/nds32-isr.c: Likewise.
4441 * config/nds32/nds32-md-auxiliary.c: Likewise.
4442 * config/nds32/nds32-multiple.md: Likewise.
4443 * config/nds32/nds32-predicates.c: Likewise.
4445 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4446 Jakub Jelinek <jakub@redhat.com>
4448 Add support for -std=c++2a.
4449 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4451 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4453 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4456 * doc/extend.texi (Common Function Attributes): Add
4457 references to ARM, AArch64, and S/390 specific attributes.
4458 (Function Specific Option Pragmas): Add AArch64 and S/390
4459 to list of back ends that support the target pragma.
4461 2017-09-15 Nathan Sidwell <nathan@acm.org>
4463 * doc/standards.texi: Fix C++17 description. Update URLs for
4466 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4468 * common.opt (Wcast-align=strict): New warning option.
4469 * doc/invoke.texi: Document -Wcast-align=strict.
4471 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4473 * cgraph.h (cgraph_thunk_info): Add comments.
4474 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4475 assert for VIRTUAL_* arguments stricter.
4477 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4479 PR tree-optimization/71026
4480 * match.pd: Move RDIV patterns from fold-const.c
4481 * fold-const.c (distribute_real_division): Removed.
4482 (fold_binary_loc): Remove calls to distribute_real_divison.
4484 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4486 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4487 c++1z and gnu++1z as deprecated. Change other references to
4488 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4489 Change -Wc++1z-compat to -Wc++17-compat.
4490 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4491 * dwarf2out.c (highest_c_language): Handle C++17.
4492 (gen_compile_unit_die): Likewise.
4494 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4496 PR rtl-optimization/82192
4497 * combine.c (make_extraction): Don't look through non-paradoxical
4498 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4501 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4502 Alan Hayward <alan.hayward@arm.com>
4503 David Sherwood <david.sherwood@arm.com>
4505 * target.def (function_arg_offset): New hook.
4506 * targhooks.h (default_function_arg_offset): Declare.
4507 * targhooks.c (default_function_arg_offset): New function.
4508 * function.c (locate_and_pad_parm): Use
4509 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4510 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4511 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4512 * doc/tm.texi: Regenerate.
4513 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4514 * config/spu/spu.c (spu_function_arg_offset): New function.
4515 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4516 * system.h (FUNCTION_ARG_OFFSET): Poison.
4518 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4519 Alan Hayard <alan.hayward@arm.com>
4520 David Sherwood <david.sherwood@arm.com>
4522 * target.def (truly_noop_truncation): New hook.
4523 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4524 than TRULY_NOOP_TRUNCATION.
4525 * hooks.h (hook_bool_uint_uint_true): Declare.
4526 * hooks.c (hook_bool_uint_uint_true): New function.
4527 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4528 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4529 * doc/tm.texi: Regenerate.
4530 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4531 rather than TRULY_NOOP_TRUNCATION in comments.
4532 (simplify_comparison): Likewise.
4533 (record_truncated_value): Likewise.
4534 * expmed.c (extract_bit_field_1): Likewise.
4535 (extract_split_bit_field): Likewise.
4536 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4537 instead of TRULY_NOOP_TRUNCATION.
4538 * function.c (assign_parm_setup_block): Likewise.
4539 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4540 * rtlhooks.c: Include target.h.
4541 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4542 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4543 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4544 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4545 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4546 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4547 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4548 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4549 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4550 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4551 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4552 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4553 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4554 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4555 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4556 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4557 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4558 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4559 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4560 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4561 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4562 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4563 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4564 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4565 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4566 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4567 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4568 rather than TRULY_NOOP_TRUNCATION in comments.
4569 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4570 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4571 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4572 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4573 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4574 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4575 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4576 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4577 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4578 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4579 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4580 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4581 rather than TRULY_NOOP_TRUNCATION in comments.
4582 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4583 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4584 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4585 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4586 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4587 TRULY_NOOP_TRUNCATION condition.
4588 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4589 (TRULY_NOOP_TRUNCATION): Delete.
4590 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4591 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4592 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4593 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4594 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4595 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4596 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4597 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4598 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4599 rather than TRULY_NOOP_TRUNCATION in comments.
4600 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4601 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4602 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4603 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4604 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4605 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4607 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4610 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4611 (*cmp_ior): Likewise.
4612 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4613 (*ior_scc_scc_cmp): Likewise.
4614 (*and_scc_scc): Likewise.
4615 (*and_scc_scc_cmp): Likewise.
4617 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4618 Alan Hayard <alan.hayward@arm.com>
4619 David Sherwood <david.sherwood@arm.com>
4621 * target.def (can_change_mode_class): New hook.
4622 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4623 (hard_regno_nregs): Likewise.
4624 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4625 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4626 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4627 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4628 (LOAD_EXTEND_OP): Update accordingly.
4629 * doc/tm.texi: Regenerate.
4630 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4631 CANNOT_CHANGE_MODE_CLASS.
4632 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4633 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4634 * combine.c (simplify_set): Update accordingly.
4635 * emit-rtl.c (validate_subreg): Likewise.
4636 * recog.c (general_operand): Likewise.
4637 * regcprop.c (mode_change_ok): Likewise.
4638 * reload1.c (choose_reload_regs): Likewise.
4639 (inherit_piecemeal_p): Likewise.
4640 * rtlanal.c (simplify_subreg_regno): Likewise.
4641 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4642 instead of CANNOT_CHANGE_MODE_CLASS.
4643 (reload_cse_simplify_operands): Likewise.
4644 * reload.c (push_reload): Use targetm.can_change_mode_class
4645 instead of CANNOT_CHANGE_MODE_CLASS.
4646 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4647 REG_CANNOT_CHANGE_MODE_P.
4648 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4649 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4650 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4651 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4652 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4653 (arm_can_change_mode_class): New function.
4654 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4655 than CANNOT_CHANGE_MODE_CLASS in comments.
4656 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4657 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4658 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4659 (ix86_can_change_mode_class): ...this new function, inverting the
4660 sense of the return value.
4661 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4662 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4663 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4664 (ia64_can_change_mode_class): New function.
4665 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4666 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4667 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4668 (m32c_can_change_mode_class): ...this new function, inverting the
4669 sense of the return value.
4670 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4671 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4672 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4673 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4674 (mips_can_change_mode_class): ...this new function, inverting the
4675 sense of the return value.
4676 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4677 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4678 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4679 (msp430_can_change_mode_class): New function.
4680 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4681 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4682 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4683 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4684 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4685 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4686 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4687 (pa_cannot_change_mode_class): Replace with...
4688 (pa_can_change_mode_class): ...this new function, inverting the
4689 sense of the return value.
4690 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4691 than CANNOT_CHANGE_MODE_CLASS in comments.
4692 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4693 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4694 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4695 (pdp11_cannot_change_mode_class): Replace with...
4696 (pdp11_can_change_mode_class): ...this new function, inverting the
4697 sense of the return value.
4698 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4699 * config/powerpcspe/powerpcspe-protos.h
4700 (rs6000_cannot_change_mode_class_ptr): Delete.
4701 * config/powerpcspe/powerpcspe.c
4702 (rs6000_cannot_change_mode_class_ptr): Delete.
4703 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4704 (rs6000_option_override_internal): Assign to
4705 targetm.can_change_mode_class instead of
4706 rs6000_cannot_change_mode_class_ptr.
4707 (rs6000_cannot_change_mode_class): Replace with...
4708 (rs6000_can_change_mode_class): ...this new function, inverting the
4709 sense of the return value.
4710 (rs6000_debug_cannot_change_mode_class): Replace with...
4711 (rs6000_debug_can_change_mode_class): ...this new function.
4712 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4713 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4714 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4715 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4716 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4718 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4719 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4720 (rs6000_option_override_internal): Assign to
4721 targetm.can_change_mode_class instead of
4722 rs6000_cannot_change_mode_class_ptr.
4723 (rs6000_cannot_change_mode_class): Replace with...
4724 (rs6000_can_change_mode_class): ...this new function, inverting the
4725 sense of the return value.
4726 (rs6000_debug_cannot_change_mode_class): Replace with...
4727 (rs6000_debug_can_change_mode_class): ...this new function.
4728 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4729 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4730 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4731 (s390_can_change_mode_class): ...this new function, inverting the
4732 sense of the return value.
4733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4734 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4735 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4736 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4737 (sh_cannot_change_mode_class): Replace with...
4738 (sh_can_change_mode_class): ...this new function, inverting the
4739 sense of the return value.
4740 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4741 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4742 (sparc_can_change_mode_class): New function.
4743 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4744 * config/spu/spu.c (spu_can_change_mode_class): New function.
4745 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4746 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4747 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4748 (visium_can_change_mode_class): New function.
4749 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4751 2017-09-15 Richard Biener <rguenther@suse.de>
4753 PR tree-optimization/82217
4754 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4755 but not undefined case.
4757 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4760 * postreload.c (reload_cse_simplify_operands): Skip
4761 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4763 2017-09-15 Richard Biener <rguenther@suse.de>
4765 PR tree-optimization/68823
4766 * graphite-scop-detection.c (build_alias_set): If we have a
4767 possible dependence check whether we can handle them by just
4768 looking at the DRs DR_ACCESS_FNs.
4769 (build_scops): If build_alias_set fails, fail the SCOP.
4771 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4773 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4774 to support float128 built-in functions that require the ISA 3.0
4776 (BU_FLOAT128_3_HW): Likewise.
4777 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4779 (FMAF128): Likewise.
4781 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4782 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4783 floating point instructions.
4784 (rs6000_invalid_builtin): Likewise.
4785 (rs6000_builtin_mask_names): Likewise.
4786 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4787 (RS6000_BTM_FLOAT128_HW): Likewise.
4788 (RS6000_BTM_COMMON): Likewise.
4789 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4791 * doc/extend.texi (RS/6000 built-in functions): Document the
4792 IEEE 128-bit floating point square root and fused multiply-add
4795 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4797 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4798 reg (r2) isn't in the set of registers defined in the prologue.
4800 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4801 Alan Hayward <alan.hayward@arm.com>
4802 David Sherwood <david.sherwood@arm.com>
4804 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4805 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4806 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4807 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4808 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4810 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4811 max_vectorization_factor.
4812 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4814 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4815 Alan Hayward <alan.hayward@arm.com>
4816 David Sherwood <david.sherwood@arm.com>
4818 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4819 (vect_worthwhile_without_simd_p): Declare.
4820 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4821 (vectorizable_reduction): Use it.
4822 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4823 (vectorizable_operation): Likewise.
4825 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4826 Alan Hayward <alan.hayward@arm.com>
4827 David Sherwood <david.sherwood@arm.com>
4829 * tree-vectorizer.h (vect_get_num_copies): New function.
4830 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4831 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4832 (vectorizable_induction): Likewise.
4833 (vectorizable_live_operation): Likewise.
4834 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4835 (vectorizable_bswap): Likewise.
4836 (vectorizable_call): Likewise.
4837 (vectorizable_conversion): Likewise.
4838 (vectorizable_assignment): Likewise.
4839 (vectorizable_shift): Likewise.
4840 (vectorizable_operation): Likewise.
4841 (vectorizable_store): Likewise.
4842 (vectorizable_load): Likewise.
4843 (vectorizable_condition): Likewise.
4844 (vectorizable_comparison): Likewise.
4845 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4847 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4848 Alan Hayward <alan.hayward@arm.com>
4849 David Sherwood <david.sherwood@arm.com>
4851 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4852 of vect_init_vector.
4854 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4855 Alan Hayward <alan.hayward@arm.com>
4856 David Sherwood <david.sherwood@arm.com>
4858 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4859 an inline wrapper that provides a location.
4860 (gimple_build_vector): Likewise.
4861 * gimple-fold.c (gimple_build_vector_from_val): New function.
4862 (gimple_build_vector): Likewise.
4863 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4864 functions to build the initial value. Always return a gimple value.
4865 (get_initial_defs_for_reduction): Likewise. Only compute
4867 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4868 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4869 (vectorizable_induction): Use gimple_build_vector rather than
4872 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4873 Alan Hayward <alan.hayward@arm.com>
4874 David Sherwood <david.sherwood@arm.com>
4876 * target.h (vec_perm_indices): New typedef.
4877 (auto_vec_perm_indices): Likewise.
4878 * optabs-query.h: Include target.h
4879 (can_vec_perm_p): Take a vec_perm_indices *.
4880 * optabs-query.c (can_vec_perm_p): Likewise.
4881 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4882 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4883 * tree-vect-generic.c (lower_vec_perm): Likewise.
4884 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4885 (vect_grouped_load_supported): Likewise.
4886 (vect_shift_permute_load_chain): Likewise.
4887 (vect_permute_store_chain): Use auto_vec_perm_indices.
4888 (vect_permute_load_chain): Likewise.
4889 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4890 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4891 Update uses of can_vec_perm_p.
4892 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4893 mode with a number of elements. Take a vec_perm_indices *.
4894 (vect_create_epilog_for_reduction): Update accordingly.
4895 Use auto_vec_perm_indices.
4896 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4897 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4898 (vect_transform_slp_perm_load): Likewise.
4899 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4900 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4901 (vect_gen_perm_mask_checked): Likewise.
4902 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4903 (vect_gen_perm_mask_checked): Likewise.
4904 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4905 (vectorizable_store): Likewise.
4906 (vectorizable_load): Likewise.
4907 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4908 (vectorizable_bswap): Likewise.
4910 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4911 Alan Hayward <alan.hayward@arm.com>
4912 David Sherwood <david.sherwood@arm.com>
4914 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4915 * tree.c (build_vector): Likewise.
4916 (build_vector_from_ctor): Update accordingly.
4917 (build_vector_from_val): Likewise.
4918 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4919 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4920 * tree-vect-generic.c (add_rshift): Likewise.
4921 (expand_vector_divmod): Likewise.
4922 (optimize_vector_constructor): Likewise.
4923 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4924 (vect_transform_slp_perm_load): Likewise.
4925 (vect_schedule_slp_instance): Likewise.
4926 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4927 (vectorizable_call): Likewise.
4928 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4929 * expmed.c (make_tree): Likewise.
4930 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4931 a vector passed to build_vector.
4932 (fold_convert_const): Likewise.
4933 (exact_inverse): Likewise.
4934 (fold_ternary_loc): Likewise.
4935 (fold_relational_const): Likewise.
4936 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4937 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4938 (const_unop): Likewise. Store the reduction accumulator in a
4939 variable rather than an array.
4940 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4941 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4942 the new vector, rather than constructing it after the input arrays.
4943 (native_interpret_vector): Use auto_vec<tree> when building
4944 a vector passed to build_vector. Add elements in order.
4945 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4946 auto_vec<tree> when building a vector passed to build_vector.
4947 (vect_create_epilog_for_reduction): Likewise.
4948 (vectorizable_induction): Likewise.
4949 (get_initial_def_for_reduction): Likewise. Fix indentation of
4951 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4953 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4954 passed to build_vector.
4956 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4957 Alan Hayward <alan.hayward@arm.com>
4958 David Sherwood <david.sherwood@arm.com>
4960 * tree-core.h (tree_base::u): Add an "nelts" field.
4961 (tree_vector): Use VECTOR_CST_NELTS as the length.
4962 * tree.c (tree_size): Likewise.
4963 (make_vector): Initialize VECTOR_CST_NELTS.
4964 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4965 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4966 TYPE_VECTOR_SUBPARTS.
4967 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4968 as the number of units, setting it from VECTOR_CST_NELTS.
4969 (const_vector_from_tree): Likewise.
4970 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4971 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4972 (fold_negate_expr_1): Likewise.
4973 (fold_convert_const): Likewise.
4974 (const_binop): Likewise. Differentiate the number of output and
4976 (const_unop): Likewise.
4977 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4978 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4979 in cases that did the opposite.
4981 2017-09-14 Richard Biener <rguenther@suse.de>
4983 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4986 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4988 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4990 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4993 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4994 if and where to split a bb, except for splitting before debug insn
4995 sequences followed by non-label real insn. Delete debug insns
4996 in between basic blocks.
4998 * combine.c (make_compound_operation_int): Formatting fixes.
5000 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5001 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5002 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5003 * config/sol2.h (LINK_EH_SPEC): Likewise.
5004 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5005 * config/s390/linux.h (LINK_SPEC): Likewise.
5006 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5007 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5008 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5009 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5010 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5011 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5012 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5013 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5015 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5017 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5019 (ENDFILE_LINUX_SPEC): Likewise.
5020 (LINK_EH_SPEC): Likewise.
5021 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5022 (LINK_OS_LINUX_SPEC32): Likewise.
5023 (LINK_OS_LINUX_SPEC64): Likewise.
5024 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5025 (LINK_OS_LINUX_SPEC): Likewise.
5027 2017-09-13 Martin Liska <mliska@suse.cz>
5030 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5031 CASE_HIGH is NULL_TREE.
5033 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5034 Alan Hayward <alan.hayward@arm.com>
5035 David Sherwood <david.sherwood@arm.com>
5037 * target.def (secondary_memory_needed): New hook.
5038 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5039 instead of SECONDARY_MEMORY_NEEDED.
5040 (secondary_memory_needed_mode): Likewise.
5041 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5042 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5043 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5044 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5045 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5046 * doc/tm.texi: Regenerate.
5047 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5048 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5049 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5050 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5051 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5052 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5053 mode argument first and change the reg_class arguments to reg_class_t.
5054 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5055 Make static. Update the call to inline_secondary_memory_needed.
5056 (ix86_register_move_cost): Update the call to
5057 inline_secondary_memory_needed.
5058 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5059 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5061 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5062 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5064 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5065 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5066 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5067 and match hook interface. Add comment from mips.h.
5068 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5069 * config/mmix/mmix.md (truncdfsf2): Refer to
5070 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5072 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5073 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5074 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5075 (pa_secondary_memory_needed): New function.
5076 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5077 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5078 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5079 (pdp11_secondary_memory_needed): Make static and match hook interface.
5080 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5081 * config/powerpcspe/powerpcspe-protos.h
5082 (rs6000_secondary_memory_needed_ptr): Delete.
5083 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5085 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5086 (rs6000_option_override_internal): Assign to
5087 targetm.secondary_memory_needed rather than
5088 rs6000_secondary_memory_needed_ptr.
5089 (rs6000_secondary_memory_needed): Match hook interface.
5090 (rs6000_debug_secondary_memory_needed): Likewise.
5091 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5092 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5093 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5094 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5095 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5096 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5098 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5099 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5100 (rs6000_option_override_internal): Assign to
5101 targetm.secondary_memory_needed rather than
5102 rs6000_secondary_memory_needed_ptr.
5103 (rs6000_secondary_memory_needed): Match hook interface.
5104 (rs6000_debug_secondary_memory_needed): Likewise.
5105 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5106 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5107 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5108 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5109 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5110 (sparc_secondary_memory_needed): New function.
5111 * lra-constraints.c (check_and_process_move): Refer to
5112 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5114 (curr_insn_transform): Likewise.
5115 (process_alt_operands): Use targetm.secondary_memory_needed
5116 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5117 (check_secondary_memory_needed_p): Likewise.
5118 (choose_split_class): Likewise.
5119 * reload.c: Unconditionally include code that was previously
5120 conditional on SECONDARY_MEMORY_NEEDED.
5121 (push_secondary_reload): Use targetm.secondary_memory_needed
5122 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5123 (push_reload): Likewise.
5124 * reload1.c: Unconditionally include code that was previously
5125 conditional on SECONDARY_MEMORY_NEEDED.
5126 (choose_reload_regs): Use targetm.secondary_memory_needed
5127 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5128 (gen_reload): Likewise.
5129 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5131 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5132 Alan Hayward <alan.hayward@arm.com>
5133 David Sherwood <david.sherwood@arm.com>
5135 * target.def (secondary_memory_needed_mode): New hook:
5136 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5137 * targhooks.h (default_secondary_memory_needed_mode): New function.
5138 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5139 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5140 * doc/tm.texi: Regenerate.
5141 * lra-constraints.c (check_and_process_move): Use
5142 targetm.secondary_memory_needed_mode instead of
5143 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5144 (curr_insn_transform): Likewise.
5145 * reload.c (get_secondary_mem): Likewise.
5146 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5147 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5149 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5150 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5151 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5152 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5153 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5155 * config/powerpcspe/powerpcspe-protos.h
5156 (rs6000_secondary_memory_needed_mode): Delete.
5157 * config/powerpcspe/powerpcspe.c
5158 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5159 (rs6000_secondary_memory_needed_mode): Make static.
5160 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5161 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5163 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5165 (rs6000_secondary_memory_needed_mode): Make static.
5166 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5167 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5168 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5169 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5170 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5172 (sparc_secondary_memory_needed_mode): New function.
5173 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5175 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5177 * config/aarch64/constraints.md (Umq): New constraint.
5178 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5180 (mov<mode>): Update condition.
5182 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5184 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5185 when bitposition is the same.
5187 2017-09-13 Richard Biener <rguenther@suse.de>
5189 * dwarf2out.c (output_die_symbol): Remove.
5190 (output_die): Do not output a DIEs symbol.
5192 2017-09-13 Richard Biener <rguenther@suse.de>
5195 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5196 default-def to avoid breaking iterator update with the weird
5197 interaction with cgraph_update_edges_for_call_stmt_node.
5199 2017-09-13 Richard Biener <rguenther@suse.de>
5201 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5202 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5203 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5204 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5206 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5208 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5209 Disable pc relative literal load irrespective of
5210 TARGET_FIX_ERR_A53_84341 for default.
5212 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5214 * config/sparc/sparc.c (output_return): Output the source location of
5215 the insn in the delay slot, if any.
5216 (output_sibcall): Likewise.
5218 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5221 * common.opt (-static-pie): New alias.
5222 (shared): Negate static-pie.
5223 (-no-pie): Update help text.
5225 (static-pie): New option.
5226 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5227 -static-pie support.
5228 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5229 (LINK_EH_SPEC): Likewise.
5230 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5231 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5232 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5233 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5234 (init_gcc_specs): Likewise.
5235 (init_spec): Likewise.
5236 (display_help): Update help message for -pie.
5237 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5240 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5242 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5243 (movdi_aarch64): Likewise.
5244 (movti_aarch64): Likewise.
5246 2017-09-12 Simon Wright <simon@pushface.org>
5249 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5250 calculation of the minor version, always output as 0.
5252 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5255 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5256 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5257 on it early, rather than manual conversion late. For
5258 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5259 instead of performing manual conversion.
5261 2017-09-12 Carl Love <cel@us.ibm.com>
5263 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5264 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5266 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5267 VMULOSW): Add definitions.
5268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5269 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5270 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5272 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5274 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5276 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5278 (movti_aarch64): Likewise.
5279 (movdf_aarch64): Likewise.
5280 (movtf_aarch64): Likewise.
5281 (load_pairdi): Likewise.
5282 (store_pairdi): Likewise.
5283 (load_pairdf): Likewise.
5284 (store_pairdf): Likewise.
5285 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5286 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5287 (ldr_got_small_<mode>): Likewise.
5288 (ldr_got_small_28k_<mode>): Likewise.
5289 (ldr_got_tiny): Likewise.
5290 * config/aarch64/iterators.md (ldst_sz): New.
5291 (ldpstp_sz): Likewise.
5292 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5294 (thunderx_load): Split load_8 to load_16.
5295 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5297 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5298 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5299 (xgene1_store_pair): Split store_8 to store_16.
5300 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5301 (falkor_st_0_st_sd): Split store_8 to store_16.
5303 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5305 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5306 and store1/2/3/4 to store_4/8/12/16.
5307 * config/aarch64/aarch64.md: Update for rename.
5308 * config/arm/arm.md: Likewise.: Likewise.
5309 * config/arm/arm.c: Likewise.
5310 * config/arm/thumb1.md: Likewise.
5311 * config/arm/thumb2.md: Likewise.
5312 * config/arm/vfp.md: Likewise.
5313 * config/arm/arm-generic.md: Likewise.
5314 * config/arm/arm1020e.md: Likewise.
5315 * config/arm/arm1026ejs.md: Likewise.
5316 * config/arm/arm1136jfs.md: Likewise.
5317 * config/arm/arm926ejs.md: Likewise.
5318 * config/arm/cortex-a15.md: Likewise.
5319 * config/arm/cortex-a17.md: Likewise.
5320 * config/arm/cortex-a5.md: Likewise.
5321 * config/arm/cortex-a53.md: Likewise.
5322 * config/arm/cortex-a57.md: Likewise.
5323 * config/arm/cortex-a7.md: Likewise.
5324 * config/arm/cortex-a8.md: Likewise.
5325 * config/arm/cortex-a9.md: Likewise.
5326 * config/arm/cortex-m4.md: Likewise.
5327 * config/arm/cortex-m7.md: Likewise.
5328 * config/arm/cortex-r4.md: Likewise.
5329 * config/arm/exynos-m1.md: Likewise.
5330 * config/arm/fa526.md: Likewise.
5331 * config/arm/fa606te.md: Likewise.
5332 * config/arm/fa626te.md: Likewise.
5333 * config/arm/fa726te.md: Likewise.
5334 * config/arm/fmp626.md: Likewise.
5335 * config/arm/iwmmxt.md: Likewise.
5336 * config/arm/ldmstm.md: Likewise.
5337 * config/arm/marvell-pj4.md: Likewise.
5338 * config/arm/xgene1.md: Likewise.
5339 * config/aarch64/thunderx.md: Likewise.
5340 * config/aarch64/thunderx2t99.md: Likewise.
5341 * config/aarch64/falkor.md: Likewise.
5343 2017-09-12 Martin Liska <mliska@suse.cz>
5345 * attribs.c (private_lookup_attribute): New function.
5346 * attribs.h (private_lookup_attribute): Declared here.
5347 (lookup_attribute): Called from this place.
5349 2017-09-12 Richard Biener <rguenther@suse.de>
5351 PR tree-optimization/82157
5352 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5353 stmts with side-effects.
5355 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5356 Alan Hayward <alan.hayward@arm.com>
5357 David Sherwood <david.sherwood@arm.com>
5359 * target.def (hard_regno_nregs): New hook.
5360 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5361 * targhooks.h (default_hard_regno_nregs): Declare.
5362 * targhooks.c (default_hard_regno_nregs): New function.
5363 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5364 (TARGET_HARD_REGNO_NREGS): ...this hook.
5365 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5366 (CLASS_MAX_NREGS): Likewise.
5367 * doc/tm.texi: Regenerate.
5368 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5369 instead of HARD_REGNO_NREGS.
5370 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5371 HARD_REGNO_NREGS in the comment.
5372 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5373 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5374 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5375 Return an unsigned int.
5376 (TARGET_HARD_REGNO_NREGS): Redefine.
5377 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5378 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5379 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5380 (arc_hard_regno_nregs): New function.
5381 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5382 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5383 (arm_hard_regno_nregs): New function.
5384 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5385 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5386 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5387 (TARGET_HARD_REGNO_NREGS): Redefine.
5388 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5389 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5390 (HARD_REGNO_NREGS): Delete.
5391 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5392 (cr16_hard_regno_nregs): New function.
5393 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5394 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5395 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5396 (cris_hard_regno_nregs): New function.
5397 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5398 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5399 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5400 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5401 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5402 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5403 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5404 (frv_hard_regno_nregs): Make static. Take and return an
5406 (frv_class_max_nregs): Remove outdated copy of documentation.
5407 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5408 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5409 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5410 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5411 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5412 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5413 (TARGET_HARD_REGNO_NREGS): Redefine.
5414 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5415 (CLASS_MAX_NREGS): Update comment.
5416 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5417 (ia64_hard_regno_nregs): New function.
5418 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5419 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5420 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5421 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5422 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5424 (m32c_hard_regno_nregs): Likewise. Make static.
5425 (TARGET_HARD_REGNO_NREGS): Redefine.
5426 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5427 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5428 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5429 (m68k_hard_regno_nregs): New function.
5430 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5431 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5432 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5433 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5434 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5435 Take and return an unsigned int.
5436 (TARGET_HARD_REGNO_NREGS): Redefine.
5437 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5438 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5439 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5440 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5441 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5442 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5443 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5444 (msp430_hard_regno_nregs): Make static. Take and return an
5446 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5447 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5448 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5449 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5450 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5451 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5452 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5453 (TARGET_HARD_REGNO_NREGS): Redefine.
5454 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5455 (PA_HARD_REGNO_NREGS): ...this.
5456 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5457 (PA_HARD_REGNO_NREGS): ...this.
5458 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5459 (pa_hard_regno_nregs): New function.
5460 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5461 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5462 (pdp11_hard_regno_nregs): New function.
5463 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5464 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5465 (rs6000_hard_regno_nregs_hook): New function.
5466 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5467 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5468 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5469 Take and return an unsigned int. Move earlier in file.
5470 (TARGET_HARD_REGNO_NREGS): Redefine.
5471 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5472 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5473 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5474 (rl78_hard_regno_nregs): Make static. Take and return an
5476 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5477 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5478 (rs6000_hard_regno_nregs_hook): New function.
5479 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5480 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5481 (TARGET_HARD_REGNO_NREGS): Redefine.
5482 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5483 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5484 instead of HARD_REGNO_NREGS.
5485 (s390_hard_regno_nregs): New function.
5486 (s390_hard_regno_mode_ok): Add comment from s390.h.
5487 (TARGET_HARD_REGNO_NREGS): Redefine.
5488 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5489 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5490 (sh_hard_regno_nregs): New function.
5491 (sh_pass_in_reg_p): Use it.
5492 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5493 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5494 (sparc_hard_regno_nregs): New function.
5495 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5496 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5497 (spu_function_arg_advance): Use it, supplying a valid register number.
5498 (TARGET_HARD_REGNO_NREGS): Redefine.
5499 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5500 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5501 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5502 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5503 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5504 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5505 (CLASS_MAX_NREGS): Remove copy of old documentation.
5506 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5507 (visium_hard_regno_nregs): New function.
5508 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5509 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5510 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5511 xtensa_hard_regno_nregs): New function.
5512 * system.h (HARD_REGNO_NREGS): Poison.
5514 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5516 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5517 hard_regno_nregs instead of HARD_REGNO_NREGS.
5518 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5519 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5520 (c6x_expand_epilogue): Likewise.
5521 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5522 (frv_read_iacc_argument): Likewise.
5523 * config/sh/sh.c: Include regs.h.
5524 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5525 (regs_used): Likewise.
5526 (output_stack_adjust): Likewise.
5527 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5528 * expmed.c: Include regs.h.
5529 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5530 * ree.c: Include regs.h.
5531 (combine_reaching_defs): Use hard_regno_nregs instead of
5533 (add_removable_extension): Likewise.
5535 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5537 * regs.h (hard_regno_nregs): Turn into a function.
5538 (end_hard_regno): Update accordingly.
5539 * caller-save.c (setup_save_areas): Likewise.
5540 (save_call_clobbered_regs): Likewise.
5541 (replace_reg_with_saved_mem): Likewise.
5542 (insert_restore): Likewise.
5543 (insert_save): Likewise.
5544 * combine.c (can_change_dest_mode): Likewise.
5545 (move_deaths): Likewise.
5546 (distribute_notes): Likewise.
5547 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5548 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5549 (rs6000_split_multireg_move): Likewise.
5550 (rs6000_register_move_cost): Likewise.
5551 (rs6000_memory_move_cost): Likewise.
5552 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5553 (rs6000_split_multireg_move): Likewise.
5554 (rs6000_register_move_cost): Likewise.
5555 (rs6000_memory_move_cost): Likewise.
5556 * cselib.c (cselib_reset_table): Likewise.
5557 (cselib_lookup_1): Likewise.
5558 * emit-rtl.c (set_mode_and_regno): Likewise.
5559 * function.c (aggregate_value_p): Likewise.
5560 * ira-color.c (setup_profitable_hard_regs): Likewise.
5561 (check_hard_reg_p): Likewise.
5562 (calculate_saved_nregs): Likewise.
5563 (assign_hard_reg): Likewise.
5564 (improve_allocation): Likewise.
5565 (calculate_spill_cost): Likewise.
5566 * ira-emit.c (modify_move_list): Likewise.
5567 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5568 (ira_hard_reg_in_set_p): Likewise.
5569 * ira.c (setup_reg_mode_hard_regset): Likewise.
5570 (clarify_prohibited_class_mode_regs): Likewise.
5571 (check_allocation): Likewise.
5572 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5573 (lra_setup_reg_renumber): Likewise.
5574 (setup_try_hard_regno_pseudos): Likewise.
5575 (spill_for): Likewise.
5576 (assign_hard_regno): Likewise.
5577 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5578 * lra-constraints.c (in_class_p): Likewise.
5579 (lra_constraint_offset): Likewise.
5580 (simplify_operand_subreg): Likewise.
5581 (lra_constraints): Likewise.
5582 (split_reg): Likewise.
5583 (split_if_necessary): Likewise.
5584 (invariant_p): Likewise.
5585 (inherit_in_ebb): Likewise.
5586 * lra-lives.c (process_bb_lives): Likewise.
5587 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5588 (get_hard_regs): Likewise.
5589 (do_remat): Likewise.
5590 * lra-spills.c (assign_spill_hard_regs): Likewise.
5591 * mode-switching.c (create_pre_exit): Likewise.
5592 * postreload.c (reload_combine_recognize_pattern): Likewise.
5593 * recog.c (peep2_find_free_register): Likewise.
5594 * regcprop.c (kill_value_regno): Likewise.
5595 (set_value_regno): Likewise.
5596 (copy_value): Likewise.
5597 (maybe_mode_change): Likewise.
5598 (find_oldest_value_reg): Likewise.
5599 (copyprop_hardreg_forward_1): Likewise.
5600 * regrename.c (check_new_reg_p): Likewise.
5601 (regrename_do_replace): Likewise.
5602 * reload.c (push_reload): Likewise.
5603 (combine_reloads): Likewise.
5604 (find_dummy_reload): Likewise.
5605 (operands_match_p): Likewise.
5606 (find_reloads): Likewise.
5607 (find_equiv_reg): Likewise.
5608 (reload_adjust_reg_for_mode): Likewise.
5609 * reload1.c (count_pseudo): Likewise.
5610 (count_spilled_pseudo): Likewise.
5611 (find_reg): Likewise.
5612 (clear_reload_reg_in_use): Likewise.
5613 (free_for_value_p): Likewise.
5614 (allocate_reload_reg): Likewise.
5615 (choose_reload_regs): Likewise.
5616 (reload_adjust_reg_for_temp): Likewise.
5617 (emit_reload_insns): Likewise.
5618 (delete_output_reload): Likewise.
5619 * rtlanal.c (subreg_get_info): Likewise.
5620 * sched-deps.c (sched_analyze_reg): Likewise.
5621 * sel-sched.c (init_regs_for_mode): Likewise.
5622 (mark_unavailable_hard_regs): Likewise.
5623 (choose_best_reg_1): Likewise.
5624 (verify_target_availability): Likewise.
5625 * valtrack.c (dead_debug_insert_temp): Likewise.
5626 * var-tracking.c (track_loc_p): Likewise.
5627 (emit_note_insn_var_location): Likewise.
5628 * varasm.c (make_decl_rtl): Likewise.
5629 * reginfo.c (choose_hard_reg_mode): Likewise.
5630 (init_reg_modes_target): Refer directly to
5631 this_target_regs->x_hard_regno_nregs.
5633 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5635 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5636 instead of hard_regno_nregs.
5638 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5640 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5641 end_hard_regno instead of hard_regno_nregs.
5642 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5643 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5644 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5645 * ira-color.c (improve_allocation): Likewise.
5646 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5647 * lra-lives.c (mark_regno_live): Likewise.
5648 (mark_regno_dead): Likewise.
5649 * lra-remat.c (operand_to_remat): Likewise.
5650 * lra.c (collect_non_operand_hard_regs): Likewise.
5651 * postreload.c (reload_combine_note_store): Likewise.
5652 (move2add_valid_value_p): Likewise.
5653 * reload.c (regno_clobbered_p): Likewise.
5655 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5657 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5659 * config/v850/v850.c (v850_reorg): Likewise.
5660 * reload.c (refers_to_regno_for_reload_p): Likewise.
5661 (find_equiv_reg): Likewise.
5662 * reload1.c (reload_reg_reaches_end_p): Likewise.
5664 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5666 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5668 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5669 * config/arm/arm.c (output_move_neon): Likewise.
5670 (arm_attr_length_move_neon): Likewise.
5671 (neon_split_vcombine): Likewise.
5672 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5673 (c6x_mark_reg_written): Likewise.
5674 (c6x_dwarf_register_span): Likewise.
5675 * config/i386/i386.c (ix86_save_reg): Likewise.
5676 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5677 (rws_access_reg): Likewise.
5678 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5679 * mode-switching.c (create_pre_exit): Likewise.
5680 * ree.c (combine_reaching_defs): Likewise.
5681 (add_removable_extension): Likewise.
5682 * regcprop.c (find_oldest_value_reg): Likewise.
5683 (copyprop_hardreg_forward_1): Likewise.
5684 * reload.c (reload_inner_reg_of_subreg): Likewise.
5685 (push_reload): Likewise.
5686 (combine_reloads): Likewise.
5687 (find_dummy_reload): Likewise.
5688 (reload_adjust_reg_for_mode): Likewise.
5689 * reload1.c (find_reload_regs): Likewise.
5690 (forget_old_reloads_1): Likewise.
5691 (reload_reg_free_for_value_p): Likewise.
5692 (reload_adjust_reg_for_temp): Likewise.
5693 (emit_reload_insns): Likewise.
5694 (delete_output_reload): Likewise.
5695 * sel-sched.c (choose_best_reg_1): Likewise.
5696 (choose_best_pseudo_reg): Likewise.
5698 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5699 Alan Hayward <alan.hayward@arm.com>
5700 David Sherwood <david.sherwood@arm.com>
5702 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5703 * target.def (slow_unaligned_access): New hook.
5704 * targhooks.h (default_slow_unaligned_access): Declare.
5705 * targhooks.c (default_slow_unaligned_access): New function.
5706 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5707 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5708 * doc/tm.texi: Regenerate.
5709 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5710 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5711 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5713 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5714 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5716 (rs6000_slow_unaligned_access): New function.
5717 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5718 (expand_block_compare): Likewise.
5719 (expand_strn_compare): Likewise.
5720 (rs6000_rtx_costs): Likewise.
5721 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5722 (riscv_slow_unaligned_access): Likewise.
5723 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5724 (riscv_slow_unaligned_access_p): ...this and make static.
5725 (riscv_option_override): Update accordingly.
5726 (riscv_slow_unaligned_access): New function.
5727 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5728 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5729 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5730 (rs6000_slow_unaligned_access): New function.
5731 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5732 (rs6000_rtx_costs): Likewise.
5733 * config/rs6000/rs6000-string.c (expand_block_compare)
5734 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5735 of SLOW_UNALIGNED_ACCESS.
5736 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5737 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5738 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5739 of SLOW_UNALIGNED_ACCESS.
5740 * expmed.c (simple_mem_bitfield_p): Likewise.
5741 * expr.c (alignment_for_piecewise_move): Likewise.
5742 (emit_group_load_1): Likewise.
5743 (emit_group_store): Likewise.
5744 (copy_blkmode_from_reg): Likewise.
5745 (emit_push_insn): Likewise.
5746 (expand_assignment): Likewise.
5747 (store_field): Likewise.
5748 (expand_expr_real_1): Likewise.
5749 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5750 * lra-constraints.c (simplify_operand_subreg): Likewise.
5751 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5752 * gimple-ssa-store-merging.c: Likewise in block comment at start
5754 * tree-ssa-strlen.c: Include target.h.
5755 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5756 of SLOW_UNALIGNED_ACCESS.
5757 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5759 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5761 PR rtl-optimization/82185
5762 * expmed.c (emit_store_flag_int): Only test tem if it has been
5765 2017-09-12 Richard Biener <rguenther@suse.de>
5768 * match.pd ((FTYPE) N CMP CST): Fix typo.
5770 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5772 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5774 (mips_near_type_p): Add 'short_call' attribute as a synonym
5776 * doc/extend.texi (short_call): Document new function attribute.
5778 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5781 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5782 assertion check that in the condition.
5783 (get_atomic_generic_size): Likewise. Before testing if parameter
5784 has pointer type, if it has array type, call for C++
5785 default_conversion to perform array-to-pointer conversion.
5787 2017-09-12 Richard Biener <rguenther@suse.de>
5789 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5790 for operations we cannot scalarize.
5792 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5794 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5795 vectors heap vectors. Clean up comments.
5796 Make visited_bbs a reference.
5797 (profitable_jump_thread_path): Make GC
5798 vectors heap vectors. Clean up comments.
5800 (convert_and_register_jump_thread_path): Make GC vectors heap
5802 (check_subpath_and_update_thread_path): Same. Clean up comments.
5803 Make visited_bbs a reference.
5804 (handle_phi): Abstract common code to to
5805 register_jump_thread_path_if_profitable.
5806 Rename VAR_BB to DEF_BB.
5808 Make GC vectors heap vectors.
5809 Make visited_bbs a reference.
5810 (handle_assignment): Same.
5811 (register_jump_thread_path_if_profitable): New.
5812 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5814 Make GC vectors heap vectors. Clean up comments.
5815 Make visited_bbs a reference.
5816 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5817 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5820 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5823 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5824 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5826 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5828 Revert r251800 and r251799.
5830 2017-09-11 Martin Jambor <mjambor@suse.cz>
5833 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5834 arguments in advance.
5835 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5836 use it to find predecessor edges.
5837 (naive_outof_ssa): Collect vector of predecessors.
5839 2017-09-08 Jason Merrill <jason@redhat.com>
5841 PR c++/70029 - ICE with ref-qualifier and -flto
5842 * langhooks.h (struct lang_hooks_for_types): Add
5843 copy_lang_qualifiers.
5844 * attribs.c (build_type_attribute_qual_variant): Use it.
5845 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5847 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5848 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5850 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5853 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5854 (*mulsi3_sp64): New instruction.
5855 (mulsi3): New expander.
5857 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5859 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5861 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5863 * sancov.c: Include memmodel.h.
5865 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5868 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5871 2017-09-07 Carl Love <cel@us.ibm.com>
5873 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5874 the sldi instruction.
5876 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5878 * sancov.c: Include tm_p.h.
5880 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5883 * output.h (switch_to_other_text_partition): New declaration.
5884 * varasm.c (switch_to_other_text_partition): New function.
5885 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5886 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5887 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5888 to the other text partition before emitting LCL label and switch back
5889 after emitting the word after it.
5891 2017-09-07 Richard Biener <rguenther@suse.de>
5893 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5894 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5895 critical edges here, after loop init.
5896 (pass_data_pre): Remove PROP_no_crit_edges flags.
5897 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5898 for valueization of call args to avoid leaking VN_TOP.
5899 (visit_use): Assert we do not visit default defs.
5900 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5901 Use error_mark_node to more easily detect leaking VN_TOP.
5902 All default-defs are varying, not VN_TOP. Mark them visited.
5903 (run_scc_vn): Make code match comment.
5905 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5907 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5908 OPTION_MASK_FLOAT128_KEYWORD.
5909 (POWERPC_MASKS): Likewise.
5910 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5911 support for the -mfloat128-type option, and make -mfloat128
5912 default on PowerPC Linux systems. Define or undefine
5913 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5914 Define __float128 to be __ieee128 if IEEE 128-bit support is
5915 enabled, or undefine it.
5916 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5917 Delete defining __FLOAT128_TYPE__.
5918 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5919 -mfloat128-type option and make -mfloat128 default on PowerPC
5921 (TARGET_FLOAT128_TYPE): Likewise.
5922 (-mfloat128-type): Likewise.
5923 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5924 Delete the -mfloat128-type option and make -mfloat128 default on
5925 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5926 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5927 128-bit floating point is enabled. Change tests from using
5928 -mfloat128-type to -mfloat128.
5929 (rs6000_mangle_type): Use the correct mangling for the __float128
5930 type even if normal long double is restricted to 64-bits.
5931 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5933 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5934 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5935 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5937 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5938 documentation for -mfloat128.
5940 2017-09-06 Olivier Hainque <hainque@adacore.com>
5942 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5944 2017-09-06 Wish Wu <wishwu007@gmail.com>
5945 Jakub Jelinek <jakub@redhat.com>
5947 * asan.c (initialize_sanitizer_builtins): Add
5948 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5949 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5950 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5951 BT_FN_VOID_UINT64_PTR variables.
5952 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5953 (BT_FN_VOID_UINT16_UINT16): Likewise.
5954 (BT_FN_VOID_UINT32_UINT32): Likewise.
5955 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5956 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5957 (BT_FN_VOID_UINT64_PTR): Likewise.
5958 * common.opt (flag_sanitize_coverage): New variable.
5959 (fsanitize-coverage=trace-pc): Remove.
5960 (fsanitize-coverage=): Add.
5961 * flag-types.h (enum sanitize_coverage_code): New enum.
5962 * fold-const.c (fold_range_test): Disable non-short-circuit
5963 optimization if flag_sanitize_coverage.
5964 (fold_truth_andor): Likewise.
5965 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5966 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5967 (coverage_sanitizer_opts): New array.
5968 (get_closest_sanitizer_option): Add OPTS argument, handle also
5969 OPT_fsanitize_coverage_.
5970 (parse_sanitizer_options): Adjusted to also handle
5971 OPT_fsanitize_coverage_.
5972 (common_handle_option): Add OPT_fsanitize_coverage_.
5973 * sancov.c (instrument_comparison, instrument_switch): New function.
5974 (sancov_pass): Add trace-cmp support.
5975 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5976 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5977 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5978 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5979 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5980 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5981 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5982 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5983 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5984 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5986 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5988 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5989 error. Only quit immediately if parsing is complete.
5990 (BEGIN): Initialize fatal_err and parse_done.
5991 (begin fpu, end fpu): Check number of arguments.
5992 (begin arch, end arch): Likewise.
5993 (begin cpu, end cpu): Likewise.
5994 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5995 (optalias): Likewise.
5997 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5999 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6000 * config/arm/arm-isa.h: Delete. Move definitions to ...
6001 * arm-cpus.in: ... here. Use new feature and fgroup values.
6002 * config/arm/arm.c (arm_option_override): Use lower case for feature
6004 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6005 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6006 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6008 (print_isa_bits_for): New function.
6009 (gen_isa): New function.
6010 (gen_comm_data): Use print_isa_bits_for.
6011 (define feature): New keyword.
6012 (define fgroup): New keyword.
6013 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6014 (arm-isa.h): Add rule to generate file.
6015 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6016 case for feature bit names.
6018 2017-09-06 Richard Biener <rguenther@suse.de>
6020 * tree-ssa-pre.c (NECESSARY): Remove.
6021 (create_expression_by_pieces): Do not touch pass-local flags.
6022 (insert_into_preds_of_block): Likewise.
6023 (do_pre_regular_insertion): Likewise.
6024 (eliminate_insert): Likewise.
6025 (eliminate_dom_walker::before_dom_children): Likewise.
6026 (fini_eliminate): Do not look at inserted_exprs.
6027 (mark_operand_necessary): Remove.
6028 (remove_dead_inserted_code): Replace with simple work-list
6029 algorithm based on inserted_exprs and SSA uses.
6030 (pass_pre::execute): Re-order fini_eliminate and
6031 remove_dead_inserted_code.
6033 2017-09-06 Olivier Hainque <hainque@adacore.com>
6035 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6036 for VxWorks 7. Adjust surrounding comments.
6038 2017-09-06 Richard Biener <rguenther@suse.de>
6040 * gimple-ssa-strength-reduction.c
6041 (find_candidates_dom_walker::before_dom_children): Also allow
6044 2017-09-06 Richard Biener <rguenther@suse.de>
6046 PR tree-optimization/82108
6047 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6048 for gap in the non-permutation SLP case.
6050 2017-09-06 Martin Jambor <mjambor@suse.cz>
6052 PR tree-optimization/82078
6053 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6054 add_access_to_work_queue...
6055 (build_accesses_from_assign): ...here.
6056 (propagate_all_subaccesses): Make sure racc is the group
6057 representative, if there is one.
6059 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6062 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6065 2017-09-06 Richard Biener <rguenther@suse.de>
6067 * gimple-ssa-strength-reduction.c
6068 (find_candidates_dom_walker::before_doom_children): Use a
6069 type and not a mode check.
6071 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6074 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6076 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6078 2017-09-05 Jeff Law <law@redhat.com>
6080 PR tree-optimization/64910
6081 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6082 cases where we have 3 or more operands.
6084 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6087 * omp-low.c (lower_omp_for): Recompute tree invariant if
6088 gimple_omp_for_initial/final is ADDR_EXPR.
6091 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6092 into gimple val before gimplification fo the COND_EXPR.
6094 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6096 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6097 REGION_COPY argument.
6098 (thread_through_all_blocks): Remove unused argument to
6099 duplicate_thread_path.
6101 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6102 Alan Hayward <alan.hayward@arm.com>
6103 David Sherwood <david.sherwood@arm.com>
6105 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6106 Take a scalar_mode rather than a machine_mode.
6107 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6108 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6109 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6110 (aarch64_gen_adjusted_ldpstp): Likewise.
6111 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6113 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6114 Alan Hayward <alan.hayward@arm.com>
6115 David Sherwood <david.sherwood@arm.com>
6117 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6118 Take a scalar_int_mode instead of a machine_mode.
6119 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6120 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6121 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6122 (aarch64_simd_attr_length_rglist): Delete.
6123 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6124 a scalar_int_mode instead of a machine_mode.
6125 (aarch64_add_offset): Likewise.
6126 (aarch64_internal_mov_immediate): Likewise
6127 (aarch64_add_constant_internal): Likewise.
6128 (aarch64_add_constant): Likewise.
6129 (aarch64_movw_imm): Likewise.
6130 (aarch64_rtx_arith_op_extract_p): Likewise.
6131 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6132 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6133 Remove assert that the mode isn't a vector.
6134 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6135 (aarch64_expand_mov_immediate): Update calls after above changes.
6136 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6137 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6138 (aarch64_move_imm): Likewise.
6139 (aarch64_can_const_movi_rtx_p): Likewise.
6140 (aarch64_strip_extend): Likewise.
6141 (aarch64_extr_rtx_p): Likewise.
6142 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6143 a CONST_INT when the mode parameter is VOIDmode.
6144 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6146 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6148 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6149 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6150 (bitwise_type_for_mode): Update accordingly.
6152 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6154 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6155 * stor-layout.c (mode_for_size_tree): Likewise.
6156 (mode_for_array): Update accordingly.
6157 (layout_decl): Likewise.
6158 (compute_record_mode): Likewise. Only set the mode once.
6160 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6162 * target.def (get_mask_mode): Change return type to opt_mode.
6164 * doc/tm.texi: Regenerate.
6165 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6166 * targhooks.c (default_get_mask_mode): Likewise.
6167 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6168 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6169 targetm.get_mask_mode.
6170 * tree.c (build_truth_vector_type): Likewise.
6172 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6174 * machmode.h (mode_for_vector): Return an opt_mode.
6175 * stor-layout.c (mode_for_vector): Likewise.
6176 (mode_for_int_vector): Update accordingly.
6177 (layout_type): Likewise.
6178 * config/i386/i386.c (emit_memmov): Likewise.
6179 (ix86_expand_set_or_movmem): Likewise.
6180 (ix86_expand_vector_init): Likewise.
6181 (ix86_get_mask_mode): Likewise.
6182 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6184 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6185 * expmed.c (extract_bit_field_1): Likewise.
6186 * expr.c (expand_expr_real_2): Likewise.
6187 * optabs-query.c (can_vec_perm_p): Likewise.
6188 (can_vec_mask_load_store_p): Likewise.
6189 * optabs.c (expand_vec_perm): Likewise.
6190 * targhooks.c (default_get_mask_mode): Likewise.
6191 * tree-vect-stmts.c (vectorizable_store): Likewise.
6192 (vectorizable_load): Likewise.
6193 (get_vectype_for_scalar_type_and_size): Likewise.
6195 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6197 * machmode.h (mode_for_int_vector): New function.
6198 * stor-layout.c (mode_for_int_vector): Likewise.
6199 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6200 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6201 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6202 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6203 (s390_expand_vcond): Likewise.
6205 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6207 * machmode.h (opt_machine_mode): New type.
6208 (opt_mode<T>): Allow construction from anything that can be
6210 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6211 (mode_for_size): Return an opt_machine_mode.
6212 * stor-layout.c (mode_for_size): Likewise.
6213 (mode_for_size_tree): Update call accordingly.
6214 (bitwise_mode_for_mode): Likewise.
6215 (make_fract_type): Likewise.
6216 (make_accum_type): Likewise.
6217 * caller-save.c (replace_reg_with_saved_mem): Update call
6219 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6220 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6221 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6222 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6223 * expmed.c (extract_bit_field_1): Likewise.
6224 * reload.c (get_secondary_mem): Likewise.
6225 * varasm.c (assemble_integer): Likewise.
6226 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6229 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6231 * machmode.h (decimal_float_mode_for_size): New function.
6232 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6233 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6234 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6235 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6236 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6237 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6239 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6241 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6242 (get_builtin_sync_mode): Likewise.
6243 (expand_ifn_atomic_compare_exchange): Likewise.
6244 (expand_builtin_atomic_clear): Likewise.
6245 (expand_builtin_atomic_test_and_set): Likewise.
6246 (fold_builtin_atomic_always_lock_free): Likewise.
6247 * calls.c (compute_argument_addresses): Likewise.
6248 (emit_library_call_value_1): Likewise.
6249 (store_one_arg): Likewise.
6250 * combine.c (combine_instructions): Likewise.
6251 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6252 * config/arm/arm.c (arm_function_value): Likewise.
6253 (aapcs_allocate_return_reg): Likewise.
6254 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6255 * config/i386/i386.c (construct_container): Likewise.
6256 (ix86_gimplify_va_arg): Likewise.
6257 (ix86_expand_sse_cmp): Likewise.
6258 (emit_memmov): Likewise.
6259 (emit_memset): Likewise.
6260 (expand_small_movmem_or_setmem): Likewise.
6261 (ix86_expand_pextr): Likewise.
6262 (ix86_expand_pinsr): Likewise.
6263 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6264 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6266 * config/mips/mips.c (mips_function_value_1) Likewise.
6267 (mips_block_move_straight): Likewise.
6268 (mips_expand_ins_as_unaligned_store): Likewise.
6269 * config/powerpcspe/powerpcspe.c
6270 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6271 (rs6000_darwin64_record_arg_flush): Likewise.
6272 * config/rs6000/rs6000.c
6273 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6274 (rs6000_darwin64_record_arg_flush): Likewise.
6275 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6276 (sparc_function_value_1): Likewise.
6277 * config/spu/spu.c (adjust_operand): Likewise.
6278 (spu_emit_branch_or_set): Likewise.
6279 (arith_immediate_p): Likewise.
6280 * emit-rtl.c (gen_lowpart_common): Likewise.
6281 * expr.c (expand_expr_real_1): Likewise.
6282 * function.c (assign_parm_setup_block): Likewise.
6283 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6284 * reload1.c (alter_reg): Likewise.
6285 * stor-layout.c (mode_for_vector): Likewise.
6286 (layout_type): Likewise.
6288 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6290 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6291 (spu_convert_move): Likewise.
6292 * lower-subreg.c (resolve_simple_move): Likewise.
6294 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6297 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6298 define_insn to a define_expand.
6299 (altivec_vsum2sws_direct): New define_insn.
6300 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6302 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6304 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6307 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6312 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6313 realign stack if stack alignment needed is less than incoming
6316 2017-09-05 Marek Polacek <polacek@redhat.com>
6319 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6322 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6324 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6326 2017-09-05 Richard Biener <rguenther@suse.de>
6328 PR tree-optimization/82084
6329 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6331 2017-09-05 Richard Biener <rguenther@suse.de>
6333 PR tree-optimization/82102
6334 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6336 2017-09-05 Martin Liska <mliska@suse.cz>
6338 PR tree-optimization/82032
6339 * tree-cfg.c (generate_range_test): New function.
6340 * tree-cfg.h (generate_range_test): Declared here.
6341 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6342 (cleanup_control_expr_graph): Use it.
6343 * tree-switch-conversion.c (try_switch_expansion): Remove
6345 (emit_case_nodes): Use generate_range_test.
6347 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6350 * config/i386/i386.md (*<btsc><mode>_mask): Add
6351 TARGET_USE_BT to insn constraint.
6352 (*btr<mode>_mask): Ditto.
6354 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6356 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6357 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6359 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6362 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6363 TARGET_NEON and TARGET_IWMMXT.
6364 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6365 TARGET_NEON and TARGET_IWMMXT.
6366 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6368 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6370 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6371 (ix86_rewrite_tls_address): Ditto.
6372 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6373 (ix86_rewrite_tls_address_1): Ditto.
6374 (ix86_rewrite_tls_address): Ditto.
6375 * config/i386/predicates.md (tls_address_pattern): New predicate.
6376 * config/i386/i386.md (TLS address splitter): New splitter.
6378 2017-09-04 Richard Biener <rguenther@suse.de>
6380 PR tree-optimization/82084
6381 * fold-const.h (can_native_encode_string_p): Declare.
6382 * fold-const.c (can_native_encode_string_p): Factor out from ...
6383 (native_encode_string): ... here.
6384 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6385 vectorizing stores from constants we later cannot handle.
6387 2017-09-04 Marek Polacek <polacek@redhat.com>
6390 * doc/invoke.texi: Update -Wtautological-compare documentation.
6392 2017-09-04 Jeff Law <law@redhat.com>
6394 PR tree-optimization/64910
6395 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6396 swap the first and last operand if the last is a constant.
6398 2017-09-04 Marek Polacek <polacek@redhat.com>
6401 * convert.c (do_narrow): When sanitizing signed integer overflows,
6402 bail out for signed types.
6403 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6405 2017-09-04 Richard Biener <rguenther@suse.de>
6407 PR tree-optimization/82060
6408 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6409 Move devirtualization after stmt folding and before EH/AB/noreturn
6410 cleanup to get the stmt refs canonicalized. Use a bool instead
6411 of gimple_modified_p since that doesn't work for NOPs. Schedule
6412 NOPs generated by folding for removal.
6414 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6415 Alan Hayward <alan.hayward@arm.com>
6416 David Sherwood <david.sherwood@arm.com>
6418 * coretypes.h (pad_direction): New enum.
6419 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6420 (FUNCTION_ARG_PADDING): Likewise.
6421 * target.def (function_arg_padding): New hook.
6422 * targhooks.h (default_function_arg_padding): Declare.
6423 * targhooks.c (default_function_arg_padding): New function.
6424 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6425 (TARGET_FUNCTION_ARG_PADDING): ...this.
6426 * doc/tm.texi: Regenerate.
6427 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6428 instead of direction.
6429 (compute_argument_addresses): Likewise.
6430 (load_register_parameters): Likewise.
6431 (emit_library_call_value_1): Likewise.
6432 (store_one_arg): Use targetm.calls.function_arg_padding instead
6433 of FUNCTION_ARG_PADDING.
6434 (must_pass_in_stack_var_size_or_pad): Likewise.
6435 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6436 (emit_group_store): Likewise.
6437 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6438 instead of FUNCTION_ARG_PADDING.
6439 (emit_push_insn): Likewise, and propagate enum change throughout
6441 * function.h (direction): Delete.
6442 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6444 * function.c (assign_parm_find_stack_rtl): Likewise.
6445 (assign_parm_setup_block_p): Likewise.
6446 (assign_parm_setup_block): Likewise.
6447 (gimplify_parameters): Likewise.
6448 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6449 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6451 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6452 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6453 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6454 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6455 (aarch64_function_arg_padding): ...this new function.
6456 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6457 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6458 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6459 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6460 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6461 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6462 (arm_pad_arg_upward): Replace with...
6463 (arm_function_arg_padding): ...this new function.
6464 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6466 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6467 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6468 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6469 (ia64_hpux_function_arg_padding): Replace with...
6470 (ia64_function_arg_padding): ...this new function. Use pad_direction
6471 instead of direction. Check for TARGET_HPUX.
6472 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6473 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6474 (iq2000_function_arg_padding): New function.
6475 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6476 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6477 (mips_function_arg_padding): ...this new function.
6478 (mips_pad_reg_upward): Update accordingly.
6479 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6480 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6481 targetm.calls.function_arg_padding.
6482 (FUNCTION_ARG_PADDING): Delete.
6483 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6484 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6485 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6486 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6487 (nios2_block_reg_padding): Return pad_direction instead of direction.
6488 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6489 instead of direction.
6490 (nios2_function_arg_padding): Likewise. Make static.
6491 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6492 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6493 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6494 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6495 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6496 (pa_function_arg_padding): Make static. Return pad_direction instead
6498 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6499 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6500 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6501 instead of direction. Use targetm.calls.function_arg_padding.
6502 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6503 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6504 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6505 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6506 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6508 (function_arg_padding): Rename to...
6509 (rs6000_function_arg_padding): ...this. Make static. Return
6510 pad_direction instead of direction.
6511 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6512 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6513 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6514 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6515 instead of direction. Use targetm.calls.function_arg_padding.
6516 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6517 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6518 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6519 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6520 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6521 (function_arg_padding): Rename to...
6522 (rs6000_function_arg_padding): ...this. Make static. Return
6523 pad_direction instead of direction.
6524 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6525 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6526 * config/s390/s390.c (s390_function_arg_padding): New function.
6527 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6528 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6529 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6530 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6531 (function_arg_padding): Rename to...
6532 (sparc_function_arg_padding): ...this. Make static. Return
6533 pad_direction instead of direction.
6534 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6535 * config/spu/spu.c (spu_function_arg_padding): New function.
6536 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6537 * system.h (FUNCTION_ARG_PADDING): Poison.
6539 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6540 Alan Hayward <alan.hayward@arm.com>
6541 David Sherwood <david.sherwood@arm.com>
6543 * target.def (modes_tieable_p): New hook.
6544 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6545 (TARGET_MODES_TIEABLE_P): ...this.
6546 * doc/tm.texi.in: Regenerate.
6547 * hooks.h (hook_bool_mode_mode_true): Declare.
6548 * hooks.c (hook_bool_mode_mode_true): New function.
6549 * combine.c (subst): Use targetm.modes_tieable_p instead of
6551 * dse.c (find_shift_sequence): Likewise.
6552 * expmed.c (extract_low_bits): Likewise.
6553 * lower-subreg.c: Include target.h.
6554 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6556 * rtlanal.c (rtx_cost): Likewise.
6557 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6558 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6559 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6560 (TARGET_MODES_TIEABLE_P): Redefine.
6561 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6562 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6563 (TARGET_MODES_TIEABLE_P): Redefine.
6564 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6565 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6566 (arc_modes_tieable_p): New function.
6567 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6568 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6569 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6570 (arm_modes_tieable_p): Make static.
6571 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6572 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6573 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6574 (TARGET_MODES_TIEABLE_P): Redefine.
6575 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6576 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6577 (TARGET_MODES_TIEABLE_P): Redefine.
6578 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6579 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6580 (cr16_modes_tieable_p): New function.
6581 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6582 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6583 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6584 (TRULY_NOOP_TRUNCATION): Update comment.
6585 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6586 (TRULY_NOOP_TRUNCATION): Update comment.
6587 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6588 (frv_modes_tieable_p): New function.
6589 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6590 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6591 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6592 (TARGET_MODES_TIEABLE_P): Redefine.
6593 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6594 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6595 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6596 (TARGET_MODES_TIEABLE_P): Redefine.
6597 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6598 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6599 (ia64_modes_tieable_p): New function.
6600 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6601 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6602 (iq2000_modes_tieable_p): New function.
6603 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6604 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6605 (lm32_modes_tieable_p): New function.
6606 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6607 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6608 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6609 (TARGET_MODES_TIEABLE_P): Redefine.
6610 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6611 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6612 (m32r_modes_tieable_p): New function.
6613 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6614 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6615 (m68k_modes_tieable_p): New function.
6616 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6617 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6618 (mcore_modes_tieable_p): New function.
6619 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6620 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6622 (TARGET_MODES_TIEABLE_P): Redefine.
6623 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6624 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6625 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6626 (TARGET_MODES_TIEABLE_P): Redefine.
6627 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6628 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6629 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6630 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6631 (mn10300_modes_tieable_p): ...this and make static.
6632 (TARGET_MODES_TIEABLE_P): Redefine.
6633 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6634 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6635 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6636 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6637 (msp430_modes_tieable_p): Make static.
6638 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6639 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6640 (TARGET_MODES_TIEABLE_P): Redefine.
6641 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6642 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6643 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6644 (TARGET_MODES_TIEABLE_P): Redefine.
6645 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6646 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6647 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6648 (TARGET_MODES_TIEABLE_P): Redefine.
6649 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6650 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6651 (pdp11_modes_tieable_p): New function.
6652 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6653 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6654 (rs6000_modes_tieable_p): New function.
6655 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6656 * config/powerpcspe/powerpcspe.md: Update comment.
6657 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6658 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6659 (TARGET_MODES_TIEABLE_P): Redefine.
6660 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6661 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6662 (rl78_modes_tieable_p): New function.
6663 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6664 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6665 (rs6000_modes_tieable_p): New function.
6666 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6667 * config/rs6000/rs6000.md: Update comment.
6668 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6669 * config/rx/rx.c (rx_modes_tieable_p): New function.
6670 (TARGET_MODES_TIEABLE_P): Redefine.
6671 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6672 * config/s390/s390.c (s390_modes_tieable_p): New function.
6673 (TARGET_MODES_TIEABLE_P): Redefine.
6674 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6675 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6676 (sh_modes_tieable_p): New function.
6677 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6678 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6679 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6680 (sparc_modes_tieable_p): Make static.
6681 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6682 * config/spu/spu.c (spu_modes_tieable_p): New function.
6683 (TARGET_MODES_TIEABLE_P): Redefine.
6684 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6685 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6686 (TARGET_MODES_TIEABLE_P): Redefine.
6687 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6688 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6689 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6690 * config/v850/v850.c (v850_modes_tieable_p): New function.
6691 (TARGET_MODES_TIEABLE_P): Redefine.
6692 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6693 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6694 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6695 (visium_modes_tieable_p): New function.
6696 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6697 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6698 (xtensa_modes_tieable_p): New function.
6699 * system.h (MODES_TIEABLE_P): Poison.
6701 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6702 Alan Hayward <alan.hayward@arm.com>
6703 David Sherwood <david.sherwood@arm.com>
6705 * target.def (hard_regno_mode_ok): New hook.
6706 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6707 (TARGET_HARD_REGNO_MODE_OK): ...this.
6708 * doc/tm.texi.in: Regenerate.
6709 * hooks.h (hook_bool_uint_mode_true): Declare.
6710 * hooks.c (hook_bool_uint_mode_true): New function.
6711 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6713 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6714 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6715 instead of HARD_REGNO_MODE_OK.
6716 * caller-save.c: Include target.h.
6717 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6719 * combine.c (can_combine_p): Likewise.
6720 (combinable_i3pat): Likewise.
6721 (can_change_dest_mode): Likewise.
6722 * expr.c (init_expr_target): Likewise.
6723 (convert_move): Likewise.
6724 (convert_modes): Likewise.
6725 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6726 (setup_prohibited_mode_move_regs): Likewise.
6727 * ira.h (target_ira): Likewise.
6728 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6729 * lra-constraints.c (process_alt_operands): Likewise.
6730 (split_reg): Likewise.
6731 * recog.c (peep2_find_free_register): Likewise.
6732 * ree.c (combine_reaching_defs): Likewise.
6733 * regcprop.c (maybe_mode_change): Likewise.
6734 * reginfo.c (init_reg_sets_1): Likewise.
6735 (choose_hard_reg_mode): Likewise.
6736 (simplifiable_subregs): Likewise.
6737 * regrename.c (check_new_reg_p): Likewise.
6738 * reload.c (find_valid_class): Likewise.
6739 (find_valid_class_1): Likewise.
6740 (reload_inner_reg_of_subreg): Likewise.
6741 (push_reload): Likewise.
6742 (combine_reloads): Likewise.
6743 (find_dummy_reload): Likewise.
6744 (find_reloads): Likewise.
6745 * reload1.c (find_reg): Likewise.
6746 (set_reload_reg): Likewise.
6747 (allocate_reload_reg): Likewise.
6748 (choose_reload_regs): Likewise.
6749 (reload_adjust_reg_for_temp): Likewise.
6750 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6751 (simplify_subreg_regno): Likewise.
6752 * sel-sched.c (init_regs_for_mode): Likewise.
6753 * varasm.c (make_decl_rtl): Likewise.
6754 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6755 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6757 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6758 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6759 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6760 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6761 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6763 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6764 (arc_mode_class): Delete.
6765 (HARD_REGNO_MODE_OK): Delete.
6766 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6767 (arc_hard_regno_mode_ok): Rename old array to...
6768 (arc_hard_regno_mode_ok_modes): ...this.
6769 (arc_conditional_register_usage): Update accordingly.
6770 (arc_mode_class): Make static.
6771 (arc_hard_regno_mode_ok): New function.
6772 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6773 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6774 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6775 (arm_hard_regno_mode_ok): Make static.
6776 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6778 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6779 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6780 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6783 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6784 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6785 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6786 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6787 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6788 * config/bfin/predicates.md (valid_reg_operand): Use
6789 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6790 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6791 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6793 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6794 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6795 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6796 (cr16_hard_regno_mode_ok): Make static and return a bool.
6797 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6798 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6799 (cris_hard_regno_mode_ok): New function.
6800 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6801 (epiphany_mode_class): Delete.
6802 (HARD_REGNO_MODE_OK): Delete.
6803 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6804 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6805 (hard_regno_mode_ok): Rename to...
6806 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6807 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6808 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6810 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6811 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6812 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6813 (frv_hard_regno_mode_ok): Make static and return a bool.
6814 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6816 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6817 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6818 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6819 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6821 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6822 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6823 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6824 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6826 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6827 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6828 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6829 (ia64_hard_regno_mode_ok): New function.
6830 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6831 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6832 (iq2000_hard_regno_mode_ok): New function.
6833 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6834 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6835 (lm32_hard_regno_mode_ok): New function.
6836 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6837 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6838 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6839 instead of HARD_REGNO_MODE_OK.
6840 (m32c_hard_regno_ok): Rename to...
6841 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6842 (m32c_cannot_change_mode_class): Update accordingly.
6843 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6844 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6845 (m32r_mode_class): Delete.
6846 (HARD_REGNO_MODE_OK): Delete.
6847 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6848 (m32r_hard_regno_mode_ok): Rename to...
6849 (m32r_hard_regno_modes): ...this.
6850 (m32r_mode_class): Make static.
6851 (m32r_hard_regno_mode_ok): New function.
6852 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6853 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6854 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6855 (m68k_hard_regno_mode_ok): Make static.
6856 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6857 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6858 (mcore_hard_regno_mode_ok): New function.
6859 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6860 (HARD_REGNO_MODE_OK): Delete.
6861 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6863 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6864 (microblaze_hard_regno_mode_ok): New function.
6865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6866 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6867 (mips_hard_regno_mode_ok): Delete.
6868 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6869 (mips_hard_regno_mode_ok_p): ...this and make static.
6870 (mips_hard_regno_mode_ok_p): Rename to...
6871 (mips_hard_regno_mode_ok_uncached): ...this.
6872 (mips_hard_regno_mode_ok): New function.
6873 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6874 of HARD_REGNO_MODE_OK.
6875 (mips_option_override): Update after above name changes.
6876 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6877 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6878 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6879 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6880 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6881 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6882 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6883 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6884 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6885 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6886 (msp430_hard_regno_mode_ok): Make static and return a bool.
6887 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6888 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6889 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6892 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6893 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6894 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6895 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6896 (PA_HARD_REGNO_MODE_OK): ...this
6897 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6898 (PA_HARD_REGNO_MODE_OK): ...this.
6899 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6900 (pa_hard_regno_mode_ok): New function.
6901 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6902 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6903 (pdp11_hard_regno_mode_ok): New function.
6904 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6905 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6907 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6909 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6910 (rs6000_hard_regno_mode_ok): Rename to...
6911 (rs6000_hard_regno_mode_ok_uncached): ...this.
6912 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6913 (rs6000_hard_regno_mode_ok): New function.
6914 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6915 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6916 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6917 (riscv_hard_regno_mode_ok): ...this and make static.
6918 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6919 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6920 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6921 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6922 (rl78_hard_regno_mode_ok): Make static and return bool.
6923 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6924 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6926 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6928 (rs6000_hard_regno_mode_ok): Rename to...
6929 (rs6000_hard_regno_mode_ok_uncached): ...this.
6930 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6931 (rs6000_hard_regno_mode_ok): New function.
6932 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6933 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6934 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6935 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6936 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6937 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6939 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6940 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6941 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6942 (sh_hard_regno_mode_ok): Make static.
6943 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6944 instead of HARD_REGNO_MODE_OK.
6945 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6946 (sparc_mode_class): Delete.
6947 (HARD_REGNO_MODE_OK): Delete.
6948 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6949 (hard_regno_mode_classes): Make static.
6950 (sparc_mode_class): Likewise.
6951 (sparc_hard_regno_mode_ok): New function.
6952 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6953 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6954 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6956 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6957 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6958 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6959 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6960 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6961 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6962 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6963 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6964 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6965 (visium_hard_regno_mode_ok): New function.
6966 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6967 instead of HARD_REGNO_MODE_OK.
6968 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6969 (HARD_REGNO_MODE_OK): Delete.
6970 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6971 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6972 (xtensa_option_override): Update accordingly.
6973 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6974 (xtensa_hard_regno_mode_ok): New function.
6975 * system.h (HARD_REGNO_MODE_OK): Poison.
6977 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6978 Alan Hayward <alan.hayward@arm.com>
6979 David Sherwood <david.sherwood@arm.com>
6981 * target.def (hard_regno_call_part_clobbered): New hook.
6982 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6983 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6984 * doc/tm.texi: Regenerate.
6985 * hooks.h (hook_bool_uint_mode_false): Declare.
6986 * hooks.c (hook_bool_uint_mode_false): New function.
6987 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6988 * cselib.c (cselib_process_insn): Use
6989 targetm.hard_regno_call_part_clobbered instead of
6990 HARD_REGNO_CALL_PART_CLOBBERED.
6991 * ira-conflicts.c (ira_build_conflicts): Likewise.
6992 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6993 * lra-constraints.c (need_for_call_save_p): Likewise.
6994 * lra-lives.c: Include target.h.
6995 (check_pseudos_live_through_calls): Use
6996 targetm.hard_regno_call_part_clobbered instead of
6997 HARD_REGNO_CALL_PART_CLOBBERED.
6998 * regcprop.c: Include target.h.
6999 (copyprop_hardreg_forward_1): Use
7000 targetm.hard_regno_call_part_clobbered instead of
7001 HARD_REGNO_CALL_PART_CLOBBERED.
7002 * reginfo.c (choose_hard_reg_mode): Likewise.
7003 * regrename.c (check_new_reg_p): Likewise.
7004 * reload.c (find_equiv_reg): Likewise.
7005 * reload1.c (emit_reload_insns): Likewise.
7006 * sched-deps.c (deps_analyze_insn): Likewise.
7007 * sel-sched.c (init_regs_for_mode): Likewise.
7008 (mark_unavailable_hard_regs): Likewise.
7009 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7010 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7011 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7013 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7014 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7015 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7017 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7019 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7020 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7021 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7023 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7024 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7025 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7027 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7028 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7030 * config/powerpcspe/powerpcspe.c
7031 (rs6000_hard_regno_call_part_clobbered): New function.
7032 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7033 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7034 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7036 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7037 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7038 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7040 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7041 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7042 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7044 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7045 Alan Hayward <alan.hayward@arm.com>
7046 David Sherwood <david.sherwood@arm.com>
7048 * rtl.h (subreg_memory_offset): Declare.
7049 * emit-rtl.c (subreg_memory_offset): New function.
7050 * expmed.c (store_bit_field_1): Use it.
7051 * expr.c (undefined_operand_subword_p): Likewise.
7052 * simplify-rtx.c (simplify_subreg): Likewise.
7054 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7056 PR rtl-optimization/57448
7059 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7060 using atomic_load pattern.
7061 (expand_atomic_store): Likewise.
7063 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7066 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7067 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7068 handling. Use replace_call_with_value with NULL instead of
7069 gsi_replace, unlink_stmt_vdef and release_defs.
7071 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7074 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7076 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7079 * rtl.h (emit_library_call_value_1): Declare.
7080 (emit_library_call): Replace declaration with a series of overloads.
7081 Remove the parameter count argument.
7082 (emit_library_call_value): Likewise.
7083 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7084 with an "rtx_mode_t *".
7085 (emit_library_call_value): Delete.
7086 (emit_library_call): Likewise.
7087 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7088 (asan_emit_allocas_unpoison): Likewise.
7089 * builtins.c (expand_builtin_powi): Likewise.
7090 (expand_asan_emit_allocas_unpoison): Likewise.
7091 * cfgexpand.c (expand_main_function): Likewise.
7092 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7093 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7094 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7095 * config/arm/arm.c (arm_trampoline_init): Likewise.
7096 (arm_call_tls_get_addr): Likewise.
7097 (arm_expand_divmod_libfunc): Likewise.
7098 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7099 (smulsi3_highpart): Likewise.
7100 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7101 (c6x_expand_compare): Likewise.
7102 (c6x_expand_movmem): Likewise.
7103 * config/frv/frv.c (frv_trampoline_init): Likewise.
7104 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7105 (ix86_expand_divmod_libfunc): Likewise.
7106 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7107 (ia64_expand_compare): Likewise.
7108 (ia64_profile_hook): Likewise.
7109 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7110 (nonlocal_goto): Likewise.
7111 (restore_stack_nonlocal): Likewise.
7112 * config/m32r/m32r.c (block_move_call): Likewise.
7113 (m32r_trampoline_init): Likewise.
7114 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7115 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7116 (m68k_call_m68k_read_tp): Likewise.
7117 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7118 (microblaze_expand_divide): Likewise.
7119 * config/mips/mips.h (mips_args): Likewise.
7120 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7121 (MIPS_ICACHE_SYNC): Likewise.
7122 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7123 (nios2_trampoline_init): Likewise.
7124 * config/pa/pa.c (hppa_tls_call): Likewise.
7125 (pa_trampoline_init): Likewise.
7126 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7127 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7128 (expand_strn_compare): Likewise.
7129 (rs6000_generate_compare): Likewise.
7130 (rs6000_expand_float128_convert): Likewise.
7131 (output_profile_hook): Likewise.
7132 (rs6000_trampoline_init): Likewise.
7133 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7134 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7135 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7136 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7137 (rs6000_generate_compare): Likewise.
7138 (rs6000_expand_float128_convert): Likewise.
7139 (output_profile_hook): Likewise.
7140 (rs6000_trampoline_init): Likewise.
7141 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7142 * config/sh/sh.c (sh_trampoline_init): Likewise.
7143 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7144 (sparc_emit_float_lib_cmp): Likewise.
7145 (sparc32_initialize_trampoline): Likewise.
7146 (sparc64_initialize_trampoline): Likewise.
7147 (sparc_profile_hook): Likewise.
7148 * config/spu/spu.c (ea_load_store): Likewise.
7149 * config/spu/spu.md (floatunssidf2): Likewise.
7150 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7151 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7152 * config/visium/visium.c (expand_block_move_4): Likewise.
7153 (expand_block_move_2): Likewise.
7154 (expand_block_move_1): Likewise.
7155 (expand_block_set_4): Likewise.
7156 (expand_block_set_2): Likewise.
7157 (expand_block_set_1): Likewise.
7158 (visium_trampoline_init): Likewise.
7159 (visium_profile_hook): Likewise.
7160 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7161 (xtensa_setup_frame_addresses): Likewise.
7162 (xtensa_trampoline_init): Likewise.
7163 * except.c (sjlj_emit_function_enter): Likewise.
7164 (sjlj_emit_function_exit): Likewise.
7165 * explow.c (allocate_dynamic_stack_space): Likewise.
7166 (probe_stack_range): Likewise.
7167 * expr.c (convert_mode_scalar): Likewise.
7168 * optabs.c (expand_binop): Likewise.
7169 (expand_twoval_binop_libfunc): Likewise.
7170 (expand_unop): Likewise.
7171 (prepare_cmp_insn): Likewise.
7172 (prepare_float_lib_cmp): Likewise.
7173 (expand_float): Likewise.
7174 (expand_fix): Likewise.
7175 (expand_fixed_convert): Likewise.
7176 (maybe_emit_sync_lock_test_and_set): Likewise.
7177 (expand_atomic_compare_and_swap): Likewise.
7178 (expand_mem_thread_fence): Likewise.
7179 (expand_atomic_fetch_op): Likewise.
7181 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7183 * doc/generic.texi (OpenACC): Adjust URL.
7184 * doc/invoke.texi (C Dialect Options): Ditto.
7186 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7188 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7189 predicate for operand 1. Add (m,<S>) constraint.
7190 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7191 Prevent memory operand 1 with register operand 2.
7193 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7195 PR rtl-optimization/82024
7196 * combine.c (try_combine): If the combination result is a PARALLEL,
7197 and we only need to retain the SET in there that would be placed
7198 at I2, check that we can place that at I3 instead, before doing so.
7200 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7203 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7205 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7206 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7209 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7210 Jeff Law <law@redhat.com>
7212 * varasm.c (bss_initializer_p): Do not put constants into .bss
7213 (categorize_decl_for_section): Handle bss_initializer_p returning
7214 false when DECL_INITIAL is NULL.
7216 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7219 * config/s390/s390.c (s390_can_inline_p): New function.
7221 2017-09-01 Jeff Law <law@redhat.com>
7223 PR tree-optimization/82052
7224 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7225 Always initialize the returned slot after a hash table miss
7226 when INSERT is true.
7228 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7230 * config/s390/s390.md (mem_signal_fence): Remove.
7231 * doc/md.texi (mem_signal_fence): Remove.
7232 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7234 * target-insns.def (mem_signal_fence): Remove.
7236 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7239 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7242 * asan.c (create_odr_indicator): Strip name encoding from assembler
7243 name before appending it after __odr_asan_.
7245 2017-09-01 Martin Liska <mliska@suse.cz>
7247 PR tree-optimization/82059
7248 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7249 frequency only when an edge is redirected.
7251 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7253 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7254 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7255 (arc_conditional_register_usage): Remove ARC600 lp_count
7257 (arc_file_start): Emit Tag_ARC_CPU_variation.
7258 (arc_can_use_doloop_p): New conditions to use ZOLs.
7259 (hwloop_fail): New function.
7260 (hwloop_optimize): Likewise.
7261 (hwloop_pattern_reg): Likewise.
7262 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7263 (arc_reorg_loops): New function, calls reorg_loops.
7264 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7265 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7267 (arc_loop_hazard): Remove function, functionality moved into
7269 (arc_hazard): Remove arc_loop_hazard call.
7270 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7271 into hwloop_optimize.
7272 (arc_label_align): Remove ZOL handling.
7273 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7274 * config/arc/arc.md (doloop_begin): Remove pattern.
7275 (doloop_begin_i): Likewise.
7276 (doloop_end_i): Likewise.
7277 (doloop_fallback): Likewise.
7278 (doloop_fallback_m): Likewise.
7279 (doloop_end): Reimplement expand.
7280 (arc_lp): New pattern for LP instruction.
7281 (loop_end): New pattern.
7282 (loop_fail): Likewise.
7283 (decrement_and_branch_until_zero): Likewise.
7284 * config/arc/arc.opt (mlpc-width): New option.
7285 * doc/invoke.texi (mlpc-width): Document option.
7287 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7289 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7290 (arc_ccfsm_advance): Fix checking for delay slots.
7291 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7293 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7295 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7297 (movhi_insn): Update store instruction constraint which are saving
7298 6-bit short immediates.
7299 (movsi_insn): Consider also short scaled load operations.
7300 (zero_extendhisi2_i): Use Usd constraint instead of T.
7301 (extendhisi2_i): Add q constraint.
7302 (arc_clzsi2): Add type and length attributes.
7303 (arc_ctzsi2): Likewise.
7304 * config/arc/constraints.md (Usc): Update constraint, the
7305 assembler can parse two relocations for a single instruction.
7307 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7309 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7310 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7312 2017-08-31 Olivier Hainque <hainque@adacore.com>
7314 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7315 match as powerpc-wrs-vxworks*.
7317 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7319 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7320 register constraint for by-element operand.
7321 (aarch64_mls_elt_merge<mode>): Likewise.
7323 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7325 * config/arc/arc.c (arc_can_follow_jump): Check for short
7328 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7330 * config.gcc: Use g.opt for arc.
7331 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7332 functionality moved to ...
7333 (legitimate_scaled_address_p): New function, ...here.
7334 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7335 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7336 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7338 (arc_override_options): Handle G option.
7339 (arc_output_pic_addr_const): Correct function definition.
7340 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7341 (arc_decl_anon_ns_mem_p): Delete.
7342 (arc_in_small_data_p): Overhaul this function to take into
7343 consideration the value given via G option.
7344 (arc_rewrite_small_data_1): Renamed and corrected old
7345 arc_rewrite_small_data function.
7346 (arc_rewrite_small_data): New function.
7347 (small_data_pattern): Don't use pic_offset_table_rtx.
7348 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7349 * config/arc/simdext.md (movmisalignv2hi): Use
7350 prepare_move_operands function.
7352 (movmisalign*): Likewise.
7353 * doc/invoke.texi (ARC options): Document -G option.
7355 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7357 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7359 * config/arc/arc.c (arc_print_operand): Output scalled address for
7360 sdata whenever is possible.
7361 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7362 load/stores are available.
7363 (compact_sda_memory_operand): Check for the alignment required by
7364 code density instructions.
7365 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7367 * config/arc/constraints.md (Usd): Update constraint.
7368 (Us0): New constraint.
7369 (Usc): Update constraint.
7371 2017-08-31 Richard Biener <rguenther@suse.de>
7374 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7377 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7379 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7380 Resize type_signature.
7382 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7383 Alan Hayward <alan.hayward@arm.com>
7384 David Sherwood <david.sherwood@arm.com>
7386 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7387 subregs whose inner modes can be stored in GPRs.
7388 (aarch64_classify_index): Likewise.
7390 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7391 Alan Hayward <alan.hayward@arm.com>
7392 David Sherwood <david.sherwood@arm.com>
7394 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7395 (V_INT_EQUIV): ...this.
7396 (v_cmp_result): Rename to...
7397 (v_int_equiv): ...this.
7398 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7399 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7400 (copysign<mode>3): Likewise.
7401 (aarch64_simd_bsl<mode>_internal): Likewise.
7402 (aarch64_simd_bsl<mode>): Likewise.
7403 (vec_cmp<mode><mode>): Likewise.
7404 (vcond<mode><mode>): Likewise.
7405 (vcond<v_cmp_mixed><mode>): Likewise.
7406 (vcondu<mode><v_cmp_mixed>): Likewise.
7407 (aarch64_cm<optab><mode>): Likewise.
7408 (aarch64_cmtst<mode>): Likewise.
7409 (aarch64_fac<optab><mode>): Likewise.
7410 (vec_perm_const<mode>): Likewise.
7411 (vcond_mask_<mode><v_cmp_result>): Rename to...
7412 (vcond_mask_<mode><v_int_equiv>): ...this.
7413 (vec_cmp<mode><v_cmp_result>): Rename to...
7414 (vec_cmp<mode><v_int_equiv>): ...this.
7416 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7417 Alan Hayward <alan.hayward@arm.com>
7418 David Sherwood <david.sherwood@arm.com>
7420 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7422 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7423 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7424 (UNSPEC_LD4_DREG): New unspecs.
7425 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7426 (aarch64_ld2<mode>_dreg_be): Replace with...
7427 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7429 (aarch64_ld3<mode>_dreg_le)
7430 (aarch64_ld3<mode>_dreg_be): Replace with...
7431 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7433 (aarch64_ld4<mode>_dreg_le)
7434 (aarch64_ld4<mode>_dreg_be): Replace with...
7435 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7438 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7440 PR tree-optimization/81987
7441 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7442 insert an initializer in a location not dominated by the stride
7445 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7447 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7448 on the entire header of the finally block in the fallthru case.
7450 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7452 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7454 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7456 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7457 rs6000_emit_move_from_cr and call renamed function.
7458 (rs6000_emit_prologue): Call renamed functions.
7459 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7460 movesi_from_cr, remove volatile CRs.
7462 2017-08-30 Jon Beniston <jon@beniston.com>
7463 Richard Biener <rguenther@suse.de>
7465 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7466 of VECTOR_MODE_P check.
7467 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7468 element vector types.
7470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7472 * df.h (df_read_modify_subreg_p): Remove in favor of...
7473 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7474 const_rtx instead of an rtx.
7475 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7476 * df-problems.c (df_word_lr_mark_ref): Likewise.
7477 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7478 (mark_pseudo_reg_dead): Likewise.
7479 (mark_ref_dead): Likewise.
7480 * reginfo.c (init_subregs_of_mode): Likewise.
7481 * sched-deps.c (sched_analyze_1): Likewise.
7482 * df-scan.c (df_def_record_1): Likewise.
7483 (df_uses_record): Likewise.
7484 (df_read_modify_subreg_p): Remove in favor of...
7485 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7486 const_rtx instead of an rtx.
7488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7489 Alan Hayward <alan.hayward@arm.com>
7490 David Sherwood <david.sherwood@arm.com>
7492 * rtl.h (partial_subreg_p): New function.
7493 * caller-save.c (save_call_clobbered_regs): Use it.
7494 * calls.c (expand_call): Likewise.
7495 * combine.c (combinable_i3pat): Likewise.
7496 (simplify_set): Likewise.
7497 (make_extraction): Likewise.
7498 (make_compound_operation_int): Likewise.
7499 (gen_lowpart_or_truncate): Likewise.
7500 (force_to_mode): Likewise.
7501 (make_field_assignment): Likewise.
7502 (reg_truncated_to_mode): Likewise.
7503 (record_truncated_value): Likewise.
7504 (move_deaths): Likewise.
7505 * cse.c (record_jump_cond): Likewise.
7506 (cse_insn): Likewise.
7507 * cselib.c (cselib_lookup_1): Likewise.
7508 * expmed.c (extract_bit_field_using_extv): Likewise.
7509 * function.c (assign_parm_setup_reg): Likewise.
7510 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7511 * ira-build.c (create_insn_allocnos): Likewise.
7512 * lra-coalesce.c (merge_pseudos): Likewise.
7513 * lra-constraints.c (match_reload): Likewise.
7514 (simplify_operand_subreg): Likewise.
7515 (curr_insn_transform): Likewise.
7516 * lra-lives.c (process_bb_lives): Likewise.
7517 * lra.c (new_insn_reg): Likewise.
7518 (lra_substitute_pseudo): Likewise.
7519 * regcprop.c (mode_change_ok): Likewise.
7520 (maybe_mode_change): Likewise.
7521 (copyprop_hardreg_forward_1): Likewise.
7522 * reload.c (push_reload): Likewise.
7523 (find_reloads): Likewise.
7524 (find_reloads_subreg_address): Likewise.
7525 * reload1.c (alter_reg): Likewise.
7526 (eliminate_regs_1): Likewise.
7527 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7529 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7531 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7532 back to if statements, including unpack.
7534 2017-08-30 Martin Liska <mliska@suse.cz>
7537 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7539 (func_checker::compare_asm_inputs_outputs): ... this function.
7540 (func_checker::compare_gimple_asm): Use the function to compare
7541 also ASM constrains.
7542 * ipa-icf-gimple.h: Rename the function.
7544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7545 Alan Hayward <alan.hayward@arm.com>
7546 David Sherwood <david.sherwood@arm.com>
7548 * coretypes.h (complex_mode): New type.
7549 * gdbhooks.py (build_pretty_printer): Handle it.
7550 * machmode.h (complex_mode): New class.
7551 (complex_mode::includes_p): New function.
7552 (is_complex_int_mode): Likewise.
7553 (is_complex_float_mode): Likewise.
7554 * genmodes.c (get_mode_class): Handle complex mode classes.
7555 * function.c (expand_function_end): Use is_complex_int_mode.
7557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * coretypes.h (scalar_mode_pod): New typedef.
7562 * gdbhooks.py (build_pretty_printer): Handle it.
7563 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7564 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7565 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7566 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7570 Alan Hayward <alan.hayward@arm.com>
7571 David Sherwood <david.sherwood@arm.com>
7573 * machmode.h (mode_for_vector): Take a scalar_mode instead
7575 * stor-layout.c (mode_for_vector): Likewise.
7576 * explow.c (promote_mode): Use as_a <scalar_mode>.
7577 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7580 Alan Hayward <alan.hayward@arm.com>
7581 David Sherwood <david.sherwood@arm.com>
7583 * target.def (preferred_simd_mode): Take a scalar_mode
7584 instead of a machine_mode.
7585 * targhooks.h (default_preferred_simd_mode): Likewise.
7586 * targhooks.c (default_preferred_simd_mode): Likewise.
7587 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7588 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7589 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7590 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7591 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7592 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7593 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7594 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7596 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7597 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7598 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7599 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7600 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7601 * doc/tm.texi: Regenerate.
7602 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7606 Alan Hayward <alan.hayward@arm.com>
7607 David Sherwood <david.sherwood@arm.com>
7609 * target.def (scalar_mode_supported_p): Take a scalar_mode
7610 instead of a machine_mode.
7611 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7612 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7613 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7614 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7615 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7616 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7617 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7618 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7619 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7620 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7621 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7622 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7623 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7624 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7626 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7627 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7628 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7629 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7630 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7632 * doc/tm.texi: Regenerate.
7634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7635 Alan Hayward <alan.hayward@arm.com>
7636 David Sherwood <david.sherwood@arm.com>
7638 * coretypes.h (opt_scalar_mode): New typedef.
7639 * gdbhooks.py (build_pretty_printers): Handle it.
7640 * machmode.h (mode_iterator::get_2xwider): Add overload for
7642 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7644 * expr.c (convert_mode_scalar): Likewise.
7645 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7646 * optabs.c (expand_float): Likewise.
7647 (expand_fix): Likewise.
7648 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7651 Alan Hayward <alan.hayward@arm.com>
7652 David Sherwood <david.sherwood@arm.com>
7654 * optabs.c (expand_float): Explicitly check for scalars before
7655 using a branching expansion.
7656 (expand_fix): Likewise.
7658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7659 Alan Hayward <alan.hayward@arm.com>
7660 David Sherwood <david.sherwood@arm.com>
7662 * expr.c (convert_mode): Split scalar handling out into...
7663 (convert_mode_scalar): ...this new function. Treat the modes
7666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7667 Alan Hayward <alan.hayward@arm.com>
7668 David Sherwood <david.sherwood@arm.com>
7670 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7672 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7675 Alan Hayward <alan.hayward@arm.com>
7676 David Sherwood <david.sherwood@arm.com>
7678 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7679 rather than a machine_mode.
7680 (fixed_from_string): Likewise.
7681 (fixed_convert): Likewise.
7682 (fixed_convert_from_int): Likewise.
7683 (fixed_convert_from_real): Likewise.
7684 (real_convert_from_fixed): Likewise.
7685 * fixed-value.c (fixed_from_double_int): Likewise.
7686 (fixed_from_string): Likewise.
7687 (fixed_convert): Likewise.
7688 (fixed_convert_from_int): Likewise.
7689 (fixed_convert_from_real): Likewise.
7690 (real_convert_from_fixed): Likewise.
7691 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7694 Alan Hayward <alan.hayward@arm.com>
7695 David Sherwood <david.sherwood@arm.com>
7697 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7698 of separate mode class checks. Do not allow vector modes here.
7699 (immed_wide_int_const): Use as_a <scalar_mode>.
7700 * explow.c (trunc_int_for_mode): Likewise.
7701 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7702 (wi::shwi): Likewise.
7703 (wi::min_value): Likewise.
7704 (wi::max_value): Likewise.
7705 * dwarf2out.c (loc_descriptor): Likewise.
7706 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7710 Alan Hayward <alan.hayward@arm.com>
7711 David Sherwood <david.sherwood@arm.com>
7713 * tree.h (SCALAR_TYPE_MODE): New macro.
7714 * expr.c (expand_expr_addr_expr_1): Use it.
7715 (expand_expr_real_2): Likewise.
7716 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7717 (fold_convert_const_fixed_from_int): Likewise.
7718 (fold_convert_const_fixed_from_real): Likewise.
7719 (native_encode_fixed): Likewise
7720 (native_encode_complex): Likewise
7721 (native_encode_vector): Likewise.
7722 (native_interpret_fixed): Likewise.
7723 (native_interpret_real): Likewise.
7724 (native_interpret_complex): Likewise.
7725 (native_interpret_vector): Likewise.
7726 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7727 (simd_clone_adjust_argument_types): Likewise.
7728 (simd_clone_init_simd_arrays): Likewise.
7729 (simd_clone_adjust): Likewise.
7730 * stor-layout.c (layout_type): Likewise.
7731 * tree.c (build_minus_one_cst): Likewise.
7732 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7733 * tree-inline.c (estimate_move_cost): Likewise.
7734 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7735 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7736 (vectorizable_reduction): Likewise.
7737 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7738 (vect_recog_mixed_size_cond_pattern): Likewise.
7739 (check_bool_pattern): Likewise.
7740 (adjust_bool_pattern): Likewise.
7741 (search_type_for_mask_1): Likewise.
7742 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7743 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7744 (vectorizable_load): Likewise.
7745 (vectorizable_store): Likewise.
7746 * ubsan.c (ubsan_encode_value): Likewise.
7747 * varasm.c (output_constant): Likewise.
7749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7750 Alan Hayward <alan.hayward@arm.com>
7751 David Sherwood <david.sherwood@arm.com>
7753 * coretypes.h (scalar_mode): New class.
7754 * machmode.h (scalar_mode): Likewise.
7755 (scalar_mode::includes_p): New function.
7756 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7757 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7758 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7759 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7760 * expmed.c (store_bit_field_1): Likewise.
7761 (extract_bit_field_1): Likewise.
7762 * expr.c (write_complex_part): Likewise.
7763 (read_complex_part): Likewise.
7764 (emit_move_complex_push): Likewise.
7765 (expand_expr_real_2): Likewise.
7766 * function.c (assign_parm_setup_reg): Likewise.
7767 (assign_parms_unsplit_complex): Likewise.
7768 * optabs.c (expand_binop): Likewise.
7769 * rtlanal.c (subreg_get_info): Likewise.
7770 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7771 * varasm.c (output_constant_pool_2): Likewise.
7773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7774 Alan Hayward <alan.hayward@arm.com>
7775 David Sherwood <david.sherwood@arm.com>
7777 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7779 (expmed_mult_highpart_optab): Likewise.
7780 (expmed_mult_highpart): Likewise.
7782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7783 Alan Hayward <alan.hayward@arm.com>
7784 David Sherwood <david.sherwood@arm.com>
7786 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7787 instead of a machine_mode.
7788 (builtin_memset_read_str): Likewise.
7789 * builtins.c (c_readstr): Likewise.
7790 (builtin_memcpy_read_str): Likewise.
7791 (builtin_strncpy_read_str): Likewise.
7792 (builtin_memset_read_str): Likewise.
7793 (builtin_memset_gen_str): Likewise.
7794 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7795 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7796 instead of a machine_mode.
7797 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7799 (make_extraction): Likewise.
7800 (try_widen_shift_mode): Take and return scalar_int_modes instead
7802 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7803 a scalar_int_mode instead of a machine_mode.
7804 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7805 (avr_addr_space_pointer_mode): Likewise.
7806 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7807 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7808 (msp430_unwind_word_mode): Likewise.
7809 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7810 (spu_addr_space_pointer_mode): Likewise.
7811 (spu_addr_space_address_mode): Likewise.
7812 (spu_libgcc_cmp_return_mode): Likewise.
7813 (spu_libgcc_shift_count_mode): Likewise.
7814 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7815 (rl78_addr_space_pointer_mode): Likewise.
7816 (fl78_unwind_word_mode): Likewise.
7817 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7819 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7820 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7821 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7822 (mips_valid_pointer_mode): Likewise.
7823 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7824 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7825 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7827 (ft32_addr_space_address_mode): Likewise.
7828 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7829 scalar_int_mode instead of a machine_mode.
7830 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7832 (m32c_addr_space_address_mode): Likewise.
7833 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7834 (rs6000_eh_return_filter_mode): Likewise.
7835 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7836 (rs6000_eh_return_filter_mode): Likewise.
7837 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7838 (s390_libgcc_shift_count_mode): Likewise.
7839 (s390_unwind_word_mode): Likewise.
7840 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7842 * target.def (mode_rep_extended): Likewise.
7843 (valid_pointer_mode): Likewise.
7844 (addr_space.valid_pointer_mode): Likewise.
7845 (eh_return_filter_mode): Return a scalar_int_mode rather than
7847 (libgcc_cmp_return_mode): Likewise.
7848 (libgcc_shift_count_mode): Likewise.
7849 (unwind_word_mode): Likewise.
7850 (addr_space.pointer_mode): Likewise.
7851 (addr_space.address_mode): Likewise.
7852 * doc/tm.texi: Regenerate.
7853 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7855 (do_jump): Use scalar_int_mode for local variables.
7856 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7857 rather than a machine_mode.
7858 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7859 (scompare_loc_descriptor_wide): Likewise.
7860 (scompare_loc_descriptor_narrow): Likewise.
7861 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7863 * except.c (sjlj_emit_dispatch_table): Likewise.
7864 (expand_builtin_eh_copy_values): Likewise.
7865 * explow.c (convert_memory_address_addr_space_1): Likewise.
7866 Take a scalar_int_mode rather than a machine_mode.
7867 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7868 than a machine_mode.
7869 (memory_address_addr_space): Use scalar_int_mode for local variables.
7870 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7871 rather than a machine_mode.
7872 * expmed.c (mask_rtx): Likewise.
7873 (init_expmed_one_conv): Likewise.
7874 (expand_mult_highpart_adjust): Likewise.
7875 (extract_high_half): Likewise.
7876 (expmed_mult_highpart_optab): Likewise.
7877 (expmed_mult_highpart): Likewise.
7878 (expand_smod_pow2): Likewise.
7879 (expand_sdiv_pow2): Likewise.
7880 (emit_store_flag_int): Likewise.
7881 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7883 (extract_low_bits): Likewise.
7884 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7886 * expr.c (pieces_addr::adjust): Likewise.
7887 (can_store_by_pieces): Likewise.
7888 (store_by_pieces): Likewise.
7889 (clear_by_pieces_1): Likewise.
7890 (expand_expr_addr_expr_1): Likewise.
7891 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7892 (expand_expr_real_1): Likewise.
7893 (try_casesi): Likewise.
7894 * final.c (shorten_branches): Likewise.
7895 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7896 type of "mode" to machine_mode.
7897 * internal-fn.c (expand_arith_overflow_result_store): Take a
7898 scalar_int_mode rather than a machine_mode.
7899 (expand_mul_overflow): Use scalar_int_mode for local variables.
7900 * loop-doloop.c (doloop_modify): Likewise.
7901 (doloop_optimize): Likewise.
7902 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7903 than a machine_mode.
7904 (expand_doubleword_shift_condmove): Likewise.
7905 (expand_doubleword_shift): Likewise.
7906 (expand_doubleword_clz): Likewise.
7907 (expand_doubleword_popcount): Likewise.
7908 (expand_doubleword_parity): Likewise.
7909 (expand_absneg_bit): Use scalar_int_mode for local variables.
7910 (prepare_float_lib_cmp): Likewise.
7911 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7912 rather than a machine_mode.
7913 (convert_memory_address_addr_space): Likewise.
7914 (get_mode_bounds): Likewise.
7915 (get_address_mode): Return a scalar_int_mode rather than a
7917 * rtlanal.c (get_address_mode): Likewise.
7918 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7919 than a machine_mode.
7920 * targhooks.c (default_mode_rep_extended): Likewise.
7921 (default_valid_pointer_mode): Likewise.
7922 (default_addr_space_valid_pointer_mode): Likewise.
7923 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7924 than a machine_mode.
7925 (default_libgcc_cmp_return_mode): Likewise.
7926 (default_libgcc_shift_count_mode): Likewise.
7927 (default_unwind_word_mode): Likewise.
7928 (default_addr_space_pointer_mode): Likewise.
7929 (default_addr_space_address_mode): Likewise.
7930 * targhooks.h (default_eh_return_filter_mode): Likewise.
7931 (default_libgcc_cmp_return_mode): Likewise.
7932 (default_libgcc_shift_count_mode): Likewise.
7933 (default_unwind_word_mode): Likewise.
7934 (default_addr_space_pointer_mode): Likewise.
7935 (default_addr_space_address_mode): Likewise.
7936 (default_mode_rep_extended): Take a scalar_int_mode rather than
7938 (default_valid_pointer_mode): Likewise.
7939 (default_addr_space_valid_pointer_mode): Likewise.
7940 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7942 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7943 rather than a machine_mode.
7944 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7945 for local variables.
7946 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7947 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7948 than a machine_mode.
7950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7955 the mode argument to scalar_int_mode.
7956 (do_jump_by_parts_zero_rtx): Likewise.
7957 (do_jump_by_parts_equality_rtx): Likewise.
7958 (do_jump_by_parts_greater): Take a mode argument.
7959 (do_jump_by_parts_equality): Likewise.
7960 (do_jump_1): Update calls accordingly.
7962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7963 Alan Hayward <alan.hayward@arm.com>
7964 David Sherwood <david.sherwood@arm.com>
7966 * is-a.h (safe_dyn_cast): New function.
7967 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7968 (jump_table_for_label): Likewise.
7969 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7970 instead of an rtx_insn *.
7971 (shorten_branches): Use dyn_cast instead of LABEL_P and
7972 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7973 rtx_jump_table_data::get_data_mode.
7974 (final_scan_insn): Likewise.
7976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7977 Alan Hayward <alan.hayward@arm.com>
7978 David Sherwood <david.sherwood@arm.com>
7980 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7981 trying to combine a full-register integer set with a subreg
7984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7985 Alan Hayward <alan.hayward@arm.com>
7986 David Sherwood <david.sherwood@arm.com>
7988 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7989 that is always either address_mode or pointer_mode.
7991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7992 Alan Hayward <alan.hayward@arm.com>
7993 David Sherwood <david.sherwood@arm.com>
7995 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7996 when the two are known to be equal.
7998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7999 Alan Hayward <alan.hayward@arm.com>
8000 David Sherwood <david.sherwood@arm.com>
8002 * simplify-rtx.c (simplify_const_unary_operation): Use
8003 is_a <scalar_int_mode> instead of checking for a nonzero
8004 precision. Forcibly convert op_mode to a scalar_int_mode
8005 in that case. More clearly differentiate the operand and
8006 result modes and use the former when deciding what the value
8007 of a count-bits operation should be. Use is_int_mode instead
8008 of checking for a MODE_INT. Remove redundant check for whether
8009 this mode has a zero precision.
8011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8012 Alan Hayward <alan.hayward@arm.com>
8013 David Sherwood <david.sherwood@arm.com>
8015 * optabs.c (widen_leading): Change the type of the mode argument
8016 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8017 (widen_bswap): Likewise.
8018 (expand_parity): Likewise.
8019 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8020 (expand_ffs): Likewise.
8021 (epand_unop): Check for scalar integer modes before calling the
8024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8025 Alan Hayward <alan.hayward@arm.com>
8026 David Sherwood <david.sherwood@arm.com>
8028 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8030 (expand_expr_real_1): Update call accordingly.
8032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8033 Alan Hayward <alan.hayward@arm.com>
8034 David Sherwood <david.sherwood@arm.com>
8036 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8037 value_mode arguments. Use scalar_int_mode internally.
8038 (store_bit_field_1): Rename the new integer mode from imode
8039 to op0_mode and use it instead of GET_MODE (op0). Update calls
8040 to store_split_bit_field, store_bit_field_using_insv and
8041 store_fixed_bit_field.
8042 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8043 Use scalar_int_mode internally. Use a bit count rather than a mode
8044 when calculating the largest bit size for get_best_mode.
8045 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8046 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8047 Remove assertion that OP0 has a scalar integer mode.
8048 (store_split_bit_field): Add op0_mode and value_mode arguments.
8049 Update calls to extract_fixed_bit_field.
8050 (extract_bit_field_using_extv): Add an op0_mode argument.
8051 Use scalar_int_mode internally.
8052 (extract_bit_field_1): Rename the new integer mode from imode to
8053 op0_mode and use it instead of GET_MODE (op0). Update calls to
8054 extract_split_bit_field, extract_bit_field_using_extv and
8055 extract_fixed_bit_field.
8056 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8057 to extract_split_bit_field and extract_fixed_bit_field_1.
8058 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8059 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8061 (extract_split_bit_field): Add an op0_mode argument. Update call
8062 to extract_fixed_bit_field.
8064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8065 Alan Hayward <alan.hayward@arm.com>
8066 David Sherwood <david.sherwood@arm.com>
8068 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8069 * explow.c (hard_function_value): Likewise.
8070 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8071 convert_to_mode call outside the loop.
8072 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8073 for the mode iterator. Require the mode specified by max_pieces
8075 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8077 (copy_blkmode_to_reg): Likewise.
8078 (set_storage_via_setmem): Likewise.
8079 * optabs.c (prepare_cmp_insn): Likewise.
8080 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8081 * stor-layout.c (finish_bitfield_representative): Likewise.
8083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8088 * expr.c (convert_move): Use them.
8089 (convert_modes): Likewise.
8090 (store_expr_with_bounds): Likewise.
8092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8093 Alan Hayward <alan.hayward@arm.com>
8094 David Sherwood <david.sherwood@arm.com>
8096 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8097 parameter for the mode of "x". Remove the "known_x", "known_mode"
8098 and "known_ret" arguments. Change the type of the mode argument
8100 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8101 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8102 (reg_num_sign_bit_copies_for_combine): Likewise.
8103 * rtlanal.c (nonzero_bits1): Likewise.
8104 (num_sign_bit_copies1): Likewise.
8105 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8106 (reg_num_sign_bit_copies_general): Likewise.
8107 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8108 (reg_nonzero_bits_general): Likewise.
8110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8111 Alan Hayward <alan.hayward@arm.com>
8112 David Sherwood <david.sherwood@arm.com>
8114 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8115 than in subroutines. Return 1 for non-integer modes.
8116 (cached_num_sign_bit_copies): Change the type of the mode parameter
8118 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8119 classes. Handle CONST_INT_P first and then check whether X also
8120 has a scalar integer mode. Check the same thing for inner registers
8121 of a SUBREG and for values that are being extended or truncated.
8123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8124 Alan Hayward <alan.hayward@arm.com>
8125 David Sherwood <david.sherwood@arm.com>
8127 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8128 in subroutines. Return the mode mask for non-integer modes.
8129 (cached_nonzero_bits): Change the type of the mode parameter
8131 (nonzero_bits1): Likewise. Remove early exit for other mode
8132 classes. Handle CONST_INT_P first and then check whether X
8133 also has a scalar integer mode.
8135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8136 Alan Hayward <alan.hayward@arm.com>
8137 David Sherwood <david.sherwood@arm.com>
8139 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8140 Return a scalar_int_mode and assert that the size is greater than
8142 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8143 (op_by_pieces_d::op_by_pieces_d): Likewise.
8144 (op_by_pieces_d::run): Likewise.
8145 (can_store_by_pieces): Likewise.
8147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8148 Alan Hayward <alan.hayward@arm.com>
8149 David Sherwood <david.sherwood@arm.com>
8151 * combine.c (extract_left_shift): Add a mode argument and update
8153 (make_compound_operation_int): Change the type of the mode parameter
8154 to scalar_int_mode and update the call to extract_left_shift.
8156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8157 Alan Hayward <alan.hayward@arm.com>
8158 David Sherwood <david.sherwood@arm.com>
8160 * combine.c (simplify_and_const_int): Change the type of the mode
8161 parameter to scalar_int_mode.
8162 (simplify_and_const_int_1): Likewise. Update recursive call.
8164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8165 Alan Hayward <alan.hayward@arm.com>
8166 David Sherwood <david.sherwood@arm.com>
8168 * combine.c (simplify_compare_const): Check that the mode is a
8169 scalar_int_mode (rather than VOIDmode) before testing its
8171 (simplify_comparison): Move COMPARISON_P handling out of the
8172 loop and restrict the latter part of the loop to scalar_int_modes.
8173 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8174 and when considering SUBREG_REGs. Use is_int_mode instead of
8175 checking GET_MODE_CLASS against MODE_INT.
8177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8178 Alan Hayward <alan.hayward@arm.com>
8179 David Sherwood <david.sherwood@arm.com>
8181 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8182 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8183 shift_unit_mode and for modes involved in scalar shifts.
8185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8186 Alan Hayward <alan.hayward@arm.com>
8187 David Sherwood <david.sherwood@arm.com>
8189 * combine.c (force_int_to_mode): New function, split out from...
8190 (force_to_mode): ...here. Keep xmode up-to-date and use it
8191 instead of GET_MODE (x).
8193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8194 Alan Hayward <alan.hayward@arm.com>
8195 David Sherwood <david.sherwood@arm.com>
8197 * optabs-query.h (extraction_insn::struct_mode): Change type to
8198 opt_scalar_int_mode and update comment.
8199 (extraction_insn::field_mode): Change type to scalar_int_mode.
8200 (extraction_insn::pos_mode): Likewise.
8201 * combine.c (make_extraction): Update accordingly.
8202 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8203 (get_optab_extraction_insn): Likewise.
8204 * recog.c (simplify_while_replacing): Likewise.
8205 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8206 parameter to opt_scalar_int_mode.
8208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8209 Alan Hayward <alan.hayward@arm.com>
8210 David Sherwood <david.sherwood@arm.com>
8212 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8213 to a scalar_int_mode instead of a machine_mode.
8214 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8215 (get_best_mode): Return a boolean and use a pointer argument to store
8216 the selected mode. Replace the limit mode parameter with a bit limit.
8217 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8218 for the values returned by bit_field_mode_iterator::next_mode.
8219 (store_bit_field): Update call to get_best_mode.
8220 (store_fixed_bit_field): Likewise.
8221 (extract_fixed_bit_field): Likewise.
8222 * expr.c (optimize_bitfield_assignment_op): Likewise.
8223 * fold-const.c (optimize_bit_field_compare): Likewise.
8224 (fold_truth_andor_1): Likewise.
8225 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8226 Update for new type of m_mode.
8227 (get_best_mode): As above.
8229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8230 Alan Hayward <alan.hayward@arm.com>
8231 David Sherwood <david.sherwood@arm.com>
8233 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8234 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8235 (store_bit_field): Check is_a <scalar_int_mode> before calling
8236 strict_volatile_bitfield_p.
8237 (extract_bit_field): Likewise.
8239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8240 Alan Hayward <alan.hayward@arm.com>
8241 David Sherwood <david.sherwood@arm.com>
8243 * target.def (cstore_mode): Return a scalar_int_mode.
8244 * doc/tm.texi: Regenerate.
8245 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8246 * targhooks.h (default_cstore_mode): Likewise.
8247 * targhooks.c (default_cstore_mode): Likewise, using a forced
8249 * expmed.c (emit_cstore): Expect the target of the cstore to be
8252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8253 Alan Hayward <alan.hayward@arm.com>
8254 David Sherwood <david.sherwood@arm.com>
8256 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8258 (niter_desc): Likewise mode.
8259 (iv_analyze): Add a mode parameter.
8261 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8262 and change its type to scalar_int_mode.
8263 * loop-iv.c: Update commentary at head of file.
8264 (iv_constant): Pass the mode paraeter before the rtx it describes
8265 and change its type to scalar_int_mode. Remove VOIDmode handling.
8266 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8267 (iv_extend): Likewise.
8268 (shorten_into_mode): Likewise.
8269 (iv_add): Use scalar_int_mode.
8270 (iv_mult): Likewise.
8271 (iv_shift): Likewise.
8272 (canonicalize_iv_subregs): Likewise.
8273 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8274 it describes and change its mode to scalar_int_mode. Also change
8275 the type of the returned inner_mode to scalar_int_mode.
8276 (get_biv_step): Likewise, turning outer_mode from a pointer
8277 into a direct parameter. Update call to get_biv_step_1.
8278 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8279 iv_constant and get_biv_step.
8280 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8281 and change its type to scalar_int_mode. Don't initialise iv->mode
8282 to VOIDmode and remove later checks for its still being VOIDmode.
8283 Update calls to iv_analyze_op and iv_analyze_expr. Check
8284 is_a <scalar_int_mode> when changing the mode under consideration.
8285 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8286 Update call to iv_analyze_expr.
8287 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8288 inner register is not also a scalar_int_mode. Update call to
8290 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8291 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8292 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8293 separate mode class checks. Update calls to iv_analyze. Remove
8294 fix-up of VOIDmodes after iv_analyze_biv.
8295 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8296 don't have a scalar_int_mode. Update call to biv_p.
8298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8299 Alan Hayward <alan.hayward@arm.com>
8300 David Sherwood <david.sherwood@arm.com>
8302 * cfgexpand.c (convert_debug_memory_address): Use
8303 as_a <scalar_int_mode>.
8304 * combine.c (expand_compound_operation): Likewise.
8305 (make_extraction): Likewise.
8306 (change_zero_ext): Likewise.
8307 (simplify_comparison): Likewise.
8308 * cse.c (cse_insn): Likewise.
8309 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8310 (mem_loc_descriptor): Likewise.
8311 (loc_descriptor): Likewise.
8312 * expmed.c (init_expmed_one_mode): Likewise.
8313 (synth_mult): Likewise.
8314 (emit_store_flag_1): Likewise.
8315 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8316 of a comparison with size.
8317 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8318 (reduce_to_bit_field_precision): Likewise.
8319 * function.c (expand_function_end): Likewise.
8320 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8321 * loop-doloop.c (doloop_modify): Likewise.
8322 * optabs.c (expand_binop): Likewise.
8323 (expand_unop): Likewise.
8324 (expand_copysign_absneg): Likewise.
8325 (prepare_cmp_insn): Likewise.
8326 (maybe_legitimize_operand): Likewise.
8327 * recog.c (const_scalar_int_operand): Likewise.
8328 * rtlanal.c (get_address_mode): Likewise.
8329 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8330 (simplify_cond_clz_ctz): Likewise.
8331 * tree-nested.c (get_nl_goto_field): Likewise.
8332 * tree.c (build_vector_type_for_mode): Likewise.
8333 * var-tracking.c (use_narrower_mode): Likewise.
8335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8336 Alan Hayward <alan.hayward@arm.com>
8337 David Sherwood <david.sherwood@arm.com>
8339 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8340 * builtins.c (expand_builtin_signbit): Use it.
8341 * cfgexpand.c (expand_debug_expr): Likewise.
8342 * dojump.c (do_jump): Likewise.
8343 (do_compare_and_jump): Likewise.
8344 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8345 * expmed.c (make_tree): Likewise.
8346 * expr.c (expand_expr_real_2): Likewise.
8347 (expand_expr_real_1): Likewise.
8348 (try_casesi): Likewise.
8349 * fold-const-call.c (fold_const_call_ss): Likewise.
8350 * fold-const.c (unextend): Likewise.
8351 (extract_muldiv_1): Likewise.
8352 (fold_single_bit_test): Likewise.
8353 (native_encode_int): Likewise.
8354 (native_encode_string): Likewise.
8355 (native_interpret_int): Likewise.
8356 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8357 * internal-fn.c (expand_addsub_overflow): Likewise.
8358 (expand_neg_overflow): Likewise.
8359 (expand_mul_overflow): Likewise.
8360 (expand_arith_overflow): Likewise.
8361 * match.pd: Likewise.
8362 * stor-layout.c (layout_type): Likewise.
8363 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8364 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8365 * tree-ssanames.c (get_range_info): Likewise.
8366 * tree-switch-conversion.c (array_value_type) Likewise.
8367 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8368 (vect_recog_divmod_pattern): Likewise.
8369 (vect_recog_mixed_size_cond_pattern): Likewise.
8370 * tree-vrp.c (extract_range_basic): Likewise.
8371 (simplify_float_conversion_using_ranges): Likewise.
8372 * tree.c (int_fits_type_p): Likewise.
8373 * ubsan.c (instrument_bool_enum_load): Likewise.
8374 * varasm.c (mergeable_string_section): Likewise.
8375 (narrowing_initializer_constant_valid_p): Likewise.
8376 (output_constant): Likewise.
8378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8379 Alan Hayward <alan.hayward@arm.com>
8380 David Sherwood <david.sherwood@arm.com>
8382 * machmode.h (NARROWEST_INT_MODE): New macro.
8383 * expr.c (alignment_for_piecewise_move): Use it instead of
8384 GET_CLASS_NARROWEST_MODE (MODE_INT).
8385 (push_block): Likewise.
8386 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8388 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8391 Alan Hayward <alan.hayward@arm.com>
8392 David Sherwood <david.sherwood@arm.com>
8394 * postreload.c (move2add_valid_value_p): Change the type of the
8395 mode parameter to scalar_int_mode.
8396 (move2add_use_add2_insn): Add a mode parameter and use it instead
8398 (move2add_use_add3_insn): Likewise.
8399 (reload_cse_move2add): Update accordingly.
8401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8402 Alan Hayward <alan.hayward@arm.com>
8403 David Sherwood <david.sherwood@arm.com>
8405 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8407 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8408 * optabs.c (expand_unop): Likewise.
8410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8411 Alan Hayward <alan.hayward@arm.com>
8412 David Sherwood <david.sherwood@arm.com>
8414 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8415 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8416 (popcount_loc_descriptor): Likewise.
8417 (bswap_loc_descriptor): Likewise.
8418 (rotate_loc_descriptor): Likewise.
8419 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8420 calling the functions above.
8422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8423 Alan Hayward <alan.hayward@arm.com>
8424 David Sherwood <david.sherwood@arm.com>
8426 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8428 (try_combine): Likewise.
8429 (simplify_if_then_else): Likewise.
8430 * cse.c (cse_insn): Likewise.
8431 * dwarf2out.c (mem_loc_descriptor): Likewise.
8432 * emit-rtl.c (gen_lowpart_common): Likewise.
8433 * simplify-rtx.c (simplify_truncation): Likewise.
8434 (simplify_binary_operation_1): Likewise.
8435 (simplify_const_relational_operation): Likewise.
8436 (simplify_ternary_operation): Likewise.
8437 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8440 Alan Hayward <alan.hayward@arm.com>
8441 David Sherwood <david.sherwood@arm.com>
8443 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8444 * reload.c (push_reload): Likewise.
8445 (find_reloads): Likewise.
8447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8448 Alan Hayward <alan.hayward@arm.com>
8449 David Sherwood <david.sherwood@arm.com>
8451 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8452 (make_compound_operation_int): Likewise.
8453 (change_zero_ext): Likewise.
8454 * expr.c (convert_move): Likewise.
8455 (convert_modes): Likewise.
8456 * fwprop.c (forward_propagate_subreg): Likewise.
8457 * loop-iv.c (get_biv_step_1): Likewise.
8458 * optabs.c (widen_operand): Likewise.
8459 * postreload.c (move2add_valid_value_p): Likewise.
8460 * recog.c (simplify_while_replacing): Likewise.
8461 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8462 (simplify_binary_operation_1): Likewise. Remove redundant
8463 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
8470 is_a <scalar_int_mode>.
8471 (simplify_if_then_else): Likewise.
8472 (make_field_assignment): Likewise.
8473 (simplify_comparison): Likewise.
8474 * ifcvt.c (noce_try_bitop): Likewise.
8475 * loop-invariant.c (canonicalize_address_mult): Likewise.
8476 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8479 Alan Hayward <alan.hayward@arm.com>
8480 David Sherwood <david.sherwood@arm.com>
8482 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8483 is_a <scalar_int_mode> instead of != BLKmode.
8485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8486 Alan Hayward <alan.hayward@arm.com>
8487 David Sherwood <david.sherwood@arm.com>
8489 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8490 instead of != VOIDmode.
8491 * combine.c (if_then_else_cond): Likewise.
8492 (change_zero_ext): Likewise.
8493 * dwarf2out.c (mem_loc_descriptor): Likewise.
8494 (loc_descriptor): Likewise.
8495 * rtlanal.c (canonicalize_condition): Likewise.
8496 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8499 Alan Hayward <alan.hayward@arm.com>
8500 David Sherwood <david.sherwood@arm.com>
8502 * simplify-rtx.c (simplify_binary_operation_1): Use
8503 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8506 Alan Hayward <alan.hayward@arm.com>
8507 David Sherwood <david.sherwood@arm.com>
8509 * wide-int.h (int_traits<unsigned char>) New class.
8510 (int_traits<unsigned short>) Likewise.
8511 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8512 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8514 * combine.c (set_nonzero_bits_and_sign_copies): Use
8515 is_a <scalar_int_mode>.
8516 (find_split_point): Likewise.
8517 (combine_simplify_rtx): Likewise.
8518 (simplify_logical): Likewise.
8519 (expand_compound_operation): Likewise.
8520 (expand_field_assignment): Likewise.
8521 (make_compound_operation): Likewise.
8522 (extended_count): Likewise.
8523 (change_zero_ext): Likewise.
8524 (simplify_comparison): Likewise.
8525 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8526 (ucompare_loc_descriptor): Likewise.
8527 (minmax_loc_descriptor): Likewise.
8528 (mem_loc_descriptor): Likewise.
8529 (loc_descriptor): Likewise.
8530 * expmed.c (init_expmed_one_mode): Likewise.
8531 * lra-constraints.c (lra_constraint_offset): Likewise.
8532 * optabs.c (prepare_libcall_arg): Likewise.
8533 * postreload.c (move2add_note_store): Likewise.
8534 * reload.c (operands_match_p): Likewise.
8535 * rtl.h (load_extend_op): Likewise.
8536 * rtlhooks.c (gen_lowpart_general): Likewise.
8537 * simplify-rtx.c (simplify_truncation): Likewise.
8538 (simplify_unary_operation_1): Likewise.
8539 (simplify_binary_operation_1): Likewise.
8540 (simplify_const_binary_operation): Likewise.
8541 (simplify_const_relational_operation): Likewise.
8542 (simplify_subreg): Likewise.
8543 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8544 * var-tracking.c (adjust_mems): Likewise.
8545 (prepare_call_arguments): Likewise.
8547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8548 Alan Hayward <alan.hayward@arm.com>
8549 David Sherwood <david.sherwood@arm.com>
8551 * machmode.h (is_int_mode): New fuction.
8552 * combine.c (find_split_point): Use it.
8553 (combine_simplify_rtx): Likewise.
8554 (simplify_if_then_else): Likewise.
8555 (simplify_set): Likewise.
8556 (simplify_shift_const_1): Likewise.
8557 (simplify_comparison): Likewise.
8558 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8559 * cse.c (notreg_cost): Likewise.
8560 (cse_insn): Likewise.
8561 * cselib.c (cselib_lookup_1): Likewise.
8562 * dojump.c (do_jump_1): Likewise.
8563 (do_compare_rtx_and_jump): Likewise.
8564 * dse.c (get_call_args): Likewise.
8565 * dwarf2out.c (rtl_for_decl_init): Likewise.
8566 (native_encode_initializer): Likewise.
8567 * expmed.c (emit_store_flag_1): Likewise.
8568 (emit_store_flag): Likewise.
8569 * expr.c (convert_modes): Likewise.
8570 (store_field): Likewise.
8571 (expand_expr_real_1): Likewise.
8572 * fold-const.c (fold_read_from_constant_string): Likewise.
8573 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8574 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8575 * optabs.c (expand_binop): Likewise.
8576 (expand_unop): Likewise.
8577 (expand_abs_nojump): Likewise.
8578 (expand_one_cmpl_abs_nojump): Likewise.
8579 * simplify-rtx.c (mode_signbit_p): Likewise.
8580 (val_signbit_p): Likewise.
8581 (val_signbit_known_set_p): Likewise.
8582 (val_signbit_known_clear_p): Likewise.
8583 (simplify_relational_operation_1): Likewise.
8584 * tree.c (vector_type_mode): Likewise.
8586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8587 Alan Hayward <alan.hayward@arm.com>
8588 David Sherwood <david.sherwood@arm.com>
8590 * machmode.h (smallest_mode_for_size): Fix formatting.
8591 (smallest_int_mode_for_size): New function.
8592 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8593 instead of smallest_mode_for_size.
8594 * combine.c (make_extraction): Likewise.
8595 * config/arc/arc.c (arc_expand_movmem): Likewise.
8596 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8597 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8598 * config/s390/s390.c (s390_expand_insv): Likewise.
8599 * config/sparc/sparc.c (assign_int_registers): Likewise.
8600 * config/spu/spu.c (spu_function_value): Likewise.
8601 (spu_function_arg): Likewise.
8602 * coverage.c (get_gcov_type): Likewise.
8603 (get_gcov_unsigned_t): Likewise.
8604 * dse.c (find_shift_sequence): Likewise.
8605 * expmed.c (store_bit_field_1): Likewise.
8606 * expr.c (convert_move): Likewise.
8607 (store_field): Likewise.
8608 * internal-fn.c (expand_arith_overflow): Likewise.
8609 * optabs-query.c (get_best_extraction_insn): Likewise.
8610 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8611 * stor-layout.c (layout_type): Likewise.
8612 (initialize_sizetypes): Likewise.
8613 * targhooks.c (default_get_mask_mode): Likewise.
8614 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8617 Alan Hayward <alan.hayward@arm.com>
8618 David Sherwood <david.sherwood@arm.com>
8620 * machmode.h (opt_mode::else_blk): New function.
8621 (int_mode_for_mode): Declare.
8622 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8623 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8625 * cfgexpand.c (expand_debug_expr): Likewise.
8626 * combine.c (gen_lowpart_or_truncate): Likewise.
8627 (gen_lowpart_for_combine): Likewise.
8628 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8629 * config/avr/avr.c (avr_to_int_mode): Likewise.
8630 (avr_out_plus_1): Likewise.
8631 (avr_out_plus): Likewise.
8632 (avr_out_round): Likewise.
8633 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8634 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8635 (s390_expand_vcond): Likewise.
8636 * config/spu/spu.c (spu_split_immediate): Likewise.
8637 (spu_expand_mov): Likewise.
8638 * dse.c (get_stored_val): Likewise.
8639 * expmed.c (store_bit_field_1): Likewise.
8640 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8642 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8643 (extract_low_bits): Likewise.
8644 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8645 handling rather than repeating the check.
8646 (emit_group_store): Likewise.
8647 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8648 * optabs.c (expand_absneg_bit): Likewise.
8649 (expand_copysign_absneg): Likewise.
8650 (expand_copysign_bit): Likewise.
8651 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8652 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8653 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8654 * var-tracking.c (prepare_call_arguments): Likewise.
8655 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8656 int_mode_for_mode instead of mode_for_size.
8657 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8660 Alan Hayward <alan.hayward@arm.com>
8661 David Sherwood <david.sherwood@arm.com>
8663 * machmode.h (int_mode_for_size): New function.
8664 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8665 instead of mode_for_size.
8666 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8668 * combine.c (expand_field_assignment): Use int_mode_for_size
8669 instead of mode_for_size.
8670 (make_extraction): Likewise.
8671 (simplify_shift_const_1): Likewise.
8672 (simplify_comparison): Likewise.
8673 * dojump.c (do_jump): Likewise.
8674 * dwarf2out.c (mem_loc_descriptor): Likewise.
8675 * emit-rtl.c (init_derived_machine_modes): Likewise.
8676 * expmed.c (flip_storage_order): Likewise.
8677 (convert_extracted_bit_field): Likewise.
8678 * expr.c (copy_blkmode_from_reg): Likewise.
8679 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8680 * internal-fn.c (expand_mul_overflow): Likewise.
8681 * lower-subreg.c (simple_move): Likewise.
8682 * optabs-libfuncs.c (init_optabs): Likewise.
8683 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8684 * tree.c (vector_type_mode): Likewise.
8685 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8686 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8687 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8688 * tree-vect-stmts.c (vectorizable_load): Likewise.
8689 (vectorizable_store): Likewise.
8691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8692 Alan Hayward <alan.hayward@arm.com>
8693 David Sherwood <david.sherwood@arm.com>
8695 * coretypes.h (pod_mode): New type.
8696 (scalar_int_mode_pod): New typedef.
8697 * machmode.h (pod_mode): New class.
8698 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8699 * genmodes.c (emit_mode_int_n): Update accordingly.
8700 * lower-subreg.h (target_lower_subreg): Change type to
8701 scalar_int_mode_pod.
8702 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8703 scalar_int_mode_pod.
8705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8706 Alan Hayward <alan.hayward@arm.com>
8707 David Sherwood <david.sherwood@arm.com>
8709 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8710 machine_mode to scalar_int_mode.
8711 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8712 (rs6000_option_override_internal): Remove cast to int.
8713 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8714 machine_mode to scalar_int_mode.
8715 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8716 (rs6000_option_override_internal): Remove cast to int.
8717 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8718 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8720 * config/s390/s390.c (s390_expand_builtin): Likewise.
8721 * coretypes.h (scalar_int_mode): New type.
8722 (opt_scalar_int_mode): New typedef.
8723 * machmode.h (scalar_int_mode): New class.
8724 (scalar_int_mode::includes_p): New function.
8725 (byte_mode): Change type to scalar_int_mode.
8726 (word_mode): Likewise.
8727 (ptr_mode): Likewise.
8728 * emit-rtl.c (byte_mode): Likewise.
8729 (word_mode): Likewise.
8730 (ptr_mode): Likewise.
8731 (init_derived_machine_modes): Update accordingly.
8732 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8733 and MODE_PARTIAL_INT.
8734 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8735 opt_scalar_int_mode.
8737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8738 Alan Hayward <alan.hayward@arm.com>
8739 David Sherwood <david.sherwood@arm.com>
8741 * target.def (libgcc_floating_mode_supported_p): Take a
8743 * doc/tm.texi: Regenerate.
8744 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8746 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8747 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8751 Alan Hayward <alan.hayward@arm.com>
8752 David Sherwood <david.sherwood@arm.com>
8754 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8755 * doc/tm.texi: Regenerate.
8756 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8757 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8758 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8759 * targhooks.h (default_floatn_mode): Likewise.
8760 * targhooks.c (default_floatn_mode): Likewise.
8761 * tree.c (build_common_tree_nodes): Update accordingly.
8763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8764 Alan Hayward <alan.hayward@arm.com>
8765 David Sherwood <david.sherwood@arm.com>
8767 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8768 (mode_iterator::iterate_p): Likewise.
8769 (mode_iterator::get_wider): Likewise.
8770 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8773 Alan Hayward <alan.hayward@arm.com>
8774 David Sherwood <david.sherwood@arm.com>
8776 * coretypes.h (opt_scalar_float_mode): New typedef.
8777 * machmode.h (float_mode_for_size): New function.
8778 * emit-rtl.c (double_mode): Delete.
8779 (init_emit_once): Use float_mode_for_size.
8780 * stor-layout.c (layout_type): Likewise.
8781 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
8788 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8789 * config/arm/arm.md (consttable_4): Likewise.
8790 (consttable_8): Likewise.
8791 (consttable_16): Likewise.
8792 * config/mips/mips.md (consttable_float): Likewise.
8793 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8794 * varasm.c (assemble_real): Take a scalar_float_mode.
8795 (output_constant_pool_2): Update accordingly.
8796 (output_constant): Likewise.
8798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8799 Alan Hayward <alan.hayward@arm.com>
8800 David Sherwood <david.sherwood@arm.com>
8802 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8803 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8804 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8805 (native_encode_real): Likewise.
8806 (native_interpret_real): Likewise.
8807 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8808 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8811 Alan Hayward <alan.hayward@arm.com>
8812 David Sherwood <david.sherwood@arm.com>
8814 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8815 <scalar_float_mode>. Simplify.
8816 (gen_extend_conv_libfunc): Likewise.
8818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8819 Alan Hayward <alan.hayward@arm.com>
8820 David Sherwood <david.sherwood@arm.com>
8822 * coretypes.h (scalar_float_mode): New type.
8823 * machmode.h (mode_traits::from_int): Use machine_mode if
8824 USE_ENUM_MODES is defined.
8825 (is_a): New function.
8827 (dyn_cast): Likewise.
8828 (scalar_float_mode): New class.
8829 (scalar_float_mode::includes_p): New function.
8830 (is_float_mode): Likewise.
8831 * gdbhooks.py (MachineModePrinter): New class.
8832 (build_pretty_printer): Use it for scalar_float_mode.
8833 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8834 (format_helper::format_helper): Turn into a template.
8835 * genmodes.c (get_mode_class): New function.
8836 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8837 or machine_mode if none.
8838 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8839 as_a <scalar_float_mode>.
8840 * dwarf2out.c (mem_loc_descriptor): Likewise.
8841 (insert_float): Likewise.
8842 (add_const_value_attribute): Likewise.
8843 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8844 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8845 (expand_unop): Update accordingly.
8846 (expand_abs_nojump): Likewise.
8847 (expand_copysign_absneg): Take a scalar_float_mode.
8848 (expand_copysign_bit): Likewise.
8849 (expand_copysign): Update accordingly.
8851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8852 Alan Hayward <alan.hayward@arm.com>
8853 David Sherwood <david.sherwood@arm.com>
8855 * coretypes.h (opt_mode): New class.
8856 * machmode.h (opt_mode): Likewise.
8857 (opt_mode::else_void): New function.
8858 (opt_mode::require): Likewise.
8859 (opt_mode::exists): Likewise.
8860 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8861 (GET_MODE_2XWIDER_MODE): Likewise.
8862 (mode_iterator::get_wider): Update accordingly.
8863 (mode_iterator::get_2xwider): Likewise.
8864 (mode_iterator::get_known_wider): Likewise, turning into a template.
8865 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8866 forcing a wider mode to exist.
8867 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8868 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8869 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8870 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8871 * lower-subreg.c (init_lower_subreg): Likewise.
8872 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8873 on the final iteration.
8874 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8875 a wider mode exists before asking for a move pattern.
8876 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8877 forcing a wider mode to exist.
8878 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8879 returning false if no such mode exists.
8880 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8881 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8882 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8883 Avoid checking for a MODE_INT if we already know the mode is not a
8885 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8886 forcing a wider mode to exist.
8887 (expmed_mult_highpart_optab): Likewise.
8888 (expmed_mult_highpart): Likewise.
8889 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8891 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8892 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8893 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8894 * internal-fn.c (expand_mul_overflow): Update use of
8895 GET_MODE_2XWIDER_MODE.
8896 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8897 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8898 GET_MODE_WIDER_MODE.
8899 (convert_plusminus_to_widen): Likewise.
8900 * tree-switch-conversion.c (array_value_type): Likewise.
8901 * var-tracking.c (emit_note_insn_var_location): Likewise.
8902 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8903 Return false inside rather than outside the loop if no wider mode
8905 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8906 and GET_MODE_2XWIDER_MODE
8907 (can_compare_p): Use else_void.
8908 * gdbhooks.py (OptMachineModePrinter): New class.
8909 (build_pretty_printer): Use it for opt_mode.
8911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8912 Alan Hayward <alan.hayward@arm.com>
8913 David Sherwood <david.sherwood@arm.com>
8915 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8916 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8919 Alan Hayward <alan.hayward@arm.com>
8920 David Sherwood <david.sherwood@arm.com>
8922 * machmode.h (mode_traits): New structure.
8923 (get_narrowest_mode): New function.
8924 (mode_iterator::start): Likewise.
8925 (mode_iterator::iterate_p): Likewise.
8926 (mode_iterator::get_wider): Likewise.
8927 (mode_iterator::get_known_wider): Likewise.
8928 (mode_iterator::get_2xwider): Likewise.
8929 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8930 (FOR_EACH_MODE): Likewise.
8931 (FOR_EACH_MODE_FROM): Likewise.
8932 (FOR_EACH_MODE_UNTIL): Likewise.
8933 (FOR_EACH_WIDER_MODE): Likewise.
8934 (FOR_EACH_2XWIDER_MODE): Likewise.
8935 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8936 * combine.c (simplify_comparison): Likewise
8937 * config/i386/i386.c (type_natural_mode): Likewise.
8938 * cse.c (cse_insn): Likewise.
8939 * dse.c (find_shift_sequence): Likewise.
8940 * emit-rtl.c (init_derived_machine_modes): Likewise.
8941 (init_emit_once): Likewise.
8942 * explow.c (hard_function_value): Likewise.
8943 * expmed.c (extract_fixed_bit_field_1): Likewise.
8944 (extract_bit_field_1): Likewise.
8945 (expand_divmod): Likewise.
8946 (emit_store_flag_1): Likewise.
8947 * expr.c (init_expr_target): Likewise.
8948 (convert_move): Likewise.
8949 (alignment_for_piecewise_move): Likewise.
8950 (widest_int_mode_for_size): Likewise.
8951 (emit_block_move_via_movmem): Likewise.
8952 (copy_blkmode_to_reg): Likewise.
8953 (set_storage_via_setmem): Likewise.
8954 (compress_float_constant): Likewise.
8955 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8956 * optabs-query.c (get_best_extraction_insn): Likewise.
8957 * optabs.c (expand_binop): Likewise.
8958 (expand_twoval_unop): Likewise.
8959 (expand_twoval_binop): Likewise.
8960 (widen_leading): Likewise.
8961 (widen_bswap): Likewise.
8962 (expand_parity): Likewise.
8963 (expand_unop): Likewise.
8964 (prepare_cmp_insn): Likewise.
8965 (prepare_float_lib_cmp): Likewise.
8966 (expand_float): Likewise.
8967 (expand_fix): Likewise.
8968 (expand_sfix_optab): Likewise.
8969 * postreload.c (move2add_use_add2_insn): Likewise.
8970 * reg-stack.c (reg_to_stack): Likewise.
8971 * reginfo.c (choose_hard_reg_mode): Likewise.
8972 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8973 * stor-layout.c (mode_for_size): Likewise.
8974 (smallest_mode_for_size): Likewise.
8975 (mode_for_vector): Likewise.
8976 (finish_bitfield_representative): Likewise.
8977 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8978 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8979 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8980 * var-tracking.c (prepare_call_arguments): Likewise.
8982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8983 Alan Hayward <alan.hayward@arm.com>
8984 David Sherwood <david.sherwood@arm.com>
8986 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8987 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8988 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8989 * machmode.h (mode_size): Move earlier in file.
8990 (mode_precision): Likewise.
8991 (mode_inner): Likewise.
8992 (mode_nunits): Likewise.
8993 (mode_unit_size): Likewise.
8994 (unit_unit_precision): Likewise.
8995 (mode_wider): Likewise.
8996 (mode_2xwider): Likewise.
8997 (machine_mode): New class.
8998 (mode_to_bytes): New function.
8999 (mode_to_bits): Likewise.
9000 (mode_to_precision): Likewise.
9001 (mode_to_inner): Likewise.
9002 (mode_to_unit_size): Likewise.
9003 (mode_to_unit_precision): Likewise.
9004 (mode_to_nunits): Likewise.
9005 (GET_MODE_SIZE): Use mode_to_bytes.
9006 (GET_MODE_BITSIZE): Use mode_to_bits.
9007 (GET_MODE_PRECISION): Use mode_to_precision.
9008 (GET_MODE_INNER): Use mode_to_inner.
9009 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9010 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9011 (GET_MODE_NUNITS): Use mode_to_nunits.
9012 * system.h (ALWAYS_INLINE): New macro.
9013 * config/powerpcspe/powerpcspe-c.c
9014 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9015 int for arg1_mode and arg2_mode.
9017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9018 Alan Hayward <alan.hayward@arm.com>
9019 David Sherwood <david.sherwood@arm.com>
9021 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9022 Prefix mode names with E_ in case statements.
9023 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9024 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9025 (aarch64_split_simd_move): Likewise.
9026 (aarch64_gen_storewb_pair): Likewise.
9027 (aarch64_gen_loadwb_pair): Likewise.
9028 (aarch64_gen_store_pair): Likewise.
9029 (aarch64_gen_load_pair): Likewise.
9030 (aarch64_get_condition_code_1): Likewise.
9031 (aarch64_constant_pool_reload_icode): Likewise.
9032 (get_rsqrte_type): Likewise.
9033 (get_rsqrts_type): Likewise.
9034 (get_recpe_type): Likewise.
9035 (get_recps_type): Likewise.
9036 (aarch64_gimplify_va_arg_expr): Likewise.
9037 (aarch64_simd_container_mode): Likewise.
9038 (aarch64_emit_load_exclusive): Likewise.
9039 (aarch64_emit_store_exclusive): Likewise.
9040 (aarch64_expand_compare_and_swap): Likewise.
9041 (aarch64_gen_atomic_cas): Likewise.
9042 (aarch64_emit_bic): Likewise.
9043 (aarch64_emit_atomic_swap): Likewise.
9044 (aarch64_emit_atomic_load_op): Likewise.
9045 (aarch64_evpc_trn): Likewise.
9046 (aarch64_evpc_uzp): Likewise.
9047 (aarch64_evpc_zip): Likewise.
9048 (aarch64_evpc_ext): Likewise.
9049 (aarch64_evpc_rev): Likewise.
9050 (aarch64_evpc_dup): Likewise.
9051 (aarch64_gen_ccmp_first): Likewise.
9052 (aarch64_gen_ccmp_next): Likewise.
9053 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9054 (alpha_emit_xfloating_libcall): Likewise.
9055 (emit_insxl): Likewise.
9056 (alpha_arg_type): Likewise.
9057 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9058 (arc_preferred_simd_mode): Likewise.
9059 (arc_secondary_reload): Likewise.
9060 (get_arc_condition_code): Likewise.
9061 (arc_print_operand): Likewise.
9062 (arc_legitimate_constant_p): Likewise.
9063 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9064 * config/arc/arc.md (casesi_load): Likewise.
9065 (casesi_compact_jump): Likewise.
9066 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9067 (cc_use_register): Likewise.
9068 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9069 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9070 (arm_init_iwmmxt_builtins): Likewise.
9071 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9072 (neon_expand_vector_init): Likewise.
9073 (arm_attr_length_move_neon): Likewise.
9074 (maybe_get_arm_condition_code): Likewise.
9075 (arm_emit_vector_const): Likewise.
9076 (arm_preferred_simd_mode): Likewise.
9077 (arm_output_iwmmxt_tinsr): Likewise.
9078 (thumb1_output_casesi): Likewise.
9079 (thumb2_output_casesi): Likewise.
9080 (arm_emit_load_exclusive): Likewise.
9081 (arm_emit_store_exclusive): Likewise.
9082 (arm_expand_compare_and_swap): Likewise.
9083 (arm_evpc_neon_vuzp): Likewise.
9084 (arm_evpc_neon_vzip): Likewise.
9085 (arm_evpc_neon_vrev): Likewise.
9086 (arm_evpc_neon_vtrn): Likewise.
9087 (arm_evpc_neon_vext): Likewise.
9088 (arm_validize_comparison): Likewise.
9089 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9090 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9091 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9092 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9093 (c6x_preferred_simd_mode): Likewise.
9094 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9095 (epiphany_rtx_costs): Likewise.
9096 * config/epiphany/predicates.md (proper_comparison_operator):
9098 * config/frv/frv.c (condexec_memory_operand): Likewise.
9099 (frv_emit_move): Likewise.
9100 (output_move_single): Likewise.
9101 (output_condmove_single): Likewise.
9102 (frv_hard_regno_mode_ok): Likewise.
9103 (frv_matching_accg_mode): Likewise.
9104 * config/h8300/h8300.c (split_adds_subs): Likewise.
9105 (h8300_rtx_costs): Likewise.
9106 (h8300_print_operand): Likewise.
9107 (compute_mov_length): Likewise.
9108 (output_logical_op): Likewise.
9109 (compute_logical_op_length): Likewise.
9110 (compute_logical_op_cc): Likewise.
9111 (h8300_shift_needs_scratch_p): Likewise.
9112 (output_a_shift): Likewise.
9113 (compute_a_shift_length): Likewise.
9114 (compute_a_shift_cc): Likewise.
9115 (expand_a_rotate): Likewise.
9116 (output_a_rotate): Likewise.
9117 * config/i386/i386.c (classify_argument): Likewise.
9118 (function_arg_advance_32): Likewise.
9119 (function_arg_32): Likewise.
9120 (function_arg_64): Likewise.
9121 (function_value_64): Likewise.
9122 (ix86_gimplify_va_arg): Likewise.
9123 (ix86_legitimate_constant_p): Likewise.
9124 (put_condition_code): Likewise.
9125 (split_double_mode): Likewise.
9126 (ix86_avx256_split_vector_move_misalign): Likewise.
9127 (ix86_expand_vector_logical_operator): Likewise.
9128 (ix86_split_idivmod): Likewise.
9129 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9130 (ix86_build_const_vector): Likewise.
9131 (ix86_build_signbit_mask): Likewise.
9132 (ix86_match_ccmode): Likewise.
9133 (ix86_cc_modes_compatible): Likewise.
9134 (ix86_expand_branch): Likewise.
9135 (ix86_expand_sse_cmp): Likewise.
9136 (ix86_expand_sse_movcc): Likewise.
9137 (ix86_expand_int_sse_cmp): Likewise.
9138 (ix86_expand_vec_perm_vpermi2): Likewise.
9139 (ix86_expand_vec_perm): Likewise.
9140 (ix86_expand_sse_unpack): Likewise.
9141 (ix86_expand_int_addcc): Likewise.
9142 (ix86_split_to_parts): Likewise.
9143 (ix86_vectorize_builtin_gather): Likewise.
9144 (ix86_vectorize_builtin_scatter): Likewise.
9145 (avx_vpermilp_parallel): Likewise.
9146 (inline_memory_move_cost): Likewise.
9147 (ix86_tieable_integer_mode_p): Likewise.
9148 (x86_maybe_negate_const_int): Likewise.
9149 (ix86_expand_vector_init_duplicate): Likewise.
9150 (ix86_expand_vector_init_one_nonzero): Likewise.
9151 (ix86_expand_vector_init_one_var): Likewise.
9152 (ix86_expand_vector_init_concat): Likewise.
9153 (ix86_expand_vector_init_interleave): Likewise.
9154 (ix86_expand_vector_init_general): Likewise.
9155 (ix86_expand_vector_set): Likewise.
9156 (ix86_expand_vector_extract): Likewise.
9157 (emit_reduc_half): Likewise.
9158 (ix86_emit_i387_round): Likewise.
9159 (ix86_mangle_type): Likewise.
9160 (ix86_expand_round_sse4): Likewise.
9161 (expand_vec_perm_blend): Likewise.
9162 (canonicalize_vector_int_perm): Likewise.
9163 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9164 (expand_vec_perm_1): Likewise.
9165 (expand_vec_perm_interleave3): Likewise.
9166 (expand_vec_perm_even_odd_pack): Likewise.
9167 (expand_vec_perm_even_odd_1): Likewise.
9168 (expand_vec_perm_broadcast_1): Likewise.
9169 (ix86_vectorize_vec_perm_const_ok): Likewise.
9170 (ix86_expand_vecop_qihi): Likewise.
9171 (ix86_expand_mul_widen_hilo): Likewise.
9172 (ix86_expand_sse2_abs): Likewise.
9173 (ix86_expand_pextr): Likewise.
9174 (ix86_expand_pinsr): Likewise.
9175 (ix86_preferred_simd_mode): Likewise.
9176 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9177 * config/i386/sse.md (*andnot<mode>3): Likewise.
9178 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9179 (*<code><mode>3): Likewise.
9180 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9181 (ia64_expand_atomic_op): Likewise.
9182 (ia64_arg_type): Likewise.
9183 (ia64_mode_to_int): Likewise.
9184 (ia64_scalar_mode_supported_p): Likewise.
9185 (ia64_vector_mode_supported_p): Likewise.
9186 (expand_vec_perm_broadcast): Likewise.
9187 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9188 (iq2000_function_arg_advance): Likewise.
9189 (iq2000_function_arg): Likewise.
9190 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9191 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9192 (m68k_libcall_value): Likewise.
9193 (m68k_function_value): Likewise.
9194 (sched_attr_op_type): Likewise.
9195 * config/mcore/mcore.c (mcore_output_move): Likewise.
9196 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9198 (microblaze_function_arg): Likewise.
9199 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9200 (mips_print_operand): Likewise.
9201 (mips_mode_ok_for_mov_fmt_p): Likewise.
9202 (mips_vector_mode_supported_p): Likewise.
9203 (mips_preferred_simd_mode): Likewise.
9204 (mips_expand_vpc_loongson_even_odd): Likewise.
9205 (mips_expand_vec_unpack): Likewise.
9206 (mips_expand_vi_broadcast): Likewise.
9207 (mips_expand_vector_init): Likewise.
9208 (mips_expand_vec_reduc): Likewise.
9209 (mips_expand_msa_cmp): Likewise.
9210 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9211 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9212 (cc_flags_for_mode): Likewise.
9213 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9214 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9215 (nds32_output_casesi_pc_relative): Likewise.
9216 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9217 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9218 (nvptx_gen_unpack): Likewise.
9219 (nvptx_gen_pack): Likewise.
9220 (nvptx_gen_shuffle): Likewise.
9221 (nvptx_gen_wcast): Likewise.
9222 (nvptx_preferred_simd_mode): Likewise.
9223 * config/pa/pa.c (pa_secondary_reload): Likewise.
9224 * config/pa/predicates.md (base14_operand): Likewise.
9225 * config/powerpcspe/powerpcspe-c.c
9226 (altivec_resolve_overloaded_builtin): Likewise.
9227 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9229 (rs6000_preferred_simd_mode): Likewise.
9230 (output_vec_const_move): Likewise.
9231 (rs6000_expand_vector_extract): Likewise.
9232 (rs6000_split_vec_extract_var): Likewise.
9233 (reg_offset_addressing_ok_p): Likewise.
9234 (rs6000_legitimate_offset_address_p): Likewise.
9235 (rs6000_legitimize_address): Likewise.
9236 (rs6000_emit_set_const): Likewise.
9237 (rs6000_const_vec): Likewise.
9238 (rs6000_emit_move): Likewise.
9239 (spe_build_register_parallel): Likewise.
9240 (rs6000_darwin64_record_arg_recurse): Likewise.
9241 (swap_selector_for_mode): Likewise.
9242 (spe_init_builtins): Likewise.
9243 (paired_init_builtins): Likewise.
9244 (altivec_init_builtins): Likewise.
9245 (do_load_for_compare): Likewise.
9246 (rs6000_generate_compare): Likewise.
9247 (rs6000_expand_float128_convert): Likewise.
9248 (emit_load_locked): Likewise.
9249 (emit_store_conditional): Likewise.
9250 (rs6000_output_function_epilogue): Likewise.
9251 (rs6000_handle_altivec_attribute): Likewise.
9252 (rs6000_function_value): Likewise.
9253 (emit_fusion_gpr_load): Likewise.
9254 (emit_fusion_p9_load): Likewise.
9255 (emit_fusion_p9_store): Likewise.
9256 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9257 (fusion_gpr_mem_load): Likewise.
9258 (fusion_addis_mem_combo_load): Likewise.
9259 (fusion_addis_mem_combo_store): Likewise.
9260 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9261 (fusion_gpr_mem_load): Likewise.
9262 (fusion_addis_mem_combo_load): Likewise.
9263 (fusion_addis_mem_combo_store): Likewise.
9264 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9266 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9267 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9268 (rs6000_preferred_simd_mode): Likewise.
9269 (output_vec_const_move): Likewise.
9270 (rs6000_expand_vector_extract): Likewise.
9271 (rs6000_split_vec_extract_var): Likewise.
9272 (reg_offset_addressing_ok_p): Likewise.
9273 (rs6000_legitimate_offset_address_p): Likewise.
9274 (rs6000_legitimize_address): Likewise.
9275 (rs6000_emit_set_const): Likewise.
9276 (rs6000_const_vec): Likewise.
9277 (rs6000_emit_move): Likewise.
9278 (rs6000_darwin64_record_arg_recurse): Likewise.
9279 (swap_selector_for_mode): Likewise.
9280 (paired_init_builtins): Likewise.
9281 (altivec_init_builtins): Likewise.
9282 (rs6000_expand_float128_convert): Likewise.
9283 (emit_load_locked): Likewise.
9284 (emit_store_conditional): Likewise.
9285 (rs6000_output_function_epilogue): Likewise.
9286 (rs6000_handle_altivec_attribute): Likewise.
9287 (rs6000_function_value): Likewise.
9288 (emit_fusion_gpr_load): Likewise.
9289 (emit_fusion_p9_load): Likewise.
9290 (emit_fusion_p9_store): Likewise.
9291 * config/rx/rx.c (rx_gen_move_template): Likewise.
9292 (flags_from_mode): Likewise.
9293 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9294 (s390_slb_comparison): Likewise.
9295 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9296 (s390_vector_mode_supported_p): Likewise.
9297 (s390_cc_modes_compatible): Likewise.
9298 (s390_match_ccmode_set): Likewise.
9299 (s390_canonicalize_comparison): Likewise.
9300 (s390_emit_compare_and_swap): Likewise.
9301 (s390_branch_condition_mask): Likewise.
9302 (s390_rtx_costs): Likewise.
9303 (s390_secondary_reload): Likewise.
9304 (__SECONDARY_RELOAD_CASE): Likewise.
9305 (s390_expand_cs): Likewise.
9306 (s390_preferred_simd_mode): Likewise.
9307 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9308 * config/sh/sh.c (sh_print_operand): Likewise.
9309 (dump_table): Likewise.
9310 (sh_secondary_reload): Likewise.
9311 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9312 * config/sh/sh.md (casesi_worker_1): Likewise.
9313 (casesi_worker_2): Likewise.
9314 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9315 (fcc_comparison_operator): Likewise.
9316 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9317 (emit_soft_tfmode_cvt): Likewise.
9318 (sparc_preferred_simd_mode): Likewise.
9319 (output_cbranch): Likewise.
9320 (sparc_print_operand): Likewise.
9321 (sparc_expand_vec_perm_bmask): Likewise.
9322 (vector_init_bshuffle): Likewise.
9323 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9324 (spu_vector_mode_supported_p): Likewise.
9325 (spu_expand_insv): Likewise.
9326 (spu_emit_branch_or_set): Likewise.
9327 (spu_handle_vector_attribute): Likewise.
9328 (spu_builtin_splats): Likewise.
9329 (spu_builtin_extract): Likewise.
9330 (spu_builtin_promote): Likewise.
9331 (spu_expand_sign_extend): Likewise.
9332 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9333 (tilegx_simd_int): Likewise.
9334 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9335 (tilepro_simd_int): Likewise.
9336 * config/v850/v850.c (const_double_split): Likewise.
9337 (v850_print_operand): Likewise.
9338 (ep_memory_offset): Likewise.
9339 * config/vax/vax.c (vax_rtx_costs): Likewise.
9340 (vax_output_int_move): Likewise.
9341 (vax_output_int_add): Likewise.
9342 (vax_output_int_subtract): Likewise.
9343 * config/visium/predicates.md (visium_branch_operator): Likewise.
9344 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9345 (visium_print_operand_address): Likewise.
9346 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9347 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9348 (xtensa_expand_conditional_branch): Likewise.
9349 (xtensa_copy_incoming_a7): Likewise.
9350 (xtensa_output_literal): Likewise.
9351 * dfp.c (decimal_real_maxval): Likewise.
9352 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9355 Alan Hayward <alan.hayward@arm.com>
9356 David Sherwood <david.sherwood@arm.com>
9358 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9359 (mode_nunits_inline): Likewise.
9360 (mode_inner_inline): Likewise.
9361 (mode_unit_size_inline): Likewise.
9362 (mode_unit_precision_inline): Likewise.
9363 (emit_insn_modes_h): Likewise. Also emit a #define of the
9365 (emit_mode_wider): Add an E_ prefix to mode names.
9366 (emit_mode_complex): Likewise.
9367 (emit_mode_inner): Likewise.
9368 (emit_mode_adjustments): Likewise.
9369 (emit_mode_int_n): Likewise.
9370 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9371 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9372 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9373 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9374 (CRC32_BUILTIN, ENTRY): Likewise.
9375 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9376 (aarch64_pop_regs): Likewise.
9377 (aarch64_process_components): Likewise.
9378 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9379 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9380 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9381 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9382 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9383 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9385 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9386 (pcmpistr): Likewise.
9387 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9388 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9389 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9391 * config/rl78/rl78.c (mduc_regs): Likewise.
9392 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9393 (htm_expand_builtin): Likewise.
9394 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9395 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9396 * config/xtensa/xtensa.c (print_operand): Likewise.
9397 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9398 (NUM_MODE_VECTOR_INT): Likewise.
9399 * genoutput.c (null_operand): Likewise.
9400 (output_operand_data): Likewise.
9401 * genrecog.c (print_parameter_value): Likewise.
9402 * lra.c (debug_operand_data): Likewise.
9404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9406 * dwarf2out.c (scompare_loc_descriptor_wide)
9407 (scompare_loc_descriptor_narrow): New functions, split out from...
9408 (scompare_loc_descriptor): ...here.
9409 * expmed.c (emit_store_flag_int): New function, split out from...
9410 (emit_store_flag): ...here.
9412 2017-08-30 Richard Biener <rguenther@suse.de>
9414 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9415 (dwarf2out_early_finish): Move setting of AT_pubnames from
9416 early debug output to early finish.
9418 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9420 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9421 and -mdata-region to the assembler.
9423 2017-08-30 Richard Biener <rguenther@suse.de>
9425 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9427 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9428 (dwarf2out_early_global_decl): Only generate a DIE for the
9429 abstract origin if it doesn't already exist or is a declaration DIE.
9430 (resolve_addr): Do not add the linkage name twice when
9431 generating a stub DIE for the DW_TAG_GNU_call_site target.
9433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9435 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9436 Use machine_mode rather than int for arg1_mode.
9438 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9441 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9442 that the second argument of the built-in functions to unpack
9443 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9444 switch statement instead a lot of if statements.
9445 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9446 Allow 64-bit values to be in Altivec registers as well as
9447 traditional floating point registers.
9448 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9450 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9452 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9453 MAX_REGS_PER_ADDRESS == 1.
9455 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9457 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9458 * config/i386/i386.c (ix86_option_override_internal): Simplify
9459 setting of opts->x_flag_entry.
9461 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9462 Jakub Jelinek <jakub@redhat.com>
9463 Richard Biener <rguenther@suse.de>
9465 PR tree-optimization/81503
9466 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9467 folded constant fits in the target type; reorder tests for clarity.
9469 2017-08-29 Martin Liska <mliska@suse.cz>
9471 * passes.def: Include pass_lower_switch.
9472 * stmt.c (dump_case_nodes): Remove and move to
9473 tree-switch-conversion.
9474 (case_values_threshold): Likewise.
9475 (expand_switch_as_decision_tree_p): Likewise.
9476 (emit_case_decision_tree): Likewise.
9477 (expand_case): Likewise.
9478 (balance_case_nodes): Likewise.
9479 (node_has_low_bound): Likewise.
9480 (node_has_high_bound): Likewise.
9481 (node_is_bounded): Likewise.
9482 (emit_case_nodes): Likewise.
9483 (struct simple_case_node): New struct.
9484 (add_case_node): Remove.
9485 (emit_case_dispatch_table): Use vector instead of case_list.
9486 (reset_out_edges_aux): Remove.
9487 (compute_cases_per_edge): Likewise.
9488 (expand_case): Build list of simple_case_node.
9489 (expand_sjlj_dispatch_table): Use it.
9490 * tree-switch-conversion.c (struct case_node): Moved from
9491 stmt.c and adjusted.
9492 (emit_case_nodes): Likewise.
9493 (node_has_low_bound): Likewise.
9494 (node_has_high_bound): Likewise.
9495 (node_is_bounded): Likewise.
9496 (case_values_threshold): Likewise.
9497 (reset_out_edges_aux): Likewise.
9498 (compute_cases_per_edge): Likewise.
9499 (add_case_node): Likewise.
9500 (dump_case_nodes): Likewise.
9501 (balance_case_nodes): Likewise.
9502 (expand_switch_as_decision_tree_p): Likewise.
9503 (emit_jump): Likewise.
9504 (emit_case_decision_tree): Likewise.
9505 (try_switch_expansion): Likewise.
9506 (do_jump_if_equal): Likewise.
9507 (emit_cmp_and_jump_insns): Likewise.
9508 (fix_phi_operands_for_edge): New function.
9509 (record_phi_operand_mapping): Likewise.
9510 (class pass_lower_switch): New pass.
9511 (pass_lower_switch::execute): New function.
9512 (make_pass_lower_switch): Likewise.
9513 (conditional_probability):
9514 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9515 * tree-pass.h: Add make_pass_lower_switch.
9517 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9520 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9523 2017-08-29 Richard Biener <rguenther@suse.de>
9525 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9526 we do not add a DW_AT_inline attribute twice.
9527 (gen_subprogram_die): Remove code setting DW_AT_inline on
9528 DECL_ABSTRACT_P nodes.
9530 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9532 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9533 calls to internal functions.
9534 (gimplify_modify_expr): Likewise.
9535 * tree-call-cdce.c (use_internal_fn): Likewise.
9536 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9537 (convert_to_divmod): Set the nothrow flag.
9538 * tree-if-conv.c (predicate_mem_writes): Likewise.
9539 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9540 (vectorizable_call): Likewise.
9541 (vectorizable_store): Likewise.
9542 (vectorizable_load): Likewise.
9543 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9544 (vect_recog_mask_conversion_pattern): Likewise.
9546 2017-08-29 Martin Liska <mliska@suse.cz>
9549 * gcc.c (driver_handle_option): Add new argument.
9550 * opts-common.c (handle_option): Pass
9551 target_option_override_hook.
9552 * opts-global.c (lang_handle_option): Add new option.
9553 (set_default_handlers): Add new argument.
9554 (decode_options): Likewise.
9555 * opts.c (target_handle_option): Likewise.
9556 (common_handle_option): Call target_option_override_hook.
9557 * opts.h (struct cl_option_handler_func): Add hook for
9558 target option override.
9559 (struct cl_option_handlers): Likewise.
9560 (set_default_handlers): Add new argument.
9561 (decode_options): Likewise.
9562 (common_handle_option): Likewise.
9563 (target_handle_option): Likewise.
9564 * toplev.c (toplev::main): Pass targetm.target_option.override
9567 2017-08-29 Richard Biener <rguenther@suse.de>
9568 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9570 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9571 life to the active subtree.
9573 2017-08-28 Jeff Law <law@redhat.com>
9575 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9576 derive_equivalences.
9577 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9579 (edge_info::derive_equivalences): New private member function
9581 * tree-ssa-dom.c (class edge_info): Changed from a struct
9582 to a class. Add ctor/dtor, methods and data members.
9583 (edge_info::edge_info): Renamed from allocate_edge_info.
9584 Initialize additional members.
9585 (edge_info::~edge_info): New.
9586 (free_dom_edge_info): Delete the edge info.
9587 (record_edge_info): Use new class & associated member functions.
9588 Tighten forms for testing for edge equivalences.
9589 (record_temporary_equivalences): Iterate over the simple
9590 equivalences rather than assuming there's only one per edge.
9591 (cprop_into_successor_phis): Iterate over the simple
9592 equivalences rather than assuming there's only one per edge.
9593 (optimize_stmt): Use operand_equal_p rather than pointer
9594 equality for mini-DSE code.
9596 2017-08-28 Nathan Sidwell <nathan@acm.org>
9598 * gcc.c (execute): Fold SIGPIPE handling into switch
9599 statement. Adjust internal error message.
9601 2017-08-28 Richard Biener <rguenther@suse.de>
9604 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9605 Do nothing for removed DIEs.
9607 2017-08-28 Richard Biener <rguenther@suse.de>
9609 PR tree-optimization/81977
9610 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9613 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9616 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9617 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9618 using targetm.gen_mem_thread_fence.
9620 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9623 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9624 predicate to register_operand. Reorder operands.
9625 (*btr<mode>): Ditto.
9626 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9627 (*btr<mode>_mask): Ditto.
9629 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9631 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9632 * config/rs6000/xmmintrin.h: New file.
9633 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9635 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9638 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9639 parameter and_insn and return it.
9640 (recombine_lvx_pattern): Insert a copy to ensure availability of
9641 the base register of the copied masking operation at the point of
9642 the instruction replacement.
9643 (recombine_stvx_pattern): Likewise.
9645 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9647 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9648 undocumented switches.
9649 (-mpower9-dform-vector): Likewise.
9650 (-mpower9-dform): Likewise.
9651 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9652 comments to delete references to -mpower9-dform* switches.
9653 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9654 Delete reference to -mpower9-dform* switches, test for
9655 -mpower9-vector instead.
9656 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9657 (OTHER_P9_VECTOR_MASKS): Likewise.
9658 (POWERPC_MASKS): Likewise.
9659 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9660 tests against -mpower9-dform* to -mpower9-vector. Delete code
9661 that checked for -mpower9-dform* consistancy with other options.
9662 Add test for -mpower9-misc to enable other power9 switches.
9663 (rs6000_init_hard_regno_mode_ok): Likewise.
9664 (rs6000_option_override_internal): Likewise.
9665 (rs6000_emit_prologue): Likewise.
9666 (rs6000_emit_epilogue): Likewise.
9667 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9668 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9669 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9671 (emit_fusion_p9_store): Likewise.
9672 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9673 resetting these macros if the assembler does not support ISA 3.0
9675 (TARGET_P9_DFORM_VECTOR): Likewise.
9676 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9677 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9679 2017-08-25 Alan Modra <amodra@gmail.com>
9681 PR rtl-optimization/81747
9682 * cse.c (cse_extended_basic_block): Don't attempt to record
9683 equivalences for degenerate conditional branches that branch
9684 to their fall-through.
9686 2017-08-24 Martin Sebor <msebor@redhat.com>
9689 * gimple-fold.c (size_must_be_zero_p): New function.
9690 (gimple_fold_builtin_memory_op): Call it.
9692 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9694 * config/rs6000/mm_malloc.h: New file.
9696 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9698 PR tree-optimization/81913
9699 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9700 analysis when either IVs in condition can wrap.
9702 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9704 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9705 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9707 2017-08-24 Richard Biener <rguenther@suse.de>
9710 * targhooks.c (default_target_can_inline_p): Properly
9711 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9712 is present and always compare.
9713 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9714 infer -mfpmath=sse from TARGET_SSE_P.
9715 (ix86_can_inline_p): Properly use target_option_default_node when
9716 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9718 2017-08-24 Richard Biener <rguenther@suse.de>
9721 * dwarf2out.c (output_die): Handle flag_generate_offload like
9723 (output_comp_unit): Likewise.
9724 (gen_array_type_die): Likewise.
9725 (dwarf2out_early_finish): Likewise.
9726 (note_variable_value_in_expr): Likewise.
9727 (dwarf2out_finish): Likewise. Adjust assert.
9728 * cgraphunit.c (symbol_table::compile): Move setting of
9729 flag_generate_offload earlier ...
9730 (symbol_table::finalize_compilation_unit): ... here, before
9731 early debug finalization.
9733 2017-08-24 Richard Biener <rguenther@suse.de>
9735 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9736 and ipa-fnsummary.h.
9737 (ix86_can_inline_p): When ix86_fpmath flags do not match
9738 check whether the callee uses FP math at all.
9740 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9743 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9744 instead of TYPE_PRECISION.
9746 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9747 Alan Hayward <alan.hayward@arm.com>
9748 David Sherwood <david.sherwood@arm.com>
9750 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9752 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9754 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9756 (ix86_frame::stack_realign_allocate): New field.
9757 (struct machine_frame_state): Modify comments.
9758 (machine_frame_state::sp_realigned_fp_end): New field.
9759 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9761 (sp_valid_at): Add assertion to assure no attempt to access invalid
9762 offset of a realigned stack.
9763 (fp_valid_at): Likewise.
9764 (choose_baseaddr): Modify comments.
9765 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9766 ix86_expand_prologue.
9767 (ix86_expand_prologue): Modify stack realignment and allocation.
9768 (ix86_expand_epilogue): Modify comments.
9769 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9770 avx2_runtime, avx512f, and avx512f_runtime.
9772 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9774 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9775 (mstackrealign): Do not init to -1.
9776 * config/i386/i386.c (ix86_option_override_internal):
9777 Check opts_set, not opts when setting default value of
9778 opts->x_ix86_force_align_arg_pointer.
9780 2017-08-23 Richard Biener <rguenther@suse.de>
9782 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9783 lang_hooks.decl_printable_name.
9784 * print-rtl-function.c (print_rtx_function): Likewise.
9785 * tree-pretty-print.c (dump_function_header): Likewise.
9787 2017-08-23 Richard Biener <rguenther@suse.de>
9790 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9791 -g0 at compile-time.
9793 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9796 * doc/sourcebuild.texi (Other hardware attributes):
9799 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9802 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9803 Fix single-use check.
9805 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9807 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9809 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9811 * doc/install.texi: Modify to add more details on running selected
9814 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9816 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9817 is combined with -mabi=ms.
9818 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9821 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9823 PR tree-optimization/81488
9824 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9825 and cached_basis fields.
9826 (MAX_SPREAD): New constant.
9827 (alloc_cand_and_find_basis): Initialize new fields.
9828 (clear_visited): New function.
9829 (create_phi_basis_1): Rename from create_phi_basis, set visited
9830 and cached_basis fields.
9831 (create_phi_basis): New wrapper function.
9832 (phi_add_costs_1): Rename from phi_add_costs, add spread
9833 parameter, set visited field, short-circuit when limits reached.
9834 (phi_add_costs): New wrapper function.
9835 (record_phi_increments_1): Rename from record_phi_increments, set
9837 (record_phi_increments): New wrapper function.
9838 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9839 (phi_incr_cost): New wrapper function.
9840 (all_phi_incrs_profitable_1): Rename from
9841 all_phi_incrs_profitable, set visited field.
9842 (all_phi_incrs_profitable): New wrapper function.
9844 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9845 Alan Hayward <alan.hayward@arm.com>
9846 David Sherwood <david.sherwood@arm.com>
9848 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9849 that takes the outer and inner modes.
9850 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9851 comparison as the canonical test for a paradoxical subreg.
9852 * combine.c (simplify_set): Use paradoxical_subreg_p.
9853 (make_extraction): Likewise.
9854 (force_to_mode): Likewise.
9855 (rtx_equal_for_field_assignment_p): Likewise.
9856 (gen_lowpart_for_combine): Likewise.
9857 (simplify_comparison): Likewise.
9858 * cse.c (equiv_constant): Likewise.
9859 * expmed.c (store_bit_field_1): Likewise.
9860 * final.c (alter_subreg): Likewise.
9861 * fwprop.c (propagate_rtx): Likewise.
9862 (forward_propagate_subreg): Likewise.
9863 * ira-conflicts.c (ira_build_conflicts): Likewise.
9864 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9865 * lra-constraints.c (curr_insn_transform): Likewise.
9866 (split_reg): Likewise.
9867 * lra-eliminations.c (move_plus_up): Likewise.
9868 (lra_eliminate_regs_1): Likewise.
9869 * recog.c (general_operand): Likewise.
9870 * ree.c (combine_reaching_defs): Likewise.
9871 * reload.c (push_reload): Likewise.
9872 (find_reloads): Likewise.
9873 * reload1.c (elimination_effects): Likewise.
9874 (compute_reload_subreg_offset): Likewise.
9875 (choose_reload_regs): Likewise.
9876 * rtlanal.c (subreg_lsb_1): Likewise.
9877 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9878 (simplify_subreg): Likewise.
9879 * var-tracking.c (track_loc_p): Likewise.
9880 * emit-rtl.c (byte_lowpart_offset): Likewise.
9881 (paradoxical_subreg_p): Delete out-of-line definition.
9883 2017-08-22 Jeff Law <law@redhat.com>
9885 PR tree-optimization/81741
9886 PR tree-optimization/71947
9887 * tree-ssa-dom.c: Include tree-inline.h.
9888 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9889 equivalences if one is more expensive to compute than the other.
9890 * tree-ssa-scopedtables.h (class const_or_copies): Make
9891 record_const_or_copy_raw method private.
9892 (class avail_exprs_stack): New method simplify_binary_operation.
9893 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9894 avail_exprs_stack::simplify_binary_operation as needed.
9895 (avail_exprs_stack::simplify_binary_operation): New function.
9897 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9899 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9900 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9901 (DOT_SYMBOLS): Likewise.
9902 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9903 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9904 (RS6000_ABI_NAME): Likewise.
9905 (TARGET_CMODEL): Likewise.
9906 (TOC_SECTION_ASM_OP): Likewise.
9907 (SET_CMODEL): New macro.
9908 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9910 2017-08-22 Richard Biener <rguenther@suse.de>
9912 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9913 to allow for free-lang-data replacements similar to verify_type_variant.
9915 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9916 Alan Hayward <alan.hayward@arm.com>
9917 David Sherwood <david.sherwood@arm.com>
9919 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9920 VOIDmode for the LABEL_REF.
9922 2017-08-22 Richard Biener <rguenther@suse.de>
9924 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9926 2017-08-22 Richard Biener <rguenther@suse.de>
9928 * common.opt (feliminate-dwarf2-dups): Ignore.
9929 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9930 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9931 same_die_p_wrap, compute_section_prefix,
9932 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9933 (comdat_symbol_id, comdat_symbol_number): Likewise.
9934 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9936 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9937 (output_die): Mark unreachable path unreachable.
9938 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9939 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9940 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9941 (dwarf2out_early_finish): Likewise.
9943 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9945 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9947 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9950 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9951 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9952 (avr_attribute_table) <io, io_low, address>: Initialize
9953 .decl_required with true.
9955 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9957 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9958 undocumented debugging options.
9959 (-mvsx-scalar-double): Likewise.
9960 (-mallow-df-permute): Likewise.
9961 (-mvectorize-builtins): Likewise.
9962 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9963 (rs6000_builtin_vectorized_function): Likewise.
9964 (rs6000_builtin_md_vectorized_function): Likewise.
9965 (rs6000_opt_vars): Likewise.
9967 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9970 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9971 (*btrq_imm): Rename from *btrq.
9972 (*btcq_imm): Rename from *btcq.
9973 (btsc): New code attribute.
9974 (*<btsc><mode>): New insn pattern.
9975 (*btr<mode>): Ditto.
9976 (*<btsc><mode>_mask): New insn_and_split pattern.
9977 (*btr<mode>_mask): Ditto.
9979 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9980 Alan Hayward <alan.hayward@arm.com>
9981 David Sherwood <david.sherwood@arm.com>
9983 * function.c (pad_below): Simplify padding calculation.
9985 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9986 Alan Hayward <alan.hayward@arm.com>
9987 David Sherwood <david.sherwood@arm.com>
9989 * target.def (function_prologue): Remove frame size argument.
9990 (function_epilogue): Likewise.
9991 * doc/tm.texi: Regenerate.
9992 * final.c (final_start_function): Update call to function_prologue.
9993 (final_end_function): Update call to function_epilogue.
9994 (default_function_pro_epilogue): Remove frame size argument.
9995 * output.h (default_function_pro_epilogue): Likewise.
9996 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9997 (arm_output_function_prologue): Likewise.
9998 * config/frv/frv.c (frv_function_prologue): Likewise.
9999 (frv_function_epilogue): Likewise.
10000 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10001 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10002 (ia64_output_function_epilogue): Likewise.
10003 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10004 (m32r_output_function_epilogue): Likewise.
10005 * config/microblaze/microblaze.c (microblaze_function_prologue)
10006 (microblaze_function_epilogue): Likewise.
10007 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10008 (mips_output_function_epilogue): Likewise.
10009 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10010 (mmix_target_asm_function_epilogue): Likewise.
10011 * config/msp430/msp430.c (msp430_start_function): Likewise.
10012 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10013 (nds32_asm_function_epilogue): Likewise.
10014 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10015 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10016 (pa_output_function_epilogue): Likewise.
10017 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10018 (rs6000_output_function_epilogue): Likewise.
10019 * config/rl78/rl78.c (rl78_start_function): Likewise.
10020 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10021 (rs6000_output_function_epilogue): Likewise.
10022 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10023 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10024 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10025 (sparc_asm_function_epilogue): Likewise.
10027 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10029 * tree.h (type_has_mode_precision_p): New function.
10030 * convert.c (convert_to_integer_1): Use it.
10031 * expr.c (expand_expr_real_2): Likewise.
10032 (expand_expr_real_1): Likewise.
10033 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10034 * match.pd: Likewise.
10035 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10036 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10037 * tree-tailcall.c (process_assignment): Likewise.
10038 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10039 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10040 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10041 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10042 (vectorizable_assignment): Likewise.
10043 (vectorizable_shift): Likewise.
10044 (vectorizable_operation): Likewise.
10045 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10047 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10049 * match.pd: Add pow (C, x) simplification.
10051 2017-08-21 Richard Biener <rguenther@suse.de>
10053 PR tree-optimization/81900
10054 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10055 for blocks with abnormal predecessors.
10056 (compute_antic): Do not set visited flag prematurely.
10058 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10061 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10063 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10065 * stor-layout.h (vector_type_mode): Move to...
10066 * tree.h (vector_type_mode): ...here.
10067 * stor-layout.c (vector_type_mode): Move to...
10068 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10070 2017-08-21 Richard Biener <rguenther@suse.de>
10072 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10073 register_external_die hooks.
10074 (debug_false_tree_charstarstar_uhwistar): Declare.
10075 (debug_nothing_tree_charstar_uhwi): Likewise.
10076 * debug.c (do_nothing_debug_hooks): Adjust.
10077 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10078 (debug_nothing_tree_charstar_uhwi): Likewise.
10079 * dbxout.c (dbx_debug_hooks): Adjust.
10080 (xcoff_debug_hooks): Likewise.
10081 * sdbout.c (sdb_debug_hooks): Likewise.
10082 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10083 * dwarf2out.c (macinfo_label_base): New global.
10084 (dwarf2out_register_external_die): New function for the
10085 register_external_die hook.
10086 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10087 (dwarf2_debug_hooks): Use them.
10088 (dwarf2_lineno_debug_hooks): Adjust.
10089 (struct die_struct): Add with_offset flag.
10090 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10091 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10092 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10093 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10094 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10095 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10096 defining section names for the early LTO debug variants.
10097 (reset_indirect_string): New helper.
10098 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10099 (print_dw_val): Add support for offsetted symbol references.
10100 (get_ultimate_context): Split out from is_cxx.
10101 (is_cxx): Use get_ultimate_context.
10102 (is_fortran): Add decl overload.
10103 (compute_comp_unit_symbol): Split out worker from
10104 compute_section_prefix.
10105 (compute_section_prefix): Call compute_comp_unit_symbol and
10106 set comdat_type_p here.
10107 (output_die): Skip DIE symbol output for the LTO added one.
10108 Handle DIE symbol references with offset.
10109 (output_comp_unit): Guard section name mangling properly.
10110 For LTO debug sections emit a symbol at the section beginning
10111 which we use to refer to its DIEs.
10112 (add_abstract_origin_attribute): For DIEs registered via
10113 dwarf2out_register_external_die directly refer to the early
10114 DIE rather than indirectly through the shadow one we created.
10115 Remove obsolete call to dwarf2out_abstract_function for
10116 non-function/block origins.
10117 (gen_array_type_die): When generating early LTO debug do
10118 not emit DW_AT_string_length.
10119 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10120 late when in LTO. As suggested place a gcc_unreachable for
10121 the DECL_ABSTRACT_P case.
10122 (gen_subprogram_die): Avoid another specification DIE
10123 for early built declarations/definitions for the late LTO case.
10124 (gen_variable_die): Add type references for late duplicated VLA dies
10126 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10127 we have the abstract instance already.
10128 (process_scope_var): Adjust decl DIE contexts in LTO which
10129 first puts them in limbo.
10130 (gen_decl_die): Do not generate type DIEs late apart from
10131 types for VLAs or for decls we do not yet have a DIE. Do not
10132 call dwarf2out_abstract_function late.
10133 (dwarf2out_early_global_decl): Make sure to create DIEs
10134 for abstract instances of a decl first.
10135 (dwarf2out_late_global_decl): Adjust comment.
10136 (output_macinfo_op): With multiple macro sections use
10137 macinfo_label_base to distinguish labels.
10138 (output_macinfo): Likewise. Update macinfo_label_base.
10139 Pass in the line info label.
10140 (note_variable_value_in_expr): When generating LTO resolve
10141 all variable values here by generating DIEs as needed.
10142 (init_sections_and_labels): Add early LTO debug flag parameter
10143 and generate different sections and names if set. Add generation
10144 counter for the labels so we can have multiple of them.
10145 (reset_dies): Helper to allow DIEs to be output multiple times.
10146 (dwarf2out_finish): When outputting DIEs to the fat part of an
10147 LTO object first reset DIEs.
10148 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10149 (modified_type_die): Check for decl_ultimate_origin being self
10151 (gen_type_die_with_usage): Likewise.
10152 (gen_typedef_die): Allow decl_ultimate_origin being self.
10153 (set_decl_abstract_flags): Remove.
10154 (set_block_abstract_flags): Likewise.
10155 (dwarf2out_abstract_function): Treat the early generated DIEs
10156 as the abstract copy and only add DW_AT_inline and
10157 DW_AT_artificial here and call set_decl_origin_self.
10158 If the DIE has an abstract origin don't do anything.
10159 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10160 if we have none yet (Go fails to build one, PR78628).
10161 (variably_modified_type_p): Prevent endless recursion for Ada
10162 cyclic pointer types.
10163 * lto-streamer-in.c: Include debug.h.
10164 (dref_queue): New global.
10165 (lto_read_tree_1): Stream in DIE references.
10166 (lto_input_tree): Register DIE references.
10167 (input_function): Stream DECL_DEBUG_ARGS.
10168 * lto-streamer-out.c: Include debug.h.
10169 (lto_write_tree_1): Output DIE references.
10170 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10171 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10172 (output_function): Stream DECL_DEBUG_ARGS.
10173 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10174 Stream DECL_ABSTRACT_ORIGIN.
10175 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10176 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10177 DECL_CONTEXT for file-scope decls.
10178 * lto-streamer.h (struct dref_entry): Declare.
10179 (dref_queue): Likewise.
10180 * cfgexpand.c (pass_expand::execute): Do not call the
10181 outlining_inline_function hook here.
10182 * lto-wrapper.c (debug_obj): New global.
10183 (tool_cleanup): Unlink it if required.
10184 (debug_objcopy): New function.
10185 (run_gcc): Handle early debug sections in the IL files by
10186 extracting them to separate files, partially linkin them and
10187 feeding the result back as result to the linker.
10188 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10189 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10190 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10191 sections into a separate segment.
10192 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10194 (darwin_asm_dwarf_section): Likewise.
10195 (darwin_asm_output_dwarf_offset): Likewise.
10196 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10198 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10199 Alan Hayward <alan.hayward@arm.com>
10200 David Sherwood <david.sherwood@arm.com>
10202 * read-md.h (md_reader::record_potential_iterator_use): Replace
10203 pointer argument with an rtx and an index.
10204 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10205 (apply_mode_iterator): Likewise.
10206 (apply_code_iterator): Likewise.
10207 (apply_int_iterator): Likewise.
10208 (apply_subst_iterator): Likewise.
10209 (record_iterator_use): Likewise.
10210 (record_attribute_use): Likewise.
10211 (md_reader::record_potential_iterator_use): Likewise. Update calls
10212 to record_iterator_use and apply_iterator.
10213 (iterator_use): Replace ptr with x and index.
10214 (attribute_use): Likewise.
10215 (apply_attribute_uses): Update calls to apply_iterator.
10216 (apply_iterators): Likewise. Update initialization of iterator_use.
10217 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10218 and record_potential_iterator_use.
10219 (rtx_reader::read_rtx_operand): Likewise.
10221 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10222 Alan Hayward <alan.hayward@arm.com>
10223 David Sherwood <david.sherwood@arm.com>
10225 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10228 2017-08-21 Richard Biener <rguenther@suse.de>
10230 PR middle-end/81884
10231 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10232 at struct end conservatively when comparing common bases.
10234 2017-08-21 Richard Biener <rguenther@suse.de>
10236 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10237 (mem_ref_in_stmt): Remove.
10238 (determine_max_movement): Use ref index to get at the reference.
10239 (invariantness_dom_walker::before_dom_children): Deal with
10240 lim data already initialized.
10241 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10243 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10245 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10246 (TARGET_ISA_ROUND): Ditto.
10247 (TARGET_ROUND): Ditto.
10248 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10249 * config/i386/i386.md: Ditto.
10250 * config/i386/sse.md: Ditto.
10251 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10252 with OPTION_MASK_ISA_SSE4_1.
10254 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10257 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10258 __builtin_ia32_lzcnt_u16.
10260 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10263 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10264 (rs6000_set_current_function): Rewrite function to use it.
10266 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10269 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10270 and TYPE_WARN_IF_NOT_ALIGN.
10271 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10272 (handle_warn_if_not_align): New.
10273 (place_union_field): Call handle_warn_if_not_align.
10274 (place_field): Call handle_warn_if_not_align.
10275 Copy TYPE_WARN_IF_NOT_ALIGN.
10276 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10277 (layout_type): Likewise.
10278 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10280 (tree_decl_common): Add warn_if_not_align.
10281 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10282 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10283 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10284 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10285 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10286 * doc/extend.texi: Document warn_if_not_aligned attribute.
10287 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10289 2017-08-17 Martin Liska <mliska@suse.cz>
10292 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10293 (get_data_dependence): Use it as pointer type.
10294 (distribute_loop): Likewise.
10296 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10298 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10299 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10300 (p8_vmrgow_<mode>_direct): New define_insn.
10301 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10302 handle endianness for vmrgew and vmrgow permute patterns.
10304 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10306 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10307 * config/rs6000/rs6000-cpus.def: Remove comment.
10308 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10309 (POWERPC_MASKS): Likewise.
10310 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10311 use of TARGET_VSX_TIMODE.
10312 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10313 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10314 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10315 (rs6000_option_override_internal): Remove dead code.
10316 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10317 (rs6000_legitimize_reload_address): Likewise.
10318 (rs6000_legitimate_address_p): Likewise.
10319 (rs6000_opt_masks): Delete "vsx-timode".
10320 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10321 from function comment.
10322 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10323 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10324 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10326 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10327 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10328 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10329 (VSX_TI): Likewise.
10331 (define_peephole2): Likewise.
10333 2017-08-17 Martin Sebor <msebor@redhat.com>
10336 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10337 past the end of an array.
10338 (test_pp_format): Add test cases.
10340 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10342 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10343 missing ECF_NOTHROW flags.
10345 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10348 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10349 operands residing in integer registers.
10350 (*vsx_le_perm_load_<mode>): Likewise.
10351 (*vsx_le_perm_store_<mode>): Likewise.
10352 (define_peephole2): Add peepholes to optimize the above.
10354 2017-08-17 Marek Polacek <polacek@redhat.com>
10356 PR middle-end/81814
10357 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10358 to mimic what shorten_compare did. Change the return type to bool.
10359 (fold_cond_expr_with_comparison): Update call to
10360 operand_equal_for_comparison_p.
10361 (fold_ternary_loc): Likewise.
10363 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10365 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10367 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10369 2017-08-17 Richard Biener <rguenther@suse.de>
10371 * tree-ssa-structalias.c (solve_graph): When propagating
10372 to successors update the graphs succ edges and avoid duplicate work.
10374 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10377 * config/i386/i386.c (ix86_option_override_internal): Save target
10378 specific options after ix86_stack_protector_guard_reg was changed.
10380 2017-08-17 Richard Biener <rguenther@suse.de>
10382 PR tree-optimization/81827
10383 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10384 (new_var_info): Initialize it conservatively.
10385 (get_call_vi): Mark register vars.
10386 (new_scalar_tmp_constraint_exp): Likewise.
10387 (handle_rhs_call): Likewise.
10388 (handle_const_call): Likewise.
10389 (create_function_info_for): Likewise.
10390 (solve_constraints): Sort varinfos to separate register from
10391 non-register vars to pack points-to solution bitmaps during
10394 2017-08-17 Marek Polacek <polacek@redhat.com>
10396 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10398 2017-08-17 Richard Biener <rguenther@suse.de>
10400 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10401 to true when overflow is undefined and we saturated the result.
10403 2017-08-17 Alan Modra <amodra@gmail.com>
10406 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10407 Don't use store multiple if only one reg needs saving.
10408 (interesting_frame_related_regno): New function.
10409 (rs6000_frame_related): Don't emit frame info for regs that
10411 (rs6000_emit_epilogue): Likewise.
10413 2017-08-16 Nathan Sidwell <nathan@acm.org>
10415 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10416 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10417 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10418 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10419 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10420 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10421 (verify_type): Adjust for TYPE_BINFO move.
10422 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10423 process TYPE_BINFO directly.
10424 (hash_tree): Likewise.
10425 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10427 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10430 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10432 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10434 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10437 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10438 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10439 Add ix86_binary_operator_ok to insn constraint.
10440 (*iordi_1_bts): Ditto.
10441 (*xordi_1_btc): Ditto.
10442 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10443 Update corresponding peephole2 pattern.
10447 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10449 PR tree-optimization/81832
10450 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10451 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10453 2017-08-16 Marek Polacek <polacek@redhat.com>
10456 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10457 regarding size_zero_node.
10459 2017-08-16 Martin Liska <mliska@suse.cz>
10462 * config.gcc: Respect previously set extra_objs in case
10465 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10467 PR tree-optimization/81835
10468 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10469 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10470 not depend on the phi.
10472 2017-08-16 Alan Modra <amodra@gmail.com>
10474 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10477 2017-08-16 Alan Modra <amodra@gmail.com>
10479 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10480 (save_reg_p): ..into this. Update all callers.
10481 (first_reg_to_save): Simplify.
10483 2017-08-16 Alan Modra <amodra@gmail.com>
10485 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10488 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10492 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10493 constant count if that count is less than 32.
10495 2017-08-15 Nathan Sidwell <nathan@acm.org>
10497 * gcc.c (execute): Emit friendlier message if inferior is killed
10498 by an external cause.
10500 2017-08-15 Richard Biener <rguenther@suse.de>
10502 PR tree-optimization/81790
10503 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10504 CONSTRUCTORs from simplifying and VN.
10506 2017-08-14 Martin Sebor <msebor@redhat.com>
10508 * builtin-attrs.def: Add comments.
10510 2017-08-14 Martin Sebor <msebor@redhat.com>
10513 * doc/extend.texi (attribute nonstring): Document new attribute.
10515 2017-08-14 Martin Sebor <msebor@redhat.com>
10518 * tree-diagnostic.c (default_tree_printer): Handle %G.
10519 * gimple-pretty-print.h (percent_G_format): Declare new function.
10520 * gimple-pretty-print.c (percent_G_format): Define.
10521 * tree-pretty-print.c (percent_K_format): Add argument.
10523 2017-08-14 Martin Sebor <msebor@redhat.com>
10525 PR translation/79998
10526 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10527 Remove a stray space.
10529 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10532 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10533 (*iordi_1_bts): Ditto.
10534 (*xordi_1_btc): Ditto.
10536 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10539 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10542 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10543 quoted strings, and make more translator-friendly.
10544 (darwin_rs6000_override_options): Likewise.
10545 (rs6000_option_override_internal): Likewise.
10546 (rs6000_return_in_memory): Fix overlong line.
10547 (init_cmulative_args): Use quoted strings, and make more
10548 translator-friendly.
10549 (rs6000_pass_by_reference): Fix overlong line.
10550 (def_builtin): Use quoted strings.
10551 (altivec_expand_predicate_builtin): Use quoted strings, and make
10552 more translator-friendly.
10553 (htm_expand_builtin): Use quoted strings.
10554 (cpu_expand_builtin): Use quoted strings, and make more
10555 translator-friendly.
10556 (altivec_expand_builtin): Likewise.
10557 (paired_expand_predicate_builtin): Likewise.
10558 (rs6000_invalid_builtin): Likewise.
10559 (builtin_function_type): Use quoted strings.
10560 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10561 more translator-friendly.
10562 (rs6000_trampoline_init): Likewise.
10563 (rs6000_handle_altivec_attribute): Likewise.
10564 (rs6000_inner_target_options): Use quoted strings.
10565 (rs6000_disable_incompatible_switches): Likewise.
10566 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10567 strings, and make more translator-friendly.
10568 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10570 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10572 PR tree-optimization/81799
10573 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10574 cond_expr to simple gimple operand.
10576 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10578 PR middle-end/46932
10579 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10581 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10585 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10586 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10587 TARGET_GASISR_PROLOGUES.
10588 * config/avr/avr.c (avr_option_override): Same.
10589 (avr_pass_pre_proep::execute): Same.
10591 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10594 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10595 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10597 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10599 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10600 define_insn_and_split. Split to a memory load from 0 in
10601 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10602 using PTR mode iterator.
10603 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10604 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10605 (*add_tp_<mode>): Redefine as define_insn_and_split.
10606 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10607 address space. Merge with *add_tp_x32 using PTR mode iterator.
10608 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10609 Split to an add with a memory load from 0 in
10610 DEFAULT_TLS_SEG_REG address space.
10612 2017-08-12 Andrew Pinski <apinski@cavium.com>
10614 * config/aarch64/aarch64-option-extensions.def (rdma):
10615 Fix feature string to what Linux prints out in /proc/cpuinfo.
10617 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10620 * dwarf2out.c (modified_type_die): For C typedef types that have
10621 an ultimate origin, process the ultimate origin instead of the
10623 (gen_typedef_die): Assert that input DECLs have no ultimate
10625 (gen_type_die_with_usage): For typedef variants that have an
10626 ultimate origin, just call gen_decl_die on the original DECL.
10627 (process_scope_var): Avoid creating DIEs for local typedefs and
10628 concrete static variables.
10630 2017-08-12 Alan Modra <amodra@gmail.com>
10634 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10635 match gnu-user.h startfile.
10636 (ENDFILE_LINUX_SPEC): Similarly.
10638 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10641 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10643 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10645 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10646 * config/aarch64/aarch64.md (mov<mode>): Change.
10647 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10648 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10649 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10651 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10653 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10654 for storage order barriers.
10656 2017-08-11 Martin Liska <mliska@suse.cz>
10659 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10660 variables of void type.
10662 2017-08-11 Martin Liska <mliska@suse.cz>
10664 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10665 TARGET_SUPPORTS_ALIASES.
10666 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10667 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10668 (optimize_weakref): Likewise.
10669 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10670 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10671 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10673 2017-08-11 Martin Liska <mliska@suse.cz>
10676 * config/i386/i386.c (make_resolver_func): Do complete
10677 refactoring of the function.
10679 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10682 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10683 * config/i386/i386.c (ix86_stack_protect_guard): Use
10684 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10685 * doc/invoke.texi (x86 Options): Document
10686 -mstack-protector-guard-symbol= option.
10688 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10690 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10691 * config/i386/i386.c (ix86_split_stack_guard): New function.
10692 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10693 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10694 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10695 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10696 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10697 (split_stack_space_check): Call ix86_split_stack_guard.
10699 2017-08-10 Martin Sebor <msebor@redhat.com>
10701 * print-tree.c (print_node): Print location using the established
10703 Replace spaces with colons.
10704 (debug_raw, debug): Ditto.
10706 2017-08-10 Martin Sebor <msebor@redhat.com>
10709 * pretty-print.c (pp_format): Correct the handling of %s precision.
10711 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10714 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10716 (ix86_finalize_stack_frame_flags): This. Also clear
10717 frame_pointer_needed if -fno-omit-frame-pointer is used without
10719 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10720 with ix86_finalize_stack_frame_flags.
10721 (ix86_expand_epilogue): Likewise.
10722 (ix86_expand_split_stack_prologue): Likewise.
10723 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10725 2017-08-10 Martin Liska <mliska@suse.cz>
10728 * c-attribs.c (handle_target_attribute):
10729 Report warning for an empty string argument of target attribute.
10731 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10734 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10736 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10737 or DECL_NONLOCAL labels.
10738 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10739 or DECL_NONLOCAL labels here.
10741 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10744 to indicate when early gimple folding has been disabled.
10745 (rs6000_gimple_fold_builtin): Add debug content.
10746 (rs6000_invalid_builtin): Fix whitespace.
10747 (rs6000_expand_builtin): Fix whitespace.
10748 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10750 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10753 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10754 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10756 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10758 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10759 (qdf24xx): Likewise.
10760 * config/aarch64/aarch64.md: Include falkor.md.
10761 * config/aarch64/falkor.md: New.
10763 2017-08-09 Marek Polacek <polacek@redhat.com>
10766 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10767 * diagnostic.c (emit_diagnostic): Add a comment.
10768 (emit_diagnostic_valist): New function.
10770 2017-08-09 Marek Polacek <polacek@redhat.com>
10773 * input.c (make_location): New overload.
10774 * input.h (make_location): Declare.
10776 2017-08-08 Alan Modra <amodra@gmail.com>
10777 H.J. Lu <hongjiu.lu@intel.com>
10780 * gcc.c (NO_PIE_SPEC): Delete.
10781 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10783 (LINK_PIE_SPEC): ..to here.
10784 (LINK_COMMAND_SPEC): Support -no-pie.
10785 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10786 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10788 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10789 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10790 (ENDFILE_CRTEND_SPEC): Similarly.
10792 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10795 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10796 (mstack-protector-guard-offset=): Ditto.
10797 * config/i386/i386.c (ix86_option_override): Handle
10798 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10800 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10801 ix86_stack_protect_guard_offset variables.
10802 (TARGET_STACK_PROTECT_GUARD): Always define.
10803 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10804 and -mstack-protector-guard-offset= options.
10806 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10808 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10809 boundary case for the last candidate.
10811 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10813 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10814 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10816 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10818 PR middle-end/19706
10819 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10820 * config/aarch64/aarch64-builtins.c
10821 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10822 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10823 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10825 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10826 Andrew Pinski <pinskia@gmail.com>
10828 PR middle-end/19706
10829 * internal-fn.def (XORSIGN): New.
10830 * optabs.def (xorsign_optab): New.
10831 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10832 (convert_expand_mult_copysign): New.
10833 (pass_optimize_widening_mul::execute): Call
10834 convert_expand_mult_copysign.
10836 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10838 PR tree-optimization/81354
10839 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10840 Insert on edges rather than explicitly creating landing pads.
10841 (analyze_candidates_and_replace): Commit edge inserts.
10843 2017-08-08 Richard Biener <rguenther@suse.de>
10845 PR middle-end/81719
10846 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10847 (expand_simple_operations): Also look through ADDR_EXPRs with
10848 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10850 2017-08-08 Richard Biener <rguenther@suse.de>
10852 PR tree-optimization/81723
10853 * tree-vect-slp.c (struct bst_traits): New hash traits.
10854 (bst_fail): New global.
10855 (vect_build_slp_tree_2): New worker, split out from ...
10856 (vect_build_slp_tree): ... this now wrapping it with using
10857 bst_fail set to cache SLP tree build fails. Properly handle
10859 (vect_analyze_slp_instance): Allocate and free bst_fail.
10861 2017-08-08 Martin Liska <mliska@suse.cz>
10864 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10865 LABEL_DECLs that can be from a different function.
10867 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10869 PR tree-optimization/81744
10870 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10871 loop's number of iterations.
10873 2017-08-08 Martin Liska <mliska@suse.cz>
10875 * asan.c: Include header files.
10876 * attribs.c (build_decl_attribute_variant): New function moved
10878 (build_type_attribute_qual_variant): Likewise.
10879 (cmp_attrib_identifiers): Likewise.
10880 (simple_cst_list_equal): Likewise.
10881 (omp_declare_simd_clauses_equal): Likewise.
10882 (attribute_value_equal): Likewise.
10883 (comp_type_attributes): Likewise.
10884 (build_type_attribute_variant): Likewise.
10885 (lookup_ident_attribute): Likewise.
10886 (remove_attribute): Likewise.
10887 (merge_attributes): Likewise.
10888 (merge_type_attributes): Likewise.
10889 (merge_decl_attributes): Likewise.
10890 (merge_dllimport_decl_attributes): Likewise.
10891 (handle_dll_attribute): Likewise.
10892 (attribute_list_equal): Likewise.
10893 (attribute_list_contained): Likewise.
10894 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10895 (lookup_attribute_by_prefix): Likewise.
10896 * bb-reorder.c: Include header files.
10897 * builtins.c: Likewise.
10898 * calls.c: Likewise.
10899 * cfgexpand.c: Likewise.
10900 * cgraph.c: Likewise.
10901 * cgraphunit.c: Likewise.
10902 * convert.c: Likewise.
10903 * dwarf2out.c: Likewise.
10904 * final.c: Likewise.
10905 * fold-const.c: Likewise.
10906 * function.c: Likewise.
10907 * gimple-expr.c: Likewise.
10908 * gimple-fold.c: Likewise.
10909 * gimple-pretty-print.c: Likewise.
10910 * gimple.c: Likewise.
10911 * gimplify.c: Likewise.
10912 * hsa-common.c: Likewise.
10913 * hsa-gen.c: Likewise.
10914 * internal-fn.c: Likewise.
10915 * ipa-chkp.c: Likewise.
10916 * ipa-cp.c: Likewise.
10917 * ipa-devirt.c: Likewise.
10918 * ipa-fnsummary.c: Likewise.
10919 * ipa-inline.c: Likewise.
10920 * ipa-visibility.c: Likewise.
10922 * lto-cgraph.c: Likewise.
10923 * omp-expand.c: Likewise.
10924 * omp-general.c: Likewise.
10925 * omp-low.c: Likewise.
10926 * omp-offload.c: Likewise.
10927 * omp-simd-clone.c: Likewise.
10928 * opts-global.c: Likewise.
10929 * passes.c: Likewise.
10930 * predict.c: Likewise.
10931 * sancov.c: Likewise.
10932 * sanopt.c: Likewise.
10933 * symtab.c: Likewise.
10934 * toplev.c: Likewise.
10935 * trans-mem.c: Likewise.
10936 * tree-chkp.c: Likewise.
10937 * tree-eh.c: Likewise.
10938 * tree-into-ssa.c: Likewise.
10939 * tree-object-size.c: Likewise.
10940 * tree-parloops.c: Likewise.
10941 * tree-profile.c: Likewise.
10942 * tree-ssa-ccp.c: Likewise.
10943 * tree-ssa-live.c: Likewise.
10944 * tree-ssa-loop.c: Likewise.
10945 * tree-ssa-sccvn.c: Likewise.
10946 * tree-ssa-structalias.c: Likewise.
10947 * tree-ssa.c: Likewise.
10948 * tree-streamer-in.c: Likewise.
10949 * tree-vectorizer.c: Likewise.
10950 * tree-vrp.c: Likewise.
10951 * tsan.c: Likewise.
10952 * ubsan.c: Likewise.
10953 * varasm.c: Likewise.
10954 * varpool.c: Likewise.
10955 * tree.c: Remove functions moved to attribs.[ch].
10956 * tree.h: Likewise.
10957 * config/aarch64/aarch64.c: Add attrs.h header file.
10958 * config/alpha/alpha.c: Likewise.
10959 * config/arc/arc.c: Likewise.
10960 * config/arm/arm.c: Likewise.
10961 * config/avr/avr.c: Likewise.
10962 * config/bfin/bfin.c: Likewise.
10963 * config/c6x/c6x.c: Likewise.
10964 * config/cr16/cr16.c: Likewise.
10965 * config/cris/cris.c: Likewise.
10966 * config/darwin.c: Likewise.
10967 * config/epiphany/epiphany.c: Likewise.
10968 * config/fr30/fr30.c: Likewise.
10969 * config/frv/frv.c: Likewise.
10970 * config/ft32/ft32.c: Likewise.
10971 * config/h8300/h8300.c: Likewise.
10972 * config/i386/winnt.c: Likewise.
10973 * config/ia64/ia64.c: Likewise.
10974 * config/iq2000/iq2000.c: Likewise.
10975 * config/lm32/lm32.c: Likewise.
10976 * config/m32c/m32c.c: Likewise.
10977 * config/m32r/m32r.c: Likewise.
10978 * config/m68k/m68k.c: Likewise.
10979 * config/mcore/mcore.c: Likewise.
10980 * config/microblaze/microblaze.c: Likewise.
10981 * config/mips/mips.c: Likewise.
10982 * config/mmix/mmix.c: Likewise.
10983 * config/mn10300/mn10300.c: Likewise.
10984 * config/moxie/moxie.c: Likewise.
10985 * config/msp430/msp430.c: Likewise.
10986 * config/nds32/nds32-isr.c: Likewise.
10987 * config/nds32/nds32.c: Likewise.
10988 * config/nios2/nios2.c: Likewise.
10989 * config/nvptx/nvptx.c: Likewise.
10990 * config/pa/pa.c: Likewise.
10991 * config/pdp11/pdp11.c: Likewise.
10992 * config/powerpcspe/powerpcspe.c: Likewise.
10993 * config/riscv/riscv.c: Likewise.
10994 * config/rl78/rl78.c: Likewise.
10995 * config/rx/rx.c: Likewise.
10996 * config/s390/s390.c: Likewise.
10997 * config/sh/sh.c: Likewise.
10998 * config/sol2.c: Likewise.
10999 * config/sparc/sparc.c: Likewise.
11000 * config/spu/spu.c: Likewise.
11001 * config/stormy16/stormy16.c: Likewise.
11002 * config/tilegx/tilegx.c: Likewise.
11003 * config/tilepro/tilepro.c: Likewise.
11004 * config/v850/v850.c: Likewise.
11005 * config/vax/vax.c: Likewise.
11006 * config/visium/visium.c: Likewise.
11007 * config/xtensa/xtensa.c: Likewise.
11009 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11012 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11013 constraints since the -mupper-regs-* switches have been
11015 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11016 into a vector from a double word element that was extracted from
11017 another vector, and eliminate extra XXPERMDI instructions.
11018 (vsx_concat_<mode>_2): Likewise.
11019 (vsx_concat_<mode>_3): Likewise.
11020 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11021 concat to allow optimizing inserts from previous extracts.
11023 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11025 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11026 memory reference to a SSP offset in TLS address space.
11027 (ix86_print_operand) <case '@'>: Remove.
11028 (ix86_print_operand_punct_valid_p): Remove '@' code.
11029 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11030 UNSPEC_SP_TLS_TEST.
11031 (stack_tls_protect_set_<mode>): Remove.
11032 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11033 (stack_tls_protect_test_<mode>): Remove.
11034 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11036 2017-08-07 Olivier Hainque <hainque@adacore.com>
11039 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11041 2017-08-07 Douglas Rupp <rupp@adacore.com>
11043 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11044 variable was referenced as multidir in command.
11046 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11049 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11052 2017-08-07 Martin Liska <mliska@suse.cz>
11054 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11055 * config/rl78/rl78.c: Add include of attribs.h.
11056 * config/sh/sh.c: Likewise.
11057 * config/v850/v850.c: Likewise.
11059 2017-08-07 Tom de Vries <tom@codesourcery.com>
11061 PR middle-end/78266
11062 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11064 2017-08-07 Martin Liska <mliska@suse.cz>
11066 * config/mips/mips.c: Include attribs.h.
11068 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11071 * doc/invoke.texi: Document change in behvaior for -Ofast for
11074 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11076 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11078 (aarch64_pop_regs): Likewise.
11079 (aarch64_gen_load_pair): Likewise.
11080 (aarch64_save_callee_saves): Likewise.
11081 (aarch64_restore_callee_saves): Likewise.
11083 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11085 * config/i386/i386.c: Revert the last change.
11087 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11090 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11092 (ix86_finalize_stack_frame_flags): This. Also clear
11093 frame_pointer_needed if -fno-omit-frame-pointer is used without
11095 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11096 with ix86_finalize_stack_frame_flags.
11097 (ix86_expand_epilogue): Likewise.
11098 (ix86_expand_split_stack_prologue): Likewise.
11100 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11103 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11104 to P_AES for Westmere.
11106 2017-08-07 Jonathan Yong <10walls@gmail.com>
11108 * config/i386/mingw.opt (fset-stack-executable): Removed.
11109 * config/i386/cygming.opt (fset-stack-executable): Moved
11111 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11113 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11115 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11117 2017-08-07 Marek Polacek <polacek@redhat.com>
11119 PR middle-end/81737
11120 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11122 2017-08-07 Martin Liska <mliska@suse.cz>
11124 * attribs.h (canonicalize_attr_name): New function.
11125 (cmp_attribs): Move from c-format.c and adjusted.
11126 (is_attribute_p): Moved from tree.h.
11127 * tree-inline.c: Add new includes.
11128 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11129 (private_is_attribute_p): Remove.
11130 (private_lookup_attribute): Likewise.
11131 (private_lookup_attribute_by_prefix): Simplify.
11132 (remove_attribute): Use is_attribute_p.
11133 * tree.h: Remove removed declarations.
11135 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11137 PR middle-end/81698
11138 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11139 instead of computing it in the function. Formatting fix.
11140 (expand_case): Don't rely on default_edge being the first edge,
11141 clear it if removing it, pass default_edge to
11142 emit_case_dispatch_table.
11143 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11146 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11148 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11149 insn in the function, emit NOP after the insn.
11151 2017-08-06 Tom de Vries <tom@codesourcery.com>
11153 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11156 2017-08-06 Tom de Vries <tom@codesourcery.com>
11158 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11161 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11163 PR tree-optimization/57371
11164 * match.pd: New pattern.
11166 2017-08-04 Marek Polacek <polacek@redhat.com>
11168 PR middle-end/81695
11169 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11170 perform the computation in offset_int.
11172 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11174 PR tree-optimization/81136
11175 * tree-vectorizer.h: Include tree-hash-traits.h.
11176 (vec_base_alignments): New typedef.
11177 (vec_info): Add a base_alignments field.
11178 (vect_record_base_alignments): Declare.
11179 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11181 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11182 (create_data_ref): Add an is_conditional_in_stmt argument.
11183 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11184 the is_conditional_in_stmt field.
11185 (data_ref_loc): Add an is_conditional_in_stmt field.
11186 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11187 (find_data_references_in_stmt): Update call to create_data_ref.
11188 (graphite_find_data_references_in_stmt): Likewise.
11189 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11190 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11191 (vect_record_base_alignment): New function.
11192 (vect_record_base_alignments): Likewise.
11193 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11194 for nested statements even if we fail to compute a misalignment.
11195 Use pooled base alignments for unconditional references.
11196 (vect_find_same_alignment_drs): Compare base addresses instead
11198 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11199 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11201 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11203 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11204 Add an explicit name for the enum. Use auto_vec for slp_instances
11205 and grouped_stores.
11206 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11208 (_bb_vec_info): Add a constructor and destructor.
11209 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11210 (destroy_loop_vec_info): Delete.
11211 (vect_destroy_datarefs): Likewise.
11212 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11213 (vec_info::vec_info): New function.
11214 (vec_info::~vec_info): Likewise.
11215 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11216 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11217 destroy_loop_vec_info.
11218 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11219 (_loop_vec_info::_loop_vec_info): ...this.
11220 (destroy_loop_vec_info): Replace with...
11221 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11222 the stmt_vec_infos. Leave handling of vec_info information to its
11223 destructor. Remove explicit vector releases.
11224 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11225 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11226 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11227 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11228 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11229 (destroy_bb_vec_info): Replace with...
11230 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11231 information to its destructor.
11232 (vect_slp_analyze_bb_1): Use new and delete instead of
11233 new_bb_vec_info and destroy_bb_vec_info.
11234 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11237 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11239 * tree-data-ref.h (subscript): Add access_fn field.
11240 (data_dependence_relation): Add could_be_independent_p.
11241 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11242 (same_access_functions): Move to tree-data-ref.c.
11243 * tree-data-ref.c (ref_contains_union_access_p): New function.
11244 (access_fn_component_p): Likewise.
11245 (access_fn_components_comparable_p): Likewise.
11246 (dr_analyze_indices): Add a reference to access_fn_component_p.
11247 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11249 (constant_access_functions): Likewise.
11250 (add_other_self_distances): Likewise.
11251 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11252 (initialize_data_dependence_relation): Use XCNEW and remove
11253 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11254 of access functions that have the same type. Allow the
11255 subsequence to end with different bases in some circumstances.
11256 Record the chosen access functions in SUB_ACCESS_FN.
11257 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11258 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11259 (subscript_dependence_tester_1): Likewise dra and drb.
11260 (build_classic_dist_vector): Update calls accordingly.
11261 (subscript_dependence_tester): Likewise.
11262 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11263 DDR_COULD_BE_INDEPENDENT_P.
11264 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11265 comp_alias_ddrs instead of may_alias_ddrs.
11266 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11268 (vect_analyze_data_ref_dependence): Use it if
11269 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11270 distance vectors if that fails.
11271 (dependence_distance_ge_vf): New function.
11272 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11273 LOOP_VINFO_MAY_ALIAS_DDRS.
11275 2017-08-04 Richard Biener <rguenther@suse.de>
11277 PR middle-end/81705
11278 * fold-const.c (fold_binary_loc): Properly restrict
11279 minus_var0 && minus_var1 case when associating undefined overflow
11282 2017-08-04 Tom de Vries <tom@codesourcery.com>
11284 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11286 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11288 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11289 Don't start diagnostic messages with a capital letter.
11290 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11292 (rs6000_invalid_builtin): Likewise.
11293 (rs6000_trampoline_init): Likewise.
11295 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11298 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11299 after setting changeable df flags.
11301 2017-08-03 Richard Biener <rguenther@suse.de>
11303 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11304 up if the use is in USE - X.
11306 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11308 * toplev.c (dumpfile.h): New include.
11309 (internal_error_reentered): New static function. Use it...
11310 (internal_error_function): ...here to handle reentered internal_error.
11312 2017-08-03 Richard Biener <rguenther@suse.de>
11314 PR middle-end/81148
11315 * fold-const.c (split_tree): Add minus_var and minus_con
11316 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11317 here but always use minus_*.
11318 (associate_trees): Assert we never associate with MINUS_EXPR
11319 and NULL first operand. Do not recurse for PLUS_EXPR operands
11320 when associating as MINUS_EXPR either.
11321 (fold_binary_loc): Track minus_var and minus_con.
11323 2017-08-03 Tom de Vries <tom@codesourcery.com>
11326 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11327 ACCEL_COMPILER, apply finish_options on
11328 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11330 2017-08-03 Tom de Vries <tom@codesourcery.com>
11333 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11334 function_entry_patch_area_size > 0.
11336 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11339 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11340 instead of 10??LU, perform unit multiplication in wide_int,
11341 don't change alloc_object_size_limit if the limit is larger
11344 PR tree-optimization/81655
11345 PR tree-optimization/81588
11346 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11347 the case when ranges[i].low and high are 1 for unsigned type with
11350 PR middle-end/81052
11351 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11352 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11354 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11356 * tree-vrp.h: Add include guard.
11358 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11361 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11362 (ud2): New insn pattern.
11363 * config/i386/i386.c (ix86_expand_epilogue):
11364 For naked functions, generate ud2 instead of trap insn.
11366 2017-08-02 Marek Polacek <polacek@redhat.com>
11369 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11371 2017-08-02 Tom de Vries <tom@codesourcery.com>
11372 Cesar Philippidis <cesar@codesourcery.com>
11374 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11375 Add missing edge probabilities.
11377 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11379 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11380 Correct endianness.
11382 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11384 PR middle-end/79499
11385 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11386 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11387 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11388 prologue_seq sequences - if any.
11390 2017-08-02 Richard Biener <rguenther@suse.de>
11392 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11393 via vectors if supported, integer extracts via punning if supported
11394 or otherwise vector extracts.
11396 2017-08-02 Richard Biener <rguenther@suse.de>
11398 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11400 (bitmap_insert_into_set): ... this.
11402 2017-08-02 Richard Biener <rguenther@suse.de>
11404 PR tree-optimization/81633
11406 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11408 PR tree-optimization/71752
11409 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11411 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11413 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11414 (machine_function::call_ms2sysv_pad_out): Remove field.
11415 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11416 (ix86_compute_frame_layout): Likewise.
11418 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11421 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11422 attribute with interrupt attribute.
11424 2017-08-01 Andrew Pinski <apinski@cavium.com>
11426 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11427 BIT_INSERT_EXPR's operand 1
11428 to see if the types precision matches.
11430 2017-08-01 Martin Liska <mliska@suse.cz>
11432 PR middle-end/70140
11433 * builtins.c (expand_builtin_memcpy_args): Remove.
11434 (expand_builtin_memcpy): Call newly added function
11435 expand_builtin_memory_copy_args.
11436 (expand_builtin_memcpy_with_bounds): Likewise.
11437 (expand_builtin_mempcpy): Remove last argument.
11438 (expand_builtin_mempcpy_with_bounds): Likewise.
11439 (expand_builtin_memory_copy_args): New function created from
11440 expand_builtin_mempcpy_args with small modifications.
11441 (expand_builtin_mempcpy_args): Remove.
11442 (expand_builtin_stpcpy): Remove unused argument.
11443 (expand_builtin): Likewise.
11444 (expand_builtin_with_bounds): Likewise.
11446 2017-08-01 Martin Liska <mliska@suse.cz>
11449 Make mempcpy more optimal (PR middle-end/70140).
11451 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11455 __builtin_vec_cmpne verify both arguments are compatible vectors
11456 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11457 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11458 move computation of aligned to after checking the argument types.
11462 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11463 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11466 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11467 Doug Rupp <rupp@adacore.com>
11468 Olivier Hainque <hainque@adacore.com>
11470 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11471 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11473 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11474 for TARGET_OS_CPP_BUILTIN.
11475 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11476 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11478 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11479 passing required abi options to the assembler for EABI configurations.
11480 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11481 of .text.hot and .text.unlikely sections for kernel modules when
11482 using ARM style exceptions.
11483 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11484 options. Add EXTRA_CC1_SPEC.
11485 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11487 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11489 (ARM_TARGET2_DWARF_FORMAT): Define.
11490 * config/arm/t-vxworks: Adjust multilib control to removal of the
11491 Diab command line options.
11493 2017-08-01 Martin Liska <mliska@suse.cz>
11495 PR gcov-profile/81561
11496 * gcov.c (unblock): Make unblocking safe as we need to preserve
11497 index correspondence of blocks and block_lists.
11499 2017-08-01 Richard Biener <rguenther@suse.de>
11501 PR tree-optimization/81181
11502 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11503 (compute_antic): ... end of iteration here.
11505 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11507 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11508 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11509 (ftree-slp-vectorize): Likewise.
11510 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11511 can no longer be set independent of flag_tree_loop_vectorize.
11512 * omp-general.c (emp_max_vf): Likewise.
11513 * opts.c (enable_fdo_optimizations): Remove references to
11514 flag_tree_vectorize, these are now implicit.
11515 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11516 and leave it for the options machinery.
11518 2017-08-01 Martin Liska <mliska@suse.cz>
11520 PR middle-end/70140
11521 * builtins.c (expand_builtin_memcpy_args): Remove.
11522 (expand_builtin_memcpy): Call newly added function
11523 expand_builtin_memory_copy_args.
11524 (expand_builtin_memcpy_with_bounds): Likewise.
11525 (expand_builtin_mempcpy): Remove last argument.
11526 (expand_builtin_mempcpy_with_bounds): Likewise.
11527 (expand_builtin_memory_copy_args): New function created from
11528 expand_builtin_mempcpy_args with small modifications.
11529 (expand_builtin_mempcpy_args): Remove.
11530 (expand_builtin_stpcpy): Remove unused argument.
11531 (expand_builtin): Likewise.
11532 (expand_builtin_with_bounds): Likewise.
11534 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11537 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11538 print "ds:" only for immediates in generic address space.
11540 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11543 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11544 (ix86_function_ok_for_sibcall): Return false for naked functions.
11546 2017-08-01 Richard Biener <rguenther@suse.de>
11548 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11549 (compute_antic): Seed worklist with exit block predecessors.
11550 * cfganal.c (dfs_find_deadend): For a cycle return the source
11551 of the edge closing it.
11553 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11555 * config/aarch64/aarch64.c
11556 (aarch64_can_const_movi_rtx_p): Move 0 check.
11558 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11560 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11561 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11563 * match.pd: Ditto in address comparison pattern.
11565 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11567 PR tree-optimization/81627
11568 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11569 closed ssa form for store-store chain.
11571 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11573 PR tree-optimization/81620
11574 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11575 for store-store chain.
11577 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11579 PR tree-optimization/81588
11580 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11581 ranges[i].in_p, invert comparison code ccode. For >/>=,
11582 swap rhs1 and rhs2 and comparison code unconditionally,
11583 for </<= don't do that. Don't swap rhs1/rhs2 again if
11584 ranges[i].in_p, instead invert comparison code ccode if
11585 opcode or oe->rank is BIT_IOR_EXPR.
11588 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11589 a direct optab to conversion optab.
11590 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11591 with GET_MODE_INNER as last argument instead of optab_handler.
11592 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11593 vector extraction if possible and optab is available.
11594 * expr.c (store_constructor): Use convert_optab_handler instead
11595 of optab_handler. Use vector initialization from smaller
11596 vectors if possible and optab is available.
11597 * tree-vect-stmts.c (vectorizable_load): Likewise.
11598 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11599 now have two modes.
11600 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11601 of vec_init from half-sized vectors with the same element mode.
11602 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11603 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11604 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11605 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11606 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11607 after mode in gen_vec_extract* calls.
11608 (vec_extract<mode>): Renamed to ...
11609 (vec_extract<mode><ssescalarmodelower>): ... this.
11610 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11611 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11612 element mode after mode in gen_vec_init* calls.
11613 (VEC_INIT_HALF_MODE): New mode iterator.
11614 (vec_init<mode>): Renamed to ...
11615 (vec_init<mode><ssescalarmodelower>): ... this.
11616 (vec_init<mode><ssehalfvecmodelower>): New expander.
11617 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11618 (vec_extractv2sfsf): ... this.
11619 (vec_initv2sf): Renamed to ...
11620 (vec_initv2sfsf): ... this.
11621 (vec_extractv2si): Renamed to ...
11622 (vec_extractv2sisi): ... this.
11623 (vec_initv2si): Renamed to ...
11624 (vec_initv2sisi): ... this.
11625 (vec_extractv4hi): Renamed to ...
11626 (vec_extractv4hihi): ... this.
11627 (vec_initv4hi): Renamed to ...
11628 (vec_initv4hihi): ... this.
11629 (vec_extractv8qi): Renamed to ...
11630 (vec_extractv8qiqi): ... this.
11631 (vec_initv8qi): Renamed to ...
11632 (vec_initv8qiqi): ... this.
11633 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11634 (vec_init<mode>): Renamed to ...
11635 (vec_init<mode><VEC_base_l>): ... this.
11636 (vec_extract<mode>): Renamed to ...
11637 (vec_extract<mode><VEC_base_l>): ... this.
11638 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11639 (vec_initv2sfsf): ... this.
11640 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11641 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11642 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11643 element mode after mode in gen_vec_init* calls.
11644 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11645 (vec_init<mode><Vel>): ... this.
11646 (vec_extract<mode>): Renamed to ...
11647 (vec_extract<mode><Vel>): ... this.
11648 * config/aarch64/iterators.md (Vel): New mode attribute.
11649 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11650 Add element mode after mode in gen_vec_extract* calls.
11651 * config/s390/vector.md (non_vec_l): New mode attribute.
11652 (vec_extract<mode>): Renamed to ...
11653 (vec_extract<mode><non_vec_l>): ... this.
11654 (vec_init<mode>): Renamed to ...
11655 (vec_init<mode><non_vec_l>): ... this.
11656 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11657 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11659 * config/arm/iterators.md (V_elem_l): New mode attribute.
11660 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11661 (vec_extract<mode><V_elem_l>): ... this.
11662 (vec_extractv2di): Renamed to ...
11663 (vec_extractv2didi): ... this.
11664 (vec_init<mode>): Renamed to ...
11665 (vec_init<mode><V_elem_l>): ... this.
11666 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11667 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11668 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11669 Add element mode after gen_vec_extract* calls.
11670 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11671 (vec_init<mode><unitmode>): ... this.
11672 (vec_extract<mode>): Renamed to ...
11673 (vec_extract<mode><unitmode>): ... this.
11674 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11675 (vec_init<mode><unitmode>): ... this.
11676 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11677 (vec_initv2sfsf): ... this.
11678 (vec_extractv2sf): Renamed to ...
11679 (vec_extractv2sfsf): ... this.
11680 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11681 Add element mode after gen_vec_extract* calls.
11682 * config/mips/mips.md (unitmode): New mode iterator.
11683 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11684 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11685 * config/spu/spu.md (inner_l): New mode attribute.
11686 (vec_init<mode>): Renamed to ...
11687 (vec_init<mode><inner_l>): ... this.
11688 (vec_extract<mode>): Renamed to ...
11689 (vec_extract<mode><inner_l>): ... this.
11690 * config/sparc/sparc.md (veltmode): New mode iterator.
11691 (vec_init<VMALL:mode>): Renamed to ...
11692 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11693 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11694 (vec_initv2sisi): ... this.
11695 (vec_initv2sf): Renamed to ...
11696 (vec_initv2sfsf): ... this.
11697 (vec_extractv2sf): Renamed to ...
11698 (vec_extractv2sfsf): ... this.
11699 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11700 (vec_init<mode>): Renamed to ...
11701 (vec_init<mode><VEC_base_l>): ... this.
11702 (vec_extract<mode>): Renamed to ...
11703 (vec_extract<mode><VEC_base_l>): ... this.
11704 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11705 (vec_initv2sfsf): ... this.
11706 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11707 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11708 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11709 gen_vec_init* calls.
11711 2017-08-01 Richard Biener <rguenther@suse.de>
11713 PR tree-optimization/81297
11714 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11715 TREE_OVERFLOW from INTEGER_CSTs.
11717 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11719 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11721 2017-07-31 Carl Love <cel@us.ibm.com>
11723 * config/rs6000/rs6000-c: Add support for built-in functions
11724 vector signed char vec_xl_be (signed long long, signed char *);
11725 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11726 vector signed int vec_xl_be (signed long long, signed int *);
11727 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11728 vector signed long long vec_xl_be (signed long long, signed long long *);
11729 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11730 vector signed short vec_xl_be (signed long long, signed short *);
11731 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11732 vector double vec_xl_be (signed long long, double *);
11733 vector float vec_xl_be (signed long long, float *);
11734 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11735 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11736 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11738 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11739 (altivec_expand_builtin): Add switch statement to call
11740 altivec_expand_xl_be for each builtin.
11741 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11742 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11743 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11744 __builtin_vsx_le_be_v16qi.
11745 * doc/extend.texi: Update the built-in documentation file for the
11746 new built-in functions.
11748 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11751 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11753 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11755 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11757 * config.gcc: Add z14.
11758 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11759 CPU model numbers for z13s and z14.
11760 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11762 * config/s390/s390-opts.h (enum processor_type): Rename
11763 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11764 * config/s390/s390.c (processor_table): Add field for CPU name to
11765 be passed to Binutils.
11766 (s390_asm_output_machine_for_arch): Use the new field in
11767 processor_table for Binutils.
11768 (s390_expand_builtin): Replace arch12 with z14.
11769 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11770 (s390_get_sched_attrmask): Likewise.
11771 (s390_get_unit_mask): Likewise.
11772 * config/s390/s390.opt: Add z14 to processor_type enum.
11774 2017-07-31 Martin Jambor <mjambor@suse.cz>
11777 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11778 regardless of optimization level.
11780 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11781 Martin Liska <mliska@suse.cz>
11783 * predict.def: Remove old comment and adjust probability.
11784 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11785 PREDICT statements.
11787 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11790 * config/i386/i386.c (ix86_function_naked): New function.
11791 (ix86_can_use_return_insn_p): Return false for naked functions.
11792 (ix86_expand_prologue): Skip prologue for naked functions.
11793 (ix86_expand_epilogue): Skip epilogue for naked functions
11794 and emit trap instruction.
11795 (ix86_warn_func_return): New function.
11796 (ix86_attribute_table): Add "naked" attribute specification.
11797 (TARGET_WARN_FUNC_RETURN): Define.
11798 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11800 2017-07-31 Martin Liska <mliska@suse.cz>
11802 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11803 (dump_gimple_bb_header): Always dump BB info.
11804 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11806 2017-07-31 Martin Liska <mliska@suse.cz>
11809 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11810 also with current_function_decl non-null equality.
11812 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11815 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11816 change type to the element type, instead add eltype variable and
11817 use it where we are interested in the element type.
11819 PR tree-optimization/81603
11820 * ipa-polymorphic-call.c
11821 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11822 offset arithmetic in offset_int, bail out if the resulting bit offset
11823 doesn't fit into shwi.
11825 2017-07-31 Martin Liska <mliska@suse.cz>
11827 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11828 (gimplify_save_expr): Fix comment.
11830 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11833 * config/i386/i386.c (ix86_function_arg): Update arguments for
11835 (ix86_compute_frame_layout): Set the initial stack offset to
11836 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11837 INCOMING_FRAME_SP_OFFSET.
11838 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11839 stack before exception handler returns.
11840 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11841 the 'ERROR_CODE' for exception handler.
11843 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11845 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11846 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11847 (ASM_OUTPUT_REG_POP): Ditto.
11848 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11849 instead of asm_fprintf to output pure string.
11851 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11853 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11854 to imported_module_or_decl hook.
11855 (debug_nothing_tree_tree_tree_bool): Remove.
11856 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11857 * debug.c (do_nothing_debug_hooks): Use
11858 debug_nothing_tree_tree_tree_bool_bool instead of
11859 debug_nothing_tree_tree_tree_bool.
11860 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11861 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11862 * sdbout.c (sdb_debug_hooks): Likewise.
11863 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11864 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11866 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11867 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11868 attribute, don't add anything.
11870 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11872 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11873 (fold_build2_stat_loc): Likewise.
11874 (fold_build3_stat_loc): Likewise.
11875 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11876 (fold_build1_loc): Remove macro.
11877 (fold_build2_loc): Likewise.
11878 (fold_build3_loc): Likewise.
11880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11882 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11883 (gimple_build_debug_bind_source_stat): Likewise.
11884 * gimple.h (gimple_build_debug_bind): Remove macro.
11885 (gimple_build_debug_bind_source): Likewise.
11887 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11889 * bitmap.c (bitmap_alloc): Adjust.
11890 (bitmap_gc_alloc): Likewise.
11891 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11893 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11895 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11896 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11897 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11898 (bitmap_gc_alloc_stat): Likewise.
11899 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11901 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11903 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11904 * rtl.h (shallow_copy_rtx): Remove macro.
11906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11908 * emit-rtl.c (gen_raw_REG): Adjust.
11909 * gengenrtl.c (gendef): Likewise.
11910 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11911 * rtl.h (rtx_alloc): Remove macro.
11913 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11915 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11916 (build_tree_list_stat): Likewise.
11917 * tree.h (build_tree_list): Remove macro.
11918 (build_tree_list_vec): Likewise.
11920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11922 * tree.c (make_vector_stat): Remove _stat from name.
11923 (build_vector_stat): Likewise.
11924 * tree.h (make_vector_stat): Remove macro.
11925 (build_vector_stat): Likewise.
11927 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11929 * tree.h (build_var_debug_value): Remove prototype.
11931 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11933 * tree.c (tree_cons_stat): Remove _stat from name.
11934 * tree.h (tree_cons): Remove macro.
11936 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11938 * tree.c (build_vl_exp_stat): Remove _stat from name.
11939 * tree.h (build_vl_exp): Remove macro.
11941 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11943 * tree.c (build_decl_stat): Remove _stat from name.
11944 * tree.h (build_decl): Remove macro.
11946 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11948 * gimple.c (gimple_build_with_ops_stat): Adjust.
11949 (gimple_alloc_stat): Remove _stat from name.
11950 * gimple.h (gimple_alloc): Remove macro.
11952 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11954 * tree.c (make_tree_vec_stat): Remove _stat from name.
11955 (grow_tree_vec_stat): Likewise.
11956 * tree.h (make_tree_vec_stat): Adjust prototype.
11957 (grow_tree_vec_stat): Likewise.
11958 (make_tree_vec): Remove macro.
11959 (grow_tree_vec): Likewise.
11961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11963 * fold-const.c (fold_build1_stat_loc): Adjust.
11964 (fold_build2_stat_loc): Likewise.
11965 (fold_build3_stat_loc): Likewise.
11966 * tree.c (build0_stat): Remove _stat from name.
11967 (build1_stat): Likewise.
11968 (build2_stat): Likewise.
11969 (build3_stat): Likewise.
11970 (build4_stat): Likewise.
11971 (build5_stat): Likewise.
11972 * tree.h (build1_loc): Remove macro, and rename _stat function
11974 (build2_loc): Likewise.
11975 (build3_loc): Likewise.
11976 (build4_loc): Likewise.
11977 (build5_loc): Likewise.
11979 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11981 * tree.c (make_int_cst_stat): Remove _stat from name.
11982 * tree.h (make_int_cst_stat): Adjust prototype.
11983 (make_int_cst): Remove macro.
11985 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11987 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11988 * tree.h (make_tree_binfo_stat): Adjust prototype.
11989 (make_tree_binfo): Remove.
11991 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11993 * tree.c (copy_node_stat): Rename to copy_node.
11994 (build_distinct_type_copy): Adjust.
11995 * tree.h (copy_node_stat): Adjust prototype.
11996 (copy_node): Remove macro.
11998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12000 * tree.c (make_node_stat): rename to make_node.
12001 (build_tree_list_stat): Adjust.
12002 (build0_stat): Likewise.
12003 (build2_stat): Likewise.
12004 (build3_stat): Likewise.
12005 (build4_stat): Likewise.
12006 (build5_stat): Likewise.
12007 (build_decl_stat): Likewise.
12008 * tree.h (make_node_stat): Adjust prototype.
12009 (make_node): remove macro.
12011 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12013 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12014 (PPC_FEATURE2_SCV): Likewise.
12015 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12017 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12019 * config/aarch64/aarch64.c
12020 (aarch64_internal_mov_immediate): Add new special pattern.
12021 * config/aarch64/aarch64.md (*movdi_aarch64):
12022 Add reg/32bit const mov case.
12024 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12025 Richard Sandiford <richard.sandiford@linaro.org>
12027 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12028 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12029 Add integer and movi cases.
12030 (movi-split-hf-df-sf split, fp16): New.
12031 (enabled): Added TARGET_FP_F16INST.
12032 * config/aarch64/iterators.md (GPF_HF): New.
12033 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12035 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12037 * config/aarch64/aarch64.c
12038 (aarch64_simd_container_mode): Add prototype.
12039 (aarch64_expand_mov_immediate): Add HI support.
12040 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12041 (aarch64_can_const_movi_rtx_p): New.
12042 (aarch64_preferred_reload_class):
12043 Remove restrictions of using FP registers for certain SIMD operations.
12044 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12045 (aarch64_valid_floating_const): Add integer move validation.
12046 (aarch64_simd_imm_scalar_p): Remove.
12047 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12048 (aarch64_legitimate_constant_p): Expand list of supported cases.
12049 * config/aarch64/aarch64-protos.h
12050 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12051 (aarch64_reinterpret_float_as_int): New.
12052 (aarch64_simd_imm_scalar_p): Remove.
12053 * config/aarch64/constraints.md (Uvi): New.
12054 (Dd): Split into Ds and new Dd.
12055 * config/aarch64/aarch64.md (*movsi_aarch64):
12057 (*movdi_aarch64): Add SIMD mov case.
12059 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12061 * tree-predcom.c: (struct chain): Handle store-store chain in which
12062 stores for elimination only store loop invariant values.
12063 (execute_pred_commoning_chain): Ditto.
12064 (prepare_initializers_chain_store_elim): Ditto.
12065 (prepare_finalizers): Ditto.
12066 (is_inv_store_elimination_chain): New function.
12067 (initialize_root_vars_store_elim_1): New function.
12069 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12071 * tree-predcom.c: Revise general description of the pass.
12072 (enum chain_type): New enum type for store elimination.
12073 (struct chain): New field supporting store elimination.
12074 (struct component): Ditto.
12075 (dump_chain): Dump store-stores chain.
12076 (release_chain): Release resources.
12077 (split_data_refs_to_components): Compute and create component
12078 contains only stores for elimination.
12079 (get_chain_last_ref_at): New function.
12080 (make_invariant_chain): Initialization.
12081 (make_rooted_chain): Specify chain type in parameter and record it.
12082 (add_looparound_copies): Skip for store-stores chain.
12083 (determine_roots_comp): Compute type of chain and pass it to
12085 (initialize_root_vars_store_elim_2): New function.
12086 (finalize_eliminated_stores): New function.
12087 (remove_stmt): Handle store for elimination.
12088 (execute_pred_commoning_chain): Execute predictive commoning on
12089 store-store chains.
12090 (determine_unroll_factor): Skip unroll for store-stores chain.
12091 (prepare_initializers_chain_store_elim): New function.
12092 (prepare_initializers_chain): Hanlde store-store chain.
12093 (prepare_finalizers_chain, prepare_finalizers): New function.
12094 (tree_predictive_commoning_loop): Return integer value indicating
12095 if loop is unrolled or lcssa form is corrupted.
12096 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12098 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12100 * tree-predcom.c (initialize_root): Delete.
12101 (execute_pred_commoning_chain): Initialize root vars and replace
12102 reference of non-combined chain directly, rather than call above
12105 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12107 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12108 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12110 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12112 * tree-predcom.c (struct chain): New field init_seq.
12113 (release_chain): Release init_seq.
12114 (prepare_initializers_chain): Record intialization stmts in above
12116 (insert_init_seqs): New function.
12117 (tree_predictive_commoning_loop): Call insert_init_seqs.
12119 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12121 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12123 2017-07-28 Richard Biener <rguenther@suse.de>
12125 * match.pd: Remove superfluous :c.
12126 * genmatch.c (simplify::id): Add member.
12127 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12129 (current_id): New global.
12130 (dt_node::parent): Move from ...
12131 (dt_operand::parent): ... here. Add for_id member.
12132 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12133 (decision_tree::find_node): Relax order requirement when
12134 merging DT_TRUE nodes to ones inbetween the current simplify
12135 and the one we try to merge with. Add diagnostic whenever
12136 we need to enforce pattern order by not merging.
12137 (decision_tree::insert): Set current_id.
12138 (decision_tree::print_node): Dump parent node and for_id.
12139 (parser::last_id): Add member.
12140 (parser::push_simplify): Assign unique id.
12141 (parser::parser): Initialize last_id.
12143 2017-07-28 Martin Liska <mliska@suse.cz>
12146 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12147 gimple_build_debug_bind.
12149 2017-07-28 Richard Biener <rguenther@suse.de>
12151 PR tree-optimization/81502
12152 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12154 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12156 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12157 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12158 for BIT_FIELD_REF args.
12159 * fold-const.c (make_bit_field_ref): Likewise.
12160 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12162 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12165 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12166 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12167 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12168 Or it into SANITIZER_UNDEFINED.
12169 * ubsan.c: Include gimple-fold.h and varasm.h.
12170 (ubsan_expand_ptr_ifn): New function.
12171 (instrument_pointer_overflow): New function.
12172 (maybe_instrument_pointer_overflow): New function.
12173 (instrument_object_size): Formatting fix.
12174 (pass_ubsan::execute): Call instrument_pointer_overflow
12175 and maybe_instrument_pointer_overflow.
12176 * internal-fn.c (expand_UBSAN_PTR): New function.
12177 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12178 * sanitizer.def (__ubsan_handle_pointer_overflow,
12179 __ubsan_handle_pointer_overflow_abort): New builtins.
12180 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12181 * internal-fn.def (UBSAN_PTR): New internal function.
12182 * opts.c (sanitizer_opts): Add pointer-overflow.
12183 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12184 * fold-const.c (build_range_check): Compute pointer range check in
12185 integral type if pointer arithmetics would be needed. Formatting
12188 2017-07-28 Martin Liska <mliska@suse.cz>
12191 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12192 parameters that are of a variable-length.
12194 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12196 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12198 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12199 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12200 (CRT_CALL_STATIC_FUNCTION): Likewise.
12201 (ASM_DEFAULT_SPEC): New define.
12202 (ASM_SPEC32): Likewise.
12203 (ASM_SPEC64): Likewise.
12204 (ASM_SPEC_COMMON): Likewise.
12205 (ASM_SPEC): Likewise.
12206 (INVALID_64BIT): Likewise.
12207 (LINK_OS_DEFAULT_SPEC): Likewise.
12208 (LINK_OS_SPEC32): Likewise.
12209 (LINK_OS_SPEC64): Likewise.
12210 (POWERPC_LINUX): Likewise.
12211 (PTRDIFF_TYPE): Likewise.
12212 (RESTORE_FP_PREFIX): Likewise.
12213 (RESTORE_FP_SUFFIX): Likewise.
12214 (SAVE_FP_PREFIX): Likewise.
12215 (SAVE_FP_SUFFIX): Likewise.
12216 (SIZE_TYPE): Likewise.
12217 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12218 (TARGET_64BIT): Likewise.
12219 (TARGET_64BIT): Likewise.
12220 (TARGET_AIX): Likewise.
12221 (WCHAR_TYPE_SIZE): Likewise.
12222 (WCHAR_TYPE): Undefine.
12223 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12224 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12225 (CPP_OS_RTEMS_SPEC): Delete.
12226 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12227 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12229 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12231 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12233 PR tree-optimization/81578
12234 * tree-parloops.c (build_new_reduction): Bail out if
12235 reduction_code isn't one of the standard OpenMP reductions.
12236 Move the details printing after that decision.
12238 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12240 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12241 related to reload_in_progress.
12242 (splat_input_operand): Likewise.
12243 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12245 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12247 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12248 (TARGET_INSTANTIATE_DECLS): Likewise.
12249 (legitimate_indexed_address_p): Delete reload_in_progress code.
12250 (rs6000_debug_legitimate_address_p): Likewise.
12251 (rs6000_eliminate_indexed_memrefs): Likewise.
12252 (rs6000_emit_le_vsx_store): Likewise.
12253 (rs6000_emit_move_si_sf_subreg): Likewise.
12254 (rs6000_emit_move): Likewise.
12255 (register_to_reg_type): Likewise.
12256 (rs6000_pre_atomic_barrier): Likewise.
12257 (rs6000_machopic_legitimize_pic_address): Likewise.
12258 (rs6000_allocate_stack_temp): Likewise.
12259 (rs6000_address_for_fpconvert): Likewise.
12260 (rs6000_address_for_altivec): Likewise.
12261 (rs6000_secondary_memory_needed_rtx): Delete function.
12262 (rs6000_check_sdmode): Likewise.
12263 (rs6000_alloc_sdmode_stack_slot): Likewise.
12264 (rs6000_instantiate_decls): Likewise.
12265 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12266 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12267 Delete reload_in_progress.
12268 (*vec_reload_and_plus_<mptrsize>): Likewise.
12269 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12270 (vsx_div_v2di): Likewise.
12271 (vsx_udiv_v2di): Likewise.
12273 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12275 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12276 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12277 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12278 (rs6000_debug_reg_global): Delete print of LRA status.
12279 (rs6000_option_override_internal): Delete dead LRA related code.
12280 (rs6000_lra_p): Delete function.
12281 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12283 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12285 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12286 * config/riscv/rtems.h: New file.
12288 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12289 Sudakshina Das <sudi.das@arm.com>
12291 * config/aarch64/aarch64.md
12292 (define_split for and<mode>3nr_compare): Move
12293 non aarch64_logical_operand to a register.
12294 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12295 register immediate operand to a register.
12296 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12298 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12300 PR middle-end/81564
12301 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12303 2017-07-27 Richard Biener <rguenther@suse.de>
12305 PR tree-optimization/81573
12306 PR tree-optimization/81494
12307 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12308 multi defuse cycle case.
12310 2017-07-27 Richard Biener <rguenther@suse.de>
12312 PR tree-optimization/81571
12313 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12316 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12318 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12319 earlier and only if MASK_FPU is set. Adjust formatting.
12321 2017-07-27 Martin Liska <mliska@suse.cz>
12323 * opt-functions.awk: Add validation of value of Init.
12324 * optc-gen.awk: Pass new argument.
12326 2017-07-27 Martin Liska <mliska@suse.cz>
12328 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12329 Fix wrong condition.
12331 2017-07-27 Martin Liska <mliska@suse.cz>
12333 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12334 BBs and edges seen by autoFDO.
12336 2017-07-27 Richard Biener <rguenther@suse.de>
12338 PR tree-optimization/81502
12339 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12340 with incompatible but same sized type.
12341 (execute_update_addresses_taken): Likewise.
12343 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12345 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12346 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12348 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12351 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12352 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12353 Change s_operand to memory_operand.
12355 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12357 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12358 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12359 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12360 Emit instructions rather than returning an expression. Handle TFmode
12361 and KFmode by casting to TImode.
12362 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12363 (rs6000_emit_le_vsx_store): Likewise.
12364 * config/rs6000/vsx.md (VSX_TI): New iterator.
12365 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12366 (*vsx_le_undo_permute_<mode>): Likewise.
12367 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12368 emit the split sequence.
12369 (*vsx_le_perm_store_<mode>): Likewise.
12371 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12373 PR tree-optimization/81555
12374 PR tree-optimization/81556
12375 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12376 if true, force CHANGED for the recursive invocation.
12377 (reassociate_bb): Remember original length of ops array, pass
12378 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12380 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12381 attributes for noipa attribute. For naked attribute use
12382 lookup_attribute first before lookup_attribute_spec.
12383 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12385 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12386 for functions with noipa attribute.
12387 (cgraph_externally_visible_p): Return true for functions with noipa
12389 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12390 for functions with noipa attribute.
12391 * doc/extend.texi: Document noipa function attribute.
12392 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12393 also for functions with noipa attribute.
12394 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12396 2017-07-26 Andrew Pinski <apinski@cavium.com>
12398 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12399 vec_unalign_load_cost and vec_unalign_store_cost.
12401 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12403 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12404 -mvsx-small-integer option.
12405 (ISA_3_0_MASKS_IEEE): Likewise.
12406 (OTHER_VSX_VECTOR_MASKS): Likewise.
12407 (POWERPC_MASKS): Likewise.
12408 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12409 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12410 code, only testing for DImode being allowed in non-VSX floating
12412 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12413 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12415 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12416 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12417 TARGET_P8_VECTOR test.
12418 (rs6000_secondary_reload_simple_move): Likewise.
12419 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12420 since TARGET_P9_VECTOR was already tested.
12421 (rs6000_opt_masks): Remove -mvsx-small-integer.
12422 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12423 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12425 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12426 test for TARGET_VEXTRACTUB was used, and that uses
12428 (p9 extract splitter): Likewise.
12429 (vsx_extract_<mode>_di_p9): Likewise.
12430 (vsx_extract_<mode>_store_p9): Likewise.
12431 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12432 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12433 the elimination of TARGET_VSX_SMALL_INTEGER.
12434 (vsx_extract_<mode>_p8): Likewise.
12435 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12436 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12437 (vsx_set_<mode>_p9): Likewise.
12438 (vsx_set_v4sf_p9): Likewise.
12439 (vsx_set_v4sf_p9_zero): Likewise.
12440 (vsx_insert_extract_v4sf_p9): Likewise.
12441 (vsx_insert_extract_v4sf_p9_2): Likewise.
12442 * config/rs6000/rs6000.md (sign extend splitter): Change
12443 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12444 (floatsi<mode>2_lfiwax_mem): Likewise.
12445 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12446 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12447 since a test for TARGET_P9_VECTOR was used.
12448 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12449 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12450 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12451 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12452 TARGET_P8_VECTOR test.
12453 (fix_trunc<mode>si2_stfiwx): Likewise.
12454 (fix_trunc<mode>si2_internal): Likewise.
12455 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12456 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12458 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12459 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12460 TARGET_P8_VECTOR test.
12461 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12462 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12463 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12465 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12466 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12467 since a test for TARGET_P9_VECTOR was used.
12468 (splitter for loading small constants): Likewise.
12470 2017-07-26 Andrew Pinski <apinski@cavium.com>
12472 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12475 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12478 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12479 (fp_valid_at): Likewise.
12481 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12483 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12484 (qdf24xx_addrcost_table): Likewise.
12485 (cortexa57_tunings): Update to use generic_branch_cost.
12486 (cortexa72_tunings): Likewise.
12487 (cortexa73_tunings): Likewise.
12488 (qdf24xx_tunings): Likewise.
12490 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12492 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12493 (thunderx2t99_branch_cost): Likewise.
12494 (cortexa35_tunings): Update to use generic_branch_cost.
12495 (cortexa53_tunings): Likewise.
12496 (cortexa57_tunings): Likewise.
12497 (cortexa72_tunings): Likewise.
12498 (cortexa73_tunings): Likewise.
12499 (thunderx2t99_tunings): Likewise.
12501 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12503 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12504 (sparc_option_override): Honour MASK_FSMULD.
12505 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12506 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12507 * config/sparc/sparc.opt (mfsmuld): New option.
12508 * doc/invoke.texi (mfsmuld): Document option.
12510 2017-07-26 Marek Polacek <polacek@redhat.com>
12512 PR middle-end/70992
12513 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12515 2017-07-26 Richard Biener <rguenther@suse.de>
12517 * gimple-match-head.c (do_valueize): Return OP if valueize
12519 (get_def): New helper to get at the def stmt of a SSA name
12520 if valueize allows.
12521 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12522 do_valueize to get at the def stmt.
12523 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12525 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12527 PR middle-end/46932
12528 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12530 2017-07-26 Martin Liska <mliska@suse.cz>
12533 * function.c (expand_function_start): Make expansion of
12534 nonlocal_goto_save_area after parm_birth_insn.
12536 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12538 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12539 from all CPU target flags enable members.
12541 2017-07-26 Richard Biener <rguenther@suse.de>
12543 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12544 (decision_tree::gen): Make 'type' const.
12545 (write_predicate): Likewise.
12547 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12549 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12550 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12551 (rs6000_option_override_internal): Likewise.
12552 (rs6000_expand_vector_set): Likewise.
12553 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12554 (TARGET_UPPER_REGS_SF): Likewise.
12555 (TARGET_UPPER_REGS_DI): Likewise.
12556 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12557 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12558 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12559 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12560 (Splitters for DI constants in Altivec registers): Likewise.
12561 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12562 (vsx_set_v4sf_p9): Likewise.
12563 (vsx_set_v4sf_p9_zero): Likewise.
12564 (vsx_insert_extract_v4sf_p9): Likewise.
12565 (vsx_insert_extract_v4sf_p9_2): Likewise.
12567 2017-07-25 Carl Love <cel@us.ibm.com>
12569 * doc/extend.texi: Update the built-in documentation file for the
12570 existing built-in functions
12571 vector signed char vec_cnttz (vector signed char);
12572 vector unsigned char vec_cnttz (vector unsigned char);
12573 vector signed short vec_cnttz (vector signed short);
12574 vector unsigned short vec_cnttz (vector unsigned short);
12575 vector signed int vec_cnttz (vector signed int);
12576 vector unsigned int vec_cnttz (vector unsigned int);
12577 vector signed long long vec_cnttz (vector signed long long);
12578 vector unsigned long long vec_cnttz (vector unsigned long long);
12580 2017-07-25 Andrew Pinski <apinski@cavium.com>
12582 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12583 accesses where the use is for the first operand of a BIT_INSERT.
12585 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12588 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12589 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12591 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12593 * config/i386/gstabs.h: Delete.
12594 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12596 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12598 * config/i386/i386.c (ix86_decompose_address): Do not check for
12599 register RTX when looking at index_reg or base_reg.
12600 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12602 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12604 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12605 to update EH info here.
12607 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12609 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12611 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12613 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12615 2017-07-25 Torsten Duwe <duwe@suse.de>
12617 * common.opt: Introduce -fpatchable-function-entry
12618 command line option, and its variables function_entry_patch_area_size
12619 and function_entry_patch_area_start.
12620 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12621 including a two-value parser.
12622 * target.def (print_patchable_function_entry): New target hook.
12623 * targhooks.h (default_print_patchable_function_entry): New function.
12624 * targhooks.c (default_print_patchable_function_entry): Likewise.
12625 * toplev.c (process_options): Switch off IPA-RA if
12626 patchable function entries are being generated.
12627 * varasm.c (assemble_start_function): Look at the
12628 patchable-function-entry command line switch and current
12629 function attributes and maybe generate NOP instructions by
12630 calling the print_patchable_function_entry hook.
12631 * doc/extend.texi: Document patchable_function_entry attribute.
12632 * doc/invoke.texi: Document -fpatchable_function_entry
12633 command line option.
12634 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12636 * doc/tm.texi: Re-generate.
12638 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12641 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12642 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12644 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12646 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12648 2017-07-25 Richard Biener <rguenther@suse.de>
12650 PR tree-optimization/81455
12651 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12652 not walk in cycles when looking for guards.
12654 2017-07-25 Richard Biener <rguenther@suse.de>
12656 PR tree-optimization/81529
12657 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12658 when optimizing backedge uses.
12660 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12662 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12664 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12665 to dl_section_ref. On AIX, append an expression to subtract
12666 the size of the section length to dl_section_ref.
12668 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12670 * configure.ac: If any of the config.* scripts fail, exit 1.
12671 * configure: Regenerate.
12673 2017-07-25 Richard Biener <rguenther@suse.de>
12675 PR middle-end/81546
12676 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12677 of immediate uses, be more verbose on errors.
12679 2017-07-25 Richard Biener <rguenther@suse.de>
12681 PR tree-optimization/81510
12682 * tree-vect-loop.c (vect_is_simple_reduction): When the
12683 reduction stmt is not inside the loop bail out.
12685 2017-07-25 Richard Biener <rguenther@suse.de>
12687 PR tree-optimization/81303
12688 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12689 profitability check against LOOP_VINFO_NITERSM1.
12691 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12693 * domwalk.c (cmp_bb_postorder): Simplify.
12694 (sort_bbs_postorder): New function. Use it...
12695 (dom_walker::walk): ...here to optimize common cases.
12697 2017-07-25 Martin Liska <mliska@suse.cz>
12700 * ipa-visibility.c (function_and_variable_visibility): Make the
12701 redirection just on target that supports aliasing.
12702 Fix GNU coding style.
12704 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12707 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12709 * config/arm/rtems.h: Likewise.
12710 * config/bfin/rtems.h: Likewise.
12711 * config/i386/rtemself.h: Likewise.
12712 * config/lm32/rtems.h: Likewise.
12713 * config/m32c/rtems.h: Likewise.
12714 * config/m68k/rtemself.h: Likewise.
12715 * config/microblaze/rtems.h: Likewise.
12716 * config/mips/rtems.h: Likewise.
12717 * config/moxie/rtems.h: Likewise.
12718 * config/nios2/rtems.h: Likewise.
12719 * config/powerpcspe/rtems.h: Likewise.
12720 * config/rs6000/rtems.h: Likewise.
12721 * config/rtems.h: Likewise.
12722 * config/sh/rtems.h: Likewise.
12723 * config/sh/rtemself.h: Likewise.
12724 * config/sparc/rtemself.h: Likewise.
12726 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12729 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12730 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12731 * tree-ssa-structalias.c (alias_get_name): Same.
12733 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12736 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12737 instructions if no du chain is found.
12739 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12741 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12743 2017-07-25 Richard Biener <rguenther@suse.de>
12745 PR middle-end/81505
12746 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12749 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12751 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12752 upper-regs options.
12753 (ISA_2_7_MASKS_SERVER): Likewise.
12754 (ISA_3_0_MASKS_IEEE): Likewise.
12755 (OTHER_P8_VECTOR_MASKS): Likewise.
12756 (OTHER_VSX_VECTOR_MASKS): Likewise.
12757 (POWERPC_MASKS): Likewise.
12758 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12759 duplicate list of options.
12760 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12761 explicit -mupper-regs options.
12762 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12763 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12764 alias for -mupper-regs-df.
12765 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12766 (rs6000_init_hard_regno_mode_ok): Likewise.
12767 (rs6000_option_override_internal): Likewise.
12768 (rs6000_opt_masks): Likewise.
12769 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12770 options in terms of whether -mvsx or -mpower8-vector was used.
12771 (TARGET_UPPER_REGS_DI): Likewise.
12772 (TARGET_UPPER_REGS_SF): Likewise.
12773 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12774 -mupper-regs-* options.
12776 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12778 * passes.c (emergency_dump_function): Print some empty lines and a
12779 header before the RTL dump.
12781 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12783 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12785 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12788 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12789 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12791 2017-07-24 Carl Love <cel@us.ibm.com>
12793 * config/rs6000/rs6000-c.c: Add support for built-in functions
12794 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12795 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12796 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12797 vec_extract_fp_from_shortl): Add defines for the two builtins.
12798 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12799 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12801 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12802 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12803 * doc/extend.texi: Update the built-in documentation file for the
12804 new built-in function.
12806 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12809 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12811 * doc/generic.texi: Likewise.
12812 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12813 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12815 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12817 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12818 (aarch64_mls_elt_merge<mode>): Likewise.
12820 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12822 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12823 having __cxa_atexit.
12825 2017-07-23 Michael Collison <michael.collison@arm.com>
12827 * config/arm/arm.c (arm_option_override): Deprecate
12828 use of -mstructure-size-boundary.
12829 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12830 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12832 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12835 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12836 Reduce cost estimate for direct moves.
12838 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12841 * config/i386/i386.c (ix86_option_override_internal): Disable
12842 BMI, BMI2 and TBM instructions for -m16.
12844 2017-07-21 Carl Love <cel@us.ibm.com>
12846 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12847 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12848 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12849 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12850 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12851 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12852 VMULOSW): New enum "unspec" values.
12853 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12854 altivec_vmulosw): New patterns.
12855 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12856 VMULOSW): Add definitions.
12858 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12860 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12861 (qdf24xx): Likewise.
12862 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12863 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12864 (AARCH64_FL_V8_1): Renumber.
12865 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12866 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12867 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12868 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12869 rdma to feature modifiers list.
12871 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12873 PR middle-end/56727
12874 * ipa-visibility (function_and_variable_visibility): Convert
12875 recursive PLT call to direct call if appropriate.
12877 2017-07-21 Andrew Pinski <apinski@cavium.com>
12879 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12880 operand 1 to see if the types precision matches.
12881 * fold-const.c (operand_equal_p): Likewise.
12883 2017-07-21 Richard Biener <rguenther@suse.de>
12885 PR tree-optimization/81303
12886 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12887 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12888 (vect_peeling_hash_get_lowest_cost): Adjust.
12889 (vect_enhance_data_refs_alignment): Likewise. Use
12890 vect_get_peeling_costs_all_drs to compute the penalty for no
12891 peeling to match up costs.
12893 2017-07-21 Richard Biener <rguenther@suse.de>
12895 PR tree-optimization/81500
12896 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12897 we didn't identify a reduction path.
12899 2017-07-21 Tom de Vries <tom@codesourcery.com>
12900 Cesar Philippidis <cesar@codesourcery.com>
12902 PR gcov-profile/81442
12903 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12906 2017-07-21 Tom de Vries <tom@codesourcery.com>
12909 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12911 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12912 nvptx_override_options_after_change.
12914 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12916 * dwarf2out.c (output_file_names): Avoid double testing for
12917 dwarf_version >= 5.
12919 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12921 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12923 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12925 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12927 (try_crossjump_to_edge): Do not punt on partitioned functions.
12929 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12931 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12932 Put all BBs reachable only via paths crossing cold region to cold
12934 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12936 2016-07-21 Richard Biener <rguenther@suse.de>
12938 PR tree-optimization/81303
12939 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12940 into account prologue and epilogue iterations when raising
12941 min_profitable_iters to sth at least covering one vector iteration.
12943 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12945 * config/arm/arm.c (arm_test_cpu_arch_dat):
12948 2017-07-20 Nathan Sidwell <nathan@acm.org>
12950 Remove TYPE_METHODS.
12951 * tree.h (TYPE_METHODS): Delete.
12952 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12953 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12954 (dbxout_type_methods): Scan TYPE_FIELDS.
12955 (dbxout_type): Don't check TYPE_METHODS here.
12956 * function.c (use_register_for_decl): Always ignore register for
12957 class types when not optimizing.
12958 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12959 * tree.c (free_lang_data_in_type): Stitch out member functions and
12960 templates from TYPE_FIELDS.
12961 (build_distinct_type_copy, verify_type_variant,
12962 verify_type): Member fns are on TYPE_FIELDS.
12963 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12964 * tree-pretty-print.c (dump_generic_node): Likewise.
12966 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12969 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12970 V2TImode and V4TImode.
12971 (ix86_expand_vector_extract): Likewise.
12972 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12973 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12974 (ssescalarmode): Handle V4TImode and V2TImode.
12975 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12976 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12977 (VEXTRACTI128_MODE): New mode iterator.
12978 (splitter for *vec_extractv?ti first element): New.
12979 (VEC_INIT_MODE): New mode iterator.
12980 (vec_init<mode>): Consolidate 3 expanders into one using
12981 VEC_INIT_MODE mode iterator.
12983 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12985 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12986 non_spilled_static_chain_regno_p.
12988 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12990 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12992 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12994 * bb-reorder.c (connect_traces): Allow copying of blocks within
12997 2017-07-20 Richard Biener <rguenther@suse.de>
12999 * gimple.h (gimple_phi_result): Add gphi * overload.
13000 (gimple_phi_result_ptr): Likewise.
13001 (gimple_phi_arg): Likewise. Adjust index assert to only
13002 allow actual argument accesses rather than all slots available
13004 (gimple_phi_arg_def): Add gphi * overload.
13005 * tree-phinodes.c (make_phi_node): Initialize only actual
13007 (resize_phi_node): Clear memory not covered by old node,
13008 do not initialize excess argument slots.
13009 (reserve_phi_args_for_new_edge): Initialize new argument slot
13012 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13014 PR tree-optimization/81388
13016 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13018 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13019 by removing computation of may_be_zero.
13021 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13022 Tom de Vries <tom@codesourcery.com>
13024 PR middle-end/81030
13025 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13026 when gimple level profile disagrees with what RTL expander did.
13028 2017-07-20 Richard Biener <rguenther@suse.de>
13030 PR tree-optimization/61171
13031 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13032 (vect_analyze_stmt): Add slp instance parameter.
13033 (vectorizable_reduction): Likewise.
13034 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13035 (vect_is_simple_reduction): Deal with chains not detected
13036 as SLP reduction chain, specifically not properly associated
13037 chains containing a mix of plus/minus.
13038 (get_reduction_op): Remove.
13039 (get_initial_defs_for_reduction): Simplify, pass in whether
13040 this is a reduction chain, pass in the SLP node for the PHIs.
13041 (vect_create_epilog_for_reduction): Get the SLP instance as
13043 (vectorizable_reduction): Get the SLP instance as arg.
13044 During analysis remember the SLP node with the PHIs in the
13045 instance. Simplify getting at the vectorized reduction PHIs.
13046 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13047 through SLP instance.
13048 (vect_slp_analyze_operations): Likewise.
13049 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13050 (vect_transform_stmt): Likewise.
13052 2017-07-20 Tom de Vries <tom@codesourcery.com>
13054 PR tree-optimization/81489
13055 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13056 read of phi arg location to before loop that modifies phi.
13058 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13060 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13063 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13065 PR middle-end/81331
13066 * except.c (execute): Fix ordering issue.
13068 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13070 PR rtl-optimization/81423
13071 * combine.c (make_compound_operation_int): Don't try to optimize
13072 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13074 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13076 PR rtl-optimization/81423
13077 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13078 with a constant that is -1 in the truncated to mode.
13080 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13082 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13083 (determine_unlikely_bbs): ... here.
13084 * predict.h (propagate_unlikely_bbs_forward): Declare.
13085 * cfgexpand.c (pass_expand::execute): Use it.
13086 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13088 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13089 propagate_unlikely_bbs_forward.
13091 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13093 PR middle-end/81331
13094 * except.c (maybe_add_nop_after_section_switch): New function.
13097 2017-07-19 Tom de Vries <tom@codesourcery.com>
13099 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13101 2017-07-19 Tom de Vries <tom@codesourcery.com>
13103 * gimple.h (gimple_phi_arg): Make assert more strict.
13105 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13107 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13108 * config/rs6000/mmintrin.h: New file.
13109 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13111 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13113 PR tree-optimization/81346
13114 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13116 2017-07-19 Tom de Vries <tom@codesourcery.com>
13118 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13120 2017-07-19 Tom de Vries <tom@codesourcery.com>
13122 * config/nvptx/nvptx-modes.def: Add V2DImode.
13123 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13124 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13125 (nvptx_output_mov_insn): Handle lack of mov.b128.
13126 (nvptx_print_operand): Handle 'H' and 'L' codes.
13127 (nvptx_vector_mode_supported): Allow V2DImode.
13128 (nvptx_preferred_simd_mode): New function.
13129 (nvptx_data_alignment): New function.
13130 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13131 nvptx_preferred_simd_mode.
13132 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13134 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13136 2017-07-19 Tom de Vries <tom@codesourcery.com>
13138 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13139 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13140 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13141 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13142 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13143 (mov<VECIM>_insn): New define_insn.
13144 (define_expand "mov<VECIM>): New define_expand.
13146 2017-07-19 Tom de Vries <tom@codesourcery.com>
13148 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13150 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13152 PR tree-optimization/81346
13153 * fold-const.h (fold_div_compare, range_check_type): Declare.
13154 * fold-const.c (range_check_type): New function.
13155 (build_range_check): Use range_check_type.
13156 (fold_div_compare): No longer static, rewritten into
13157 a match.pd helper function.
13158 (fold_comparison): Don't call fold_div_compare here.
13159 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13160 as helper function.
13162 2017-07-19 Nathan Sidwell <nathan@acm.org>
13164 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13165 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13166 * tree.c (find_decls_types_r, verify_type): Use
13167 TYPE_{MIN,MAX}_VALUE_RAW.
13168 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13169 (hash_tree): Likewise.
13170 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13172 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13175 2017-07-18 Tom de Vries <tom@codesourcery.com>
13177 PR middle-end/81464
13178 * omp-expand.c (expand_omp_for_static_chunk): Handle
13179 equal-argument loop exit phi.
13181 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13184 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13185 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13186 operand 2 predicate.
13187 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13188 operand 2 predicate.
13189 (ror,rol -> rorx splitters): Use const_int_operand as
13190 operand 2 predicate.
13192 2017-06-18 Richard Biener <rguenther@suse.de>
13194 PR tree-optimization/81410
13195 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13196 the gap in the ! slp_perm SLP case after each group.
13198 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13200 PR middle-end/81463
13201 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13204 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13206 PR middle-end/81462
13207 * predict.c (set_even_probabilities): Cleanup; do not affect
13208 probabilities that are already known.
13209 (combine_predictions_for_bb): Call even when count is set.
13211 2017-07-18 Nathan Sidwell <nathan@acm.org>
13213 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13216 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13219 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13220 optimization for loop niter analysis.
13222 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13225 * config/avr/avr.c (avr_optimize_casesi): Don't use
13226 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13228 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13230 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13231 body_cost_vec from _vect_peel_extended_info.
13232 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13233 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13236 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13238 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13240 2017-07-18 Richard Biener <rguenther@suse.de>
13242 PR tree-optimization/80620
13243 PR tree-optimization/81403
13244 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13245 info when re-using a VN table entry.
13247 2017-07-18 Richard Biener <rguenther@suse.de>
13249 PR tree-optimization/81418
13250 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13251 vectype_in. Verify that with lane-reducing reduction operations
13252 we have a single def-use cycle.
13254 2017-07-17 Carl Love <cel@us.ibm.com>
13256 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13258 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13259 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13260 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13261 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13262 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13263 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13264 VMULOSW): New enum "unspec" values.
13265 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13266 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13267 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13268 altivec_vmulosw): New patterns.
13269 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13270 VMULOSW): Add definitions.
13271 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13273 * config/alpha/alpha.c: Include predict.h.
13275 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13277 * tree-vrp.c (compare_assert_loc): Fix comparison function
13278 to return predictable results.
13280 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13282 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13284 (subdi3): Likewise.
13285 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13286 * doc/invoke.texi (mexpand-adddi): Update text.
13288 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13290 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13291 that also clobbers the CC register. The old expand code is moved
13293 (*arc_clzsi2): ... here.
13294 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13295 the CC register. The old expand code is moved to ...
13296 (arc_ctzsi2): ... here.
13298 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13300 * config/arc/arc.opt (mindexed-loads): Use initial value
13301 TARGET_INDEXED_LOADS_DEFAULT.
13302 (mauto-modify-reg): Use initial value
13303 TARGET_AUTO_MODIFY_REG_DEFAULT.
13304 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13305 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13306 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13307 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13309 2017-07-17 Martin Liska <mliska@suse.cz>
13312 * opts.c (finish_options): Do not allow -fgnu-tm
13313 w/ -fsanitize={kernel-,}address. Say sorry.
13315 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13318 * tree-loop-distribution.c (classify_partition): Only assert on
13319 numer of iterations.
13320 (merge_dep_scc_partitions): Delete prameter. Update function call.
13321 (distribute_loop): Remove code handling loop with unknown niters.
13322 (pass_loop_distribution::execute): Skip loop with unknown niters.
13324 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13327 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13328 function sort_partitions_by_post_order.
13330 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13332 PR tree-optimization/81374
13333 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13334 the max index of basic blocks, rather than number of basic blocks.
13336 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13338 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13340 (arc_legitimate_pic_operand_p): Likewise.
13341 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13343 (arc_needs_pcl_p): Likewise.
13344 (arc_legitimate_pc_offset_p): Likewise.
13345 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13346 function is also used in constrains.md.
13347 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13348 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13349 PLUS. Only return true/false in known cases, otherwise assert.
13350 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13351 is already called in arc_legitimate_constant_p.
13352 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13354 (LEGITIMATE_PIC_OPERAND_P): Use
13355 arc_raw_symbolic_reference_mentioned_p function.
13356 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13361 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13362 Andrew Burgess <andrew.burgess@embecosm.com>
13364 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13365 (arc_return_address_register): New function.
13366 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13367 (arc_handle_fndecl_attribute): Add naked attribute.
13368 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13369 (TARGET_WARN_FUNC_RETURN): Likewise.
13370 (arc_allocate_stack_slots_for_args): New function.
13371 (arc_warn_func_return): Likewise.
13372 (machine_function): Change type fn_type.
13373 (arc_compute_function_type): Consider new naked function type,
13374 change function return type.
13375 (arc_must_save_register): Adapt to handle new
13376 arc_compute_function_type's return type.
13377 (arc_expand_prologue): Likewise.
13378 (arc_expand_epilogue): Likewise.
13379 (arc_return_address_regs): Delete.
13380 (arc_return_address_register): New function.
13381 (arc_epilogue_uses): Use above function.
13382 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13383 (arc_function_type): Change encoding, add naked type.
13384 (ARC_INTERRUPT_P): Change to handle the new encoding.
13385 (ARC_FAST_INTERRUPT_P): Likewise.
13386 (ARC_NORMAL_P): Define.
13387 (ARC_NAKED_P): Likewise.
13388 (arc_compute_function_type): Delete prototype.
13389 * config/arc/arc.md (in_ret_delay_slot): Use
13390 arc_return_address_register function.
13391 (simple_return): Likewise.
13392 (p_return_i): Likewise.
13394 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13396 PR tree-optimization/81428
13397 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13398 can't be built for those types.
13400 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13402 Remove stuff dead since r239246.
13404 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13405 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13406 (avr_inform_devices): Remove dead stuff.
13408 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13410 * config/arm/arm_neon.h: Fix softp typo.
13412 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13414 PR tree-optimization/81365
13415 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13416 aggregate moves onto bb predecessor edges, make sure there are no
13417 loads that could alias the lhs in between the start of bb and the
13420 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13423 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13424 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13425 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13427 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13429 PR tree-optimization/81396
13430 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13431 (init_symbolic_number): Initialize it to 1.
13432 (perform_symbolic_merge): Add n_ops from both operands into the new
13434 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13435 without base_addr as useless if they need more than one operation.
13436 (bswap_replace): Handle !bswap case for NULL base_addr.
13438 2017-07-17 Tom de Vries <tom@codesourcery.com>
13441 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13444 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13446 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13447 conditional builtin define __FIX_LEON3FT_B2BST.
13449 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13451 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13452 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13455 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13457 PR rtl-optimization/81424
13458 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13459 to remove potential trapping from operands if -fnon-call-exceptions.
13461 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13463 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13464 profile_proability for scalling.
13465 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13467 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13469 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13471 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13473 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13474 fixpoint arithmetics.
13476 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13478 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13479 fixpoint arithmetics.
13481 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13483 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13484 fixpoint arithmetics.
13486 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13488 * profile-count.h (profile_probability::from_reg_br_prob_note,
13489 profile_probability::to_reg_br_prob_note): New functions.
13490 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13491 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13492 * predict.c (probability_reliable_p): Update.
13493 (edge_probability_reliable_p): Update.
13494 (br_prob_note_reliable_p): Update.
13495 (invert_br_probabilities): Update.
13496 (add_reg_br_prob_note): New function.
13497 (combine_predictions_for_insn): Update.
13498 * asan.c (asan_clear_shadow): Update.
13499 * cfgbuild.c (compute_outgoing_frequencies): Update.
13500 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13501 (update_br_prob_note): Update.
13502 (rtl_verify_edges): Update.
13503 (purge_dead_edges): Update.
13504 (fixup_reorder_chain): Update.
13505 * emit-rtl.c (try_split): Update.
13506 * ifcvt.c (cond_exec_process_insns): Update.
13507 (cond_exec_process_if_block): Update.
13508 (dead_or_predicable): Update.
13509 * internal-fn.c (expand_addsub_overflow): Update.
13510 (expand_neg_overflow): Update.
13511 (expand_mul_overflow): Update.
13512 * loop-doloop.c (doloop_modify): Update.
13513 * loop-unroll.c (compare_and_jump_seq): Update.
13514 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13515 * predict.h: Update.
13516 * reorg.c (mostly_true_jump): Update.
13518 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13519 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13520 * config/arc/arc.c: (emit_unlikely_jump): Update.
13521 * config/arm/arm.c: (emit_unlikely_jump): Update.
13522 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13523 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13524 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13525 (ix86_print_operand): Update.
13526 (ix86_split_fp_branch): Update.
13527 (predict_jump): Update.
13528 * config/ia64/ia64.c (ia64_print_operand): Update.
13529 * config/mmix/mmix.c (mmix_print_operand): Update.
13530 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13531 (rs6000_expand_split_stack_prologue): Update.
13532 * config/rs6000/rs6000.c: Update.
13533 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13534 (s390_expand_vec_movstr): Update.
13535 (s390_expand_cs_tdsi): Update.
13536 (s390_expand_split_stack_prologue): Update.
13537 * config/sh/sh.c (sh_print_operand): Update.
13538 (expand_cbranchsi4): Update.
13539 (expand_cbranchdi4): Update.
13540 * config/sparc/sparc.c (output_v9branch): Update.
13541 * config/spu/spu.c (get_branch_target): Update.
13542 (ea_load_store_inline): Update.
13543 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13544 * config/tilepro/tilepro.c: Update.
13546 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13548 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13549 (gimplify_save_expr): Likewise.
13551 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13553 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13555 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13557 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13559 * timevar.def (TV_IPA_FNSUMMARY): Define.
13561 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13563 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13564 to back store errata sensitive sequence from being generated.
13565 (sqrtdf2_fix): Likewise.
13567 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13569 * tree-ssa-threadupdate.c (compute_path_counts,
13570 update_joiner_offpath_counts): Use profile_probability.
13572 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13575 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13577 * config/arm/arm-c.c (arm_cpu_builtins): Define
13578 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13580 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13582 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13583 array entries to represent __ieee128 versions of the
13584 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13585 scalar_extract_sig, and scalar_insert_exp built-in functions.
13586 (altivec_resolve_overloaded_builtin): Add special case handling
13587 for the __builtin_scalar_insert_exp function, as represented by
13588 the P9V_BUILTIN_VEC_VSIEDP constant.
13589 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13590 exponent support for __ieee128 argument.
13591 (VSESQP): Add scalar extract signature support for __ieee128
13593 (VSTDCNQP): Add scalar test negative support for __ieee128
13595 (VSIEQP): Add scalar insert exponent support for __int128 argument
13596 with __ieee128 result.
13597 (VSIEQPF): Add scalar insert exponent support for __ieee128
13598 argument with __ieee128 result.
13599 (VSTDCQP): Add scalar test data class support for __ieee128
13601 (VSTDCNQP): Add overload support for scalar test negative with
13602 __ieee128 argument.
13603 (VSTDCQP): Add overload support for scalar test data class
13604 __ieee128 argument.
13605 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13606 UNSPEC_VSX_SXSIGDP.
13607 (UNSPEC_VSX_SIEXPQP): New constant.
13608 (xsxexpqp): New insn for VSX scalar extract exponent quad
13610 (xsxsigqp): New insn for VSX scalar extract significand quad
13612 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13613 precision with floating point argument.
13614 (xststdcqp): New expand for VSX scalar test data class quad
13616 (xststdcnegqp): New expand for VSX scalar test negative quad
13618 (xststdcqp): New insn to match expansions for VSX scalar test data
13619 class quad precision and VSX scalar test negative quad precision.
13620 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13621 special case operand checking to enforce that second operand of
13622 VSX scalar test data class with quad precision argument is a 7-bit
13624 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13625 prototypes and descriptions of __ieee128 versions of
13626 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13627 scalar_test_data_class, and scalar_test_neg built-in functions.
13629 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13631 PR tree-optimization/81162
13632 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13633 replace a negate with an add.
13635 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13637 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13639 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13641 * config/arm/arm-c.c (arm_cpu_builtins): Define
13642 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13644 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13646 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13647 (armv8-r): Set ARM Cortex-R52 as default CPU.
13648 * config/arm/arm-tables.opt: Regenerate.
13649 * config/arm/arm-tune.md: Regenerate.
13650 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13652 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13653 extension for -mcpu=cortex-r52.
13655 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13657 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13658 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13659 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13660 (fp-armv8): Define it as FP_ARMv8 only.
13661 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13662 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13664 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13665 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13666 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13667 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13668 than TARGET_FPU_ARMV8.
13669 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13670 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13671 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13673 * config/arm/neon.md (neon_vrint): Likewise.
13674 (neon_vcvt): Likewise.
13675 (neon_<fmaxmin_op><mode>): Likewise.
13676 (<fmaxmin><mode>3): Likewise.
13677 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13678 * config/arm/predicates.md (arm_cond_move_operator): Check against
13679 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13681 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13683 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13684 to top of function.
13686 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13688 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13689 loop in comment with memset.
13691 2017-07-14 Martin Liska <mliska@suse.cz>
13693 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13694 * dwarf2out.c (is_java): Remove the function.
13695 (output_pubname): Remove usage of the function.
13696 (lower_bound_default): Remove usage of DW_LANG_Java.
13697 (gen_compile_unit_die): Likewise.
13698 * gcc.c: Remove compiler defaults for .java and .zip files.
13699 * gimple-expr.c (remove_suffix): Change as there's no longer
13700 extension than 4-letter one.
13701 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13702 (gimplify_save_expr): Likewise.
13703 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13704 as it's possible even for other languages than Java.
13705 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13706 * lto-opts.c (lto_write_options): Remove reference to Java.
13707 * opts.c (strip_off_ending): Update file extension handling.
13708 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13709 * tree-eh.c (lower_resx): Likewise.
13710 * tree.c (free_lang_data_in_type): Remove dead code.
13711 (find_decls_types_r): Likewise.
13712 (build_common_builtin_nodes): Remove Java from a comment.
13713 (verify_type): Remove dead code.
13714 * varasm.c (assemble_external): Remove Java from a comment.
13716 2017-07-14 Martin Liska <mliska@suse.cz>
13718 * opts.c (finish_options): Add quotes.
13719 (common_handle_option): Likewise.
13721 2017-07-14 Martin Liska <mliska@suse.cz>
13723 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13724 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13725 Remove N_SO_PASCAL.
13726 * dwarf2out.c (lower_bound_default): Do not handle
13728 (gen_compile_unit_die): Likewise.
13729 * gcc.c: Remove default extension binding for GNU Pascal.
13730 * stmt.c: Remove Pascal language from a comment.
13731 * xcoffout.c: Likewise.
13733 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13736 * diagnostic-show-locus.c (fixit_cmp): New function.
13737 (layout::layout): Sort m_fixit_hints.
13738 (column_range::column_range): Assert that the values are valid.
13739 (struct char_span): New struct.
13740 (correction::overwrite): New method.
13741 (struct source_line): New struct.
13742 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13743 calls in terms of classes source_line and char_span, and
13744 correction::overwrite.
13745 (selftest::test_overlapped_fixit_printing_2): New function.
13746 (selftest::diagnostic_show_locus_c_tests): Call it.
13748 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13751 early if there is no lhs.
13753 2017-07-13 Martin Liska <mliska@suse.cz>
13755 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13756 (gen_reference_type_die): Likewise.
13757 * stor-layout.c: Remove Pascal-related comment.
13759 2017-07-13 Martin Liska <mliska@suse.cz>
13761 * opts.c (finish_options): Add quotes to error messages.
13762 (parse_sanitizer_options): Likewise.
13764 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13766 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13768 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13770 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13772 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13774 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13776 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13778 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13781 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13782 provides the hardware capability bits, define the macro
13783 __BUILTIN_CPU_SUPPORTS__.
13784 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13785 if GLIBC does not provide the hardware capability bits. Add a
13786 gcc_unreachable call if the built-in cpu function is neither
13787 __builtin_cpu_is nor __builtin_cpu_supports.
13788 (rs6000_get_function_versions_dispatcher): Change the warning
13789 that an old GLIBC is used which does not export the capability
13790 bits to be an error.
13791 * doc/extend.texi (target_clones attribute): Document the
13792 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13793 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13794 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13795 the macros defined by GCC if the newer GLIBC is available.
13797 2017-07-12 Jeff Law <law@redhat.com>
13799 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13800 remaining includes slightly.
13801 * config/riscv/riscv-builtins.c: Include profile-count.h.
13803 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13806 * config/avr/avr.c (avr_set_current_function): In diagnostic
13807 messages: Quote keywords and (parts of) identifiers.
13808 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13811 2017-07-12 Carl Love <cel@us.ibm.com>
13813 * config/rs6000/rs6000-c.c: Add support for built-in functions
13814 vector bool char vec_revb (vector bool char);
13815 vector bool short vec_revb (vector short char);
13816 vector bool int vec_revb (vector bool int);
13817 vector bool long long vec_revb (vector bool long long);
13818 * doc/extend.texi: Update the built-in documentation file for the
13819 new built-in functions.
13821 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13823 * config/s390/s390.md: Remove movcc splitter.
13825 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13827 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13828 load/store on condition.
13830 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13833 * config/avr/avr.c (avr_encode_section_info)
13834 [progmem && !TREE_READONLY]: Error if progmem object needs
13837 2017-07-11 Michael Collison <michael.collison@arm.com>
13839 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13842 2017-07-11 Carl Love <cel@us.ibm.com>
13844 * config/rs6000/rs6000-c.c: Add support for builtins
13845 vector unsigned int vec_parity_lsbb (vector signed int);
13846 vector unsigned int vec_parity_lsbb (vector unsigned int);
13847 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13848 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13849 vector unsigned long long vec_parity_lsbb (vector signed long long);
13850 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13851 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13852 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13853 * doc/extend.texi: Update the built-in documentation file for the
13854 new built-in functions.
13856 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13858 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13859 (layout::m_primary_loc): New field.
13860 (layout::layout): Initialize new field. Move location filtering
13861 logic from here to...
13862 (layout::maybe_add_location_range): ...this new method. Add
13863 support for filtering to just the lines already specified by other
13865 (layout::will_show_line_p): New method.
13866 (gcc_rich_location::add_location_if_nearby): New method.
13867 (selftest::test_add_location_if_nearby): New test function.
13868 (selftest::diagnostic_show_locus_c_tests): Call it.
13869 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13872 2017-07-11 Tom de Vries <tom@codesourcery.com>
13874 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13875 (bb_first_real_insn): New function.
13876 (nvptx_single): Add extra initialization of broadcasted condition
13879 2017-07-11 Nathan Sidwell <nathan@acm.org>
13881 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13883 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13885 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13886 Binutils doc as TEXI will mess them up.
13887 * doc/invoke.texi (AVR Options): Same here.
13889 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13891 * config/sparc/sparc.opt (mfix-ut700): New option.
13892 (mfix-gr712rc): Likewise.
13893 (sparc_fix_b2bst): New variable.
13894 * doc/invoke.texi (SPARC options): Document them.
13895 (ARM options): Fix warnings.
13896 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13897 instructions to prevent sequences that can trigger the store-store
13898 errata for certain LEON3FT processors.
13899 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13900 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13901 * config/sparc/sparc.md (fix_b2bst): New attribute.
13902 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13904 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13907 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13909 (*rsqrtsf2_sse): Ditto.
13911 (div<mode>3): Macroize insn from divdf3 and divsf3
13912 using MODEF mode iterator.
13914 2017-07-10 Martin Sebor <msebor@redhat.com>
13916 PR tree-optimization/80397
13917 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13918 instead of testing for equality to INTEGER_TYPE.
13920 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13922 * config.gcc: Remove uclibc from arc target spec.
13924 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13926 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13928 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13931 * lto-wrapper.c (remove_option): New function.
13932 (merge_and_complain): Merge PIC/PIE options more realistically.
13934 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13936 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13940 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13941 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13942 * config.in: Regenerate.
13943 * configure: Regenerate.
13944 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13945 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13946 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13947 (TARGET_GASISR_PROLOGUES): ...target mask.
13948 * common/config/avr/avr-common.c
13949 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13950 Set -mgas-isr-prologues.
13951 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13952 INSERT_PASS_BEFORE for it.
13953 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13954 * config/avr/avr.c (avr_option_override)
13955 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13956 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13957 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13958 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13959 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13961 (make_avr_pass_pre_proep): New function.
13962 (emit_push_sfr) <treg>: Add argument to function and use it
13963 instead of TMP_REG.
13964 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13965 and set machine->gasisr.yes.
13966 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13967 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13968 __gcc_isr.n_pushed to .L__stack_usage.
13969 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13970 (avr_asm_final_postscan_insn): ...this new static function.
13971 * config/avr/avr.h (machine_function)
13972 <is_no_gccisr, use_L__stack_usage>: New fields.
13973 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13974 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13975 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13976 (gasisr, *gasisr): New expander and insn.
13977 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13978 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13979 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13981 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13983 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13986 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13988 Move jump-tables out of .text again.
13991 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13992 (ASM_OUTPUT_ADDR_VEC): New function.
13993 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13994 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13995 INSN_ADDRESSes as asm comment.
13996 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13997 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13998 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13999 * config/avr/avr.md (*tablejump): Adjust comment.
14000 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14001 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14003 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14004 (avr_output_addr_vec): New proto.
14005 (avr_log_t) <insn_addresses>: New field.
14007 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14010 * config/i386/i386.c (ix86_function_arg_advance): Set
14011 outgoing_args_on_stack to true if there are outgoing arguments
14013 (ix86_function_arg): Likewise.
14014 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14015 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14016 * config/i386/i386.h (machine_function): Add
14017 outgoing_args_on_stack.
14019 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14021 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14022 supporting pthreds.
14023 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14025 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14027 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14028 (REAL_H): Remove $(MACHMODE_H).
14029 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14031 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14032 $(MACHMODE_H) and double-int.h.
14033 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14035 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14038 2017-07-07 Andrew Pinski <apinski@cavium.com>
14040 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14041 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14043 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14045 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14046 Add warning if GCC was not configured to link against a GLIBC that
14047 exports the hardware capability bits.
14048 (make_resolver_func): Make resolver function private and not a
14049 COMDAT function. Create the name with clone_function_name instead
14050 of make_unique_name.
14053 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14054 correct operand in doing the split.
14056 2017-07-07 Carl Love <cel@us.ibm.com>
14058 * config/rs6000/rs6000-c: Add support for built-in function
14059 vector unsigned short vec_pack_to_short_fp32 (vector float,
14061 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14062 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14063 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14064 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14065 (convert_4f32_8i16): Add define_expand.
14066 * doc/extend.texi: Update the built-in documentation file for the
14067 new built-in function.
14069 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14071 * config/sparc/m8.md: New file.
14072 * config/sparc/sparc.md: Include m8.md.
14074 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14076 * config/sparc/sparc.opt: New option -mvis4b.
14077 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14078 (sparc_option_override): Handle VIS4B.
14079 (enum sparc_builtins): Define
14080 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14081 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14082 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14083 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14084 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14085 (check_constant_argument): New function.
14086 (sparc_vis_init_builtins): Define builtins
14087 __builtin_vis_dictunpack{8,16,32},
14088 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14089 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14090 __builtin_vis_fpcmpde{8,16,32}shl and
14091 __builtin_vis_fpcmpur{8,16,32}shl.
14092 (sparc_expand_builtin): Check that the constant operands to
14093 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14094 constant and in range.
14095 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14097 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14098 (SPARC_IMM5_P): Likewise.
14099 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14100 (enabled): Handle vis4b.
14101 (UNSPEC_DICTUNPACK): New unspec.
14102 (UNSPEC_FPCMPSHL): Likewise.
14103 (UNSPEC_FPUCMPSHL): Likewise.
14104 (UNSPEC_FPCMPDESHL): Likewise.
14105 (UNSPEC_FPCMPURSHL): Likewise.
14106 (cpu_feature): New CPU feature `vis4b'.
14107 (dictunpack{8,16,32}): New insns.
14108 (FPCSMODE): New mode iterator.
14109 (fpcscond): New code iterator.
14110 (fpcsucond): Likewise.
14111 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14112 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14113 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14114 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14115 * config/sparc/constraints.md: Define constraints `q' for unsigned
14116 2-bit integer constants and `t' for unsigned 5-bit integer
14118 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14120 (imm5_operand_dictunpack16): Likewise.
14121 (imm5_operand_dictunpack32): Likewise.
14122 (imm2_operand): Likewise.
14123 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14124 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14125 ditunpack* and fpcmp*shl builtins.
14127 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14129 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14130 * config.in: Add HAVE_AS_SPARC6 define.
14131 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14133 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14135 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14136 (CPP_CPU_SPEC): Handle m8.
14137 (ASM_CPU_SPEC): Likewise.
14138 * config/sparc/sparc-opts.h (enum processor_type): Add
14140 * config/sparc/sparc.c (m8_costs): New struct.
14141 (sparc_option_override): Handle TARGET_CPU_m8.
14142 (sparc32_initialize_trampoline): Likewise.
14143 (sparc64_initialize_trampoline): Likewise.
14144 (sparc_issue_rate): Likewise.
14145 (sparc_register_move_cost): Likewise.
14146 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14147 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14148 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14149 (CPP_CPU_SPEC): Handle M8.
14150 (ASM_CPU_SPEC): Likewise.
14151 (AS_M8_FLAG): Define.
14152 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14153 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14154 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14156 * configure: Regenerate.
14157 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14160 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14162 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14164 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14165 ("*movdi_insn_sp32"): Do not set v3pipe.
14166 ("*movsi_insn"): Likewise.
14167 ("*movdi_insn_sp64"): Likewise.
14168 ("*movsf_insn"): Likewise.
14169 ("*movdf_insn_sp32"): Likewise.
14170 ("*movdf_insn_sp64"): Likewise.
14171 ("*zero_extendsidi2_insn_sp64"): Likewise.
14172 ("*sign_extendsidi2_insn"): Likewise.
14173 ("*mov<VM32:mode>_insn"): Likewise.
14174 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14175 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14176 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14177 ("<vlop:code><VL:mode>3"): Likewise.
14178 ("*not_<vlop:code><VL:mode>3"): Likewise.
14179 ("*nand<VL:mode>_vis"): Likewise.
14180 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14181 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14182 ("one_cmpl<VL:mode>2"): Likewise.
14183 ("faligndata<VM64:mode>_vis"): Likewise.
14184 ("alignaddrsi_vis"): Likewise.
14185 ("alignaddrdi_vis"): Likweise.
14186 ("alignaddrlsi_vis"): Likewise.
14187 ("alignaddrldi_vis"): Likewise.
14188 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14189 ("bmaskdi_vis"): Likewise.
14190 ("bmasksi_vis"): Likewise.
14191 ("bshuffle<VM64:mode>_vis"): Likewise.
14192 ("cmask8<P:mode>_vis"): Likewise.
14193 ("cmask16<P:mode>_vis"): Likewise.
14194 ("cmask32<P:mode>_vis"): Likewise.
14195 ("pdistn<P:mode>_vis"): Likewise.
14196 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14198 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14200 * config/sparc/sparc.md ("subtype"): New insn attribute.
14201 ("*wrgsr_sp64"): Set insn subtype.
14202 ("*rdgsr_sp64"): Likewise.
14203 ("alignaddrsi_vis"): Likewise.
14204 ("alignaddrdi_vis"): Likewise.
14205 ("alignaddrlsi_vis"): Likewise.
14206 ("alignaddrldi_vis"): Likewise.
14207 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14208 ("fexpand_vis"): Likewise.
14209 ("fpmerge_vis"): Likewise.
14210 ("faligndata<VM64:mode>_vis"): Likewise.
14211 ("bshuffle<VM64:mode>_vis"): Likewise.
14212 ("cmask8<P:mode>_vis"): Likewise.
14213 ("cmask16<P:mode>_vis"): Likewise.
14214 ("cmask32<P:mode>_vis"): Likewise.
14215 ("fchksm16_vis"): Likewise.
14216 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14217 ("fmean16_vis"): Likewise.
14218 ("fp<plusminus_insn>64_vis"): Likewise.
14219 ("<plusminus_insn>v8qi3"): Likewise.
14220 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14221 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14222 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14223 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14224 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14225 ("*movqi_insn"): Likewise.
14226 ("*movhi_insn"): Likewise.
14227 ("*movsi_insn"): Likewise.
14228 ("movsi_pic_gotdata_op"): Likewise.
14229 ("*movdi_insn_sp32"): Likewise.
14230 ("*movdi_insn_sp64"): Likewise.
14231 ("movdi_pic_gotdata_op"): Likewise.
14232 ("*movsf_insn"): Likewise.
14233 ("*movdf_insn_sp32"): Likewise.
14234 ("*movdf_insn_sp64"): Likewise.
14235 ("*zero_extendhisi2_insn"): Likewise.
14236 ("*zero_extendqihi2_insn"): Likewise.
14237 ("*zero_extendqisi2_insn"): Likewise.
14238 ("*zero_extendqidi2_insn"): Likewise.
14239 ("*zero_extendhidi2_insn"): Likewise.
14240 ("*zero_extendsidi2_insn_sp64"): Likewise.
14241 ("ldfsr"): Likewise.
14242 ("prefetch_64"): Likewise.
14243 ("prefetch_32"): Likewise.
14244 ("tie_ld32"): Likewise.
14245 ("tie_ld64"): Likewise.
14246 ("*tldo_ldub_sp32"): Likewise.
14247 ("*tldo_ldub1_sp32"): Likewise.
14248 ("*tldo_ldub2_sp32"): Likewise.
14249 ("*tldo_ldub_sp64"): Likewise.
14250 ("*tldo_ldub1_sp64"): Likewise.
14251 ("*tldo_ldub2_sp64"): Likewise.
14252 ("*tldo_ldub3_sp64"): Likewise.
14253 ("*tldo_lduh_sp32"): Likewise.
14254 ("*tldo_lduh1_sp32"): Likewise.
14255 ("*tldo_lduh_sp64"): Likewise.
14256 ("*tldo_lduh1_sp64"): Likewise.
14257 ("*tldo_lduh2_sp64"): Likewise.
14258 ("*tldo_lduw_sp32"): Likewise.
14259 ("*tldo_lduw_sp64"): Likewise.
14260 ("*tldo_lduw1_sp64"): Likewise.
14261 ("*tldo_ldx_sp64"): Likewise.
14262 ("*mov<VM32:mode>_insn"): Likewise.
14263 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14264 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14266 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14268 * config/sparc/sparc.md ("type"): New insn type viscmp.
14269 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14271 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14272 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14273 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14274 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14276 ("n7_vis_logical_11cycle"): Likewise.
14277 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14278 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14279 * config/sparc/niagara.md ("niag_vis"): Likewise.
14280 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14281 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14283 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14285 * config/sparc/sparc.md: New instruction type `bmask'.
14286 (bmaskdi_vis): Use the `bmask' type.
14287 (bmasksi_vis): Likewise.
14288 * config/sparc/ultra3.md (us3_array): Likewise.
14289 * config/sparc/niagara7.md (n7_array): Likewise.
14290 * config/sparc/niagara4.md (n4_array): Likewise.
14291 * config/sparc/niagara2.md (niag2_vis): Likewise.
14292 (niag3_vis): Likewise.
14293 * config/sparc/niagara.md (niag_vis): Likewise.
14295 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14297 * ipa-comdats.c: Remove optimize check from gate.
14298 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14299 for functions not optimized.
14300 (ipa_fn_summary_read): Skip optimize check.
14301 (ipa_fn_summary_write): Likewise.
14302 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14304 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14306 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14307 (check_callers): Give up on caller which is not optimized.
14308 (inline_small_functions): Likewise.
14309 (ipa_inline): Do not give up when not optimizing.
14310 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14311 away unoptimizes cdtors.
14312 (whole_program_function_and_variable_visibility): Do
14313 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14314 * ipa.c (process_references): Do not check optimize.
14315 (symbol_table::remove_unreachable_nodes): Update optimize check.
14316 (set_writeonly_bit): Update optimize check.
14317 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14318 (pass_ipa_single_use::gate): Remove.
14320 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14322 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14323 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14324 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14325 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14326 permute_load, permute_store, adjust_extract, adjust_splat,
14327 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14328 replace_swap_with_copy, dump_swap_insn_table,
14329 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14330 recombine_lvx_pattern, recombine_stvx_pattern,
14331 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14332 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14333 to file rs6000-p8swap.c.
14334 * config/rs6000/rs6000-p8swap.c: New file.
14335 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14336 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14337 and rs6000*-*-* targets.
14339 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14341 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14343 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14345 * lto-wrapper.c (merge_and_complain): Do not merge
14346 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14347 fsigned_zeros, ftrapping_math, fwrapv.
14348 (append_compiler_options): Do not track these options.
14349 (append_linker_options): Likewie
14351 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14353 * cgraphunit.c (cgraph_node::finalize_function): When
14354 !flag_toplevel_reorde set no_reorder flag.
14355 (varpool_node::finalize_decl): Likewise.
14356 (symbol_table::compile): Drop no toplevel reorder path.
14358 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14360 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14361 edges; zero probability is not better than uninitialized.
14363 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14365 * asan.h (asan_sanitize_allocas_p): Declare.
14366 * asan.c (asan_sanitize_allocas_p): New function.
14367 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14368 (handle_builtin_alloca): Likewise.
14369 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14370 if !asan_sanitize_allocas_p.
14371 * params.def (asan-instrument-allocas): Add new option.
14372 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14373 * opts.c (common_handle_option): Disable allocas sanitization for
14376 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14378 * asan.c: Include gimple-fold.h.
14379 (get_last_alloca_addr): New function.
14380 (handle_builtin_stackrestore): Likewise.
14381 (handle_builtin_alloca): Likewise.
14382 (asan_emit_allocas_unpoison): Likewise.
14383 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14384 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14385 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14386 (instrument_builtin_call): Pass gimple iterator to
14387 get_mem_refs_of_builtin_call.
14388 (last_alloca_addr): New global.
14389 * asan.h (asan_emit_allocas_unpoison): Declare.
14390 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14391 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14392 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14393 if function calls alloca.
14394 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14395 * gimple-fold.h (replace_call_with_value): Declare.
14396 * internal-fn.c: Include asan.h.
14397 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14398 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14400 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14402 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14403 (C_SELFTEST_FLAGS): New.
14404 (CPP_SELFTEST_FLAGS): New.
14405 (SELFTEST_DEPS): New, from deps of s-selftest.
14406 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14407 (CPP_SELFTEST_DEPS): New.
14408 (selftest): Add dependency on s-selftest-c++.
14409 (s-selftest): Rename to...
14410 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14411 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14412 than SELFTEST_FLAGS.
14413 (selftest-gdb): Rename to...
14414 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14416 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14417 (selftest-valgrind): Rename to...
14418 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14420 (selftest-valgrind): Reintroduce as an alias for
14421 selftest-c-valgrind.
14422 (s-selftest-c++): New.
14423 (selftest-c++-gdb): New.
14424 (selftest-c++-valgrind): New.
14426 2017-07-06 Olivier Hainque <hainque@adacore.com>
14428 * gcc.c (process_command): When deciding if undefined variables
14429 should be ignored when processing specs, accept "gcc -v" as well.
14431 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14433 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14434 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14436 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14438 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14439 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14440 * config/arm/arm-tables.opt: Regenerate.
14441 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14443 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14445 2017-07-06 Carl Love <cel@us.ibm.com>
14447 * ChangeLog: Clean up from mid air collision
14449 2017-07-06 Carl Love <cel@us.ibm.com>
14451 * config/rs6000/rs6000-c.c: Add support for built-in functions
14452 vector signed int vec_subc (vector signed int, vector signed int);
14453 vector signed __int128 vec_subc (vector signed __int128,
14454 vector signed __int128);
14455 vector unsigned __int128 vec_subc (vector unsigned __int128,
14456 vector unsigned __int128);
14457 vector signed int vec_sube (vector signed int, vector signed int,
14458 vector signed int);
14459 vector unsigned int vec_sube (vector unsigned int,
14460 vector unsigned int,
14461 vector unsigned int);
14462 vector signed __int128 vec_sube (vector signed __int128,
14463 vector signed __int128,
14464 vector signed__int128);
14465 vector unsigned __int128 vec_sube (vector unsigned __int128,
14466 vector unsigned __int128,
14467 vector unsigned __int128);
14468 vector signed int vec_subec (vector signed int, vector signed int,
14469 vector signed int);
14470 vector unsigned int vec_subec (vector unsigned int,
14471 vector unsigned int,
14472 vector unsigned int);
14473 vector signed __int128 vec_subec (vector signed __int128,
14474 vector signed __int128,
14475 vector signed__int128);
14476 vector unsigned __int128 vec_subec (vector unsigned __int128,
14477 vector unsigned __int128,
14478 vector unsigned __int128);
14479 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14480 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14481 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14482 BU_ALTIVEC_OVERLOAD_X definitions.
14483 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14484 * doc/extend.texi: Update the built-in documentation file for the new
14485 built-in functions.
14487 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14490 * diagnostic-show-locus.c (layout::layout): Use start and finish
14491 spelling location for the start and finish of each range.
14492 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14493 Add unused aspect param.
14494 * input.c (expand_location_1): Add "aspect" param, and use it
14495 to access the correct part of the location.
14496 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14498 (expand_location_to_spelling_point): Likewise.
14499 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14500 param, and pass it to expand_location_1.
14502 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14504 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14505 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14506 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14507 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14508 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14509 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14510 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14511 _mm_maskz_getmant_ss): New intrinsics.
14512 (__builtin_ia32_getexpss128_mask): Changed to ...
14513 __builtin_ia32_getexpss128_round ... this.
14514 (__builtin_ia32_getexpsd128_mask): Changed to ...
14515 __builtin_ia32_getexpsd128_round ... this.
14516 * config/i386/i386-builtin-types.def
14517 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14518 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14519 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14520 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14521 __builtin_ia32_getmantss_mask_round): New builtins.
14522 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14523 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14524 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14525 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14526 * config/i386/sse.md
14527 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14528 avx512f_sgetexp<mode><mask_scalar_name>
14529 <round_saeonly_scalar_name> ... this.
14530 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14531 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14532 vgetexp<ssescalarmodesuffix>
14533 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14534 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14535 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14536 avx512f_vgetmant<mode><mask_scalar_name>
14537 <round_saeonly_scalar_name> ... this.
14538 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14539 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14540 vgetmant<ssescalarmodesuffix>
14541 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14542 %0<mask_scalar_operand4>, %1, %2
14543 <round_saeonly_scalar_mask_op4>, %3} ... this.
14544 * config/i386/subst.md (mask_scalar_operand4,
14545 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14546 round_saeonly_scalar_nimm_predicate): New subst attributes.
14548 2017-07-06 Julia Koval <julia.koval@intel.com>
14550 * config/i386/i386.c (ix86_erase_embedded_rounding):
14551 Remove code for old rounding pattern.
14553 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14555 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14557 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14559 * doc/sourcebuild.texi (Test Directives, Variants of
14560 dg-require-support): Add documentation for dg-require-stack-check.
14562 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14564 * config/i386/subst.md (mask_scalar, round_scalar,
14565 round_saeonly_scalar): New meta-templates.
14566 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14567 round_scalar_mask_operand3, round_scalar_mask_op3,
14568 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14569 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14570 round_saeonly_scalar_constraint,
14571 round_saeonly_scalar_prefix): New subst attribute.
14572 * config/i386/sse.md
14573 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14574 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14575 <round_scalar_name> ... this.
14576 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14577 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14578 <round_scalar_name> ... this.
14579 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14580 <sse>_vm<code><mode>3<mask_scalar_name>
14581 <round_saeonly_scalar_name> ... this.
14582 (v<plusminus_mnemonic><ssescalarmodesuffix>
14583 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14584 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14585 v<plusminus_mnemonic><ssescalarmodesuffix>
14586 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14587 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14588 (v<multdiv_mnemonic><ssescalarmodesuffix>
14589 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14590 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14591 v<multdiv_mnemonic><ssescalarmodesuffix>
14592 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14593 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14594 (v<maxmin_float><ssescalarmodesuffix>
14595 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14596 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14597 v<maxmin_float><ssescalarmodesuffix>
14598 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14599 %0<mask_scalar_operand3>, %1, %<iptr>2
14600 <round_saeonly_scalar_mask_op3>} ... this.
14602 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14604 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14605 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14607 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14608 Alan Hayward <alan.hayward@arm.com>
14609 David Sherwood <david.sherwood@arm.com>
14611 * combine.c (simplify_if_then_else): Remove "enum" before
14613 * compare-elim.c (can_eliminate_compare): Likewise.
14614 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14616 (aarch64_lookup_simd_builtin_type): Likewise.
14617 (aarch64_simd_builtin_type): Likewise.
14618 (aarch64_init_simd_builtin_types): Likewise.
14619 (aarch64_simd_expand_args): Likewise.
14620 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14622 (aarch64_reverse_mask): Likewise.
14623 (aarch64_simd_emit_reg_reg_move): Likewise.
14624 (aarch64_gen_adjusted_ldpstp): Likewise.
14625 (aarch64_ccmp_mode_to_code): Likewise.
14626 (aarch64_operands_ok_for_ldpstp): Likewise.
14627 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14628 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14630 (aarch64_min_divisions_for_recip_mul): Likewise.
14631 (aarch64_reassociation_width): Likewise.
14632 (aarch64_get_condition_code_1): Likewise.
14633 (aarch64_simd_emit_reg_reg_move): Likewise.
14634 (aarch64_simd_attr_length_rglist): Likewise.
14635 (aarch64_reverse_mask): Likewise.
14636 (aarch64_operands_ok_for_ldpstp): Likewise.
14637 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14638 (aarch64_gen_adjusted_ldpstp): Likewise.
14639 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14641 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14642 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14643 (arm_lookup_simd_builtin_type): Likewise.
14644 (arm_simd_builtin_type): Likewise.
14645 (arm_init_simd_builtin_types): Likewise.
14646 (arm_expand_builtin_args): Likewise.
14647 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14648 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14649 (ft32_setup_incoming_varargs): Likewise.
14650 (ft32_function_arg): Likewise.
14651 (ft32_function_arg_advance): Likewise.
14652 (ft32_pass_by_reference): Likewise.
14653 (ft32_arg_partial_bytes): Likewise.
14654 (ft32_valid_pointer_mode): Likewise.
14655 (ft32_addr_space_pointer_mode): Likewise.
14656 (ft32_addr_space_legitimate_address_p): Likewise.
14657 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14659 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14660 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14661 (iamcu_alignment): Likewise.
14662 (canonicalize_vector_int_perm): Likewise.
14663 (ix86_noce_conversion_profitable_p): Likewise.
14664 (ix86_mpx_bound_mode): Likewise.
14665 (ix86_operands_ok_for_move_multiple): Likewise.
14666 * config/microblaze/microblaze-protos.h
14667 (microblaze_expand_conditional_branch_reg): Likewise.
14668 * config/microblaze/microblaze.c
14669 (microblaze_expand_conditional_branch_reg): Likewise.
14670 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14672 (rs6000_reassociation_width): Likewise.
14673 (rs6000_invalid_binary_op): Likewise.
14674 (fusion_p9_p): Likewise.
14675 (emit_fusion_p9_load): Likewise.
14676 (emit_fusion_p9_store): Likewise.
14677 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14679 (riscv_hard_regno_mode_ok_p): Likewise.
14680 (riscv_address_insns): Likewise.
14681 (riscv_split_symbol): Likewise.
14682 (riscv_legitimize_move): Likewise.
14683 (riscv_function_value): Likewise.
14684 (riscv_hard_regno_nregs): Likewise.
14685 (riscv_expand_builtin): Likewise.
14686 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14687 (riscv_build_integer): Likewise.
14688 (riscv_split_integer): Likewise.
14689 (riscv_legitimate_constant_p): Likewise.
14690 (riscv_cannot_force_const_mem): Likewise.
14691 (riscv_regno_mode_ok_for_base_p): Likewise.
14692 (riscv_valid_base_register_p): Likewise.
14693 (riscv_valid_offset_p): Likewise.
14694 (riscv_valid_lo_sum_p): Likewise.
14695 (riscv_classify_address): Likewise.
14696 (riscv_legitimate_address_p): Likewise.
14697 (riscv_address_insns): Likewise.
14698 (riscv_load_store_insns): Likewise.
14699 (riscv_force_binary): Likewise.
14700 (riscv_split_symbol): Likewise.
14701 (riscv_force_address): Likewise.
14702 (riscv_legitimize_address): Likewise.
14703 (riscv_move_integer): Likewise.
14704 (riscv_legitimize_const_move): Likewise.
14705 (riscv_legitimize_move): Likewise.
14706 (riscv_address_cost): Likewise.
14707 (riscv_subword): Likewise.
14708 (riscv_output_move): Likewise.
14709 (riscv_canonicalize_int_order_test): Likewise.
14710 (riscv_emit_int_order_test): Likewise.
14711 (riscv_function_arg_boundary): Likewise.
14712 (riscv_pass_mode_in_fpr_p): Likewise.
14713 (riscv_pass_fpr_single): Likewise.
14714 (riscv_pass_fpr_pair): Likewise.
14715 (riscv_get_arg_info): Likewise.
14716 (riscv_function_arg): Likewise.
14717 (riscv_function_arg_advance): Likewise.
14718 (riscv_arg_partial_bytes): Likewise.
14719 (riscv_function_value): Likewise.
14720 (riscv_pass_by_reference): Likewise.
14721 (riscv_setup_incoming_varargs): Likewise.
14722 (riscv_print_operand): Likewise.
14723 (riscv_elf_select_rtx_section): Likewise.
14724 (riscv_save_restore_reg): Likewise.
14725 (riscv_for_each_saved_reg): Likewise.
14726 (riscv_register_move_cost): Likewise.
14727 (riscv_hard_regno_mode_ok_p): Likewise.
14728 (riscv_hard_regno_nregs): Likewise.
14729 (riscv_class_max_nregs): Likewise.
14730 (riscv_memory_move_cost): Likewise.
14731 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14732 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14733 (rl78_addr_space_address_mode): Likewise.
14734 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14736 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14737 (rs6000_reassociation_width): Likewise.
14738 (rs6000_invalid_binary_op): Likewise.
14739 (fusion_p9_p): Likewise.
14740 (emit_fusion_p9_load): Likewise.
14741 (emit_fusion_p9_store): Likewise.
14742 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14743 (ok_for_simple_move_operands): Likewise.
14744 (ok_for_simple_move_strict_operands): Likewise.
14745 (ok_for_simple_arith_logic_operands): Likewise.
14746 (visium_legitimize_reload_address): Likewise.
14747 (visium_select_cc_mode): Likewise.
14748 (output_cbranch): Likewise.
14749 (visium_split_double_move): Likewise.
14750 (visium_expand_copysign): Likewise.
14751 (visium_expand_int_cstore): Likewise.
14752 (visium_expand_fp_cstore): Likewise.
14753 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14754 (visium_function_arg): Likewise.
14755 (visium_function_arg_advance): Likewise.
14756 (visium_libcall_value): Likewise.
14757 (visium_setup_incoming_varargs): Likewise.
14758 (visium_legitimate_constant_p): Likewise.
14759 (visium_legitimate_address_p): Likewise.
14760 (visium_legitimize_address): Likewise.
14761 (visium_secondary_reload): Likewise.
14762 (visium_register_move_cost): Likewise.
14763 (visium_memory_move_cost): Likewise.
14764 (prepare_move_operands): Likewise.
14765 (ok_for_simple_move_operands): Likewise.
14766 (ok_for_simple_move_strict_operands): Likewise.
14767 (ok_for_simple_arith_logic_operands): Likewise.
14768 (visium_function_value_1): Likewise.
14769 (rtx_ok_for_offset_p): Likewise.
14770 (visium_legitimize_reload_address): Likewise.
14771 (visium_split_double_move): Likewise.
14772 (visium_expand_copysign): Likewise.
14773 (visium_expand_int_cstore): Likewise.
14774 (visium_expand_fp_cstore): Likewise.
14775 (visium_split_cstore): Likewise.
14776 (visium_select_cc_mode): Likewise.
14777 (visium_split_cbranch): Likewise.
14778 (output_cbranch): Likewise.
14779 (visium_print_operand_address): Likewise.
14780 * expmed.c (flip_storage_order): Likewise.
14781 * expmed.h (emit_cstore): Likewise.
14782 (flip_storage_order): Likewise.
14783 * genrecog.c (validate_pattern): Likewise.
14784 * hsa-gen.c (gen_hsa_addr): Likewise.
14785 * internal-fn.c (expand_arith_overflow): Likewise.
14786 * ira-color.c (allocno_copy_cost_saving): Likewise.
14787 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14788 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14789 (process_invariant_for_inheritance): Likewise.
14790 * lra-eliminations.c (move_plus_up): Likewise.
14791 * omp-low.c (lower_oacc_reductions): Likewise.
14792 * simplify-rtx.c (simplify_subreg): Likewise.
14793 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14794 (TARGET_CHKP_BOUND_MODE): Likewise..
14795 * targhooks.c (default_chkp_bound_mode): Likewise.
14796 (default_setup_incoming_vararg_bounds): Likewise.
14797 * targhooks.h (default_chkp_bound_mode): Likewise.
14798 (default_setup_incoming_vararg_bounds): Likewise.
14799 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14800 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14801 (have_whole_vector_shift): Likewise.
14802 * tree-vect-stmts.c (vectorizable_load): Likewise.
14803 * doc/tm.texi: Regenerate.
14805 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14807 Graceful degrade if Binutils PR21472 is not available.
14810 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14811 .rodata in flash test fails.
14812 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14813 * confgure: Regenerate.
14814 * config.in: Regenerate.
14815 * config/avr/avr.c (avr_asm_named_section)
14816 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14817 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14818 (avr_asm_init_sections): Same.
14820 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14822 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14823 (fma<VH:mode>4_intrinsic): Likewise.
14824 (*fmsub<VCVTF:mode>4): Likewise.
14825 (*fmsub<VH:mode>4_intrinsic): Likewise.
14827 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14830 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14831 Don't depend on "optimize > 0".
14832 (out_movhi_r_mr, out_movqi_mr_r): Same.
14833 (out_movhi_mr_r, out_movqi_r_mr): Same.
14834 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14835 io_address_operand on "optimize > 0".
14837 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14839 * tree-loop-distribution.c: Add general explanantion on the pass.
14840 (generate_loops_for_partition): Mark distributed loop.
14841 (pg_add_dependence_edges): New parameter. Handle alias data
14842 dependence specially and record it in the parameter if asked.
14843 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14844 (init_partition_graph_vertices, add_partition_graph_edge): New.
14845 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14846 (free_partition_graph_vdata, build_partition_graph): New.
14847 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14848 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14849 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14850 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14851 (version_for_distribution_p, finalize_partitions): New.
14852 (distribute_loop): Handle alias data dependence specially. Factor
14853 out loop fusion code as functions and call these functions.
14855 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14857 * tree-loop-distribution.c (classify_partition): New parameter and
14858 better handle reduction statement.
14859 (rdg_build_partitions): Revise comment.
14860 (distribute_loop): Compute statements in all partitions and pass it
14861 to classify_partition.
14863 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14865 * tree-loop-distribution.c (enum partition_type): New.
14866 (struct partition): New field type.
14867 (partition_merge_into): Add parameter. Update partition type.
14868 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14869 (build_rdg_partition_for_vertex): Compute partition type.
14870 (rdg_build_partitions): Dump partition type.
14871 (distribute_loop): Update calls to partition_merge_into.
14873 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14875 * tree-loop-distribution.c (struct ddr_hasher): New.
14876 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14878 (classify_partition): Call get_data_dependence.
14879 (pg_add_dependence_edges): Ditto.
14880 (distribute_loop): Release data dependence hash table.
14882 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14884 * tree-loop-distribution.c (ref_base_address): Delete.
14885 (similar_memory_accesses): Rename ...
14886 (share_memory_accesses): ... to this. Check if partitions access
14887 the same memory reference.
14888 (distribute_loop): Call share_memory_accesses.
14890 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14892 * tree-loop-distribution.c (struct partition): New field recording
14893 its data reference.
14894 (partition_alloc, partition_free): Init and release data refs.
14895 (partition_merge_into): Merge data refs.
14896 (build_rdg_partition_for_vertex): Collect data refs for partition.
14897 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14899 (distribute_loop): Remve data refs from vertice data of partition
14902 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14904 * tree-loop-distribution.c (params.h): Include header file.
14905 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14906 (datarefs_vec): New global var.
14907 (create_rdg_vertices): Use datarefs_vec directly.
14908 (free_rdg): Don't free data references.
14909 (build_rdg): Update use. Don't free data references.
14910 (distribute_loop): Compute global variable for data references.
14911 Bail out if there are too many data references.
14913 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14915 * tree-loop-distribution.c (loop_nest): New global var.
14916 (build_rdg): Use loop directly, rather than loop nest.
14917 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14919 (distribute_loop): Compute global variable loop nest. Update use.
14921 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14923 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14924 (partition_merge_into): New parameter. Dump reason for fusion.
14925 (distribute_loop): Update use of partition_merge_into.
14927 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14929 * tree-loop-distribution.c (bb_top_order_index): New.
14930 (bb_top_order_index_size, bb_top_order_cmp): New.
14931 (stmts_from_loop): Use topological order.
14932 (pass_loop_distribution::execute): Compute and release topological
14933 order for basic blocks.
14935 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14937 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14940 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14942 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14943 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14944 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14945 * internal-fn.def (LOOP_DIST_ALIAS): New.
14946 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14947 (fold_loop_internal_call): ... this.
14948 (vect_loop_dist_alias_call): New function.
14949 (set_uid_loop_bbs): Call fold_loop_internal_call.
14950 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14953 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14956 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14957 Require dead FLAGS_REG at the beginning of a peephole.
14959 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14962 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14963 arguments in the call to __builtin_ia32_sbb_u32.
14964 (_subborrow_u64): Swap _X and _Y arguments in the call to
14965 __builtin_ia32_sbb_u64.
14967 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14970 * tree-vrp.c (compare_assert_loc): Turn into a function template
14971 with stable template parameter. Only test if a->e is NULL,
14972 !a->e == !b->e has been verified already. Use e == NULL or
14973 e != NULL instead of e or ! e tests. If stable is true, don't use
14974 iterative_hash_expr, on the other side allow a or b or both NULL
14975 and sort the NULLs last.
14976 (process_assert_insertions): Sort using compare_assert_loc<false>
14977 instead of compare_assert_loc, later sort using
14978 compare_assert_loc<true> before calling process_assert_insertions_for
14979 in a loop. Use break instead of continue once seen NULL pointer.
14981 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14983 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14984 Cortex-R7 and Cortex-R8 processors.
14986 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14988 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14989 uninitialized while src is not.
14991 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14993 * common/config/arm/arm-common.c: Adjust include path for
14995 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14996 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14997 (arm-cpu-data.h): Likewise.
14998 (arm-cpu-cdata.h): Likewise.
14999 * config/arm/arm-cpu.h: Delete.
15000 * config/arm/arm-cpu-cdata.h: Delete.
15001 * config/arm/arm-cpu-data.h: Delete.
15003 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15005 * config/arm/arm-cpus.in (cortex-a55): New.
15006 (cortex-a75): Likewise.
15007 (cortex-a75.cortex-a55): Likewise.
15008 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15010 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15011 * config/arm/arm-cpu-cdata.h: Regenerate.
15012 * config/arm/arm-cpu-data.h: Regenerate.
15013 * config/arm/arm-cpu.h: Regenerate.
15014 * config/arm/arm-tables.opt: Regenerate.
15015 * config/arm/arm-tune.md: Regenerate.
15017 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15019 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15021 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15023 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15026 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15028 PR tree-optimization/81292
15029 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15030 full_string_p, also call adjust_related_strinfos if the adjustment
15031 is simple, otherwise invalidate related strinfos.
15033 2017-07-04 Martin Liska <mliska@suse.cz>
15036 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15037 newly created variable as DECL_IGNORED_P.
15039 2017-07-04 Martin Liska <mliska@suse.cz>
15042 * ipa-inline.c (inline_small_functions):
15043 Use xstrdup_for_dump.
15045 2017-07-04 Tom de Vries <tom@codesourcery.com>
15047 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15049 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15052 * config/darwin.c (darwin_function_switched_text_sections):
15055 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15057 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15059 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15061 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15063 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15065 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15066 min_profitable_iters, and th as inclusive lower bounds.
15067 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15068 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15069 for min_profitable_iters and min_profitable_estimate.
15070 (vect_transform_loop): Treat th as an inclusive lower bound.
15071 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15073 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15076 * config/darwin.c (darwin_function_switched_text_sections):
15077 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15078 in two pieces, and suppress the use of buf.
15080 2017-07-03 Nathan Sidwell <nathan@acm.org>
15082 * hash-table.h (hash_table_mod1): Fix indentation.
15084 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15086 PR middle-end/81290
15087 * predict.c (force_edge_cold): Be more careful about propagation
15089 * profile-count.h (profile_probability::guessed,
15090 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15092 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15094 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15096 * doc/invoke.texi (rcpc architecture extension): Document it.
15098 2017-07-03 Richard Biener <rguenther@suse.de>
15100 PR tree-optimization/60510
15101 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15102 the scalar reduction PHI and use it.
15103 (vectorizable_reduction): Properly guard the single_defuse_cycle
15104 path for non-SLP reduction chains where we cannot use it.
15105 Rework reduc_def/index and vector type deduction. Rework
15106 vector operand gathering during reduction op code-gen.
15107 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15108 chains dissolve the chain and leave it to non-SLP reduction
15111 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15113 * tree-data-ref.h (dr_alignment): Declare.
15114 * tree-data-ref.c (dr_alignment): New function.
15115 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15116 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15118 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15120 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15122 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15123 and base_misalignment fields.
15124 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15125 * tree-data-ref.c: Include builtins.h.
15126 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15127 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15128 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15129 * tree-vect-data-refs.c: Include tree-cfg.h.
15130 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15131 fields instead of calculating an alignment here.
15132 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15133 innermost_loop_behavior fields.
15135 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15137 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15139 (DR_STEP_ALIGNMENT): New macro.
15140 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15141 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15142 (create_data_ref): Print it.
15143 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15144 to tell whether the step preserves vector (mis)alignment.
15145 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15146 Move the check for an integer step and generalise to all INTEGER_CST.
15147 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15148 Print the outer step alignment.
15150 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15152 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15153 with offset_alignment.
15154 (DR_ALIGNED_TO): Delete.
15155 (DR_OFFSET_ALIGNMENT): New macro.
15156 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15157 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15158 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15159 (create_data_ref): Likewise.
15160 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15161 (vect_analyze_data_refs): Likewise.
15162 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15163 creating dummy innermost behavior.
15165 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15167 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15168 with a "innermost_loop_behavior *" and refeence tree.
15169 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15170 (create_data_ref): Update call accordingly.
15171 * tree-predcom.c (find_looparound_phi): Likewise.
15173 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15175 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15176 fields with dr_wrt_vec_loop.
15177 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15178 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15179 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15180 (vect_dr_behavior): New function.
15181 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15182 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15183 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15184 track whether the step preserves the misalignment.
15185 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15186 Use vect_dr_behavior.
15187 (vect_setup_realignment): Update call accordingly.
15188 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15189 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15190 call to vect_create_addr_base_for_vector_ref.
15191 (vect_create_cond_for_align_checks): Likewise.
15192 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15193 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15194 (vect_recog_mask_conversion_pattern): Likewise.
15195 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15196 (new_stmt_vec_info): Remove redundant zeroing.
15198 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15200 * common/config/arm/arm-common.c (arm_be8_option): New function.
15201 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15202 (ISA_ARMv6): Add isa_bit_be8.
15203 * config/arm/arm.h (arm_be8_option): Add prototype.
15204 (BE8_SPEC_FUNCTION): New define.
15205 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15206 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15207 (mlittle-endian): Similarly.
15208 (mbe8, mbe32): New options.
15209 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15210 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15212 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15214 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15216 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15218 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15219 (cleanup_tree_cfg_bb): Use it.
15220 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15222 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15224 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15227 * loop-doloop.c (add_test): Update profile.
15229 2017-07-03 Martin Liska <mliska@suse.cz>
15232 * sanopt.c (rewrite_usage_of_param): New function.
15233 (sanitize_rewrite_addressable_params): Likewise.
15234 (pass_sanopt::execute): Call rewrite_usage_of_param.
15236 2017-07-03 Richard Biener <rguenther@suse.de>
15238 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15239 back to using VIEW_CONVERT_EXPR.
15241 2017-07-03 Martin Liska <mliska@suse.cz>
15244 * doc/extend.texi: Document when a resolver function is
15245 generated for target_clones.
15247 2017-07-03 Martin Liska <mliska@suse.cz>
15249 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15250 and shadow memory of auto variables which are subject of
15251 use-after-scope sanitization.
15252 (asan_expand_mark_ifn): Add do set only when is_poison.
15254 2016-07-03 Richard Biener <rguenther@suse.de>
15256 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15258 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15259 (vectorizable_reduction): Perform reduction PHI creation when
15260 visiting a reduction PHI and adjust and simplify code generation
15261 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15262 (vect_transform_loop): Visit reduction PHIs.
15263 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15264 defs into the SLP tree.
15265 (vect_build_slp_tree): Reduction defs terminate the recursion.
15266 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15268 (vect_get_vec_defs_for_stmt_copy): Export.
15269 (vect_get_vec_defs): Likewise.
15270 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15272 (vect_get_vec_defs_for_stmt_copy): Declare.
15273 (vect_get_vec_defs): Likewise.
15275 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15277 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15278 parameter with a "loop" parameter and use it instead of the
15279 loop containing DR_STMT. Don't check simple_iv when doing
15280 BB analysis. Describe the two analysis modes in the comment.
15282 2017-07-03 Tom de Vries <tom@codesourcery.com>
15284 PR tree-optimization/69468
15285 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15286 (find_same_succ_bb): Handle ignore_edge_flags.
15288 2017-07-03 Tom de Vries <tom@codesourcery.com>
15290 PR tree-optimization/81192
15291 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15293 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15295 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15297 2017-07-03 Tom de Vries <tom@codesourcery.com>
15299 PR tree-optimization/81192
15300 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15301 BB_SAME_SUCC (bb) == NULL.
15303 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15305 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15308 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15310 * dumpfile.c: Include profile-count.h
15311 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15313 (insert_cond_bb): Update profile.
15314 * tree-cfg.h (insert_cond_bb): Update prototype.
15315 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15316 * tree-dump.c: Do not include tree-cfg.
15318 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15320 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15324 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15327 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15329 * tree-complex.c (expand_complex_div_wide): update profile.
15331 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15332 Alan Hayward <alan.hayward@arm.com>
15333 David Sherwood <david.sherwood@arm.com>
15335 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15336 (CORETYPES_H): New define.
15337 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15338 (insn-modes-inline.h, s-modes-inline-h): New rules.
15339 (generated_files): Add insn-modes-inline.h.
15340 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15341 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15342 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15343 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15344 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15345 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15346 (build/gencodes.o, build/genconditions.o): Likewise.
15347 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15348 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15349 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15350 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15351 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15352 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15353 * coretypes.h: Include everything up to real.h for generators.
15354 Include insn-modes.h first. Include wide-int-print.h after
15355 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15356 * machmode.h: Don't include insn-modes.h here.
15357 * function-tests.c: Remove includes of signop.h, machmode.h,
15358 double-int.h and wide-int.h.
15360 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15362 * optc-save-gen.awk: Likewise.
15363 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15364 * godump.c: Remove include of wide-int-print.h.
15365 * pretty-print.h: Likewise.
15366 * wide-int-print.cc: Likewise.
15367 * wide-int.cc: Likewise.
15368 * hash-map-tests.c: Remove include of signop.h.
15369 * hash-set-tests.c: Likewise.
15370 * rtl-tests.c: Likewise.
15371 * mkconfig.sh: Remove include of machmode.h.
15372 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15374 (emit_insn_modes_inline_h): ...this new function. Emit the code
15375 into an insn-modes-inline.h header file, adding appropriate
15376 include guards and end comments.
15377 (emit_insn_modes_c_header): Remove include of machmode.h.
15378 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15380 (main): Handle -i flag and call emit_insn_modes_inline_h when
15383 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15385 * tree-ssa-strlen.c (strinfo): Rename the length field to
15386 nonzero_chars. Add a full_string_p field.
15387 (compare_nonzero_chars, zero_length_string_p): New functions.
15388 (get_addr_stridx): Add an offset_out parameter.
15389 Use compare_nonzero_chars.
15390 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15391 (new_strinfo): Update after above changes to strinfo.
15392 (set_endptr_and_length): Set full_string_p.
15393 (get_string_length): Update after above changes to strinfo.
15394 (unshare_strinfo): Update call to new_strinfo.
15395 (maybe_invalidate): Likewise.
15396 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15397 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15398 as a uhwi instead of an shwi. Update after above changes to
15399 strinfo and new_strinfo.
15400 (zero_length_string): Assert that chainsi contains full strings.
15401 Use zero_length_string_p. Update call to new_strinfo.
15402 (adjust_related_strinfos): Update after above changes to strinfo.
15403 Copy full_string_p from origsi.
15404 (adjust_last_stmt): Use zero_length_string_p.
15405 (handle_builtin_strlen): Update after above changes to strinfo and
15406 new_strinfo. Install the lhs as the string length if the previous
15407 entry didn't describe a full string.
15408 (handle_builtin_strchr): Update after above changes to strinfo
15410 (handle_builtin_strcpy): Likewise.
15411 (handle_builtin_strcat): Likewise.
15412 (handle_builtin_malloc): Likewise.
15413 (handle_pointer_plus): Likewise.
15414 (handle_builtin_memcpy): Likewise. Track nonzero characters
15415 that aren't necessarily followed by a nul terminator.
15416 (handle_char_store): Likewise.
15418 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15420 PR tree-optimization/80769
15421 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15422 for malloc and calloc. Document the new invariant that all related
15423 strinfos have delayed lengths or none do.
15424 (verify_related_strinfos): Move earlier in file.
15425 (set_endptr_and_length): New function, split out from...
15426 (get_string_length): ...here. Also set the lengths of related
15428 (zero_length_string): Assert that chainsi has known (rather than
15430 (adjust_related_strinfos): Likewise.
15432 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15434 PR tree-optimization/81136
15435 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15436 assert that two references with the same misalignment have the same
15437 compile-time misalignment if those compile-time misalignments
15440 2017-07-01 Andi Kleen <ak@linux.intel.com>
15442 * print-tree.c (print_node): Print all attributes.
15444 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15446 * cfg.c (scale_bbs_frequencies): New function.
15447 * cfg.h (scale_bbs_frequencies): Declare it.
15448 * cfgloopanal.c (single_likely_exit): Cleanup.
15449 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15451 (scale_loop_profile): Likewise.
15452 (loop_version): Likewise.
15453 (create_empty_loop_on_edge): Update.
15454 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15455 scale_loop_frequencies, scale_loop_profile, loopify,
15456 loop_version): Update prototypes.
15457 * modulo-sched.c (sms_schedule): Update.
15458 * predict.c (unlikely_executed_edge_p): Also check probability.
15459 (probably_never_executed_edge_p): Fix typo.
15460 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15461 * tree-parloops.c (gen_parallel_loop): Update.
15462 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15463 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15464 * tree-ssa-loop-split.c (split_loop): Update.
15465 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15466 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15467 (vect_loop_versioning): Update.
15468 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15470 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15472 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15474 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15476 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15477 to keep profile consistent.
15479 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15481 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15482 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15483 * profile-count.h (max_safe_multiplier): Make unsigned.
15484 (profile_count::guessed_zero): New.
15486 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15488 * bb-reorder.c (fix_up_crossing_landing_pad,
15489 fix_crossing_conditional_branches): Use make_single_succ_edge
15490 to keep profile consistent.
15492 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15494 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15497 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15500 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15501 the right scopes, make sure cond_jump isn't preserved between multiple
15502 iterations. Search for fallthru edge whenever there are 3+ edges and
15503 use find_fallthru_edge for it.
15505 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15507 Patch by Alexander Monakov <amonakov@ispras.ru>
15508 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15509 probabilities consistently.
15511 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15513 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15514 emit_cmp_and_jump_insns.
15516 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15519 * tree-inline.c (expand_call_inline): Combine profile statuses.
15521 2017-06-30 Andrew Pinski <apinski@cavium.com>
15523 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15524 fold_stmt returned true.
15526 2017-06-30 Nathan Sidwell <nathan@acm.org>
15528 * ggc.h (empty_string): Delete.
15529 * cfgexpand.c (expand_asm_stmt): Use plain "".
15530 * optabs.c (expand_asm_memory_barrier): Likewise.
15531 * stringpool.c (empty_string): Delete.
15532 (digit_vector, digit_string): Delete.
15533 (ggc_alloc_string): Use plain "", don't optimize single digit
15534 strings. Use ggc_alloc_atomic.
15536 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15538 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15539 comparison set and one other set, use the cost of the non-comparison
15542 2017-06-30 Nathan Sidwell <nathan@acm.org>
15544 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15547 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15549 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15550 loops. Remove now unneeded calls to gimple_switch_set_label() that
15551 just set removed labels to NULL_TREE.
15553 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15555 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15556 (set_range_info): ...here. Only call set_range_info_raw if domain
15558 (set_nonzero_bits): Call set_range_info_raw.
15559 * tree-ssanames.h (set_range_info_raw): New.
15561 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15564 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15565 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15566 of nonimmediate_operand and <store_mask_constraint> instead of m
15567 for the input operand. For V8FI iterator, always split if input
15568 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15569 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15570 <store_mask_predicate> instead of register_operand and
15571 <store_mask_constraint> instead of v for the input operand. Make
15572 sure both operands aren't MEMs for if not <mask_applied>.
15574 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15576 * lto-wrapper.c (copy_file) Close both file descriptors before
15579 2017-06-30 Martin Liska <mliska@suse.cz>
15582 * multiple_target.c (create_dispatcher_calls): Make ifunc
15583 also for function that don't have calls or are not referenced.
15585 2017-06-30 Richard Biener <rguenther@suse.de>
15587 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15588 analyze the first scalar stmt. Move vector type computation
15589 for the BB case here from ...
15590 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15591 live operation processing in the SLP case properly.
15593 2017-06-30 Richard Biener <rguenther@suse.de>
15595 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15597 2017-06-30 Martin Liska <mliska@suse.cz>
15600 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15601 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15603 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15605 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15608 2017-06-30 Richard Biener <rguenther@suse.de>
15610 PR tree-optimization/81249
15611 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15612 condition reduction result to original scalar type.
15614 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15616 * profile-count.h (enum profile_quality): Fix typos and whitespace
15619 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15621 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15622 type for branch probabilities.
15624 2017-06-29 Julian Brown <julian@codesourcery.com>
15625 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15627 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15628 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15629 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15630 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15632 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15634 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15635 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15636 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15637 CC usage from generic code to here.
15638 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15639 CC usage into the target macros.
15641 2017-06-29 Maya Rashish <coypu@sdf.org>
15643 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15646 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15648 * arm/arm-builtins.c: Include profile-count.h
15649 * except.c (sjlj_emit_function_enter): Use
15650 profile_probability::unlikely.
15652 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15654 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15655 and tocrel_offset be pointer args rather than implicitly using
15657 (legitimate_constant_pool_address_p, rs6000_emit_move,
15658 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15659 tocrel_offset and use in toc_relative_expr_p call.
15660 (print_operand, print_operand_address): Use static tocrel_base_oac
15661 and tocrel_offset_oac.
15662 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15665 2017-06-29 Maya Rashish <coypu@sdf.org>
15667 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15669 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15671 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15672 objects, take into account only the alignment of 'op0' and 'mode1' if
15675 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15677 * ccmp.c (ccmp_tree_comparison_p): New function.
15678 (ccmp_candidate_p): Update to use above function.
15679 (get_compare_parts): New function.
15680 (expand_ccmp_next): Update to use new functions.
15681 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15683 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15684 take mode as argument.
15685 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15686 * expr.c (expand_expr_real_1): Pass mode as argument.
15688 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15690 * combine.c (combine_instructions): Print insns to dump_file, together
15693 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15695 * asan.c (asan_emit_stack_protection): Update.
15696 (create_cond_insert_point): Update.
15697 * auto-profile.c (afdo_propagate_circuit): Update.
15698 * basic-block.h (struct edge_def): Turn probability to
15699 profile_probability.
15700 (EDGE_FREQUENCY): Update.
15701 * bb-reorder.c (find_traces_1_round): Update.
15702 (better_edge_p): Update.
15703 (sanitize_hot_paths): Update.
15704 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15705 (make_single_succ_edge): Update.
15706 (check_bb_profile): Update.
15707 (dump_edge_info): Update.
15708 (update_bb_profile_for_threading): Update.
15709 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15710 probabilitycount to 0.
15711 * cfgbuild.c (compute_outgoing_frequencies): Update.
15712 * cfgcleanup.c (try_forward_edges): Update.
15713 (outgoing_edges_match): Update.
15714 (try_crossjump_to_edge): Update.
15715 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15716 (expand_gimple_tailcall): Update.
15717 (construct_init_block): Use make_single_succ_edge.
15718 (construct_exit_block): Use make_single_succ_edge.
15719 * cfghooks.c (verify_flow_info): Update.
15720 (redirect_edge_succ_nodup): Update.
15721 (split_edge): Update.
15722 (account_profile_record): Update.
15723 * cfgloopanal.c (single_likely_exit): Update.
15724 * cfgloopmanip.c (scale_loop_profile): Update.
15725 (set_zero_probability): Remove.
15726 (duplicate_loop_to_header_edge): Update.
15727 * cfgloopmanip.h (loop_version): Update prototype.
15728 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15729 (force_nonfallthru_and_redirect): Update.
15730 (update_br_prob_note): Update.
15731 (rtl_verify_edges): Update.
15732 (purge_dead_edges): Update.
15733 (rtl_lv_add_condition_to_bb): Update.
15734 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15735 * cgraphunit.c (init_lowered_empty_function): Update.
15736 (cgraph_node::expand_thunk): Update.
15737 * cilk-common.c: Include profile-count.h
15738 * dojump.c (inv): Remove.
15739 (jumpifnot): Update.
15740 (jumpifnot_1): Update.
15741 (do_jump_1): Update.
15743 (do_jump_by_parts_greater_rtx): Update.
15744 (do_compare_rtx_and_jump): Update.
15745 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15746 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15747 * dwarf2cfi.c: Include profile-count.h
15748 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15749 (sjlj_emit_dispatch_table): Likewise.
15750 * explow.c: Include profile-count.h
15751 * expmed.c (emit_store_flag_force): Update.
15752 (do_cmp_and_jump): Update.
15753 * expr.c (compare_by_pieces_d::generate): Update.
15754 (compare_by_pieces_d::finish_mode): Update.
15755 (emit_block_move_via_loop): Update.
15756 (store_expr_with_bounds): Update.
15757 (store_constructor): Update.
15758 (expand_expr_real_2): Update.
15759 (expand_expr_real_1): Update.
15760 * expr.h (try_casesi, try_tablejump): Update prototypes.
15761 * gimple-pretty-print.c (dump_probability): Update.
15762 (dump_profile): New.
15763 (dump_gimple_label): Update.
15764 (dump_gimple_bb_header): Update.
15765 * graph.c (draw_cfg_node_succ_edges): Update.
15766 * hsa-gen.c (convert_switch_statements): Update.
15767 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15768 (find_if_case_1): Update.
15769 (find_if_case_2): Update.
15770 * internal-fn.c (expand_arith_overflow_result_store): Update.
15771 (expand_addsub_overflow): Update.
15772 (expand_neg_overflow): Update.
15773 (expand_mul_overflow): Update.
15774 (expand_vector_ubsan_overflow): Update.
15775 * ipa-cp.c (good_cloning_opportunity_p): Update.
15776 * ipa-split.c (split_function): Use make_single_succ_edge.
15777 * ipa-utils.c (ipa_merge_profiles): Update.
15778 * loop-doloop.c (add_test): Update.
15779 (doloop_modify): Update.
15780 * loop-unroll.c (compare_and_jump_seq): Update.
15781 (unroll_loop_runtime_iterations): Update.
15782 * lra-constraints.c (lra_inheritance): Update.
15783 * lto-streamer-in.c (input_cfg): Update.
15784 * lto-streamer-out.c (output_cfg): Update.
15785 * mcf.c (adjust_cfg_counts): Update.
15786 * modulo-sched.c (sms_schedule): Update.
15787 * omp-expand.c (expand_omp_for_init_counts): Update.
15788 (extract_omp_for_update_vars): Update.
15789 (expand_omp_ordered_sink): Update.
15790 (expand_omp_for_ordered_loops): Update.
15791 (expand_omp_for_generic): Update.
15792 (expand_omp_for_static_nochunk): Update.
15793 (expand_omp_for_static_chunk): Update.
15794 (expand_cilk_for): Update.
15795 (expand_omp_simd): Update.
15796 (expand_omp_taskloop_for_outer): Update.
15797 (expand_omp_taskloop_for_inner): Update.
15798 * omp-simd-clone.c (simd_clone_adjust): Update.
15799 * optabs.c (expand_doubleword_shift): Update.
15800 (expand_abs): Update.
15801 (emit_cmp_and_jump_insn_1): Update.
15802 (expand_compare_and_swap_loop): Update.
15803 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15804 * predict.c (predictable_edge_p): Update.
15805 (edge_probability_reliable_p): Update.
15806 (set_even_probabilities): Update.
15807 (combine_predictions_for_insn): Update.
15808 (combine_predictions_for_bb): Update.
15809 (propagate_freq): Update.
15810 (estimate_bb_frequencies): Update.
15811 (force_edge_cold): Update.
15812 * profile-count.c (profile_count::dump): Add missing space into dump.
15813 (profile_count::debug): Add newline.
15814 (profile_count::differs_from_p): Explicitly convert to unsigned.
15815 (profile_count::stream_in): Update.
15816 (profile_probability::dump): New member function.
15817 (profile_probability::debug): New member function.
15818 (profile_probability::differs_from_p): New member function.
15819 (profile_probability::differs_lot_from_p): New member function.
15820 (profile_probability::stream_in): New member function.
15821 (profile_probability::stream_out): New member function.
15822 * profile-count.h (profile_count_quality): Rename to ...
15823 (profile_quality): ... this one.
15824 (profile_probability): New.
15825 (profile_count): Update.
15826 * profile.c (compute_branch_probabilities): Update.
15827 * recog.c (peep2_attempt): Update.
15828 * sched-ebb.c (schedule_ebbs): Update.
15829 * sched-rgn.c (find_single_block_region): Update.
15830 (compute_dom_prob_ps): Update.
15831 (schedule_region): Update.
15832 * sel-sched-ir.c (compute_succs_info): Update.
15833 * stmt.c (struct case_node): Update.
15834 (do_jump_if_equal): Update.
15835 (get_outgoing_edge_probs): Update.
15836 (conditional_probability): Update.
15837 (emit_case_dispatch_table): Update.
15838 (expand_case): Update.
15839 (expand_sjlj_dispatch_table): Update.
15840 (emit_case_nodes): Update.
15841 * targhooks.c: Update.
15842 * tracer.c (better_p): Update.
15843 (find_best_successor): Update.
15844 * trans-mem.c (expand_transaction): Update.
15845 * tree-call-cdce.c: Update.
15846 * tree-cfg.c (gimple_split_edge): Upate.
15847 (move_sese_region_to_fn): Upate.
15848 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15849 * tree-eh.c (lower_resx): Upate.
15850 (cleanup_empty_eh_move_lp): Upate.
15851 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15852 * tree-inline.c (copy_edges_for_bb): Update.
15853 (copy_cfg_body): Update.
15854 * tree-parloops.c (gen_parallel_loop): Update.
15855 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15856 (gimple_gen_time_profiler): Update.
15857 * tree-ssa-dce.c (remove_dead_stmt): Update.
15858 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15859 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15860 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15861 (unloop_loops): Update.
15862 (try_peel_loop): Update.
15863 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15864 * tree-ssa-loop-split.c (connect_loops): Update.
15865 (split_loop): Update.
15866 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15867 (hoist_guard): Update.
15868 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15869 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15870 (value_replacement): Update.
15871 * tree-ssa-reassoc.c (branch_fixup): Update.
15872 * tree-ssa-tail-merge.c (replace_block_by): Update.
15873 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15874 (create_edge_and_update_destination_phis): Update.
15875 (compute_path_counts): Update.
15876 (recompute_probabilities): Update.
15877 (update_joiner_offpath_counts): Update.
15878 (freqs_to_counts_path): Update.
15879 (duplicate_thread_path): Update.
15880 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15881 (struct switch_conv_info): Update.
15882 (gen_inbound_check): Update.
15883 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15884 (vect_do_peeling): Update.
15885 (vect_loop_versioning): Update.
15886 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15887 (optimize_mask_stores): Update.
15888 * ubsan.c (ubsan_expand_null_ifn): Update.
15889 * value-prof.c (gimple_divmod_fixed_value): Update.
15890 (gimple_divmod_fixed_value_transform): Update.
15891 (gimple_mod_pow2): Update.
15892 (gimple_mod_pow2_value_transform): Update.
15893 (gimple_mod_subtract): Update.
15894 (gimple_mod_subtract_transform): Update.
15895 (gimple_ic): Update.
15896 (gimple_stringop_fixed_value): Update.
15897 (gimple_stringops_transform): Update.
15898 * value-prof.h: Update.
15900 2017-06-29 Carl Love <cel@us.ibm.com>
15902 * config/rs6000/rs6000-c.c: Add support for built-in functions
15903 vector signed int vec_signed (vector float);
15904 vector signed long long vec_signed (vector double);
15905 vector signed int vec_signed2 (vector double, vector double);
15906 vector signed int vec_signede (vector double);
15907 vector signed int vec_signedo (vector double);
15908 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15909 instruction generator.
15910 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15911 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15912 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15914 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15915 vunsignede_v2df): Add define_expands.
15916 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15917 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15918 VEC_UNSIGNEDO): Add definitions.
15919 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15920 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15921 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15922 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15923 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15924 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15925 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15926 vec_unsignede, vec_unsignedo): Add builtin defines.
15927 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15929 * doc/extend.texi: Update the built-in documentation file for the
15930 new built-in functions.
15932 2017-06-29 Richard Biener <rguenther@suse.de>
15934 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15935 reduction chains to LOOP_VINFO_REDUCTIONS.
15936 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15937 SLP reductions after processing reduction chains.
15939 2017-06-29 Nathan Sidwell <nathan@acm.org>
15941 * builtins.c (fold_builtin_FUNCTION): Use
15942 lang_hooks.decl_printable_name.
15944 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15946 PR middle-end/81194
15947 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15948 with only one label.
15949 * stmt.c (expand_case): Assert NCASES is greater than one.
15951 2017-06-29 Richard Biener <rguenther@suse.de>
15953 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15955 (group_case_labels): Likewise.
15956 (find_taken_edge): Push sanity checking on val to workers...
15957 (find_taken_edge_cond_expr): ... here
15958 (find_taken_edge_switch_expr): ... and here, handle cases
15959 with just a default label.
15960 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15961 (group_case_labels): Likewise.
15962 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15963 group_case_labels does anything cleanup the CFG again.
15965 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15967 PR tree-optimization/81196
15968 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15969 exit condition comparing two IVs.
15971 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15973 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15974 profile to the dummy entry at the end of the list of architectures.
15975 * config/arm/arm-cpu-cdata.h: Regenerated.
15977 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15978 Michael Collison <michael.collison@arm.com>
15981 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15983 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15984 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15985 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15986 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15987 of shift when the shift amount is masked with constant equal to
15988 the size of the mode.
15989 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15992 2017-06-29 Martin Liska <mliska@suse.cz>
15994 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15996 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15999 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16002 (edge_growth_cache_entry::edge_growth_cache_entry): New
16004 (reset_edge_growth_cache): Update to use constructor.
16006 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16008 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16009 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16010 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16012 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16014 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16015 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16017 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16019 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16020 (*-linux-uclibc*): Add t-uclibc tmake_file.
16021 * config/t-musl: New.
16022 * config/t-uclibc: New.
16024 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16026 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16028 (gen_comm_data): Emit architectural setting of arch_prof.
16029 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16031 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16032 (armv8-m.base, armv8-m.main): Likewise.
16033 * arm-protos.h (arm_build_target): Add profile field.
16034 (arch_option): Likewise.
16035 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16037 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16038 arm_active_target.profile.
16040 2017-06-28 Richard Biener <rguenther@suse.de>
16042 PR middle-end/81227
16043 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16044 TYPE_OVERFLOW_WRAPS.
16045 * match.pd (negate_expr_p): Likewise.
16046 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16047 fold_build2, not fold_binary.
16049 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16051 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16052 Convert memory address to Pmode.
16053 (aarch64_print_operand): Assert MEM operands are always Pmode.
16055 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16058 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16059 Remove redundant if.
16060 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16061 * config/arm/aarch-common-protos.h
16062 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16063 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16065 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16068 * multiple_target.c (create_dispatcher_calls): Set the default
16069 clone to be static, not public.
16071 2017-06-28 Richard Biener <rguenther@suse.de>
16073 * tree-vect-loop.c (vectorizable_reduction): Move special
16074 cond reduction IV var creation ...
16075 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16076 parameter. Use STMT_VINFO_VECTYPE.
16077 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16080 2017-06-28 Martin Liska <mliska@suse.cz>
16083 * ipa-visibility.c (non_local_p): Handle visibility.
16085 2017-06-28 Martin Liska <mliska@suse.cz>
16088 * common.opt: Add IntegerRange to various options.
16089 * opt-functions.awk (integer_range_info): New function.
16090 * optc-gen.awk: Add integer_range_info to cl_options struct.
16091 * opts-common.c (decode_cmdline_option): Handle
16092 CL_ERR_INT_RANGE_ARG.
16093 (cmdline_handle_error): Likewise.
16094 * opts.c (print_filtered_help): Show valid interval in
16095 when --help is provided.
16096 * opts.h (struct cl_option): Add range_min and range_max fields.
16097 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16099 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16101 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16102 (x * C EQ/NE y * C): New transformation.
16104 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16106 * genmultilib (combination_space): Accept '+' in option names.
16108 2017-06-28 Martin Liska <mliska@suse.cz>
16111 * asan.c (instrument_derefs): Bail out inner references
16112 that are hard register variables.
16114 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16117 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16118 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16120 2017-06-28 Richard Biener <rguenther@suse.de>
16122 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16123 (vect_get_slp_defs): Adjust.
16124 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16126 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16128 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16129 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16130 (vectorizable_reduction): Adjust.
16131 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16133 (vect_get_slp_defs): Likewise.
16134 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16135 (vectorizable_bswap): Adjust.
16136 (vectorizable_call): Likewise.
16137 (vectorizable_conversion): Likewise.
16138 (vectorizable_assignment): Likewise.
16139 (vectorizable_shift): Likewise.
16140 (vectorizable_operation): Likewise.
16141 (vectorizable_store): Likewise.
16142 (vectorizable_condition): Likewise.
16143 (vectorizable_comparison): Likewise.
16145 2017-06-28 Michael Collison <michael.collison@arm.com>
16148 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16150 (arm_gen_movmemqi): Removed unused variable 'i'.
16151 Convert 'for' loop into 'while' loop.
16152 (arm_expand_prologue): Remove last unnecessary set of insn.
16153 (thumb_pop): Remove unused variable 'pushed_words'.
16154 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16156 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16158 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16159 * config/s390/s390-protos.h: Add prototype of
16160 s390_rel_address_ok_p.
16161 * config/s390/s390.c (s390_got_symbol): New function.
16162 (s390_rel_address_ok_p): New function.
16163 (legitimize_pic_address): Use s390_rel_address_ok_p.
16164 (s390_load_got): Use s390_got_symbol.
16165 (s390_option_override): Issue error if
16166 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16167 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16169 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16171 2017-06-27 Andrew Pinski <apinski@cavium.com>
16173 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16174 (X * copysign (1.0, X)): New pattern.
16175 (X * copysign (1.0, -X)): New pattern.
16176 (copysign (-1.0, CST)): New pattern.
16178 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16180 * genmultilib (combination_space): Remove variable.
16181 Validate reuse rules against regular expression for any sequence
16182 of multilib options in any order.
16184 2017-06-27 Michael Collison <michael.collison@arm.com>
16186 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16187 call aarch64_split_simd_combine.
16188 * (aarch64_combine_internal<mode>): Delete pattern.
16189 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16190 Allow register and subreg operands.
16192 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16194 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16195 specific need, just fallback on defaults.
16196 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16198 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16199 Olivier Hainque <hainque@adacore.com>
16201 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16203 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16204 targets. Pick a default if no particular attempt applied.
16205 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16208 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16210 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16211 (x86_64-wrs-vxworks7): Likewise.
16213 2017-06-27 Marek Polacek <polacek@redhat.com>
16216 * ubsan.c (instrument_null): Check get_base_address's result for null.
16218 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16220 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16222 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16224 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16225 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16226 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16227 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16228 New function types.
16229 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16230 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16231 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16232 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16233 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16234 BUILT_IN_FEUPDATEENV): New builtins.
16235 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16236 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16237 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16238 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16240 (builtin_structptr_types): Adjust size.
16241 * tree.c (builtin_structptr_types): Add four entries.
16243 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16244 Olivier Hainque <hainque@adacore.com>
16246 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16247 (TLS_SYM): New local macro, forcing reference to __tls__ on
16248 link command lines for VxWorks 7 RTPs, triggering initialization
16250 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16251 OS features TLS support, true for RTPs on VxWorks 7.
16252 * config/vxworks.c (vxworks_override_options): Setup emutls
16255 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16257 * predict.c (test_prediction_value_range): Use -1U instead of -1
16258 to avoid narrowing conversion warning.
16259 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16260 to avoid narrowing conversion warning.
16261 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16263 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16265 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16267 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16268 64bit configurations.
16269 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16270 (SIZE_TYPE): Likewise.
16271 * config/vxworks.c (vxworks_emutls_var_fields): Use
16272 long_unsigned_type_node instead of unsigned_type_node as the offset
16273 field type, which is "pointer" mode in emutls.c.
16275 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16278 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16280 PR middle-end/81207
16281 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16282 gimple_vuse copying separately from gimple_vdef copying.
16284 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16286 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16289 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16290 Olivier Hainque <hainque@adacore.com>
16292 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16293 for all vxworks7 targets.
16294 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16295 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16296 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16297 variations for VX6/VX7 and 32/64bits later on in ...
16298 (VXWORKS_LIB_SPEC): Leverage new macros.
16299 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16300 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16302 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16304 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16305 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16307 2017-06-26 Carl Love <cel@us.ibm.com>
16309 * config/rs6000/rs6000-c.c: Add support for built-in functions
16310 vector bool char vec_reve (vector bool char);
16311 vector signed char vec_reve (vector signed char);
16312 vector unsigned char vec_reve (vector unsigned char);
16313 vector bool int vec_reve (vector bool int);
16314 vector signed int vec_reve (vector signed int);
16315 vector unsigned int vec_reve (vector unsigned int);
16316 vector bool long long vec_reve (vector bool long long);
16317 vector signed long long vec_reve (vector signed long long);
16318 vector unsigned long long vec_reve (vector unsigned long long);
16319 vector bool short vec_reve (vector bool short);
16320 vector signed short vec_reve (vector signed short);
16321 vector double vec_reve (vector double);
16322 vector float vec_reve (vector float);
16323 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16324 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16325 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16326 (altivec_vreve): New pattern.
16327 * config/rs6000/altivec.h (vec_reve): New define.
16328 * doc/extend.texi (vec_rev): Update the built-in documentation file
16329 for the new built-in functions.
16331 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16333 PR tree-optimization/71815
16334 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16336 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16338 (slsr_process_phi): Likewise.
16339 (replace_uncond_cands_and_profitable_phis): Don't replace a
16340 multiply candidate with a stride of 1 (copy or cast).
16341 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16343 (lowest_cost_path): Likewise.
16344 (total_savings): Likewise.
16346 2017-06-26 Richard Biener <rguenther@suse.de>
16349 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16350 Use def_builtin_pure for all gather builtins.
16352 2017-06-26 Richard Biener <rguenther@suse.de>
16354 PR tree-optimization/81203
16355 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16356 non-dominating BBs.
16358 2017-06-26 Marek Polacek <polacek@redhat.com>
16361 * doc/invoke.texi: Document -Wmultistatement-macros.
16363 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16365 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16366 arm_neon_ok_no_float_abi effective target.
16368 2017-06-26 Richard Biener <rguenther@suse.de>
16370 PR tree-optimization/80928
16371 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16372 (copy_bbs): Set BB_DUPLICATED flag early.
16373 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16375 (execute_on_shrinking_pred): Likewise.
16376 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16377 BB_DUPLICATED blocks.
16378 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16379 iterate over all PHIs considering removal of *gsi.
16381 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16383 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16386 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16388 * config/rs6000/rs6000-string.c: (expand_block_clear,
16389 do_load_for_compare, select_block_compare_mode,
16390 compute_current_alignment, expand_block_compare,
16391 expand_strncmp_align_check, expand_strn_compare,
16392 expand_block_move, rs6000_output_load_multiple)
16393 Move functions related to string/block move/compare
16394 to a separate file.
16395 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16396 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16397 for this function which is now used in two files.
16398 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16399 * config.gcc: Add rs6000-string.o to extra_objs for
16400 targets powerpc*-*-* and rs6000*-*-*.
16402 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16405 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16406 32-bit, since indexed is not valid for DImode.
16407 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16408 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16409 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16410 (define_peephole2 for Altivec d-form store): Likewise.
16413 * multiple_target.c (create_dispatcher_calls): Only create the
16414 dispatcher call if the function is the default clone of a
16415 versioned function.
16417 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16419 PR middle-end/80902
16420 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16421 a call, force the call to not be a tail call.
16423 2017-06-23 Jeff Law <law@redhat.com>
16425 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16428 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16430 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16431 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16432 handling for early expansion of vector shifts (sl,sr,sra,rl).
16433 (builtin_function_type): Add vector shift right instructions
16434 to the unsigned argument list.
16436 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16438 rtl-optimizatoin/79286
16439 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16440 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16441 trap. PIC register plus a const unspec without offset can never trap.
16443 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16445 * tree.h (builtin_structptr_type): New type.
16446 (builtin_structptr_types): Declare new array.
16447 * tree.c (builtin_structptr_types): New array.
16448 (free_lang_data, build_common_tree_nodes): Use it.
16450 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16453 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16456 2017-06-22 Matt Turner <mattst88@gmail.com>
16458 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16459 Lake models to skylake case. Assume skylake for unknown
16460 models with clflushopt.
16462 2017-06-22 Jeff Law <law@redhat.com>
16464 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16465 frame sizes that do not satisfy aarch64_uimm12_shift.
16467 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16469 * profile-count.h (apply_probability,
16470 apply_scale, probability_in): Fix checks for zero.
16472 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16474 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16475 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16477 2016-06-22 Richard Biener <rguenther@suse.de>
16479 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16480 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16481 REDUC_MAX_EXPR support.
16482 (vectorizable_reduction): Likewise.
16483 (vect_create_epilog_for_reduction): Likewise.
16485 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16487 * match.pd (A / (1 << B) -> A >> B): New.
16488 * generic-match-head.c: Include optabs-tree.h.
16489 * gimple-match-head.c: Likewise.
16490 * optabs-tree.h (target_supports_op_p): New.
16491 * optabs-tree.c (target_supports_op_p): New.
16493 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16495 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16496 $gcc_cv_ld --help output.
16497 (gcc_cv_ld_demangle): Likewise.
16498 (gcc_cv_ld_eh_frame_hdr): Likewise.
16499 (gcc_cv_ld_pie): Likewise.
16500 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16501 (gcc_cv_ld_buildid): Likewise.
16502 (gcc_cv_ld_sysroot): Likewise.
16503 (ld_bndplt_support): Likewise.
16504 (ld_pushpopstate_support): Likewise.
16505 * configure: Regenerate.
16506 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16508 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16511 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16512 operands indexes to avoid gap between operands and match_dups.
16514 2017-06-21 Andrew Pinski <apinski@cavium.com>
16516 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16517 Increment Arith_shift and Arith_shift_reg by 1.
16518 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16520 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16521 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16522 (aarch64_strip_extend): Add new argument and test for it.
16523 (aarch64_cheap_mult_shift_p): New function.
16524 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16525 add a cost if it is true.
16526 Update calls to aarch64_strip_extend.
16527 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16529 2017-06-21 Andrew Pinski <apinski@cavium.com>
16531 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16533 (thunderxt88): Likewise.
16534 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16535 (thunderx_prefetch_tune): New variable.
16536 (thunderx2t99_prefetch_tune): Update for the correct values.
16537 (thunderxt88_tunings): New variable.
16538 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16539 generic_prefetch_tune.
16540 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16542 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16544 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16545 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16546 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16547 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16548 (aarch64_atomic_cas<mode>, GPI): Likewise.
16550 2017-06-21 Martin Liska <mliska@suse.cz>
16552 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16553 statements on cold and hot labels.
16554 * predict.c (tree_estimate_probability_bb): Remove the
16555 prediction from this place.
16557 2017-06-21 Martin Liska <mliska@suse.cz>
16559 PR tree-optimization/79489
16560 * gimplify.c (maybe_add_early_return_predict_stmt): New
16562 (gimplify_return_expr): Call the function.
16563 * predict.c (tree_estimate_probability_bb): Remove handling
16565 * predict.def: Update comment about early return predictor.
16566 * gimple-predict.h (is_gimple_predict): New function.
16567 * predict.def: Change default value of early return to 66.
16568 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16570 * passes.def: Put pass_strip_predict_hints to the beginning of
16573 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16575 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16576 FUNCTION_DECL declarations.
16577 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16579 (dwaf2out_decl): Likewise.
16580 * godump.c (go_early_global_decl): Skip call to the real debug hook
16581 for FUNCTION_DECL declarations.
16582 * passes.c (rest_of_decl_compilation): Skip call to the
16583 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16584 -fdump-go-spec is passed.
16586 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16588 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16589 Reorder for compactness.
16590 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16591 (def_builtin_pure, def_builtin_pure2): New functions.
16592 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16594 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16596 * match.pd (nop_convert): New predicate.
16597 ((A +- CST1) +- CST2): Allow some NOP conversions.
16599 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16602 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16603 with ctors/dtors if GOVD_SHARED is set.
16605 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16607 * config/aarch64/aarch64.md (movti_aarch64):
16608 Emit mov rather than orr.
16609 (movtf_aarch64): Likewise.
16610 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16611 Emit mov rather than orr.
16613 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16615 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16616 Swap alternatives, make integer dup more expensive.
16618 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16620 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16621 Return true for non-tls symbols.
16623 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16625 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16626 (cortex-a75): Likewise.
16627 (cortex-a75.cortex-a55): Likewise.
16628 * config/aarch64/aarch64-tune.md: Regenerate.
16629 * doc/invoke.texi (-mtune): Document new values for -mtune.
16631 2017-06-21 Tom de Vries <tom@codesourcery.com>
16633 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16634 stack_size feature.
16635 (Effective-Target Keywords, Other attributes): Suggest using
16636 dg-add-options stack_size feature to get stack limit in stack_size
16637 effective target documentation.
16639 2017-06-21 Julian Brown <julian@codesourcery.com>
16640 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16642 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16643 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16644 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16646 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16647 attribute type list for neon_multiply.
16648 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16649 attribute type list for neon_multiply.
16650 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16651 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16652 attribute type list for neon_multiply.
16653 * config/arm/types.md (crypto_pmull): Add.
16654 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16655 attribute type list.
16657 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16659 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16662 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16664 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16665 to make sure not to dereference a NULL cost_classes_ptr pointer.
16667 2017-06-20 Carl Love <cel@us.ibm.com>
16669 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16670 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16671 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16673 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16674 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16675 VMULOSW): New enum "unspec" values.
16676 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16677 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16678 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16679 altivec_vmulosw): New patterns.
16680 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16681 VMULOSW): Add definitions.
16683 2017-06-20 Julia Koval <julia.koval@intel.com>
16685 * config/i386/i386.c: Fix rounding expand for new pattern.
16686 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16688 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16690 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16691 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16693 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16695 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16698 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16700 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16701 architecture, then by implementer ID.
16702 * config/aarch64/aarch64-tune.md: Regenerate.
16704 2017-06-20 Richard Biener <rguenther@suse.de>
16706 PR middle-end/81097
16707 * fold-const.c (split_tree): Fold to type before negating.
16709 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16711 * diagnostic-show-locus.c
16712 (selftest::test_fixit_deletion_affecting_newline): New function.
16713 (selftest::diagnostic_show_locus_c_tests): Call it.
16715 2017-06-20 Andreas Schwab <schwab@suse.de>
16718 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16721 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16723 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16724 __ARM_FEATURE_COPROC according to support.
16726 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16728 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16729 Rewritten to avoid overflow for > 32-bit pointers.
16732 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16733 by removing enum keyword.
16734 (ubsan_type_descriptor): Likewise. Formatting fix.
16737 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16738 splitter): Require TARGET_SSE2 in the condition.
16740 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16743 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16744 for doing vector set of SFmode on ISA 3.0.
16745 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16746 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16748 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16749 SFmode value into a V4SF variable that was extracted from another
16750 V4SF variable without converting the element to double precision
16751 and back to single precision vector format.
16752 (vsx_insert_extract_v4sf_p9_2): Likewise.
16754 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16756 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16757 in UWHI to avoid undefined overflow.
16760 * ubsan.h (enum ubsan_encode_value_phase): New.
16761 (ubsan_encode_value): Change second argument to
16762 enum ubsan_encode_value_phase with default value of
16763 UBSAN_ENCODE_VALUE_GENERIC.
16764 * ubsan.c (ubsan_encode_value): Change second argument to
16765 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16766 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16767 create_tmp_var_raw instead of create_tmp_var and use a
16769 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16770 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16771 ubsan_encode_value callers.
16774 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16775 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16776 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16778 2017-06-19 Richard Biener <rguenther@suse.de>
16780 PR middle-end/81118
16781 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16782 estimates if we changed anything.
16784 2017-06-19 Richard Biener <rguenther@suse.de>
16786 PR tree-optimization/80887
16787 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16788 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16789 simplified lookups, then reset mprts_hook.
16790 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16792 (try_to_simplify): Likewise.
16794 2017-06-19 Martin Liska <mliska@suse.cz>
16797 * gimplify.c (gimplify_switch_expr):
16798 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16800 2017-06-19 Martin Liska <mliska@suse.cz>
16802 * doc/install.texi: Document that PGO runs in 4 stages.
16804 2017-06-19 Martin Liska <mliska@suse.cz>
16807 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16808 to dispatcher function name.
16809 * multiple_target.c (replace_function_decl): New function.
16810 (create_dispatcher_calls): Redirect both edges and references.
16812 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16814 * profile-count.c (profile_count::dump): Dump quality.
16815 (profile_count::differs_from_p): Update for unsigned val.
16816 * profile-count.h (profile_count_quality): New enum.
16817 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16819 2017-06-19 Richard Biener <rguenther@suse.de>
16821 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16822 struct function as arg.
16823 (estimate_numbers_of_iterations): Export overload with loop arg.
16824 (free_numbers_of_iterations_estimates_loop): Use an overload of
16825 free_numbers_of_iterations_estimates instead.
16826 * tree-cfg.c (remove_bb): Adjust.
16827 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16828 * tree-parloops.c (gen_parallel_loop): Likewise.
16829 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16831 (tree_unroll_loops_completely): Likewise.
16832 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16833 Use an overload instead and export.
16834 (estimated_loop_iterations): Adjust.
16835 (max_loop_iterations): Likewise.
16836 (likely_max_loop_iterations): Likewise.
16837 (estimate_numbers_of_iterations): Take struct function as arg
16839 (loop_exits_before_overflow): Adjust.
16840 (free_numbers_of_iterations_estimates_loop): Use an overload.
16841 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16842 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16844 2017-06-19 Richard Biener <rguenther@suse.de>
16847 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16848 RANGE_EXPR conservatively.
16850 2017-06-16 Carl Love <cel@us.ibm.com>
16852 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16853 definitions for vec_float, vec_float2, vec_floato,
16854 vec_floate built-ins.
16855 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16856 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16858 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16859 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16860 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16861 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16862 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16863 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16864 vec_floato): Add builtin defines.
16865 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16866 Update the built-in documentation file for the new built-in
16869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16871 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16872 (mthumb): Mark as the negative of -marm.
16874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16876 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16878 (ARM Options, -mtune): Document that this accepts the same
16879 extension options as -mcpu.
16880 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16884 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16885 permitted extensions.
16887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16889 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16890 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16891 (armv8-m.main): Add option +nodsp.
16892 * config/arm/arm-cpu-cdata.h: Regenerated.
16894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16896 * config/arm/t-fuchsia: New file.
16897 * config.gcc (arm*-*-fuchsia*): Use it.
16899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16901 * config/arm/t-symbian: Rewrite for new option infrastructure.
16903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16905 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16906 (MULTILIB_REQUIRED): Likewise.
16908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16910 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16911 (MULTILIB_RESUE): Likewise.
16912 (MULTILIB_MATCHES): Likewise.
16913 (MULTLIB_REQUIRED): Likewise.
16915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16917 * config/arm/t-rtems: Rewrite for new option framework.
16919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16921 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16922 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16923 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16924 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16925 * config/arm/t-multilib: ... here.
16926 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16927 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16928 armv7-a and armv8*-a when A-profile libraries have not been built.
16929 * config/arm/t-rmprofile: Rewrite.
16931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16933 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16934 with a backslash. Remove the backslash after substituting unescaped
16936 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16940 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16941 the driver to pass through the default setting of -mfloat-abi.
16942 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16944 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16945 (all_feat_combs): New rule.
16946 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16948 * config/arm/t-aprofile: Rewrite.
16950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16952 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16953 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16954 fpu is not specified by the user/command-line.
16955 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16956 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16957 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16958 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16959 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16960 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16964 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16965 * config/arm/t-arm-elf: Rewritten.
16967 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16969 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16970 have some floating-point instructions.
16971 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16972 (TARGET_MAYBE_HARD_FLOAT): New macro.
16973 * config/arm/arm-builtins.c (arm_init_builtins): Use
16974 TARGET_MAYBE_HARD_FLOAT.
16975 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16979 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16980 (configargs.h): Include it.
16981 (arm_print_hint_for_fpu_option): New function.
16982 (arm_parse_fpu_option): New function.
16983 (candidate_extension): New class.
16984 (arm_canon_for_multilib): New function.
16985 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16986 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16987 (ARCH_CANONICAL_SPECS): New macro.
16988 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16992 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16993 are set after handling multilib fragments. Set target_cpu_default2
16996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16998 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17000 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17001 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17003 (check_arch): Likewise.
17004 * config/arm/arm.c (arm_configure_build_target): Handle
17005 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17006 options in the default.
17008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17010 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17011 when an option is an alias of another.
17012 * config/arm/parsecpu.awk (optalias): New parser token.
17013 (gen_comm_data): Mark non-alias options as such. Emit entries
17014 for extension aliases.
17015 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17016 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17017 (armv6kz, armv6zk, armv6t2): Likewise.
17018 (armv7): Make vfpv3-d16 an alias.
17019 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17021 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17022 Sort in canonical order.
17023 (armv8-a): Sort in canonical order.
17024 (armv8.1-a, armv8.2-a): Likewise.
17025 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17027 (cortex-a9): Sort in canonical order.
17028 * config/arm/arm.c (selftests.h): Include it.
17029 (arm_test_cpu_arch_data): New function.
17030 (arm_run_self_tests): New function.
17031 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17032 (targetm): Move declaration to the end of the file.
17033 * arm-cpu-cdata.h: Regenerated.
17035 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17037 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17038 call to target_mode_check describing the type of option passed.
17039 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17040 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17041 arm_parse_cpu_option_name to match parameters against list of
17043 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17044 arm_arch_core_flags data structure.
17045 * config/arm/arm-cpu_cdata.h: Regenerated.
17047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17049 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17051 (arm_print_hint_for_cpu_option): Likewise.
17052 (arm_print_hint_for_arch_option): Likewise.
17053 (arm_parse_cpu_option_name): Likewise.
17054 (arm_parse_arch_option_name): Likewise.
17055 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17056 of entries in the all_fpus list.
17057 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17058 (arm_parse_cpu_option_name): Declare.
17059 (arm_parse_arch_option_name): Declare.
17060 (arm_parse_option_features): Declare.
17061 (arm_intialize_isa): Declare.
17062 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17064 (gen_comm_data): ... here. Make definitions non-static.
17065 * config/arm/arm-cpu-data.h: Regenerated.
17066 * config/arm/arm-cpu-cdata.h: Regenerated.
17068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17070 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17071 (cpu_arch_extension): New structure.
17072 (cpu_arch_option, arch_option, cpu_option): New structures.
17073 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17074 architecture types.
17075 (gen_data): Generate new format data tables.
17076 * config/arm/arm.c (cpu_tune): New structure.
17077 (cpu_option, processors): Delete.
17078 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17079 (arm_print_hint_for_cpu_option): ... this and ...
17080 (arm_print_hint_for_arch_option): ... this.
17081 (arm_parse_arch_cpu_name): Delete. Replace with ...
17082 (arm_parse_cpu_option_name): ... this and ...
17083 (arm_parse_arch_option_name): ... this.
17084 (arm_unrecognized_feature): Change type of target parameter to
17086 (arm_parse_arch_cpu_features): Delete. Replace with ...
17087 (arm_parse_option_features): ... this.
17088 (arm_configure_build_target): Rework to use new configuration data
17090 (arm_print_tune_info): Rework for new configuration data tables.
17091 * config/arm/arm-cpu-data.h: Regenerated.
17092 * config/arm/arm-cpu.h: Regenerated.
17094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17096 * Makefile.in (OBJS): Move sbitmap.o from here ...
17097 (OBJS-libcommon): ... to here.
17099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17101 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17102 (ISA_ALL_CRYPTO): New macro.
17103 (ISA_ALL_SIMD): New macro
17104 (ISA_ALL_FP): New macro.
17105 * config/arm/arm.c (fpu_bitlist): Update initializer.
17106 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17108 (arm9e): Add fpu. Add option for nofp
17109 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17110 (arm926ej-s, arm1026ej-s): Likewise.
17111 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17112 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17113 neon-fp16, neon-vfpv4, nofp and nosimd.
17114 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17115 (cortex-a8): Add fpu. Add option for nofp.
17116 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17117 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17118 (cortex-r4f): Add fpu.
17119 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17120 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17122 (cortex-r8): Likewise.
17123 (cortex-m4): Add fpu. Add option for nofp.
17124 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17125 (cortex-a17.cortex-a7): Likewise.
17126 (cortex-a32): Add fpu. Add options for crypto and nofp.
17127 (cortex-a35, cortex-a53): Likewise.
17128 (cortex-a57): Add fpu. Add option for crypto.
17129 (cortex-a72, cortex-a73): Likewise.
17130 (exynos-m1): Likewise.
17131 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17132 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17133 (cortex-m33): Add fpu. Add option for nofp.
17134 * config/arm/arm-cpu-cdata.h: Regenerated
17135 * config/arm/arm-cpu-data.h: Regenerated.
17137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17139 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17140 (armv5te, armv5tej): Likewise.
17141 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17142 (armv7): Add options fp and vfpv3-d16.
17143 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17144 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17146 (armv7ve): Likewise.
17147 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17148 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17149 (armv8-a): Add nocrypto option.
17150 (armv8.1-a, armv8.2-a): Likewise.
17151 (armv8-m.main): add options fp, fp.dp and nofp.
17153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17155 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17157 (armv8-a+crc): Delete.
17158 (armv8.1-a): Add options simd, crypto and nofp.
17159 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17160 (armv8.2-a+fp16): Delete.
17161 (armv8-m.main): Add option dsp.
17162 (armv8-m.main+dsp): Delete.
17163 (cortex-a8): Add fpu. Add nofp option.
17164 (cortex-a9): Add fpu. Add nofp and nosimd options.
17165 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17166 link to main cpu and architecture data structures.
17167 (gen_comm_data): Only put isa attributes from the main architecture
17169 (option): New statement for architecture and CPU entries.
17170 * arm.c (struct cpu_option): New structure.
17171 (struct processors): Add entry for options.
17172 (arm_unrecognized_feature): New function.
17173 (arm_parse_arch_cpu_name): Ignore any characters after the first
17175 (arm_parse_arch_cpu_feature): New function.
17176 (arm_configure_build_target): Separate out any CPU and architecture
17177 features and parse separately. Don't error out if -mfpu=auto is
17178 used with only an architecture string.
17179 (arm_print_asm_arch_directives): New function.
17180 (arm_file_start): Call it.
17181 * config/arm/arm-cpu-cdata.h: Regenerated.
17182 * config/arm/arm-cpu-data.h: Likewise.
17183 * config/arm/arm-tables.opt: Likewise.
17185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17187 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17188 assembler when it is not -mfpu=auto.
17190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17192 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17193 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17194 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17195 (ASM_CPU_SPEC): Rewrite.
17196 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17197 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17198 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17199 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17200 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17201 copied string is NUL-terminated. Also strip any characters prefixed
17203 (arm_rewrite_selected_arch): New function.
17204 (arm_rewrite_march): New function.
17206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17208 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17209 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17210 (march, mcpu, mtune): Convert to string-based options.
17211 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17212 (arm_parse_arch_cpu_name): New function.
17213 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17214 identify selected architecture or CPU.
17215 (arm_option_save): New function.
17216 (TARGET_OPTION_SAVE): Redefine.
17217 (arm_option_restore): Restore string options.
17218 (arm_option_print): Print string options.
17220 2017-06-16 Martin Sebor <msebor@redhat.com>
17222 PR tree-optimization/80933
17223 PR tree-optimization/80934
17224 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17225 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17226 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17227 (gimple_fold_builtin): Call them.
17229 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17231 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17232 as unlikely; update profile.
17234 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17236 * predict.c (force_edge_cold): Handle declaring edges impossible
17239 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17241 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17243 (try_unroll_loop_completely): Fix reporting.
17245 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17247 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17249 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17252 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17253 if given a non-constant argument for an intrinsic which requires a
17256 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17258 * profile.c (compare_freqs): New function.
17259 (branch_prob): Sort edge list.
17260 (find_spanning_tree): Assume that the list is priority sorted.
17262 2017-06-16 Richard Biener <rguenther@suse.de>
17264 PR tree-optimization/81090
17265 * passes.def (pass_record_bounds): Remove.
17266 * tree-pass.h (make_pass_record_bounds): Likewise.
17267 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17268 make_pass_record_bounds): Likewise.
17269 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17270 not free niter estimates at the beginning but at the end.
17271 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17273 2017-06-16 Richard Biener <rguenther@suse.de>
17275 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17276 initializer to workaround ICE in host GCC 4.8.
17278 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17280 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17282 (clone_inlined_nodes): Update.
17284 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17286 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17287 prefetch settings, and enable prefetching by default at -O3.
17289 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17291 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17292 Set flag_prefetch_loop_arrays according to tuning data.
17294 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17296 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17297 New tune structure.
17298 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17299 [Unrelated to main purpose of the patch] Place the pointer field last
17300 to enable type checking errors when tune structure are wrongly merged.
17301 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17302 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17303 (thunderx2t99_prefetch_tune): New tune constants.
17304 (tune_params *_tunings): Update all tunings (no functional change).
17305 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17306 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17307 from tunings structures.
17309 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17312 * ubsan.c (instrument_null): Add T argument, use it instead
17313 of computing it based on IS_LHS.
17314 (instrument_object_size): Likewise.
17315 (pass_ubsan::execute): Adjust instrument_null and
17316 instrument_object_size callers to pass gimple_get_lhs or
17317 gimple_assign_rhs1 result to it. Use instrument_null instead of
17318 calling get_base_address and instrument_mem_ref. Handle
17319 aggregate call arguments for object-size sanitization.
17321 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17323 PR tree-optimization/81089
17324 * tree-vrp.c (is_masked_range_test): Validate operands of
17327 2017-06-15 Martin Sebor <msebor@redhat.com>
17330 * dumpfile.c (dump_register): Avoid calling memset to initialize
17331 a class with a default ctor.
17332 * gcc.c (struct compiler): Remove const qualification.
17333 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17334 * hash-table.h: Ditto.
17335 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17337 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17338 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17340 * params.h (struct param_info): Make struct members non-const.
17341 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17342 with default initialization.
17343 * vec.h (vec_copy_construct, vec_default_construct): New helper
17345 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17346 with vec_copy_construct.
17347 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17348 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17349 * doc/invoke.texi (-Wclass-memaccess): Document.
17351 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17353 * emit-rtl.h (is_leaf): Update comment about local
17354 register allocator.
17356 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17359 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17360 for a variable to have a section before checking if the section has a
17362 Set section to.persistent if persistent attribute is set.
17363 Warn if .persistent attribute is used on an automatic variable.
17365 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17367 PR rtl-optimization/80474
17368 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17370 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17372 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17375 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17377 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17378 (casesi): Emit bounds checking as RTL.
17379 (casesi_internal_mips16_<mode>): Remove bounds checking.
17381 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17383 * config/xtensa/xtensa.c (xtensa_option_override): Append
17384 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17385 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17386 xtensa_doloop_hooks): Define unconditionally.
17387 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17389 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17390 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17391 for it in xtensa_option_override.
17392 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17393 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17395 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17397 * doc/cppopts.texi: Document '-' special value to -MF.
17399 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17401 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17402 (cortex_a53_fconst): Likewise.
17403 (cortex_a53_fpmul): Likewise.
17404 (cortex_a53_f_load_64): Likewise.
17405 (cortex_a53_f_load_many): Likewise.
17406 (cortex_a53_advsimd_alu): Likewise.
17407 (cortex_a53_advsimd_alu_q): Likewise.
17408 (cortex_a53_advsimd_mul): Likewise.
17409 (cortex_a53_advsimd_mul_q): Likewise.
17410 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17411 Add missing fmul, r2f_cvt and fconst cases.
17413 2017-06-14 Richard Biener <rguenther@suse.de>
17415 PR middle-end/81088
17416 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17418 (fold_binary_loc): When associating do not treat pre-existing
17419 TREE_OVERFLOW on literal constants as a reason to allow
17420 TREE_OVERFLOW on associated literal constants.
17422 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17424 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17425 (MASK_FEATURES): New macro.
17426 * config/sparc/sparc.c (sparc_option_override): Remove the special
17427 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17429 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17431 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17432 a division of 0 if non-call exceptions are enabled.
17434 2017-06-14 Andrew Pinski <apinski@cavium.com>
17435 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17438 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17439 Improve vector initialization code gen for only variable case.
17441 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17443 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17445 2017-06-14 Richard Biener <rguenther@suse.de>
17447 PR tree-optimization/81083
17448 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17451 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17453 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17454 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17455 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17456 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17457 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17458 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17460 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17462 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17463 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17465 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17467 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17469 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17471 * config/rs6000/t-rtems: Don't handle SPE.
17473 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17475 * config/rs6000/t-linux: Don't handle SPE.
17477 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17479 * config/rs6000/eabispe.h: Delete file.
17481 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17483 * config/rs6000/t-spe: Delete file.
17485 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17487 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17488 (rs6000_legitimate_offset_address_p): Return false for anything in
17489 V2SImode or V2SFmode.
17491 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17493 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17494 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17495 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17497 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17498 (rs6000_legitimate_offset_address_p): Ditto.
17499 (rs6000_emit_move): Ditto.
17500 (rs6000_init_builtins): Remove V4HI_type_node.
17502 2017-06-13 Martin Liska <mliska@suse.cz>
17505 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17506 (gate_asan): Likewise.
17507 * asan.h (asan_no_sanitize_address_p): Remove the function.
17508 (sanitize_flags_p): New function.
17509 * builtins.def: Fix coding style.
17510 * common.opt: Use renamed enum value.
17511 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17512 * doc/extend.texi: Document no_sanitize attribute.
17513 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17514 to SANITIZE_UNDEFINED_NONDEFAULT.
17515 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17516 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17517 Use sanitize_flags_p.
17518 * gimplify.c (gimplify_function_tree): Likewise.
17519 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17520 * opts.c (parse_no_sanitize_attribute): New function.
17521 (common_handle_option): Use renamed enum value.
17522 * opts.h (parse_no_sanitize_attribute): Declare.
17523 * tree.c (sanitize_flags_p): New function.
17524 * tree.h: Declared here.
17525 * tsan.c: Use sanitize_flags_p.
17526 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17527 (instrument_mem_ref): Likewise.
17528 (instrument_bool_enum_load): Likewise.
17529 (do_ubsan_in_current_function): Remove the function.
17530 (pass_ubsan::execute): Use sanitize_flags_p.
17531 * ubsan.h: Remove do_ubsan_in_current_function
17532 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17533 (dump_function_to_file): Use it here.
17535 2017-06-13 Martin Jambor <mjambor@suse.cz>
17537 PR tree-optimization/80803
17538 PR tree-optimization/81063
17539 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17540 (propagate_subaccesses_across_link): Enqueue subtree whenever
17541 necessary instead of relying on the caller.
17543 2017-06-13 Martin Jambor <mjambor@suse.cz>
17545 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17546 that have a first_link.
17547 (sort_and_splice_var_accesses): Do not check first_link before
17549 (subtree_mark_written_and_enqueue): Likewise.
17550 (propagate_all_subaccesses): Likewise and do not stop at first
17551 parent with a first_link.
17553 2017-06-13 Martin Jambor <mjambor@suse.cz>
17555 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17558 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17560 * match.pd: New pattern.
17562 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17564 * tree-vrp.c (is_masked_range_test): New function.
17565 (register_edge_assert_for): Determine ranges for
17568 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17570 PR tree-optimization/67328
17571 * fold-const.c (maskable_range_p): New function.
17572 (build_range_check): Generate bittests if possible.
17574 2017-06-13 Martin Liska <mliska@suse.cz>
17576 * gimple-pretty-print.c (dump_probability): Add new argument.
17577 (dump_edge_probability): Dump both probability and count.
17578 (dump_gimple_label): Likewise.
17579 (dump_gimple_bb_header): Likewise.
17581 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17584 * config/avr/avr-devices.c: Fix indentation.
17585 * config/avr/gen-avr-mmcu-specs.c: Dito.
17587 2017-06-13 Richard Biener <rguenther@suse.de>
17589 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17590 instead get vector type from stmt_info.
17591 (vectorizable_reduction): Adjust. Remove dead code.
17593 2017-06-13 Richard Biener <rguenther@suse.de>
17595 PR middle-end/81065
17596 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17597 case of C * (x * C2 + C3).
17598 (fold_addr_of_array_ref_difference): Properly fold index difference.
17600 2017-06-12 David S. Miller <davem@davemloft.net>
17603 * config/sparc/sparc.md (return expander): Emit frame blockage if
17604 function uses alloca.
17606 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17608 * combine.c (make_field_assignment): Check len rather than the mode
17609 precision when calling force_to_mode.
17611 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17613 Support multilibs and devices that see flash in RAM address range.
17616 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17617 (avr_mcu_t) <flash_pm_offset>: New field.
17618 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17619 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17620 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17621 (AVR_TINY_PM_OFFSET): Remove macro.
17622 * config/avr/avr.opt (-mshort-calls): New option.
17623 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17624 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17625 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17626 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17627 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17628 instead of avr_arch->have_jmp_call.
17629 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17630 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17631 avr_arch->flash_pm_offset to define.
17632 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17633 new field flash_pm_offset. Add entry for avrxmega3.
17634 (avr_texinfo): Add entry for avrxmega3.
17635 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17636 attiny212, attiny214,
17637 attiny412, attiny414, attiny416, attiny417,
17638 attiny814, attiny816, attiny817,
17639 attiny1614, attiny1616, attiny1617,
17640 attiny3214, attiny3216, attiny3217.
17641 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17642 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17643 (avr_print_operand_address) [AVR_TINY]: Same.
17644 (avr_asm_init_sections) <readonly_data_section>: Only patch
17645 callback if avr_arch->flash_pm_offset = 0.
17646 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17647 for rodata if avr_arch->flash_pm_offset != 0.
17648 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17649 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17650 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17651 (m_options): Append opt_rcall.
17652 (m_dirnames): Append dir_rcall.
17653 * config/avr/t-multilib: Regenerate.
17655 * configure.ac [target=avr]: Check whether avrxmega3 default
17656 linker description file works as needed.
17657 * configure: Regenerate.
17658 * doc/avr-mmcu.texi: Regenerate.
17659 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17660 <__AVR_ARCH__>: Document avrxmega3 and 103.
17661 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17662 <__AVR_SHORT_CALLS__>: Document it.
17663 <__AVR_PM_BASE_ADDRESS__>: Document it.
17664 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17665 (AVR Variable Attributes) <progmem>: Document this is
17666 not needed for avrxmega3.
17667 (AVR Named Address Spaces) <__flash>: Dito.
17669 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17671 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17673 2017-06-12 Doug Rupp <rupp@adacore.com>
17675 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17676 Append vxworks-stdint.h to the tm_file list.
17677 * config/vxworks-stdint.h: New file.
17679 2017-06-12 Martin Liska <mliska@suse.cz>
17681 PR tree-optimization/81041
17682 * tree-profile.c (gimple_gen_ic_func_profiler):
17683 Create an extra BB in profile-generate
17684 (gimple_gen_time_profiler): Likewise.
17686 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17688 PR tree-optimization/81003
17689 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17690 (update_range_test): Use it instead of force_gimple_operand_gsi.
17692 2017-06-12 Richard Biener <rguenther@suse.de>
17694 PR tree-optimization/81053
17695 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17696 with backedge value not defined in loop. Simplify def stmt
17699 2017-06-11 Tom de Vries <tom@codesourcery.com>
17702 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17704 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17705 nvptx_cannot_force_const_mem.
17707 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17709 * opts.c (finish_options): Move test for flag_split_stack after
17710 it has been initialized.
17712 2017-06-11 Jason Merrill <jason@redhat.com>
17714 * tree.h (id_equal): New.
17715 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17716 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17717 instead of strcmp of IDENTIFIER_POINTER.
17719 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17721 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17722 (mark_all_inlined_calls_cdtor): Fix formating.
17723 (inline_transform): Rescale profile before inlining.
17725 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17727 * cgraph.h (cgraph_edge::clone): Update prototype.
17728 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17729 (cgraph_node::create_clone): Update.
17730 (cgraph_node::create_version_clone): Update.
17731 * tree-inline.c (copy_bb): Update.
17732 (expand_call_inline): Update.
17734 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17736 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17737 factored out from ...
17738 (rs6000_emit_prologue): ... here.
17740 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17742 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17743 factored out from ...
17744 (rs6000_emit_prologue): ... here.
17746 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17748 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17750 (handle_missing_profiles): Fix computation of tp_first_run.
17751 (counts_to_freqs): Do not touch freqs when count is 0.
17753 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17755 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17758 2017-06-10 Tom de Vries <tom@codesourcery.com>
17760 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17761 attributes): Document signal effective target.
17763 2017-06-10 Tom de Vries <tom@codesourcery.com>
17765 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17766 Document effective target stack_size.
17768 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17770 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17771 to the edit_context if they can be auto-applied.
17773 2017-06-9 Ian Lance Taylor <iant@golang.org>
17775 * opts.c (finish_options): If -fsplit-stack, disable implicit
17776 -forder-blocks-and-partition.
17777 * doc/invoke.texi (Optimize Options): Document that when using
17778 -fsplit-stack -forder-blocks-and-partition is not implicitly
17781 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17783 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17784 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17785 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17786 * builtins.def (abort, trap, unreachable): Declare cold.
17787 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17788 * tree-core.h (ECF_COLD): New.
17789 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17790 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17792 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17794 * predict.c (unlikely_executed_stmt_p): Cleanup.
17796 2017-06-09 Richard Biener <rguenther@suse.de>
17798 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17799 model if the ref is always written to.
17801 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17803 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17805 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17807 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17810 2017-06-09 Tom de Vries <tom@codesourcery.com>
17813 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17814 "target cannot support label values" when encountering LABEL_REF.
17816 2017-06-09 Martin Liska <mliska@suse.cz>
17818 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17819 (gimple_gen_ic_func_profiler): Emit direct comparison
17820 of __gcov_indirect_call_callee with NULL.
17821 (gimple_gen_time_profiler): Change probability from
17822 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17824 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17826 * profile.c (edge_gcov_counts): Turn to pointer.
17827 (compute_branch_probabilities, compute_branch_probabilities): Update.
17828 (branch_prob): Do not clear edge_gcov_count.
17829 * profile.h (edge_gcov_counts): Turn to pointer.
17830 (edge_gcov_count): Update.
17832 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17834 * gimple.h (gimple_check_failed): Mark cold.
17836 2017-06-09 Richard Biener <rguenther@suse.de>
17838 PR tree-optimization/66623
17839 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17840 refactor check_reduction into two parts, properly computing
17841 whether we have to check reduction validity for outer loop
17844 2017-06-09 Richard Biener <rguenther@suse.de>
17846 PR tree-optimization/79483
17847 * graphite-scop-detection.c (order): New global.
17848 (get_order): Compute bb to order mapping that satisfies code
17849 generation constraints.
17850 (cmp_pbbs): New helper.
17851 (build_scops): Start domwalk at entry block, sort generated
17854 2017-06-09 Richard Biener <rguenther@suse.de>
17856 PR middle-end/81007
17857 * ipa-polymorphic-call.c
17858 (ipa_polymorphic_call_context::restrict_to_inner_class):
17859 Skip FIELD_DECLs with error_mark_node type.
17860 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17863 2017-06-09 Martin Liska <mliska@suse.cz>
17865 * predict.c (struct branch_predictor): New struct.
17866 (test_prediction_value_range): New test.
17867 (predict_c_tests): New function.
17868 * selftest-run-tests.c (selftest::run_tests): Run the function.
17869 * selftest.h: Declare new tests.
17871 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17874 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17875 gen_add3_insn did not fail.
17876 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17877 r0, construct that number in a temporary reg and add that reg to r0.
17878 If asked to put the result in r0 as well, fail.
17880 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17882 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17883 for early expansion of vec_eqv.
17885 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17887 PR middle-end/81005
17888 * ubsan.c (instrument_null): Avoid pointless code temporary.
17889 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17891 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17895 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17898 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17901 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17903 * predict.c (unlikely_executed_edge_p): Move ahead.
17904 (probably_never_executed_edge_p): Use it.
17906 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17908 PR middle-end/79988
17909 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17910 gimple_call_builtin_p call.
17912 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17914 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17915 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17916 rtl_check_failed_type2, rtl_check_failed_code1,
17917 rtl_check_failed_code2, rtl_check_failed_code_mode,
17918 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17919 rtvec_check_failed_bounds, rtl_check_failed_flag,
17920 _fatal_insn_not_found, _fatal_insn): Likewise.
17921 * tree.h (tree_contains_struct_check_failed,
17922 tree_check_failed, tree_not_check_failed,
17923 tree_class_check_failed, tree_range_check_failed,
17924 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17925 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17926 tree_operand_check_failed, omp_clause_check_failed,
17927 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17930 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17932 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17933 flag_branch_probabilities.
17934 * ipa-inline.c (edge_badness): Likewise.
17935 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17936 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17937 * predict.c (maybe_hot_frequency_p): Likewise.
17938 (probably_never_executed): Likewise.
17939 * sched-ebb.c (schedule_ebbs): Likewise.
17940 * sched-rgn.c (find_single_block_region): Likewise.
17941 * tracer.c (tail_duplicate): Likewise.
17943 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17945 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17946 longer requires x_flag_profile_use.
17948 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17950 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17951 instead of flag_reorder_blocks_and_partition.
17952 * dbxout.c (dbxout_function_end): Likewise.
17953 * dwarf2out.c (gen_subprogram_die): Likewise.
17954 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17955 * hw-doloop.c (reorg_loops): Likewise.
17956 * varasm.c (assemble_start_function,
17957 assemble_end_function): Likewise.
17958 (decide_function_section): Do not check for
17959 flag_reorder_blocks_and_partition.
17961 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17963 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17965 (chkp_get_hard_register_fake_addr_expr): Ditto.
17966 (chkp_build_addr_expr): Add check for hard reg case.
17967 (chkp_parse_array_and_component_ref): Ditto.
17968 (chkp_find_bounds_1): Ditto.
17969 (chkp_process_stmt): Don't generate bounds store for
17972 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17974 * predict.c (maybe_hot_bb_p): Do not check profile status.
17975 (maybe_hot_edge_p): Likewise.
17976 (probably_never_executed): Check for zero counts even if profile
17978 (unlikely_executed_edge_p): New function.
17979 (unlikely_executed_stmt_p): New function.
17980 (unlikely_executed_bb_p): New function.
17981 (set_even_probabilities): Use unlikely predicates.
17982 (combine_predictions_for_bb): Likewise.
17983 (predict_paths_for_bb): Likewise.
17984 (predict_paths_leading_to_edge): Likewise.
17985 (determine_unlikely_bbs): New function.
17986 (estimate_bb_frequencies): Use it.
17987 (compute_function_frequency): Use zero counts even if profile is
17989 * profile-count.h: Fix typo.
17991 2017-08-08 Julia Koval <julia.koval@intel.com>
17993 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17994 _mm512_mask_cvtsepi16_storeu_epi8,
17995 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17996 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17997 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17998 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17999 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18000 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18001 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18002 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18003 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18004 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18005 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18006 __builtin_ia32_pmovuswb256mem_mask,
18007 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18008 __builtin_ia32_pmovwb512mem_mask): New builtins.
18010 2017-08-08 Julia Koval <julia.koval@intel.com>
18012 PR target/73350,80862
18013 * config/i386/subst.md (round): Fix round pattern.
18014 * config/i386/i386.c (ix86_erase_embedded_rounding):
18015 Fix erasing rounding for the fixed pattern.
18017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18019 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18021 2017-06-08 Martin Liska <mliska@suse.cz>
18023 PR gcov-profile/80911
18024 * gcov.c (block_info::block_info): New constructor.
18026 2017-06-07 Carl Love <cel@us.ibm.com>
18028 * config/rs6000/rs6000-c: The return type of the following
18029 built-in functions was implemented as int not long long. Fix sign
18030 of return value for the unsigned version of vec_mulo and vec_mule.
18031 vector unsigned long long vec_bperm (vector unsigned long long,
18032 vector unsigned char)
18033 vector signed long long vec_mule (vector signed int,
18035 vector unsigned long long vec_mule (vector unsigned int,
18036 vector unsigned int)
18037 vector signed long long vec_mulo (vector signed int,
18039 vector unsigned long long vec_mulo (vector unsigned int,
18040 vector unsigned int)
18041 * doc/extend.texi: Fix the documentation for the built-in
18044 2017-06-07 Carl Love <cel@us.ibm.com>
18047 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18050 2017-06-07 Carl Love <cel@us.ibm.com>
18052 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18053 support, Generate doublehv for signed int/float for BE case only.
18055 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18057 * doc/invoke.texi (mcx16): Rewrite.
18059 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18061 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18062 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18063 *mov<mode>_softfloat, and an anonymous splitter): Use
18064 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18066 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18068 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18070 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18071 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18072 (rs6000_debug_reg_global): Adjust.
18073 (rs6000_init_hard_regno_mode_ok): Adjust.
18074 (rs6000_dbx_register_number): Adjust.
18075 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18076 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18077 Remove SPE_ACC and SPEFSCR.
18078 (REG_ALLOC_ORDER): Ditto.
18079 (FRAME_POINTER_REGNUM): Change to 111.
18080 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18081 (REG_CLASS_NAMES): Ditto.
18082 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18083 (REGISTER_NAMES): Ditto.
18084 (ADDITIONAL_REG_NAMES): Ditto.
18085 (rs6000_reg_names): Ditto.
18086 * config/rs6000/rs6000.md: Renumber some register number
18089 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18091 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18093 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18094 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18096 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18097 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18098 Delete the SPE high registers.
18099 (REG_ALLOC_ORDER): Ditto.
18100 (enum reg_class): Remove SPE_HIGH_REGS.
18101 (REG_CLASS_NAMES): Ditto.
18102 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18103 (REGISTER_NAMES): Ditto.
18104 (rs6000_reg_names): Ditto.
18105 * doc/tm.texi.in: Remove SPE as example.
18106 * doc/tm.texi: Regenerate.
18108 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18110 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18111 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18112 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18113 * config/rs6000/rs6000.md (type): Remove "brinc".
18115 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18117 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18118 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18119 * config/rs6000/linuxspe.h: Delete file.
18120 * config/rs6000/rs6000.md: Don't include spe.md.
18121 * config/rs6000/spe.h: Delete file.
18122 * config/rs6000/spe.md: Delete file.
18123 * config/rs6000/t-rs6000: Remove spe.md.
18125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18127 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18128 (reg_or_none500mem_operand): Delete.
18129 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18130 instead of reg_or_none500mem_operand.
18132 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18134 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18135 handling of SPE flags.
18136 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18138 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18140 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18142 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18143 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18144 paired_divv2sf3): Similar.
18145 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18146 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18147 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18149 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18150 Rename the paired_* instruction patterns.
18151 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18153 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18154 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18155 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18156 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18157 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18158 PAIRED_VECTOR_MODE.
18159 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18160 (spe_func_has_64bit_regs_p): Delete.
18161 (spe_expand_predicate_builtin): Delete.
18162 (spe_expand_evsel_builtin): Delete.
18163 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18164 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18165 (invalid_e500_subreg): Delete.
18166 (rs6000_legitimize_address): Always force_reg op2 as well, for
18167 paired single memory accesses.
18168 (rs6000_member_type_forces_blk): Delete.
18169 (rs6000_spe_function_arg): Delete.
18170 (rs6000_expand_unop_builtin): Delete SPE handling.
18171 (rs6000_expand_binop_builtin): Ditto.
18172 (spe_expand_stv_builtin): Delete.
18173 (bdesc_2arg_spe): Delete.
18174 (spe_expand_builtin): Delete.
18175 (spe_expand_predicate_builtin): Delete.
18176 (spe_expand_evsel_builtin): Delete.
18177 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18178 (spe_init_builtins): Delete.
18179 (spe_func_has_64bit_regs_p): Delete.
18180 (savres_routine_name): Delete "info" parameter. Adjust callers.
18181 (rs6000_emit_stack_reset): Ditto.
18182 (rs6000_dwarf_register_span): Delete.
18183 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18184 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18185 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18187 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18189 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18190 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18191 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18192 mulv2sf3, divv2sf3): Delete expanders.
18194 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18196 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18198 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18200 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18201 * config/rs6000/rs6000.c: Ditto.
18203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18205 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18206 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18207 comparison_operator.
18209 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18211 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18212 * config/rs6000/rs6000.opt: Ditto.
18213 * config/rs6000/t-rtems: Ditto.
18215 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18217 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18218 TARGET_E500_SINGLE by 0, simplify.
18219 * config/rs6000/rs6000.c: Ditto.
18220 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18221 (spe_build_register_parallel): Delete.
18222 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18223 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18224 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18225 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18226 (E500_CONVERT): Delete.
18227 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18229 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18231 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18232 * config/rs6000/dfp.md: Ditto.
18233 (negdd2, *negdd2_fpr): Merge.
18234 (absdd2, *absdd2_fpr): Merge.
18235 (negtd2, *negtd2_fpr): Merge.
18236 (abstd2, *abstd2_fpr): Merge.
18237 * config/rs6000/e500.h: Delete file.
18238 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18239 TARGET_FPRS by 1 and simplify.
18240 * config/rs6000/rs6000-c.c: Ditto.
18241 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18242 TARGET_DF_SPE by 0.
18243 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18245 * config/rs6000/rs6000.md: Ditto.
18246 (floatdidf2, *floatdidf2_fpr): Merge.
18247 (move_from_CR_gt_bit): Delete.
18248 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18249 (E500_CR_IOR_COMPARE): Delete.
18250 (All patterns that require !TARGET_FPRS): Delete.
18251 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18253 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18255 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18257 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18259 * graphds.c (add_edge): Intitialize edge's attached data.
18260 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18261 pointer parameter. Call pointed function on each edge during
18262 graph traversing. Skip traversing the edge when the function
18264 (graphds_dfs, graphds_scc): Ditto.
18265 (for_each_edge): New parameter. Pass the new parameter to callback
18267 * graphds.h (skip_edge_callback): New function pointer type.
18268 (graphds_dfs, graphds_scc): New function pointer parameter.
18269 (graphds_edge_callback, for_each_edge): New parameter.
18271 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18273 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18274 out code checking if runtime alias check is possible to below ...
18275 Call the new function.
18276 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18277 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18279 2017-06-07 Marek Polacek <polacek@redhat.com>
18282 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18283 TYPE_OVERFLOW_WRAPS check.
18285 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18287 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18288 if versioning is required.
18289 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18290 peeling with the check for versioning.
18292 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18294 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18295 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18296 Set true to new parameter if new ssa variable is defined.
18297 (vect_gen_vector_loop_niters): Refactor. Set range information
18298 for the new vector loop bound variable.
18299 (vect_do_peeling): Ditto.
18301 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18303 * tree-affine.c (ssa.h): Include header file.
18304 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18305 has wrapping overflow behavior.
18307 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18309 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18313 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18314 (tree_to_aff_combination): ... here.
18316 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18318 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18319 reg_pressure model function.
18320 (ivopts_global_cost_for_size): Delete.
18321 (determine_set_costs, iv_ca_recount_cost): Call new model function
18322 ivopts_estimate_reg_pressure.
18324 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18326 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18327 expensive than udiv. Remove floating point cases from mod.
18329 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18331 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18332 Increase idiv cost.
18334 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18336 * config/aarch64/aarch64.md
18337 (copysignsf3): Fix mask generation.
18339 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18341 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18343 (class dump_manager): Add register_dumps method.
18344 * dumpfile.c: Include langhooks.h.
18345 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18346 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18347 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18348 (dump_manager::dump_register): Start with 512 entries instead of 32.
18349 (dump_manager::register_dumps): New method.
18350 * toplev.c (general_init): Instead of invoking register_dumps
18351 langhook, invoke register_dumps method on the dump manager.
18352 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18355 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18357 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18358 Say that # requires an associated define_split to exist, and that
18359 the define_split must be suitable for use after register allocation.
18361 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18363 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18364 (compute_outgoing_frequencies): Also initialize zero counts.
18365 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18366 around loops; preserve more of profile when nothing changes.
18368 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18370 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18372 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18373 * config/arm/arm-cpu-cdata.h: Regenerate.
18374 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18375 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18376 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18377 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18378 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18380 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18382 * config/arm/t-rmprofile: Likewise.
18383 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18385 2017-06-06 David S. Miller <davem@davemloft.net>
18388 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18389 blockage if function uses alloca.
18391 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18393 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18394 New "uid" fields to hold pretty-print IDs of group and ref.
18395 Memory references are now identified as <group_id>:<ref_id>
18396 instead of using [random] addresses.
18397 (dump_mem_details): Simplify, no functional change.
18398 (dump_mem_ref): Simplify and make output more concise.
18399 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18400 (find_or_create_group): Initialize group uid.
18401 (record_ref): Initialize ref uid. Improve debug output.
18402 (prune_group_by_reuse, should_issue_prefetch_p,)
18403 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18404 (mark_nontemporal_store, determine_loop_nest_reuse):
18405 Improve debug output.
18407 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18409 * dbgcnt.def (prefetch): New debug counter.
18410 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18411 (schedule_prefetches): Stop issueing prefetches if debug counter
18414 2017-06-06 Tom de Vries <tom@codesourcery.com>
18416 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18417 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18419 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18421 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18422 Use aarch64_reg_or_zero predicate for operand 4.
18423 (aarch64_compare_and_swap<mode> define_insn_and_split):
18424 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18425 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18427 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18429 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18430 (arm_compute_save_core_reg_mask): This.
18431 (thumb1_compute_save_reg_mask): Rename into ...
18432 (thumb1_compute_save_core_reg_mask): This.
18433 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18434 (arm_compute_frame_layout): Likewise.
18436 2017-06-06 Richard Biener <rguenther@suse.de>
18438 PR tree-optimization/80974
18439 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18440 keep or clear leaders SSA info.
18442 2017-06-06 Tom de Vries <tom@codesourcery.com>
18444 * config/nvptx/nvptx.c (split_mode_p): New function.
18445 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18447 2017-06-06 Tom de Vries <tom@codesourcery.com>
18449 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18451 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18454 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18457 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18459 * shrink-wrap.c (handle_simple_exit): Update profile.
18460 (try_shrink_wrapping): Upate profile.
18462 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18464 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18465 (tree_guess_outgoing_edge_probabilities): New.
18466 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18467 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18469 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18471 * ipa-split.c (split_function): Initialize return bb profile.
18473 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18475 * profile.c (compute_branch_probabilities): Also initialize
18476 EXIT_BLOCK profile.
18478 2017-06-06 Richard Biener <rguenther@suse.de>
18480 PR tree-optimization/80928
18481 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18482 (vect_analyze_loop_operations): Properly guard analysis for
18484 (vect_transform_loop): Likewise.
18485 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18486 (vect_model_induction_cost): Do not cost for pure SLP.
18487 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18488 of induction in inner loop vectorization.
18489 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18490 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18491 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18493 (vect_analyze_slp_cost_1): Cost induction.
18494 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18495 (vect_get_slp_vect_defs): Likewise.
18496 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18497 (vect_transform_stmt): Handle SLP reductions.
18498 * tree-vectorizer.h (vectorizable_induction): Adjust.
18500 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18502 * config/rs6000/rs6000.c (make_resolver_func): Update
18503 init_lowered_empty_function call.
18505 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18507 * doc/invoke.texi: Document the -fprofile-abs-path option.
18508 * common.opt (fprofile-abs-path): New option.
18509 * gcov-io.h (gcov_write_filename): Declare.
18510 * gcov-io.c (gcov_write_filename): New function.
18511 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18512 * profile.c (output_location): Likewise.
18514 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18516 * shring-wrap.c: Revert accidental commit.
18518 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18520 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18522 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18524 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18526 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18527 profile in callgraph edge.
18528 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18529 (apply_scale): Likewise.
18530 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18531 Also scale profile when inlining function with zero profile.
18532 (initialize_cfun): Update exit block profile even when it is zero.
18533 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18534 when profile is read.
18536 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18538 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18539 (CLONE_*): New constants to define the processors we can generate
18540 code for with the target_clone attribute.
18541 (rs6000_clone_map): New array to identify which clone processors
18542 the current program is running on.
18543 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18544 target_clone attribute.
18545 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18546 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18547 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18548 (cpu_expand_builtin): Add support for target_clone attribute.
18549 (rs6000_valid_attribute_p): Allow "default" attribute.
18550 (get_decl_name): New debug function to simplify printing the
18551 current function name in debugging statements.
18552 (rs6000_clone_priority): New functions to support the target_clone
18553 attribute, and be able to generate code to switch between ISA 2.05
18554 through ISA 3.0 (power6 through power9).
18555 (rs6000_compare_version_priority): Likewise.
18556 (rs6000_get_function_versions_dispatcher): Likewise.
18557 (make_resolver_func): Likewise.
18558 (add_condition_to_bb): Likewise.
18559 (dispatch_function_versions): Likewise.
18560 (rs6000_generate_version_dispatcher_body): Likewise.
18561 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18562 (fusion_gpr_load_p): Fix a spacing issue.
18563 * doc/extend.texi (Common Function Attributes): Document that the
18564 PowerPC supports the target_clone attribute.
18566 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18568 * config/arm/arm.h: explain F symbol found in description of ARM
18569 register allocation in its legend.
18571 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18573 * config/mips/frame-header-opt.c: Include profile-count.h.
18574 * config/riscv/riscv.c: Include profile-count.h
18576 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18578 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18580 (sm_set_flag_if_changed): Add bbs field.
18581 (execute_sm_if_changed_flag_set): Pass BBS.
18582 (execute_sm): Update.
18584 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18586 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18589 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18591 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18592 (peephole2): New peephole2 to emit the above.
18593 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18595 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18597 * config/aarch64/aarch64.c (define_peephole2 above
18598 *sub_<shift>_<mode>): New peephole.
18600 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18602 * config/i386/i386.c (make_resolver_func): Update.
18603 * Makefile.in: Add profile-count.h and profile-count.o
18604 * auto-profile.c (afdo_indirect_call): Update to new API.
18605 (afdo_set_bb_count): Update.
18606 (afdo_propagate_edge): Update.
18607 (afdo_propagate_circuit): Update.
18608 (afdo_calculate_branch_prob): Update.
18609 (afdo_annotate_cfg): Update.
18610 * basic-block.h: Include profile-count.h
18611 (struct edge_def): Turn count to profile_count.
18612 (struct basic_block_def): Likewie.
18613 (REG_BR_PROB_BASE): Move to profile-count.h
18614 (RDIV): Move to profile-count.h
18615 * bb-reorder.c (max_entry_count): Turn to profile_count.
18616 (find_traces): Update.
18617 (rotate_loop):Update.
18618 (connect_traces):Update.
18619 (sanitize_hot_paths):Update.
18620 * bt-load.c (migrate_btr_defs): Update.
18621 * cfg.c (RDIV): Remove.
18622 (init_flow): Use alloc_block.
18623 (alloc_block): Uninitialize count.
18624 (unchecked_make_edge): Uninitialize count.
18625 (check_bb_profile): Update.
18626 (dump_edge_info): Update.
18627 (dump_bb_info): Update.
18628 (update_bb_profile_for_threading): Update.
18629 (scale_bbs_frequencies_int): Update.
18630 (scale_bbs_frequencies_gcov_type): Update.
18631 (scale_bbs_frequencies_profile_count): New.
18632 * cfg.h (update_bb_profile_for_threading): Update.
18633 (scale_bbs_frequencies_profile_count): Declare.
18634 * cfgbuild.c (compute_outgoing_frequencies): Update.
18635 (find_many_sub_basic_blocks): Update.
18636 * cfgcleanup.c (try_forward_edges): Update.
18637 (try_crossjump_to_edge): Update.
18638 * cfgexpand.c (expand_gimple_tailcall): Update.
18639 (construct_exit_block): Update.
18640 * cfghooks.c (verify_flow_info): Update.
18641 (dump_bb_for_graph): Update.
18642 (split_edge): Update.
18643 (make_forwarder_block): Update.
18644 (duplicate_block): Update.
18645 (account_profile_record): Update.
18646 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18647 (get_estimated_loop_iterations): Update.
18648 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18649 (single_likely_exit): Update.
18650 * cfgloopmanip.c (scale_loop_profile): Update.
18652 (set_zero_probability): Update.
18653 (lv_adjust_loop_entry_edge): Update.
18654 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18655 (purge_dead_edges): Update.
18656 (rtl_account_profile_record): Update.
18657 * cgraph.c (cgraph_node::create): Uninitialize count.
18658 (symbol_table::create_edge): Uninitialize count.
18659 (cgraph_update_edges_for_call_stmt_node): Update.
18660 (cgraph_edge::dump_edge_flags): Update.
18661 (cgraph_node::dump): Update.
18662 (cgraph_edge::maybe_hot_p): Update.
18663 * cgraph.h: Include profile-count.h
18664 (create_clone), create_edge, create_indirect_edge): Update.
18665 (cgraph_node): Turn count to profile_count.
18666 (cgraph_edge0: Likewise.
18667 (make_speculative, clone): Update.
18668 (create_edge): Update.
18669 (init_lowered_empty_function): Update.
18670 * cgraphclones.c (cgraph_edge::clone): Update.
18671 (duplicate_thunk_for_node): Update.
18672 (cgraph_node::create_clone): Update.
18673 * cgraphunit.c (cgraph_node::analyze): Update.
18674 (cgraph_node::expand_thunk): Update.
18675 * final.c (dump_basic_block_info): Update.
18676 * gimple-streamer-in.c (input_bb): Update.
18677 * gimple-streamer-out.c (output_bb): Update.
18678 * graphite.c (print_global_statistics): Update.
18679 (print_graphite_scop_statistics): Update.
18680 * hsa-brig.c: Include basic-block.h.
18681 * hsa-dump.c: Include basic-block.h.
18682 * hsa-gen.c (T sum_slice): Update.
18683 (convert_switch_statements):Update.
18684 * hsa-regalloc.c: Include basic-block.h.
18685 * ipa-chkp.c (chkp_produce_thunks): Update.
18686 * ipa-cp.c (struct caller_statistics): Update.
18687 (init_caller_stats): Update.
18688 (gather_caller_stats): Update.
18689 (ipcp_cloning_candidate_p): Update.
18690 (good_cloning_opportunity_p): Update.
18691 (get_info_about_necessary_edges): Update.
18692 (dump_profile_updates): Update.
18693 (update_profiling_info): Update.
18694 (update_specialized_profile): Update.
18695 (perhaps_add_new_callers): Update.
18696 (decide_about_value): Update.
18697 (ipa_cp_c_finalize): Update.
18698 * ipa-devirt.c (struct odr_type_warn_count): Update.
18699 (struct decl_warn_count): Update.
18700 (struct final_warning_record): Update.
18701 (possible_polymorphic_call_targets): Update.
18702 (ipa_devirt): Update.
18703 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18704 * ipa-icf.c (sem_function::merge): Update.
18705 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18706 * ipa-inline.c (compute_uninlined_call_time): Update.
18707 (compute_inlined_call_time): Update.
18708 (want_inline_small_function_p): Update.
18709 (want_inline_self_recursive_call_p): Update.
18710 (edge_badness): Update.
18711 (lookup_recursive_calls): Update.
18712 (recursive_inlining): Update.
18713 (inline_small_functions): Update.
18714 (dump_overall_stats): Update.
18715 (dump_inline_stats): Update.
18716 * ipa-profile.c (ipa_profile_generate_summary): Update.
18717 (ipa_propagate_frequency): Update.
18718 (ipa_profile): Update.
18719 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18720 * ipa-utils.c (ipa_merge_profiles): Update.
18721 * loop-doloop.c (doloop_modify): Update.
18722 * loop-unroll.c (report_unroll): Update.
18723 (unroll_loop_runtime_iterations): Update.
18724 * lto-cgraph.c (lto_output_edge): Update.
18725 (lto_output_node): Update.
18726 (input_node): Update.
18727 (input_edge): Update.
18728 (merge_profile_summaries): Update.
18729 * lto-streamer-in.c (input_cfg): Update.
18730 * lto-streamer-out.c (output_cfg): Update.
18731 * mcf.c (create_fixup_graph): Update.
18732 (adjust_cfg_counts): Update.
18733 (sum_edge_counts): Update.
18734 * modulo-sched.c (sms_schedule): Update.
18735 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18736 * predict.c (maybe_hot_count_p): Update.
18737 (probably_never_executed): Update.
18738 (dump_prediction): Update.
18739 (combine_predictions_for_bb): Update.
18740 (propagate_freq): Update.
18741 (handle_missing_profiles): Update.
18742 (counts_to_freqs): Update.
18743 (rebuild_frequencies): Update.
18744 (force_edge_cold): Update.
18745 * predict.h: Include profile-count.h
18746 (maybe_hot_count_p, counts_to_freqs): UPdate.
18747 * print-rtl-function.c: Do not include cfg.h
18748 * print-rtl.c: Include basic-block.h
18749 * profile-count.c: New file.
18750 * profile-count.h: New file.
18751 * profile.c (is_edge_inconsistent): Update.
18752 (correct_negative_edge_counts): Update.
18753 (is_inconsistent): Update.
18754 (set_bb_counts): Update.
18755 (read_profile_edge_counts): Update.
18756 (compute_frequency_overlap): Update.
18757 (compute_branch_probabilities): Update; Initialize and deinitialize
18759 (branch_prob): Update.
18760 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18761 (edge_gcov_count): New.
18762 (bb_gcov_count): New.
18763 * shrink-wrap.c (try_shrink_wrapping): Update.
18764 * tracer.c (better_p): Update.
18765 * trans-mem.c (expand_transaction): Update.
18766 (ipa_tm_insert_irr_call): Update.
18767 (ipa_tm_insert_gettmclone_call): Update.
18768 * tree-call-cdce.c: Update.
18769 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18770 (gimple_duplicate_sese_tail): Update.
18771 (gimple_account_profile_record): Update.
18772 (execute_fixup_cfg): Update.
18773 * tree-inline.c (copy_bb): Update.
18774 (copy_edges_for_bb): Update.
18775 (initialize_cfun): Update.
18776 (freqs_to_counts): Update.
18777 (copy_cfg_body): Update.
18778 (expand_call_inline): Update.
18779 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18780 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18781 (try_unroll_loop_completely): Update.
18782 (try_peel_loop): Update.
18783 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18784 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18785 * tree-ssa-loop-split.c (connect_loops): Update.
18786 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18787 * tree-ssa-reassoc.c (branch_fixup): Update.
18788 * tree-ssa-tail-merge.c (replace_block_by): Update.
18789 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18790 (compute_path_counts): Update.
18791 (update_profile): Update.
18792 (recompute_probabilities): Update.
18793 (update_joiner_offpath_counts): Update.
18794 (estimated_freqs_path): Update.
18795 (freqs_to_counts_path): Update.
18796 (clear_counts_path): Update.
18797 (ssa_fix_duplicate_block_edges): Update.
18798 (duplicate_thread_path): Update.
18799 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18800 (struct switch_conv_info): Update.
18801 * tree-tailcall.c (decrease_profile): Update.
18802 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18803 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18804 * value-prof.c (check_counter): Update.
18805 (gimple_divmod_fixed_value): Update.
18806 (gimple_mod_pow2): Update.
18807 (gimple_mod_subtract): Update.
18808 (gimple_ic_transform): Update.
18809 (gimple_stringop_fixed_value): Update.
18810 * value-prof.h (gimple_ic): Update.
18812 2017-06-02 Carl Love <cel@us.ibm.com>
18814 * config/rs6000/rs6000-c: Add support for built-in functions
18815 vector double vec_doublee (vector signed int);
18816 vector double vec_doublee (vector unsigned int);
18817 vector double vec_doublee (vector float);
18818 vector double vec_doubleh (vector signed int);
18819 vector double vec_doubleh (vector unsigned int);
18820 vector double vec_doubleh (vector float);
18821 vector double vec_doublel (vector signed int);
18822 vector double vec_doublel (vector unsigned int);
18823 vector double vec_doublel (vector float);
18824 vector double vec_doubleo (vector signed int);
18825 vector double vec_doubleo (vector unsigned int);
18826 vector double vec_doubleo (vector float);.
18827 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18828 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18830 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18831 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18832 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18834 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18835 vec_doublel, vec_doubleh.
18836 * doc/extend.texi: Update the built-in documentation file for the
18837 new built-in functions.
18839 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18842 * ipa-inline-analysis.c (free_growth_caches): Set
18843 edge_removal_hook_holder to NULL after removing it.
18845 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18847 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18848 comparision with zero.
18850 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18852 for early expansion of vec_min and vec_max builtins.
18853 (builtin_function_type): Add min/max unsigned variants to those
18854 identified as having unsigned arguments.
18856 2017-06-02 Olivier Hainque <hainque@adacore.com>
18858 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18860 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18862 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18863 Use VALL_F16 iterator rather than VALL.
18865 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18867 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18868 Emit CBNZ inside loop when doing a strong exchange and comparing
18869 against zero. Generate the CC flags after the loop.
18871 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18873 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18874 (dl_section_ref): New.
18875 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18876 On AIX, append an expression to subtract the size of the
18877 section length to dl_section_ref.
18879 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18881 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18882 for early expansion of vector absolute builtins.
18884 2017-06-02 Richard Biener <rguenther@suse.de>
18886 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18887 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18889 2017-06-02 Richard Biener <rguenther@suse.de>
18891 PR tree-optimization/80948
18892 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18893 stmt order as well.
18895 2017-06-02 Richard Biener <rguenther@suse.de>
18897 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18899 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18900 for inductions as relevant.
18902 2017-06-02 Richard Biener <rguenther@suse.de>
18904 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18905 (vectorizable_induction): ... this. Remove dead code.
18907 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18909 * builtins. (expand_builtin_alloca): Remove second parameter and
18910 infer its value from the first parameter instead.
18911 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18913 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18915 PR rtl-optimization/80903
18916 * loop-doloop.c (add_test): Unshare sequence.
18918 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18920 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18922 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18924 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18926 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18927 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18928 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18929 (xlogue_layout::get_stub_rtx): Make static.
18930 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18931 (xlogue_layout::compute_stub_managed_regs): Rename to...
18932 (xlogue_layout::count_stub_managed_regs): ...this.
18933 (xlogue_layout::is_stub_managed_reg): New function.
18934 (xlogue_layout::m_stub_names): Rename to...
18935 (xlogue_layout::s_stub_names): ...this, make static.
18936 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18937 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18938 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18939 xlogue_layout::s_stub_names): Instantiate statics.
18940 (stub_managed_regs): Remove.
18941 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18942 (disable_call_ms2sysv_xlogues): Rename to...
18943 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18944 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18946 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18947 change after reload_completed.
18948 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18950 (ix86_expand_prologue): Likewise.
18951 (ix86_expand_epilogue): Likewise.
18952 (ix86_expand_split_stack_prologue): Likewise.
18953 (ix86_compute_frame_layout): Remove frame parameter ...
18954 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18955 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18957 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18958 (ix86_frame): Move from here ...
18959 * config/i386/i386.h (ix86_frame): ... to here.
18960 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18961 complete ix86_frame data structure instead. Remove some_ld_name.
18963 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18965 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18966 symbols that hold a DECL_VALUE_EXPR.
18968 2017-06-01 Martin Jambor <mjambor@suse.cz>
18970 PR tree-optimization/80898
18971 * tree-sra.c (process_subtree_disqualification): Removed.
18972 (disqualify_candidate): Do not acll
18973 process_subtree_disqualification.
18974 (subtree_mark_written_and_enqueue): New function.
18975 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18976 RHS has been disqualified and re-queue LHS if necessary. Apart
18977 from that, ignore disqualified RHS.
18979 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18981 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18982 address fetch for z10 or later.
18984 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18986 * config/arc/arc.md (tst_movb): Add guard when splitting.
18988 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18990 * config/arc/arc.c (arc_can_eliminate): Test against
18991 arc_frame_pointer_needed.
18993 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18995 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18996 to prevent store reordering.
18997 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18998 (type): Add block type.
18999 (stack_tie): Define special instruction to be used in
19002 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19004 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19005 constraint. It is not valid for the pattern.
19006 (noncommutative_binary_comparison): Likewise.
19008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19010 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19013 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19015 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19016 be used by the reg-alloc.
19018 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19020 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19021 reg-alloc when having mul64 or mul32x16 instructions.
19022 (mulsidi3): Likewise.
19023 (umulsidi3): Likewise.
19024 (mulsi32x16): New pattern.
19025 (mulsi64): Likewise.
19026 (mulsidi64): Likewise.
19027 (umulsidi64): Likewise.
19028 (MUL32x16_REG): Define.
19029 (mul64_600): Use MUL32x16_REG.
19030 (mac64_600): Likewise.
19031 (umul64_600): Likewise.
19032 (umac64_600): Likewise.
19034 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19036 * config/arc/arc.md (mulsi3_700): Make it commutative.
19038 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19040 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19042 (*sign_extendsidi2_insn): Likewise for movstosw.
19044 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19046 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19047 the type of the input discriminant value. Convert the
19048 discriminant value of signedness vary.
19050 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19052 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19053 Add to -Wall section.
19055 2017-06-01 Richard Biener <rguenther@suse.de>
19057 PR middle-end/66313
19058 * fold-const.c (fold_plusminus_mult_expr): If the factored
19059 factor may be zero use a wrapping type for the inner operation.
19060 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19061 and handle moved defs.
19062 (process_assignment): Properly guard the unary op case. Return a
19063 tri-state indicating that moving the stmt before the call may allow
19064 to continue. Pass through to_move.
19065 (find_tail_calls): Handle moving unrelated defs before
19068 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19071 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19072 splitter result in the canonical way.
19074 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19076 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19077 also for 32bit target. Update insn attributes.
19078 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19080 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19082 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19083 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19084 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19085 (_mm_maskz_min_ss): New intrinsics.
19087 2017-05-31 Martin Liska <mliska@suse.cz>
19089 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19090 Change comment style to one we normally use.
19091 (vectorizable_reduction): Likewise.
19092 (vectorizable_induction): Likewise.
19093 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19094 (vectorizable_call): Likewise.
19095 (vectorizable_simd_clone_call): Likewise.
19096 (vectorizable_conversion): Likewise.
19097 (vectorizable_assignment): Likewise.
19098 (vectorizable_shift): Likewise.
19099 (vectorizable_operation): Likewise.
19100 (vectorizable_store): Likewise.
19101 (vectorizable_load): Likewise.
19102 * tree-vectorizer.h: Likewise.
19104 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19106 * passes.c (emergency_dump_function): New.
19107 * tree-pass.h (emergency_dump_function): Declare.
19108 * plugin.c (plugins_internal_error_function): Remove.
19109 * plugin.h (plugins_internal_error_function): Remove declaration.
19110 * toplev.c (internal_error_function): New static function. Use it...
19111 (general_init): ...here.
19113 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19115 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19116 (arc_rtx_costs): Add costs for new patterns.
19117 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19118 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19120 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19122 * tree-ssa-strlen.c (get_next_strinfo): New function.
19123 (get_stridx_plus_constant): Use it.
19124 (zero_length_string): Likewise.
19125 (adjust_related_strinfos): Likewise.
19126 (adjust_last_stmt): Likewise.
19128 2017-05-31 Richard Biener <rguenther@suse.de>
19131 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19132 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19134 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19136 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19137 loop_vinfo argument and use of dependence distance vectors.
19138 Check instead whether the two references differ only in their
19139 initial value and assume that they have the same alignment if the
19140 difference is a multiple of the vector alignment.
19141 (vect_analyze_data_refs_alignment): Update call accordingly.
19143 2017-05-31 Martin Liska <mliska@suse.cz>
19146 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19148 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19150 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19151 (create_intersect_range_checks): Move from ...
19152 * tree-data-ref.c (create_intersect_range_checks_index)
19153 (create_intersect_range_checks): ... to here.
19154 (create_runtime_alias_checks): New function factored from ...
19155 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19156 here. Call above function.
19157 * tree-data-ref.h (create_runtime_alias_checks): New function.
19159 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19161 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19162 segment length for dr_b and compute it in wide_int.
19164 2017-05-31 Richard Biener <rguenther@suse.de>
19166 PR tree-optimization/80906
19167 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19168 and pass through iv_map.
19169 (copy_bb_and_scalar_dependences): Adjust.
19170 (translate_pending_phi_nodes): Likewise.
19171 (copy_loop_close_phi_args): Handle code-generating IVs instead
19174 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19176 * diagnostic-color.c (color_dict): Add "type-diff".
19177 (parse_gcc_colors): Update comment.
19178 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19179 -fdiagnostics-show-template-tree and -fno-elide-type.
19180 (GCC_COLORS): Add type-diff to example.
19182 (-fdiagnostics-show-template-tree): New.
19183 (-fno-elide-type): New.
19184 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19185 the pp_format_decoder callback. Call any m_format_postprocessor's
19187 (pretty_printer::pretty_printer): Initialize
19188 m_format_postprocessor.
19189 (pretty_printer::~pretty_printer): Delete any
19190 m_format_postprocessor.
19191 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19192 (class format_postprocessor): New class.
19193 (struct pretty_printer::format_decoder): Document the new parameters.
19194 (struct pretty_printer::m_format_postprocessor): New field.
19195 * tree-diagnostic.c (default_tree_printer): Update for new
19196 bool and const char ** params.
19197 * tree-diagnostic.h (default_tree_printer): Likewise.
19199 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19201 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19202 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19203 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19204 rs6000_gen_cell_microcode code.
19205 (rs6000_final_prescan_insn): Delete.
19206 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19207 "warn-cell-microcode" entries.
19208 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19209 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19210 throughout. Change cc_reg_not_micro_cr0_operand to
19211 cc_reg_not_cr0_operand throughout.
19212 (*extendhi<mode>2_noload): Delete.
19213 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19214 (mwarn-cell-microcode): Delete.
19215 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19216 -mgen-cell-microcode and -mwarn-cell-microcode.
19218 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19221 * config/i386/constraints.md (Yd): New constraint.
19223 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19224 and (?Yd, r) alternatives. Update insn attributes.
19225 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19226 and (?*Yd, r) alternatives. Update insn attributes.
19227 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19229 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19231 * gimplify.c (gimplify_modify_expr): Don't create a
19232 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19235 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19237 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19239 2017-05-30 Richard Biener <rguenther@suse.de>
19241 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19242 and reduc_def fields.
19243 (STMT_VINFO_REDUC_TYPE): New define.
19244 (STMT_VINFO_REDUC_DEF): Likewise.
19245 (vect_force_simple_reduction): Adjust prototype.
19246 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19247 (vect_is_simple_reduction): Remove check_reduction argument.
19248 (vect_force_simple_reduction): Adjust and set
19249 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19250 (vectorizable_reduction): Do not re-do reduction analysis
19251 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19252 * tree-parloops.c (gather_scalar_reductions): Adjust.
19254 2017-05-30 Richard Biener <rguenther@suse.de>
19256 PR middle-end/80901
19257 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19260 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19262 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19263 Introduce unknown_misalignment parameter and remove vf.
19264 (vect_peeling_hash_get_lowest_cost):
19265 Pass unknown_misalignment parameter.
19266 (vect_enhance_data_refs_alignment):
19267 Fix unsupportable data ref treatment.
19269 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19271 * tree-vect-data-refs.c (vect_get_data_access_cost):
19272 Workaround for SLP handling.
19273 (vect_enhance_data_refs_alignment):
19274 Compute costs for doing no peeling at all, compare to the best
19275 peeling costs so far and avoid peeling if cheaper.
19277 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19279 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19280 Return peeling info and set costs to zero for unlimited cost
19282 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19283 with unknown misalignment. Compute and costs for unknown
19284 misalignment, compare them to the costs for known misalignment
19285 and choose the cheapest for peeling.
19287 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19289 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19290 (vect_get_peeling_costs_all_drs): Create function.
19291 (vect_peeling_hash_get_lowest_cost):
19292 Use vect_get_peeling_costs_all_drs.
19293 (vect_peeling_supportable): Create function.
19294 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19296 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19298 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19299 DR_HAS_NEGATIVE_STEP.
19300 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19301 (vect_enhance_data_refs_alignment): Use.
19302 (vect_duplicate_ssa_name_ptr_info): Use.
19303 * tree-vectorizer.h (dr_misalignment): Use.
19304 (known_alignment_for_access_p): Use.
19306 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19309 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19311 (has_section_name): New function.
19313 2017-05-30 Martin Liska <mliska@suse.cz>
19316 * auto-profile.c (get_function_decl_from_block): Fix
19319 2017-05-30 Richard Biener <rguenther@suse.de>
19321 PR middle-end/80876
19322 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19324 2017-05-30 Martin Liska <mliska@suse.cz>
19326 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19327 * dumpfile.h (struct dump_file_info): Remove ctors.
19329 2017-05-30 Martin Liska <mliska@suse.cz>
19331 * predict.def: Fix GNU coding style.
19333 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19335 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19336 Mark 'to' argument with ATTRIBUTE_UNUSED.
19338 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19340 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19341 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19342 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19345 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19347 * doc/install.texi (Options specification): Restore entry of
19348 --enable-sjlj-exceptions.
19350 2017-05-27 Michael Eager <eager@eagercon.com>
19353 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19355 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19357 * config/microblaze/microblaze.h
19358 (FIXED_REGISTERS): Update in macro.
19359 (CALL_USED_REGISTERS): Update in macro.
19361 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19363 * doc/install.texi: Add links to macOS binary distributions.
19365 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19369 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19371 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19373 2017-05-26 Martin Liska <mliska@suse.cz>
19375 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19377 2017-05-26 Martin Liska <mliska@suse.cz>
19379 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19380 always leading ';; '.
19381 (dump_bb_info): Likewise.
19382 (brief_dump_cfg): Likewise.
19383 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19384 * dumpfile.c: Remove usage of TDF_VERBOSE.
19385 * dumpfile.h (enum dump_kind): Likewise.
19386 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19387 * print-tree.c (debug_verbose): Remove.
19388 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19389 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19390 * tree-diagnostic.c (default_tree_printer): Replace
19391 TDF_DIAGNOSTIC with TDF_SLIM.
19393 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19395 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19396 in parameter loop, rather than loop_vinfo.
19397 (create_intersect_range_checks): Ditto.
19398 (vect_create_cond_for_alias_checks): Update call to above functions.
19400 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19402 PR tree-optimization/80815
19403 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19404 for merging runtime alias checks. Handle negative DR_STEPs.
19406 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19408 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19410 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19411 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19412 out code pruning runtime alias checks.
19413 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19414 factored out from above.
19415 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19417 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19419 (prune_runtime_alias_test_list): New decalaration.
19421 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19423 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19424 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19425 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19426 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19427 (operator==, comp_dr_with_seg_len_pair): Ditto.
19428 (vect_prune_runtime_alias_test_list): Ditto.
19430 2017-05-26 Martin Liska <mliska@suse.cz>
19433 * params.def: Bound partial-inlining-entry-probability param.
19435 2017-05-26 Marek Polacek <polacek@redhat.com>
19438 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19441 2017-05-26 Richard Biener <rguenther@suse.de>
19443 PR tree-optimization/80842
19444 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19447 2017-05-26 Richard Biener <rguenther@suse.de>
19449 PR tree-optimization/80844
19450 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19452 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19454 * doc/md.texi (Machine Constraints): Update x86 family
19455 machine constraints section to match 'config/i386/constraints.md'.
19457 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19459 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19461 2017-05-25 Nathan Sidwell <nathan@acm.org>
19463 * doc/invoke.texi (--enable-languages): Update documentation.
19465 2017-05-25 Martin Liska <mliska@suse.cz>
19467 * dumpfile.c: Add TDF_FOLDING.
19468 * dumpfile.h (enum dump_kind): Likewise.
19469 * genmatch.c (dt_simplify::gen_1): Use it.
19471 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19473 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19475 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19477 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19478 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19480 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19482 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19483 * match.pd (X == C): Rewrite it here.
19484 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19485 with_certain_nonzero_bits2): New predicates.
19486 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19488 2017-05-24 Nathan Sidwell <nathan@acm.org>
19490 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19493 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19496 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19498 * config/powerpcspe: New port. Files are copied from the rs6000
19499 port, with "rs6000" in filenames replaced by "powerpcspe".
19501 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19503 PR rtl-optimization/80754
19504 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19506 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19508 * config/sparc/sparc.md (length): Return the correct value for -mflat
19509 sibcalls to match output_sibcall.
19511 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19515 * config/rs6000/rs6000.c (struct machine_function): Add new field
19517 (rs6000_get_separate_components): Init that field, use it.
19518 (rs6000_components_for_bb): Use the field.
19520 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19522 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19524 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19526 PR middle-end/80823
19527 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19529 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19532 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19533 address against address_operand predicate.
19534 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19536 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19538 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19539 parameters passed indirectly.
19541 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19543 * config/i386/i386.md (*movdi_internal): Remove SSE4
19544 alternative 18 (?r, *v). Update insn attributes.
19545 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19546 Update insn attributes.
19547 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19548 Update insn attributes.
19549 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19550 alternative 1 (r, v). Remove isa attribute.
19551 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19552 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19553 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19555 2017-05-23 Tom de Vries <tom@codesourcery.com>
19557 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19560 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19562 * cgraphunit.c (symbol_table::process_new_functions): Update.
19563 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19564 (inline_generate_summary): Rename to ...
19565 (ipa_fn_summary_generate): ... this one.
19566 (inline_read_summary): Rename to ...
19567 (ipa_fn_summary_read): ... this one.
19568 (inline_write_summary): Rename to ...
19569 (ipa_fn_summary_write): ... this one.
19570 (inline_free_summary): Rename to ...
19571 (ipa_free_fn_summary): ... this one.
19572 (pass_data_local_fn_summary, pass_local_fn_summary,
19573 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19574 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19575 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19576 make_pass_ipa_fn_summary): New.
19577 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19578 inline_write_summary, inline_free_summary): Remove.
19579 (ipa_free_fn_summary) : New.
19580 * ipa-inline.c (ipa_inline): Update.
19581 (pass_ipa_inline): Do not generate summaries.
19582 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19584 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19585 and add pass_ipa_fn_summary.
19586 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19588 (make_pass_inline_parameters): Remove.
19590 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19592 * omp-low.c (struct omp_context): Remove "default_kind" member.
19595 * omp-offload.c (execute_oacc_device_lower): Remove the
19596 parallelism dimensions function attributes for unparallelized
19597 OpenACC kernels constructs.
19599 2017-05-23 Martin Liska <mliska@suse.cz>
19601 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19603 (cgraph_edge::make_speculative): Likewise.
19604 (cgraph_edge::resolve_speculation): Likewise.
19605 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19606 (cgraph_node::dump): Likewise.
19607 * cgraph.h: Likewise.
19608 * cgraphunit.c (analyze_functions): Likewise.
19609 (symbol_table::compile): Likewise.
19610 * ipa-cp.c (print_all_lattices): Likewise.
19611 (determine_versionability): Likewise.
19612 (initialize_node_lattices): Likewise.
19613 (ipcp_verify_propagated_values): Likewise.
19614 (estimate_local_effects): Likewise.
19615 (update_profiling_info): Likewise.
19616 (create_specialized_node): Likewise.
19617 (perhaps_add_new_callers): Likewise.
19618 (decide_about_value): Likewise.
19619 (decide_whether_version_node): Likewise.
19620 (identify_dead_nodes): Likewise.
19621 (ipcp_store_bits_results): Likewise.
19622 * ipa-devirt.c (dump_targets): Likewise.
19623 (ipa_devirt): Likewise.
19624 * ipa-icf.c (sem_item::dump): Likewise.
19625 (sem_function::equals): Likewise.
19626 (sem_variable::equals): Likewise.
19627 (sem_item_optimizer::read_section): Likewise.
19628 (sem_item_optimizer::execute): Likewise.
19629 (congruence_class::dump): Likewise.
19630 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19631 (dump_inline_summary): Likewise.
19632 (estimate_node_size_and_time): Likewise.
19633 (inline_analyze_function): Likewise.
19634 * ipa-inline-transform.c (inline_call): Likewise.
19635 * ipa-inline.c (report_inline_failed_reason): Likewise.
19636 (want_early_inline_function_p): Likewise.
19637 (edge_badness): Likewise.
19638 (update_edge_key): Likewise.
19639 (inline_small_functions): Likewise.
19640 * ipa-profile.c (ipa_profile): Likewise.
19641 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19642 (ipa_make_edge_direct_to_target): Likewise.
19643 (remove_described_reference): Likewise.
19644 (ipa_impossible_devirt_target): Likewise.
19645 (propagate_controlled_uses): Likewise.
19646 (ipa_print_node_params): Likewise.
19647 (ipcp_transform_function): Likewise.
19648 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19649 (propagate_pure_const): Likewise.
19650 * ipa-reference.c (generate_summary): Likewise.
19651 (read_write_all_from_decl): Likewise.
19652 (propagate): Likewise.
19653 (ipa_reference_read_optimization_summary): Likewise.
19654 * ipa-utils.c (ipa_merge_profiles): Likewise.
19655 * ipa.c (walk_polymorphic_call_targets): Likewise.
19656 (symbol_table::remove_unreachable_nodes): Likewise.
19657 (ipa_single_use): Likewise.
19658 * passes.c (execute_todo): Likewise.
19659 * predict.c (drop_profile): Likewise.
19660 * symtab.c (symtab_node::get_dump_name): New function.
19661 (symtab_node::dump_name): Likewise.
19662 (symtab_node::dump_asm_name): Likewise.
19663 (symtab_node::dump_references): Likewise.
19664 (symtab_node::dump_referring): Likewise.
19665 (symtab_node::dump_base): Likewise.
19666 (symtab_node::debug_symtab): Likewise.
19667 * tree-sra.c (convert_callers_for_node): Likewise.
19668 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19669 * value-prof.c (init_node_map): Likewise.
19671 2017-05-23 Martin Liska <mliska@suse.cz>
19673 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19674 and symtab_node::debug_symtab to symbol_table::debug.
19675 * cgraphunit.c (analyze_functions): Use the renamed function.
19676 (symbol_table::compile): Likewise.
19677 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19678 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19679 * passes.c (execute_todo): Likewise.
19680 * symtab.c (symbol_table::dump): New function.
19681 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19683 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19685 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19686 that nonconst implies exec.
19688 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19690 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19691 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19692 (inline_edge_summary_vec): Turn into ...
19693 (ipa_call_summaries): ... this one.
19694 (redirect_to_unreachable, edge_set_predicate,
19695 evaluate_properties_for_edge, inline_summary_alloc,
19696 reset_ipa_call_summary, reset_inline_summary,
19697 inline_summary_t::duplicate): Update.
19698 (inline_edge_duplication_hook): Turn to ...
19699 (ipa_call_summary_t::duplicate): ... this one.
19700 (inline_edge_removal_hook): Turn to ...
19701 (ipa_call_summary_t::remove): ... this one.
19702 (dump_inline_edge_summary): Turn to ...
19703 (dump_ipa_call_summary): ... this one.
19704 (estimate_function_body_sizes): Update.
19705 (inline_update_callee_summaries): Update.
19706 (remap_edge_change_prob): Update.
19707 (remap_edge_summaries): Update.
19708 (inline_merge_summary): Update.
19709 (do_estimate_edge_time): Update.
19710 (inline_generate_summary): Update.
19711 (inline_read_section): Update.
19712 (inline_read_summary): Update.
19713 (inline_free_summary): Update.
19714 * ipa-inline.c (can_inline_edge_p): Update.
19715 (compute_inlined_call_time): Update.
19716 (want_inline_small_function_p): Update.
19717 (edge_badness): Update.
19718 (early_inliner): Update.
19719 * ipa-inline.h (inline_edge_summary): Turn to ...
19720 (ipa_call_summary): ... this one.
19721 (ipa_call_summary_t): New class.
19722 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19723 (ipa_call_summaries): New.
19724 (inline_edge_summary): Remove.
19725 (estimate_edge_growth): Update.
19726 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19727 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19728 * ipa-split.c (execute_split_functions): Update.
19729 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19731 2017-05-23 Tom de Vries <tom@codesourcery.com>
19733 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19734 attributes): Document rdrand effective target.
19736 2017-05-23 Tom de Vries <tom@codesourcery.com>
19738 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19739 attributes): Sort alphabetically.
19741 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19743 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19745 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19748 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19749 V2DF/V2DI splat into two separate patterns, one that handles
19750 registers, and the other that only handles memory. Drop support
19751 for splatting from a GPR on ISA 2.07 and then splitting the
19752 splat into direct move and splat.
19753 (vsx_splat_<mode>_reg): Likewise.
19754 (vsx_splat_<mode>_mem): Likewise.
19756 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19758 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19760 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19762 PR middle-end/80809
19763 * omp-low.c (finish_taskreg_remap): New function.
19764 (finish_taskreg_scan): If unit size of ctx->record_type
19765 is non-constant, unshare the size expression and replace
19766 decls in it with possible outer var refs.
19768 PR middle-end/80809
19769 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19770 GOVD_SHARED rather than GOVD_PRIVATE with it.
19771 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19772 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19774 PR middle-end/80853
19775 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19776 as last argument to build_outer_var_ref for pointer bases of array
19777 section reductions.
19779 2017-05-19 Martin Sebor <msebor@redhat.com>
19781 * print-tree.c (print_node): Print DECL_READ_P flag.
19783 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19785 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19786 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19787 * cgraph.c: Likewise.
19788 * cgraphunit.c: Likewise.
19789 * gengtype.c: Likewise.
19790 * ipa-cp.c: Likewise.
19791 * ipa-devirt.c: Likewise.
19792 * ipa-icf.c: Likewise.
19793 * ipa-predicate.c: Likewise.
19794 * ipa-profile.c: Likewise.
19795 * ipa-prop.c: Likewise.
19796 * ipa-split.c: Likewise.
19798 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19799 edge_predicate_pool, dump_inline_hints,
19800 inline_summary::account_size_time, redirect_to_unreachable,
19801 edge_set_predicate, set_hint_predicate,
19802 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19803 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19804 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19805 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19806 ipa_call_summary_t::remove, initialize_growth_caches,
19807 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19808 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19809 mark_modified, unmodified_parm_1, unmodified_parm,
19810 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19811 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19812 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19813 will_be_nonconstant_predicate, record_modified_bb_info,
19814 get_minimal_bb, record_modified, param_change_prob,
19815 phi_result_unknown_predicate, predicate_for_phi_result,
19816 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19817 estimate_function_body_sizes, compute_inline_parameters,
19818 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19819 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19820 inline_update_callee_summaries, remap_edge_change_prob,
19821 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19822 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19823 inline_analyze_function, inline_summary_t::insert,
19824 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19825 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19826 inline_free_summary): Move to ipa-fnsummary.h
19827 (predicate_t): Remove.
19828 * ipa-fnsummary.c: New file.
19829 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19830 (enum inline_hints_vals, inline_hints, agg_position_info,
19831 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19832 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19833 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19834 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19835 inline_read_summary, inline_write_summary, inline_free_summary,
19836 inline_analyze_function, initialize_inline_failed,
19837 inline_merge_summary, inline_update_overall_summary,
19838 compute_inline_parameters): Move to ipa-fnsummary.h
19839 * ipa-fnsummary.h: New file.
19840 * ipa-inline-transform.h: Include ipa-inline.h.
19841 * ipa-inline.c: LIkewise.
19843 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19845 * ipa-inline.c (edge_badness): Use inlined_time instead of
19846 inline_summaries->get.
19848 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19850 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19852 2017-05-22 Nathan Sidwell <nathan@acm.org>
19854 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19855 (fdump-lang): Document 'raw' option.
19856 * dumpfile.h (TDI_tu): Delete.
19857 * dumpfile.c (dump_files): Remove translation-unit.
19858 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19860 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19862 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19863 command option from $(AWK) call.
19864 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19865 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19866 [FORMAT]: Remove handling of variable.
19867 * config/avr/t-multilib: Regenerate.
19869 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19871 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19873 (dump_inline_summary): Do not print self_time.
19874 (estimate_function_body_sizes): Do not set self_time.
19875 (compute_inline_parameters): Likewise.
19876 (inline_read_section, inline_write_summary): Do not stream self_time.
19877 * ipa-inline.h (inline_summary): Drop self_time.
19879 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19881 * ipa-inline-analysis.c (account_size_time): Rename to ...
19882 (inline_summary::account_size_time): ... this one.
19883 (reset_ipa_call_summary): Turn to ...
19884 (ipa_call_summary::reset): ... this one.
19885 (reset_inline_summary): Turn to ...
19886 (inline_summary::reset): ... this one.
19887 (inline_summary_t::remove): Update.
19888 (inline_summary_t::duplicate): Update.
19889 (ipa_call_summary_t::remove): Update.
19890 (dump_inline_summary): Update.
19891 (estimate_function_body_sizes): Update.
19892 (compute_inline_parameters): Update.
19893 (estimate_node_size_and_time): Update.
19894 (inline_merge_summary): Update.
19895 (inline_update_overall_summary): Update.
19896 (inline_read_section): Update.
19897 (inline_write_summary): Update.
19898 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19899 add account_size_time and reset member functions.
19900 (ipa_call_summary): Add reset function.
19901 * ipa-predicate.h (predicate::operator &): Constify.
19903 2017-05-22 Richard Biener <rguenther@suse.de>
19905 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19907 2017-05-19 Jason Merrill <jason@redhat.com>
19909 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19911 2017-05-19 Marek Polacek <polacek@redhat.com>
19914 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19915 TYPE_OVERFLOW_WRAPS checks.
19917 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19919 * tree-core.h (enum omp_clause_default_kind): Add
19920 "OMP_CLAUSE_DEFAULT_PRESENT".
19921 * tree-pretty-print.c (dump_omp_clause): Handle it.
19922 * gimplify.c (enum gimplify_omp_var_data): Add
19923 "GOVD_MAP_FORCE_PRESENT".
19924 (gimplify_adjust_omp_clauses_1): Map it to
19925 "GOMP_MAP_FORCE_PRESENT".
19926 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19928 * gimplify.c (oacc_default_clause): Clarify.
19930 2017-05-19 Nathan Sidwell <nathan@acm.org>
19932 LANG_HOOK_REGISTER_DUMPS
19933 * toplev.c (general_init): Call register dump lang hook.
19934 * doc/invoke.texi: Document -fdump-lang option family.
19935 * dumpfile.c (dump_files): Remove class dump here.
19936 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19937 * dumpfile.h (tree_dump_index): Remove TDI_class.
19938 * langhooks-def.h (lhd_register_dumps): Declare.
19939 (LANG_HOOKS_REGISTER_DUMPS): Define.
19940 (LANG_HOOKS_INITIALIZER): Add it.
19941 * langhooks.c (lhd_register_dumps): Define.
19942 * langhooks.h (struct lang_hooks): Add register_dumps.
19944 2017-05-19 Nathan Sidwell <nathan@acm.org>
19946 * context.h (context::set_passes): New.
19947 * context.c (context::context): Do not create pass manager.
19948 * toplev.c (general_init): Create pass manager here.
19950 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19952 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19953 use this splitter if two add or or instructions would also work for
19954 the constant we want to generate.
19956 2017-05-19 Richard Biener <rguenther@suse.de>
19959 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19960 predicate evaluation.
19962 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19964 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19966 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19969 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19971 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19972 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19973 (inline_edge_summary_vec): Turn into ...
19974 (ipa_call_summaries): ... this one.
19975 (redirect_to_unreachable, edge_set_predicate,
19976 evaluate_properties_for_edge, inline_summary_alloc,
19977 reset_ipa_call_summary, reset_inline_summary,
19978 inline_summary_t::duplicate): Update.
19979 (inline_edge_duplication_hook): Turn to ...
19980 (ipa_call_summary_t::duplicate): ... this one.
19981 (inline_edge_removal_hook): Turn to ...
19982 (ipa_call_summary_t::remove): ... this one.
19983 (dump_inline_edge_summary): Turn to ...
19984 (dump_ipa_call_summary): ... this one.
19985 (estimate_function_body_sizes): Update.
19986 (inline_update_callee_summaries): Update.
19987 (remap_edge_change_prob): Update.
19988 (remap_edge_summaries): Update.
19989 (inline_merge_summary): Update.
19990 (do_estimate_edge_time): Update.
19991 (inline_generate_summary): Update.
19992 (inline_read_section): Update.
19993 (inline_read_summary): Update.
19994 (inline_free_summary): Update.
19995 * ipa-inline.c (can_inline_edge_p): Update.
19996 (compute_inlined_call_time): Update.
19997 (want_inline_small_function_p): Update.
19998 (edge_badness): Update.
19999 (early_inliner): Update.
20000 * ipa-inline.h (inline_edge_summary): Turn to ...
20001 (ipa_call_summary): ... this one.
20002 (ipa_call_summary_t): New class.
20003 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20004 (ipa_call_summaries): New.
20005 (inline_edge_summary): Remove.
20006 (estimate_edge_growth): Update.
20007 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20008 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20009 * ipa-split.c (execute_split_functions): Update.
20010 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20012 2017-05-19 Richard Biener <rguenther@suse.de>
20014 PR middle-end/80764
20015 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20017 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20019 * config/rs6000/rs6000.c (struct machine_function): Add field
20020 fpr_is_wrapped_separately.
20021 (rs6000_get_separate_components): Use 64 components. Handle the
20022 new FPR components.
20023 (rs6000_components_for_bb): Handle the FPR components.
20024 (rs6000_emit_prologue_components): Handle the FPR components.
20025 (rs6000_emit_epilogue_components): Handle the FPR components.
20026 (rs6000_set_handled_components): Handle the FPR components.
20027 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20028 that are already separately shrink-wrapped.
20029 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20030 that are already separately shrink-wrapped.
20032 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20035 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20038 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20039 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20040 cases where the register allocator uses a move and an offsettable
20041 memory operation to/from a FPR register on ISA 2.06/2.07.
20042 (define_peephole2 for Altivec d-form store): Likewise.
20044 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20047 * config/i386/mmx.md (*mov<mode>_internal): Enable
20048 alternatives 11, 12, 13 and 14 also for 32bit targets.
20049 Remove alternatives 15, 16, 17 and 18.
20050 * config/i386/sse.md (vec_concatv2di): Change
20051 alternative (!x, *y) to (x, ?!*Yn).
20053 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20055 * dumpfile.h (enum dump_kind): Remove stray comma.
20057 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20059 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20060 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20061 predicate::num_conditions
20062 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20063 (CHANGED): turn into predicate::changed.
20064 (agg_position_info): Move to ipa-predicate.h
20065 (add_condition, predicate::add_clause, predicate::operator &=,
20066 predicate::or_with, predicate::evaluate, predicate::probability,
20067 dump_condition, dump_clause, predicate::dump,
20068 predicate::remap_after_duplication, predicate::remap_after_inlining,
20069 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20070 (evaluate_conditions_for_known_args): Update.
20071 (set_cond_stmt_execution_predicate): Update.
20072 * ipa-inline.h: Include ipa-predicate.h
20073 (condition, inline_param_summary, conditions, agg_position_info,
20074 predicate): Move to ipa-predicate.h
20075 * ipa-predicate.c: New file.
20076 * ipa-predicate.h: New file.
20078 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20080 * final.c (leaf_function_p): Check we are not in a sequence.
20082 2017-05-18 Martin Liska <mliska@suse.cz>
20084 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20085 * dumpfile.c (dump_register): Use new enum dump_kind.
20086 (get_dump_file_name): Likewise.
20087 (dump_enable_all): Likewise.
20088 (dump_switch_p_1): Likewise.
20089 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20090 * dumpfile.h (enum dump_kind): New enum type.
20091 (struct dump_file_info): Create constructor and
20092 format fields and comments.
20093 * passes.c (pass_manager::register_one_dump_file):
20095 * statistics.c (statistics_early_init): Likewise.
20096 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20097 TDF_TREE with TDF_SLIM.
20098 (gather_memory_references_ref): Likewise.
20100 2017-05-18 Martin Liska <mliska@suse.cz>
20102 * vec.h (struct vnull): Use it.
20104 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20106 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20107 (true_predicate, false_predicate, true_predicate_p,
20108 false_predicate_p): Remove.
20109 (single_cond_predicate, not_inlined_predicate): Turn to member function
20111 (add_condition): Update.
20112 (add_clause): Turn to...
20113 (predicate::add_clause): ... this one; update; allow passing NULL
20115 (and_predicates): Turn to ...
20116 (predicate::operator &=): ... this one.
20117 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20118 (or_predicates): Turn to ...
20119 (predicate::or_with): ... this one.
20120 (evaluate_predicate): Turn to ...
20121 (predicate::evaluate): ... this one.
20122 (predicate_probability): Turn to ...
20123 (predicate::probability): ... this one.
20124 (dump_condition): Update.
20125 (dump_predicate): Turn to ...
20126 (predicate::dump): ... this one.
20127 (account_size_time): Update.
20128 (edge_set_predicate): Update.
20129 (set_hint_predicate): UPdate.
20130 (evaluate_conditions_for_known_args): Update.
20131 (evaluate_properties_for_edge): Update.
20132 (remap_predicate_after_duplication): Turn to...
20133 (predicate::remap_after_duplication): ... this one.
20134 (remap_hint_predicate_after_duplication): Update.
20135 (inline_summary_t::duplicate): UPdate.
20136 (dump_inline_edge_summary): Update.
20137 (dump_inline_summary): Update.
20138 (set_cond_stmt_execution_predicate): Update.
20139 (set_switch_stmt_execution_predicate): Update.
20140 (compute_bb_predicates): Update.
20141 (will_be_nonconstant_expr_predicate): Update.
20142 (will_be_nonconstant_predicate): Update.
20143 (phi_result_unknown_predicate): Update.
20144 (predicate_for_phi_result): Update.
20145 (array_index_predicate): Update.
20146 (estimate_function_body_sizes): Update.
20147 (estimate_node_size_and_time): Update.
20148 (estimate_ipcp_clone_size_and_time): Update.
20149 (remap_predicate): Rename to ...
20150 (predicate::remap_after_inlining): ... this one.
20151 (remap_hint_predicate): Update.
20152 (inline_merge_summary): Update.
20153 (inline_update_overall_summary): Update.
20154 (estimate_size_after_inlining): Update.
20155 (read_predicate): Rename to ...
20156 (predicate::stream_in): ... this one.
20157 (read_inline_edge_summary): Update.
20158 (write_predicate): Rename to ...
20159 (predicate::stream_out): ... this one.
20160 (write_inline_edge_summary): Update.
20161 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20162 (clause_t): Turn to uint32_t
20163 (predicate): Turn to class; implement constructor and operators
20165 (size_time_entry): Update.
20166 (inline_summary): Update.
20167 (inline_edge_summary): Update.
20169 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20171 * fold-const.c (fold_binary_loc): Move transformation...
20172 * match.pd (C - X CMP X): ... here.
20174 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20176 * config/sparc/sparc.c (sparc_option_override): Set function
20177 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20178 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20180 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20182 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20184 2017-05-18 Marek Polacek <polacek@redhat.com>
20187 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20188 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20189 gimple_assign_load_p.
20191 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20193 PR middle-end/80692
20194 * real.c (do_compare): Give decimal_do_compare preference over
20195 comparing just the signs.
20197 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20199 * doc/md.texi (Canonicalization of Instructions): Describe the
20200 canonical form of instructions that inherently set a condition
20203 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20205 PR middle-end/80775
20206 * tree-cfg.c: Move deletion of unreachable case statements to after
20207 the merging of consecutive case labels.
20209 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20211 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20212 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20213 restoring of callee-saved registers.
20215 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20217 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20218 * config/visium/visium.c (single_set_and_flags): Likewise.
20219 * config/visium/visium.md (Substitutions): Likewise.
20221 2017-05-17 Martin Liska <mliska@suse.cz>
20223 * cfg.c: Introduce dump_flags_t type and
20224 use it instead of int type.
20226 * cfghooks.c: Likewise.
20227 * cfghooks.h (struct cfg_hooks): Likewise.
20228 * cfgrtl.c: Likewise.
20229 * cfgrtl.h: Likewise.
20230 * cgraph.c (cgraph_node::get_body): Likewise.
20231 * coretypes.h: Likewise.
20232 * domwalk.c: Likewise.
20233 * domwalk.h: Likewise.
20234 * dumpfile.c (struct dump_option_value_info): Likewise.
20235 (dump_enable_all): Likewise.
20236 (dump_switch_p_1): Likewise.
20237 (opt_info_switch_p): Likewise.
20238 * dumpfile.h (enum tree_dump_index): Likewise.
20239 (struct dump_file_info): Likewise.
20240 * genemit.c: Likewise.
20241 * generic-match-head.c: Likewise.
20242 * gengtype.c (open_base_files): Likewise.
20243 * gimple-pretty-print.c: Likewise.
20244 * gimple-pretty-print.h: Likewise.
20245 * graph.c (print_graph_cfg): Likewise.
20246 * graphite-scop-detection.c (dot_all_sese): Likewise.
20247 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20248 * loop-unroll.c (report_unroll): Likewise.
20249 * passes.c (pass_manager::register_one_dump_file): Likewise.
20250 * print-tree.c: Likewise.
20251 * statistics.c: Likewise.
20252 * tree-cfg.c: Likewise.
20253 * tree-cfg.h: Likewise.
20254 * tree-dfa.c: Likewise.
20255 * tree-dfa.h: Likewise.
20256 * tree-dump.c (dump_function): Likewise.
20257 * tree-dump.h (struct dump_info): Likewise.
20258 * tree-pretty-print.c: Likewise.
20259 * tree-pretty-print.h: Likewise.
20260 * tree-ssa-live.c: Likewise.
20261 * tree-ssa-live.h: Likewise.
20262 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20263 * tree-vect-loop.c: Likewise.
20264 * tree-vect-slp.c: Likewise.
20266 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20267 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20269 PR tree-optimization/80457
20270 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20271 of all arguments to a statement as scalar_to_vec operations.
20272 (vectorizable_call): Adjust call to vect_model_simple_cost for
20274 (vectorizable_conversion): Likewise.
20275 (vectorizable_assignment): Likewise.
20276 (vectorizable_shift): Likewise.
20277 (vectorizable_operation): Likewise.
20278 (vectorizable_comparison): Likewise.
20279 (vect_is_simple_cond): Record the def types for operands.
20280 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20281 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20282 for statement argument count.
20284 2017-05-16 Carl Love <cel@us.ibm.com>
20286 * config/rs6000/rs6000-c: Add support for built-in functions
20287 vector unsigned long long vec_bperm (vector unsigned long long,
20288 vector unsigned char)
20289 vector signed long long vec_mule (vector signed int,
20291 vector unsigned long long vec_mule (vector unsigned int,
20292 vector unsigned int)
20293 vector signed long long vec_mulo (vector signed int,
20295 vector unsigned long long vec_mulo (vector unsigned int,
20296 vector unsigned int)
20297 vector signed char vec_sldw (vector signed char,
20298 vector signed char,
20300 vector unsigned char vec_sldw (vector unsigned char,
20301 vector unsigned char,
20303 vector signed short vec_sldw (vector signed short,
20304 vector signed short,
20306 vector unsigned short vec_sldw (vector unsigned short,
20307 vector unsigned short,
20309 vector signed int vec_sldw (vector signed int,
20312 vector unsigned int vec_sldw (vector unsigned int,
20313 vector unsigned int,
20315 vector signed long long vec_sldw (vector signed long long,
20316 vector signed long long,
20318 vector unsigned long long vec_sldw (vector unsigned long long,
20319 vector unsigned long long,
20321 * config/rs6000/rs6000-c: Add support for built-in functions
20322 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20323 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20324 * doc/extend.texi: Update the built-in documentation for the
20325 new built-in functions.
20327 2017-05-16 Marek Polacek <polacek@redhat.com>
20331 * tree.c (save_expr): Don't fold the expression.
20333 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20335 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20336 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20337 and (?*y,m). Update insn attributes.
20339 2017-05-16 Martin Liska <mliska@suse.cz>
20341 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20342 flags argument of print_gimple_stmt, print_gimple_expr,
20343 print_generic_stmt and print_generic_expr.
20344 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20345 * coretypes.h: Likewise.
20346 * except.c (dump_eh_tree): Likewise.
20347 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20348 * gimple-pretty-print.h: Likewise.
20349 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20350 (backprop::push_to_worklist): Likewise.
20351 (backprop::pop_from_worklist): Likewise.
20352 (backprop::process_use): Likewise.
20353 (backprop::intersect_uses): Likewise.
20354 (note_replacement): Likewise.
20355 * gimple-ssa-store-merging.c
20356 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20357 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20358 (pass_store_merging::execute): Likewise.
20359 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20360 (ssa_base_cand_dump_callback): Likewise.
20361 (dump_incr_vec): Likewise.
20362 (replace_refs): Likewise.
20363 (replace_mult_candidate): Likewise.
20364 (create_add_on_incoming_edge): Likewise.
20365 (create_phi_basis): Likewise.
20366 (insert_initializers): Likewise.
20367 (all_phi_incrs_profitable): Likewise.
20368 (introduce_cast_before_cand): Likewise.
20369 (replace_one_candidate): Likewise.
20370 * gimplify.c (gimplify_expr): Likewise.
20371 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20372 (set_rename): Likewise.
20373 (rename_uses): Likewise.
20374 (copy_loop_phi_nodes): Likewise.
20375 (add_close_phis_to_merge_points): Likewise.
20376 (copy_loop_close_phi_args): Likewise.
20377 (copy_cond_phi_args): Likewise.
20378 (graphite_copy_stmts_from_block): Likewise.
20379 (translate_pending_phi_nodes): Likewise.
20380 * graphite-poly.c (print_pdr): Likewise.
20381 (dump_gbb_cases): Likewise.
20382 (dump_gbb_conditions): Likewise.
20383 (print_scop_params): Likewise.
20384 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20385 (build_cross_bb_scalars_use): Likewise.
20386 (gather_bbs::before_dom_children): Likewise.
20387 * hsa-dump.c (dump_hsa_immed): Likewise.
20388 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20389 (get_replacement_map): Likewise.
20390 * ipa-inline-analysis.c (dump_condition): Likewise.
20391 (estimate_function_body_sizes): Likewise.
20392 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20393 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20394 * ipa-prop.c (ipa_dump_param): Likewise.
20395 (ipa_print_node_jump_functions_for_edge): Likewise.
20396 (ipa_modify_call_arguments): Likewise.
20397 (ipa_modify_expr): Likewise.
20398 (ipa_dump_param_adjustments): Likewise.
20399 (ipa_dump_agg_replacement_values): Likewise.
20400 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20401 * ipa-pure-const.c (check_stmt): Likewise.
20402 (pass_nothrow::execute): Likewise.
20403 * ipa-split.c (execute_split_functions): Likewise.
20404 * omp-offload.c (dump_oacc_loop_part): Likewise.
20405 (dump_oacc_loop): Likewise.
20406 * trans-mem.c (tm_log_emit): Likewise.
20407 (tm_memopt_accumulate_memops): Likewise.
20408 (dump_tm_memopt_set): Likewise.
20409 (dump_tm_memopt_transform): Likewise.
20410 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20411 (print_loop): Likewise.
20412 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20413 (chkp_gather_checks_info): Likewise.
20414 (chkp_get_check_result): Likewise.
20415 (chkp_remove_check_if_pass): Likewise.
20416 (chkp_use_outer_bounds_if_possible): Likewise.
20417 (chkp_reduce_bounds_lifetime): Likewise.
20418 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20419 (chkp_mark_completed_bounds): Likewise.
20420 (chkp_register_incomplete_bounds): Likewise.
20421 (chkp_mark_invalid_bounds): Likewise.
20422 (chkp_maybe_copy_and_register_bounds): Likewise.
20423 (chkp_build_returned_bound): Likewise.
20424 (chkp_get_bound_for_parm): Likewise.
20425 (chkp_build_bndldx): Likewise.
20426 (chkp_get_bounds_by_definition): Likewise.
20427 (chkp_generate_extern_var_bounds): Likewise.
20428 (chkp_get_bounds_for_decl_addr): Likewise.
20429 * tree-chrec.c (chrec_apply): Likewise.
20430 * tree-data-ref.c (dump_data_reference): Likewise.
20431 (dump_subscript): Likewise.
20432 (dump_data_dependence_relation): Likewise.
20433 (analyze_overlapping_iterations): Likewise.
20434 * tree-inline.c (expand_call_inline): Likewise.
20435 (tree_function_versioning): Likewise.
20436 * tree-into-ssa.c (dump_defs_stack): Likewise.
20437 (dump_currdefs): Likewise.
20438 (dump_names_replaced_by): Likewise.
20439 (dump_update_ssa): Likewise.
20440 (update_ssa): Likewise.
20441 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20442 * tree-parloops.c (build_new_reduction): Likewise.
20443 (try_create_reduction_list): Likewise.
20444 (ref_conflicts_with_region): Likewise.
20445 (oacc_entry_exit_ok_1): Likewise.
20446 (oacc_entry_exit_single_gang): Likewise.
20447 * tree-pretty-print.h: Likewise.
20448 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20449 (get_scalar_evolution): Likewise.
20450 (add_to_evolution): Likewise.
20451 (get_loop_exit_condition): Likewise.
20452 (analyze_evolution_in_loop): Likewise.
20453 (analyze_initial_condition): Likewise.
20454 (analyze_scalar_evolution): Likewise.
20455 (instantiate_scev): Likewise.
20456 (number_of_latch_executions): Likewise.
20457 (gather_chrec_stats): Likewise.
20458 (final_value_replacement_loop): Likewise.
20459 (scev_const_prop): Likewise.
20460 * tree-sra.c (dump_access): Likewise.
20461 (disqualify_candidate): Likewise.
20462 (create_access): Likewise.
20463 (reject): Likewise.
20464 (maybe_add_sra_candidate): Likewise.
20465 (create_access_replacement): Likewise.
20466 (analyze_access_subtree): Likewise.
20467 (analyze_all_variable_accesses): Likewise.
20468 (sra_modify_assign): Likewise.
20469 (initialize_constant_pool_replacements): Likewise.
20470 (find_param_candidates): Likewise.
20471 (decide_one_param_reduction): Likewise.
20472 (replace_removed_params_ssa_names): Likewise.
20473 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20474 * tree-ssa-copy.c (dump_copy_of): Likewise.
20475 (copy_prop_visit_cond_stmt): Likewise.
20476 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20477 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20478 (record_equivalences_from_stmt): Likewise.
20479 * tree-ssa-dse.c (compute_trims): Likewise.
20480 (delete_dead_call): Likewise.
20481 (delete_dead_assignment): Likewise.
20482 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20483 (forward_propagate_into_cond): Likewise.
20484 (pass_forwprop::execute): Likewise.
20485 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20486 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20488 (move_computations_worker): Likewise.
20489 (execute_sm): Likewise.
20490 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20491 (remove_exits_and_undefined_stmts): Likewise.
20492 (remove_redundant_iv_tests): Likewise.
20493 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20494 (adjust_iv_update_pos): Likewise.
20495 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20496 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20497 (value_replacement): Likewise.
20498 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20499 * tree-ssa-pre.c (print_pre_expr): Likewise.
20500 (get_representative_for): Likewise.
20501 (create_expression_by_pieces): Likewise.
20502 (insert_into_preds_of_block): Likewise.
20503 (eliminate_insert): Likewise.
20504 (eliminate_dom_walker::before_dom_children): Likewise.
20505 (eliminate): Likewise.
20506 (remove_dead_inserted_code): Likewise.
20507 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20508 * tree-ssa-reassoc.c (get_rank): Likewise.
20509 (eliminate_duplicate_pair): Likewise.
20510 (eliminate_plus_minus_pair): Likewise.
20511 (eliminate_not_pairs): Likewise.
20512 (undistribute_ops_list): Likewise.
20513 (eliminate_redundant_comparison): Likewise.
20514 (update_range_test): Likewise.
20515 (optimize_range_tests_var_bound): Likewise.
20516 (optimize_vec_cond_expr): Likewise.
20517 (rewrite_expr_tree): Likewise.
20518 (rewrite_expr_tree_parallel): Likewise.
20519 (linearize_expr): Likewise.
20520 (break_up_subtract): Likewise.
20521 (linearize_expr_tree): Likewise.
20522 (attempt_builtin_powi): Likewise.
20523 (attempt_builtin_copysign): Likewise.
20524 (transform_stmt_to_copy): Likewise.
20525 (transform_stmt_to_multiply): Likewise.
20526 (dump_ops_vector): Likewise.
20527 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20528 (print_scc): Likewise.
20529 (set_ssa_val_to): Likewise.
20530 (visit_reference_op_store): Likewise.
20531 (visit_use): Likewise.
20532 (sccvn_dom_walker::before_dom_children): Likewise.
20533 (run_scc_vn): Likewise.
20534 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20536 (expr_hash_elt::print): Likewise.
20537 (const_and_copies::pop_to_marker): Likewise.
20538 (const_and_copies::record_const_or_copy_raw): Likewise.
20539 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20540 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20541 (dump_predicates): Likewise.
20542 (find_uninit_use): Likewise.
20543 (warn_uninitialized_phi): Likewise.
20544 (pass_late_warn_uninitialized::execute): Likewise.
20545 * tree-ssa.c (verify_vssa): Likewise.
20546 (verify_ssa): Likewise.
20547 (maybe_optimize_var): Likewise.
20548 * tree-vrp.c (dump_value_range): Likewise.
20549 (dump_all_value_ranges): Likewise.
20550 (dump_asserts_for): Likewise.
20551 (register_edge_assert_for_2): Likewise.
20552 (vrp_visit_cond_stmt): Likewise.
20553 (vrp_visit_switch_stmt): Likewise.
20554 (vrp_visit_stmt): Likewise.
20555 (vrp_visit_phi_node): Likewise.
20556 (simplify_cond_using_ranges_1): Likewise.
20557 (fold_predicate_in): Likewise.
20558 (evrp_dom_walker::before_dom_children): Likewise.
20559 (evrp_dom_walker::push_value_range): Likewise.
20560 (evrp_dom_walker::pop_value_range): Likewise.
20561 (execute_early_vrp): Likewise.
20563 2017-05-16 Richard Biener <rguenther@suse.de>
20565 * dwarf2out.c (loc_list_from_tree_1): Do not create
20566 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20568 2017-05-16 Richard Biener <rguenther@suse.de>
20570 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20572 (note_variable_value_in_expr): If we resolved the decl ref
20573 do not push to the stack.
20575 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20577 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20578 operations in fast-math mode.
20579 (vaddq_f16): Likewise.
20580 (vmul_f16): Likewise.
20581 (vmulq_f16): Likewise.
20582 (vsub_f16): Likewise.
20583 (vsubq_f16): Likewise.
20584 * config/arm/neon.md (add<mode>3): New.
20586 (fma:<VH:mode>3): New. Also remove outdated comment.
20589 2017-05-16 Martin Liska <mliska@suse.cz>
20593 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20594 (odr_types_equivalent_p): Likewise.
20596 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20598 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20600 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20603 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20604 non-interunit SSE move alternatives with '?'.
20605 (zero-extendsidi peephole2): New peephole to skip intermediate
20606 general register in SSE zero-extend sequence.
20608 2017-05-15 Jeff Law <law@redhat.com>
20610 * reorg.c (relax_delay_slots): Create a new variable to hold
20611 the temporary target rather than clobbering TARGET_LABEL.
20613 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20614 missing argument to extract_bit_field call.
20615 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20617 2017-05-15 Martin Liska <mliska@suse.cz>
20620 * gcc.c (process_command): Do not allow empty argument of -o option.
20622 2017-05-15 Renlin Li <renlin.li@arm.com>
20624 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20625 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20626 * config/aarch64/constraints.md (Usf): Add long call check.
20627 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20628 (call_value): Likewise.
20629 (sibcall): Likewise.
20630 (sibcall_value): Likewise.
20632 (call_value_insn): New.
20633 (sibcall_insn): Update rtx pattern.
20634 (sibcall_value_insn): Likewise.
20635 (call_internal): Remove.
20636 (call_value_internal): Likewise.
20637 (sibcall_internal): Likewise.
20638 (sibcall_value_internal): Likewise.
20639 (call_reg): Likewise.
20640 (call_symbol): Likewise.
20641 (call_value_reg): Likewise.
20642 (call_value_symbol): Likewise.
20644 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20647 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20649 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20651 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20652 compatible with CCGOCmode and with CCZmode.
20654 2017-05-14 Martin Sebor <msebor@redhat.com>
20656 PR middle-end/77671
20657 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20658 (gimple_fold_builtin_snprintf): Same.
20659 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20660 (gimple_fold_builtin_snprintf): Same.
20661 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20662 of character types.
20663 (is_call_safe): New function.
20664 (try_substitute_return_value): Call it.
20665 (try_simplify_call): New function.
20666 (pass_sprintf_length::handle_gimple_call): Call it.
20668 2017-05-14 Martin Sebor <msebor@redhat.com>
20670 PR middle-end/80669
20671 * builtins.c (expand_builtin_stpncpy): Simplify.
20673 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20675 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20676 * config/i386/i386.h
20677 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20678 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20679 (struct machine_function): Add new members call_ms2sysv,
20680 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20681 (struct machine_frame_state): New fields sp_realigned and
20682 sp_realigned_offset.
20683 * config/i386/i386.c
20684 (enum xlogue_stub): New enum.
20685 (enum xlogue_stub_sets): New enum.
20686 (class xlogue_layout): New class.
20687 (struct ix86_frame): New fields stack_realign_allocate_offset,
20688 stack_realign_offset and outlined_save_offset. Modify comments to
20689 detail stack layout when using out-of-line stubs.
20690 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20691 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20692 -mcall-ms2sysv-xlogues.
20693 (stub_managed_regs): New static variable.
20694 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20695 registers managed by out-of-line stub.
20696 (disable_call_ms2sysv_xlogues): New function.
20697 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20698 m->call_ms2sysv when appropriate and compute frame layout for
20700 (sp_valid_at, fp_valid_at): New inline functions.
20701 (choose_basereg): New function.
20702 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20704 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20705 Use align parameter of choose_baseaddr to generated aligned SSE movs
20707 (pro_epilogue_adjust_stack): Modify to track
20708 machine_frame_state::sp_realigned.
20709 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20710 (ix86_nsaved_sseregs): Likewise.
20711 (ix86_emit_save_regs): Likewise.
20712 (ix86_emit_save_regs_using_mov): Likewise.
20713 (ix86_emit_save_sse_regs_using_mov): Likewise.
20714 (get_scratch_register_on_entry): Likewise.
20715 (gen_frame_set): New function.
20716 (gen_frame_load): Likewise.
20717 (gen_frame_store): Likewise.
20718 (emit_outlined_ms2sysv_save): Likewise.
20719 (emit_outlined_ms2sysv_restore): Likewise.
20720 (ix86_expand_prologue): Modify stack re-alignment code and call
20721 emit_outlined_ms2sysv_save when appropriate.
20722 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20723 parameter rtx_insn *insn, which allows the function to be used to only
20724 generate the notes.
20725 (ix86_expand_epilogue): Modify validity checks of frame and stack
20726 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20727 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20728 * config/i386/predicates.md
20729 (save_multiple): New predicate.
20730 (restore_multiple): Likewise.
20731 * config/i386/sse.md
20732 (save_multiple<mode>): New pattern.
20733 (save_multiple_realign<mode>): Likewise.
20734 (restore_multiple<mode>): Likewise.
20735 (restore_multiple_and_return<mode>): Likewise.
20736 (restore_multiple_leave_return<mode>): Likewise.
20737 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20739 2017-05-14 Julia Koval <julia.koval@intel.com>
20741 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20742 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20743 (__builtin_ia32_xsetbv): New builtins.
20744 * config/i386/i386.c (ix86_expand_special_args_builtin):
20746 (ix86_expand_builtin): Special expand for new intrinsics.
20747 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20748 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20749 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20751 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20753 * cfganal.c (inverted_post_order_compute): Change argument type
20755 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20756 * df-core.c (rest_of_handle_df_initialize): Adjust.
20757 (rest_of_handle_df_finish): Likewise.
20758 (df_analyze_1): Likewise.
20759 (df_analyze): Likewise.
20760 (loop_inverted_post_order_compute): Change argument to be a vec *.
20761 (df_analyze_loop): Adjust.
20762 (df_get_n_blocks): Likewise.
20763 (df_get_postorder): Likewise.
20764 * df.h (struct df_d): Change field to be a vec.
20765 * lcm.c (compute_laterin): Adjust.
20766 (compute_available): Likewise.
20767 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20768 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20769 * tree-ssa-pre.c (compute_antic): Likewise.
20771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20773 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20774 (depth_first_search::depth_first_search): Change structure init
20775 function to this constructor.
20776 (depth_first_search::add_bb): Rename function to this member.
20777 (depth_first_search::execute): Likewise.
20778 (flow_dfs_compute_reverse_finish): Adjust.
20780 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20782 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20783 (longest_simple_path): Likewise.
20784 * shrink-wrap.c (spread_components): Likewise.
20785 (disqualify_problematic_components): Likewise.
20786 (emit_common_heads_for_components): Likewise.
20787 (emit_common_tails_for_components): Likewise.
20788 (insert_prologue_epilogue_for_components): Likewise.
20790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20792 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20795 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20797 * df-core.c (df_set_blocks): Start using auto_bitmap.
20798 (df_compact_blocks): Likewise.
20799 * df-problems.c (df_rd_confluence_n): Likewise.
20800 * df-scan.c (df_insn_rescan_all): Likewise.
20801 (df_process_deferred_rescans): Likewise.
20802 (df_update_entry_block_defs): Likewise.
20803 (df_update_exit_block_uses): Likewise.
20804 (df_entry_block_bitmap_verify): Likewise.
20805 (df_exit_block_bitmap_verify): Likewise.
20806 (df_scan_verify): Likewise.
20807 * lra-constraints.c (lra_constraints): Likewise.
20808 (undo_optional_reloads): Likewise.
20809 (lra_undo_inheritance): Likewise.
20810 * lra-remat.c (calculate_gen_cands): Likewise.
20811 (do_remat): Likewise.
20812 * lra-spills.c (assign_spill_hard_regs): Likewise.
20813 (spill_pseudos): Likewise.
20814 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20815 (bitmap_set_subtract_values): Likewise.
20817 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20819 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20820 management with auto_bitmap.
20821 (fix_inter_tick): Likewise.
20822 (fix_recovery_deps): Likewise.
20823 * ira.c (add_store_equivs): Likewise.
20824 (find_moveable_pseudos): Likewise.
20825 (split_live_ranges_for_shrink_wrap): Likewise.
20826 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20827 (rtx_reuse_manager::seen_def_p): Likewise.
20828 (rtx_reuse_manager::set_seen_def): Likewise.
20829 * print-rtl.h (class rtx_reuse_manager): Likewise.
20831 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20833 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20835 (migrate_btr_def): Likewise.
20836 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20837 * df-core.c (loop_post_order_compute): Likewise.
20838 (loop_inverted_post_order_compute): Likewise.
20839 * hsa-common.h: Likewise.
20840 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20841 * init-regs.c (initialize_uninitialized_regs): Likewise.
20842 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20843 (inline_small_functions): Likewise.
20844 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20845 * ira.c (combine_and_move_insns): Likewise.
20846 (build_insn_chain): Likewise.
20847 * loop-invariant.c (find_invariants): Likewise.
20848 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20849 * predict.c (tree_predict_by_opcode): Likewise.
20850 (predict_paths_leading_to): Likewise.
20851 (predict_paths_leading_to_edge): Likewise.
20852 (estimate_loops_at_level): Likewise.
20853 (estimate_loops): Likewise.
20854 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20855 (spread_components): Likewise.
20856 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20857 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20858 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20859 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20860 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20861 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20862 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20863 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20864 (mark_threaded_blocks): Likewise.
20865 (thread_through_all_blocks): Likewise.
20866 * tree-ssa.c (verify_ssa): Likewise.
20867 (execute_update_addresses_taken): Likewise.
20868 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20870 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20872 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20874 (post_order_compute): Likewise.
20875 (inverted_post_order_compute): Likewise.
20876 (pre_and_rev_post_order_compute_fn): Likewise.
20878 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20880 * genrecog.c (int_set::int_set): Explicitly construct our
20881 auto_vec base class.
20882 * vec.h (auto_vec::auto_vec): New constructor.
20884 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20886 * bitmap.h (class auto_bitmap): New constructor taking
20887 bitmap_obstack * argument.
20889 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20891 * bitmap.h (class auto_bitmap): Change type of m_bits to
20892 bitmap_head, and adjust ctor / dtor and member operators.
20894 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20896 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20897 when returned register mode doesn't match original mode.
20899 2017-05-12 Jeff Law <law@redhat.com>
20900 Jakub Jelinek <jakub@redhat.com>
20902 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20903 we look for cc setter after the compare-elim changes.
20904 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20905 within the vector to match what compare-elim now expects.
20906 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20907 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20909 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20910 after the compare-elim changes.
20911 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20912 the vector to match what compare-elim now expects.
20913 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20914 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20915 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20916 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20917 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20919 * config/visium/visium.c (single_set_and_flags): Fix where
20920 we look for cc setter after the compare-elim changes.
20921 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20922 with the vector to match what compare-elim now expects.
20923 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20924 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20925 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20926 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20927 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20928 (neg<mode>2_insn_set_overflow): Likewise.
20930 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20932 PR middle-end/79794
20933 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20934 maybe_expand_insn call, set ops[0].target. If still set after call,
20935 set alt_rtl. Add extra arg to recursive calls.
20936 (extract_bit_field): Add alt_rtl argument. Pass to
20938 * expmed.h (extract_bit_field): Fix prototype.
20939 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20940 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20941 to extract_bit_field_calls.
20942 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20943 Pass alt_rtl to extract_bit_field calls.
20944 * calls.c (store_unaligned_arguments_into_psuedos)
20945 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20946 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20948 * optabs.h (struct expand_operand): Add target bitfield.
20950 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20952 * compare-elim.c (try_eliminate_compare): Canonicalize
20953 operation with embedded compare to
20954 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20955 (set (reg) (operation)].
20957 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20959 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20962 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20963 cost of adding a carry flag for ADC instruction.
20964 [case MINUS]: Ignore the cost of subtracting a carry flag
20965 for SBB instruction.
20967 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20969 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20971 * config/rs6000/bmiintrin.h: New file.
20972 * config/rs6000/bmi2intrin.h: New file.
20973 * config/rs6000/x86intrin.h: New file.
20975 2017-05-12 Jeff Law <law@redhat.com>
20977 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20980 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20982 PR middle-end/80707
20983 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20985 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20987 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20988 early expansion of vector divide builtins.
20989 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20990 builtins identified as having unsigned arguments.
20992 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20994 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20995 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20996 expansion of vector logical operations (and, andc, or, xor,
20999 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21001 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21002 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21004 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21007 early GIMPLE expansion of vector multiplies.
21009 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21011 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21012 TARGET_HAVE_MOVT conditional.
21013 (movt splitter): Likewise.
21015 2017-05-12 Richard Biener <rguenther@suse.de>
21017 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21018 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21019 Fold all stmts not inplace.
21021 2017-05-12 Richard Biener <rguenther@suse.de>
21023 PR tree-optimization/80713
21024 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21025 inserted_exprs bit for not removed stmts.
21027 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21029 PR middle-end/69921
21030 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21031 parallelized" attribute for parallelized OpenACC kernels.
21032 * omp-offload.c (execute_oacc_device_lower): Use it.
21034 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21035 Set "oacc kernels" attribute.
21036 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21037 parameter. Adjust all users.
21038 (oacc_fn_attrib_kernels_p): Remove function.
21039 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21040 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21041 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21042 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21043 assert "oacc kernels" attribute is set.
21045 2017-05-11 Carl Love <cel@us.ibm.com>
21047 * config/rs6000/rs6000-c: Add support for built-in functions
21048 vector unsigned char vec_popcnt (vector signed char)
21049 vector unsigned char vec_popcnt (vector unsigned char)
21050 vector unsigned short vec_popcnt (vector signed short)
21051 vector unsigned short vec_popcnt (vector unsigned short)
21052 vector unsigned int vec_popcnt (vector signed int)
21053 vector unsigned int vec_popcnt (vector unsigned int)
21054 vector unsigned long long vec_popcnt (vector signed long long)
21055 vector unsigned long long vec_popcnt (vector unsigned long long)
21056 vector signed long long vec_slo (vector signed long long,
21057 vector signed char)
21058 vector signed long long vec_slo (vector signed long long,
21059 vector unsigned char)
21060 vector unsigned long long vec_slo (vector unsigned long long,
21061 vector signed char)
21062 vector unsigned long long vec_slo (vector unsigned long long,
21063 vector unsigned char)
21064 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21065 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21066 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21067 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21068 * doc/extend.texi: Update the built-in documentation file for the
21069 new built-in functions.
21071 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21073 * attribs.h (sorted_attr_string): Move machine independent
21074 functions for target clone support from the i386 port to common
21075 code. Rename ix86_function_versions to common_function_versions.
21076 Rename make_name to make_unique_name.
21077 (common_function_versions): Likewise.
21078 (make_unique_name): Likewise.
21079 (make_dispatcher_decl): Likewise.
21080 (is_function_default_version): Likewise.
21081 * attribs.c (attr_strcmp): Likewise.
21082 (sorted_attr_string): Likewise.
21083 (common_function_versions): Likewise.
21084 (make_unique_name): Likewise.
21085 (make_dispatcher_decl): Likewise.
21086 (is_function_default_version): Likewise.
21087 * config/i386/i386.c (attr_strcmp): Likewise.
21088 (sorted_attr_string): Likewise.
21089 (ix86_function_versions): Likewise.
21090 (make_name): Likewise.
21091 (make_dispatcher_decl): Likewise.
21092 (is_function_default_version): Likewise.
21093 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21095 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21098 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21099 Account for direct move costs for vec_construct of integer
21102 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21105 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21106 (UNSPEC_STX_ATOMIC): Ditto.
21107 (loaddi_via_sse): New insn.
21108 (storedi_via_sse): Ditto.
21109 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21110 Update corresponding peephole2 patterns.
21111 (atomic_storedi_fpu): Ditto.
21113 2017-05-11 Julia Koval <julia.koval@intel.com>
21115 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21116 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21118 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21119 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21120 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21122 2017-05-11 Nathan Sidwell <nathan@acm.org>
21124 * graphite-poly.c: Include dumpfile.h.
21126 * dumpfle.h (dump_function): Declare here ...
21127 * tree-dump.h (dump_function): ... not here.
21128 * dumpfile.c: #include tree-cfg.h.
21129 (dump_function): Move here from ...
21130 * tree-dump.c (dump_function): ... here.
21131 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21132 * graphite-poly.c: Don't include tree-dump.h.
21133 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21134 * print-tree.c: Likewise.
21135 * stor-layout.c: Likewise.
21136 * tree-nested.c: Likewise.
21138 * dumpfile.c (dump_start): Use TDF_FLAGS.
21139 (dump_enable_all): Fix TDF_KIND check thinko.
21141 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21143 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21144 array entries to represent two legal parameterizations of the
21145 overloaded __builtin_cmpb function, as represented by the
21146 P6_OV_BUILTIN_CMPB constant.
21147 (altivec_resolve_overloaded_builtin): Add special case handling
21148 for the __builtin_cmpb function, as represented by the
21149 P6_OV_BUILTIN_CMPB constant.
21150 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21151 (BU_P6_64BIT_2): New macro.
21152 (BU_P6_OVERLOAD_2): New macro
21153 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21154 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21155 (CMPB): Add overload support to represent both 32-bit and 64-bit
21156 compare-bytes function.
21157 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21158 support for TARGET_CMPB.
21159 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21160 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21161 documentation of the __builtin_cmpb overloaded built-in function.
21163 2017-05-11 Richard Biener <rguenther@suse.de>
21165 PR tree-optimization/80705
21166 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21167 bases are not vectorizable.
21169 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21171 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21172 when counting register pressure.
21174 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21176 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21177 (iv_ca_more_deps): Renamed to ...
21178 (iv_ca_compare_deps): ... this.
21179 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21183 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21185 (determine_group_iv_costs): ... here.
21186 (find_inv_vars_cb): Record inv var if it's not recorded before.
21188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21190 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21191 (get_shiftadd_cost): Ditto.
21193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21195 * tree-ssa-address.c: Include header file.
21196 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21198 (add_to_parts): Refactor.
21199 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21200 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21205 PR tree-optimization/53090
21206 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21208 (extract_cond_operands): Detect condition with IV on both sides
21209 and return COMP_IV_EXPR_2.
21210 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21211 (rewrite_use_compare): Simplify by removing call to function
21212 extract_cond_operands.
21214 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21216 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21217 (extract_cond_operands): Detect condition comparing against non-
21218 invariant bound and return appropriate enum value.
21219 (find_interesting_uses_cond): Update use of extract_cond_operands.
21220 Handle its return value accordingly.
21221 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21225 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21226 nonlinear iv_use computation in loop invariant sensitive way.
21228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21230 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21231 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21235 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21236 (dump_cand): Support iv_cand.inv_exprs.
21237 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21239 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21242 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21244 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21246 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21247 as local function. Include necessary header files.
21248 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21250 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21252 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21254 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21256 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21257 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21258 RSHIFT_EXPR and BIT_NOT_EXPR.
21260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21262 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21263 (adjust_setup_cost): New parameter supporting round up adjustment.
21264 (struct address_cost_data): Delete.
21265 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21266 (split_address_cost, ptr_difference_cost): Delete.
21267 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21268 (struct ainc_cost_data): New struct.
21269 (get_address_cost_ainc): New function.
21270 (get_address_cost, get_computation_cost): Reimplement.
21271 (determine_group_iv_cost_address): Record inv_expr for all uses of
21273 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21274 (iv_ca_has_deps): Reimplemented to ...
21275 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21277 (iv_ca_extend): Call iv_ca_more_deps.
21279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21281 * tree-ssa-address.c (struct mem_address): Move to header file.
21282 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21283 * tree-ssa-address.h (struct mem_address): Move from C file.
21284 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21286 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21288 * tree-affine.h (aff_combination_type): New interface.
21289 (aff_combination_zero_p): Remove static.
21290 (aff_combination_const_p): New interface.
21291 (aff_combination_singleton_var_p): New interfaces.
21293 2017-05-11 Richard Biener <rguenther@suse.de>
21295 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21296 Skip unreachable blocks and destinations.
21297 (eliminate): Move stmt removal and fixup ...
21298 (fini_eliminate): ... here. Skip inserted exprs.
21299 (pass_pre::execute): Move fini_pre after fini_eliminate.
21300 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21301 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21302 PRE to get rid of dead code that has invalid SSA form and
21303 split critical edges again.
21305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21307 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21309 2017-05-11 Richard Biener <rguenther@suse.de>
21311 * passes.c (execute_function_todo): Verify loops if they are
21312 said to be up-to-date.
21313 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21314 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21316 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21319 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21320 handle calling assemble_external ourself.
21323 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21324 modes with zero size. Enhance comment.
21326 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21328 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21329 built-ins for vec_xl and vec_xst with short and char pointer
21332 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21334 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21335 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21336 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21337 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21338 (_mm_maskz_min_round_ss): New intrinsics.
21339 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21340 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21341 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21342 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21343 (__builtin_ia32_minss_mask_round): New builtins.
21344 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21345 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21346 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21348 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21349 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21351 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21354 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21356 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21357 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21358 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21359 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21360 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21361 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21362 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21363 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21364 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21365 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21366 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21367 (__builtin_ia32_mulss_mask_round): New builtins.
21368 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21369 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21370 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21372 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21373 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21375 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21378 2017-05-10 Julia Koval <julia.koval@intel.com>
21380 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21381 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21382 (_mm256_setr_m128i): New intrinsics.
21384 2017-05-10 Julia Koval <julia.koval@intel.com>
21386 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21387 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21388 (_mm_maskz_rcp14_ss): New intrinsics.
21389 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21390 (__builtin_ia32_rcp14ss_mask): New builtins.
21391 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21393 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21395 PR tree-optimization/51513
21396 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21397 (assert_unreachable_fallthru_edge_p): Use it.
21398 (group_case_labels_stmt): Likewise.
21399 * tree-cfg.h: Prototype it.
21400 * stmt.c: Include cfghooks.h and tree-cfg.h.
21401 (emit_case_dispatch_table) <gap_label>: New local variable.
21402 Use it to fill dispatch table gaps.
21403 Test for default_label before updating probabilities.
21404 (expand_case) <default_label>: Remove unneeded initialization.
21405 Test for unreachable default case statement and remove its edge.
21406 Set default_label accordingly.
21407 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21409 2017-05-10 Carl Love <cel@us.ibm.com>
21411 * config/rs6000/rs6000-c: Add support for built-in functions
21412 vector signed char vec_neg (vector signed char)
21413 vector signed short int vec_neg (vector short int)
21414 vector signed int vec_neg (vector signed int)
21415 vector signed long long vec_neg (vector signed long long)
21416 vector float vec_neg (vector float)
21417 vector double vec_neg (vector double)
21418 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21420 * config/rs6000/altivec.h: Add define for vec_neg
21421 * doc/extend.texi: Update the built-in documentation for the
21422 new built-in functions.
21424 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21426 PR tree-optimization/77644
21427 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21429 2017-05-10 Nathan Sidwell <nathan@acm.org>
21431 * dumpfile.h (TDI_lang_all): New.
21432 (TDF_KIND): New. Renumber others
21433 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21435 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21437 (get_dump_file_name): Adjust suffix generation.
21438 (dump_enable_all): Use TDF_KIND.
21439 * doc/invoke.texi (-fdump-lang-all): Document.
21441 * dumpfile.h: Tabify.
21443 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21446 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21447 Move member access before delete.
21449 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21451 * tree-inline.c (expand_call_inline): Split block at stmt
21454 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21457 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21458 are now unused after splitting mov{sf,sd}_hardfloat.
21459 (f32_lr2): Likewise.
21460 (f32_lm): Likewise.
21461 (f32_lm2): Likewise.
21462 (f32_li): Likewise.
21463 (f32_li2): Likewise.
21464 (f32_lv): Likewise.
21465 (f32_sr): Likewise.
21466 (f32_sr2): Likewise.
21467 (f32_sm): Likewise.
21468 (f32_sm2): Likewise.
21469 (f32_si): Likewise.
21470 (f32_si2): Likewise.
21471 (f32_sv): Likewise.
21472 (f32_dm): Likewise.
21473 (f32_vsx): Likewise.
21474 (f32_av): Likewise.
21475 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21476 For movsf, order stores so the VSX stores occur before the GPR
21477 store which encourages the register allocator to use a traditional
21478 FPR instead of a GPR. For movsd, order the stores so that the GPR
21479 store comes before the VSX stores to allow the power6 to work.
21480 This is due to the power6 not having a 32-bit integer store
21481 instruction from a FPR.
21482 (movsf_hardfloat): Likewise.
21483 (movsd_hardfloat): Likewise.
21485 2017-05-09 Martin Sebor <msebor@redhat.com>
21487 PR translation/80280
21488 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21491 PR translation/80280
21492 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21493 data member added in r247778.
21494 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21496 2017-05-09 Nathan Sidwell <nathan@acm.org>
21498 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21500 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21503 2017-05-09 Marek Polacek <polacek@redhat.com>
21505 * doc/invoke.texi: Fix typo.
21507 2017-05-09 Richard Biener <rguenther@suse.de>
21509 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21510 (vrp_val_is_min): Likewise.
21511 (set_value_range_to_value): Likewise.
21512 (set_value_range_to_nonnegative): Likewise.
21513 (gimple_assign_nonzero_p): Likewise.
21514 (gimple_stmt_nonzero_p): Likewise.
21515 (vrp_int_const_binop): Likewise. Remove unreachable case.
21516 (adjust_range_with_scev): Adjust comments.
21517 (compare_range_with_value): Likewise.
21518 (extract_range_from_phi_node): Likewise.
21519 (test_for_singularity): Likewise.
21521 2017-05-09 Richard Biener <rguenther@suse.de>
21523 * tree-vrp.c (get_single_symbol): Add assert that we don't
21524 get overflowed constants as invariant part.
21525 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21526 checks. Use wi::cmp instead of recursing for integer constants.
21527 (compare_values): Just ignore whether we assumed undefined
21528 overflow instead of failing the compare.
21529 (extract_range_for_var_from_comparison_expr): Add comment before the
21530 TREE_NO_WARNING sets.
21531 (test_for_singularity): Likewise.
21532 (extract_range_from_comparison): Do not disable optimization
21533 when we assumed undefined overflow.
21534 (extract_range_basic): Remove init of unused var.
21536 2017-05-09 Richard Biener <rguenther@suse.de>
21538 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21539 (extract_range_from_multiplicative_op_1): Adjust.
21540 (extract_range_from_binary_expr_1): Use int_const_binop.
21542 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21545 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21546 rs6000_store_data_bypass_p in seven define_bypass directives and
21547 in several comments.
21548 * config/rs6000/rs6000-protos.h: Add prototype for
21549 rs6000_store_data_bypass_p function.
21550 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21551 function implements slightly different (rs6000-specific) semantics
21552 than store_data_bypass_p, returning false rather than aborting
21553 with assertion error when arguments do not satisfy the
21554 requirements of store data bypass.
21555 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21556 rs6000_store_data_bypass_p.
21558 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21560 * config/xtensa/xtensa-protos.h
21561 (xtensa_initial_elimination_offset): New declaration.
21562 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21563 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21564 macro definition, add case for FRAME_POINTER_REGNUM when
21565 FRAME_GROWS_DOWNWARD.
21566 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21567 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21568 xtensa_initial_elimination_offset.
21570 2017-05-08 Nathan Sidwell <nathan@acm.org>
21572 * doc/invoke.texi: Alphabetize -fdump options.
21574 2017-05-08 Martin Sebor <msebor@redhat.com>
21576 PR translation/80280
21577 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21579 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21581 * target.def (compute_frame_layout): New optional target hook.
21582 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21583 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21584 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21586 * reload1.c (verify_initial_elim_offsets): Likewise.
21587 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21588 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21589 (arm_get_frame_offsets): Split up into this ...
21590 (arm_compute_frame_layout): ... and this function.
21592 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21594 * config/aarch64/constraints.md (Usa): New constraint.
21595 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21597 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21599 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21600 with_multilib_list after it has been checked.
21602 2017-05-08 Richard Biener <rguenther@suse.de>
21604 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21605 (bitmap_set_subtract_values): Likewise.
21607 2017-05-08 Richard Biener <rguenther@suse.de>
21609 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21610 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21612 (gimple_stmt_nonzero_warnv_p): Rename to ...
21613 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21615 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21616 (extract_range_basic): Adjust, do not disable propagation on
21617 strict overflow sensitive simplification.
21618 (vrp_visit_cond_stmt): Likewise.
21620 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21622 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21623 body size unconditionally.
21625 2017-05-07 Jeff Law <law@redhat.com>
21628 2017-05-06 Jeff Law <law@redhat.com>
21629 PR tree-optimization/78496
21630 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21633 PR tree-optimization/78496
21634 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21635 (simplify_stmt_using_ranges): Call it.
21636 (vrp_dom_walker::before_dom_children): Extract equivalences
21637 from an ASSERT_EXPR with an equality comparison against a
21640 2017-05-06 Jeff Law <law@redhat.com>
21642 PR tree-optimization/78496
21643 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21646 PR tree-optimization/78496
21647 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21648 (simplify_stmt_using_ranges): Call it.
21649 (vrp_dom_walker::before_dom_children): Extract equivalences
21650 from an ASSERT_EXPR with an equality comparison against a
21653 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21655 * lra-constraints.c (lra_copy_reg_equiv): New function.
21656 (split_reg): Use it to copy equivalence information from the
21657 original register to the spill register.
21659 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21661 PR rtl-optimization/75964
21662 * simplify-rtx.c (simplify_const_relational_operation): Remove
21663 invalid handling of comparisons of integer ABS.
21665 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21667 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21668 initialize to zero.
21669 (init_regs): Remove declaration.
21670 (function_arg_advance_32): Initialize error_p as boolean variable.
21672 2017-05-05 Nathan Sidwell <nathan@acm.org>
21674 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21675 lines. Use for (;;).
21677 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21679 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21680 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21681 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21682 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21683 VF=2 that require versioning.
21685 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21687 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21690 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21692 * diagnostic.h (diagnostic_override_option_index): Convert from
21693 macro to inline function.
21695 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21697 * diagnostic.c (last_module_changed_p): New function.
21698 (set_last_module): New function.
21699 (diagnostic_report_current_module): Convert macro usage to
21700 the above functions.
21701 * diagnostic.h (diagnostic_context::last_module): Strengthen
21702 from const line_map * to const line_map_ordinary *.
21703 (diagnostic_last_module_changed): Delete macro.
21704 (diagnostic_set_last_module): Delete macro.
21706 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21708 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21709 with diagnostic_report_diagnostic.
21710 (diagnostic_n_impl_richloc): Likewise.
21711 * diagnostic.h (report_diagnostic): Delete macro.
21712 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21713 with diagnostic_report_diagnostic.
21714 * substring-locations.c (format_warning_va): Likewise.
21716 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21718 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21719 save/restor of format_spec. Move option-printing code to...
21720 (print_option_information): ...this new function, and
21721 reimplement by simply printing to the pretty_printer,
21722 rather than appending to the format string.
21724 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21726 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21727 handling logic into...
21728 (update_effective_level_from_pragmas): ...this new function.
21730 2017-05-04 Andrew Waterman <andrew@sifive.com>
21732 * config/riscv/riscv.opt (mstrict-align): New option.
21733 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21734 (SLOW_UNALIGNED_ACCESS): Define.
21735 (riscv_slow_unaligned_access): Declare.
21736 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21738 (riscv_slow_unaligned_access): New variable.
21739 (rocket_tune_info): Set slow_unaligned_access to true.
21740 (optimize_size_tune_info): Set slow_unaligned_access to false.
21741 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21742 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21743 (riscv_option_override): Set riscv_slow_unaligned_access.
21744 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21746 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21748 * config/riscv/riscv.md: Unify indentation.
21750 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21755 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21757 (extendsi<mode>2): Add support for doing sign extension via
21758 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21759 don't have ISA 3.0 instructions.
21760 (extendsi<mode>2 splitter): Likewise.
21761 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21762 generate the normal insns since SImode can now go in vector
21763 registers. Disallow the special UNSPECs needed for previous
21764 machines to hide SImode being used. Add new insns
21765 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21766 (fix_trunc<mode>si2_stfiwx): Likewise.
21767 (fix_trunc<mode>si2_internal): Likewise.
21768 (fixuns_trunc<mode>si2): Likewise.
21769 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21770 (fctiw<u>z_<mode>_smallint): Likewise.
21771 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21772 of floating point to 32-bit integer from doing a direct move to
21773 the GPR registers to do a store.
21774 (fctiwz_<mode>): Break long line.
21776 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21778 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21779 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21780 (addr_list, addr_offset_valid_p): New.
21781 (split_address_groups): Check offset validity with above function.
21782 (gt-tree-ssa-loop-ivopts.h): Include header file.
21784 2017-05-05 Nathan Sidwell <nathan@acm.org>
21786 * config.gcc (arm*-*-*): Add missing 'fi'.
21788 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21790 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21791 included in -fopt-info does not matter.
21792 * doc/optinfo.texi (-fopt-info): Fix description of default
21793 behavour. Explicitly say order of options included in -fopt-info
21796 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21798 * config.gcc: Allow combinations of aprofile and rmprofile values for
21799 --with-multilib-list.
21800 * config/arm/t-multilib: New file.
21801 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21802 variables. Remove setting of ISA and floating-point ABI in
21803 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21804 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21805 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21806 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21808 * config/arm/t-rmprofile: Likewise except for the matches changes.
21809 * doc/install.texi (--with-multilib-list): Document the combination of
21810 aprofile and rmprofile values and warn about pitfalls in doing that.
21812 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21814 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21815 (movdi_aarch64): Likewise.
21817 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21819 PR tree-optimization/80632
21820 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21822 (build_arrays): Initialize it for virtual phis.
21823 (fix_phi_nodes): Use it for virtual phis.
21825 PR tree-optimization/80558
21826 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21827 [x, y] op z into [x op, y op z] for op & or | if conditions
21830 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21831 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21834 * config/arm/arm.md (use_literal_pool): Remove.
21835 (64-bit immediate split): No longer takes cost into consideration
21836 if arm_disable_literal_pool is enabled.
21837 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21838 used when arm_disable_literal_pool is enabled.
21839 (arm_max_const_double_inline_cost): Remove use of
21840 arm_disable_literal_pool.
21841 (push_minipool_fix): Add assert.
21842 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21843 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21844 (no_literal_pool_sf_immediate): New.
21846 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21848 PR tree-optimization/80613
21849 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21850 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21852 2017-05-05 Richard Biener <rguenther@suse.de>
21854 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21856 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21858 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21859 of this flag from insn conditions due to removal from r247495.
21861 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21863 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21865 (arm_early_store_addr_dep_ptr): Likewise.
21866 * config/arm/aarch-common-protos.h
21867 (arm_early_load_addr_dep_ptr): Add prototype.
21868 (arm_early_store_addr_dep_ptr): Likewise.
21869 * config/arm/cortex-a53.md: Add new bypasses.
21871 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21873 * tree.c (next_type_uid): Change type to unsigned.
21874 (type_hash_canon): Decrement back next_type_uid if
21875 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21876 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21879 2017-05-04 Martin Sebor <msebor@redhat.com>
21881 * builtins.c: Fix a trivial typo in a comment.
21883 PR middle-end/79234
21884 * builtins.c (check_sizes): Adjust to handle reading past the end.
21885 Avoid printing excessive upper bound of ranges. Use %E to print
21886 tree nodes instead of converting them to %wu.
21887 (expand_builtin_memchr): New function.
21888 (compute_dest_size): Rename...
21889 (compute_objsize): ...to this.
21890 (expand_builtin_memcpy): Adjust.
21891 (expand_builtin_mempcpy): Adjust.
21892 (expand_builtin_strcat): Adjust.
21893 (expand_builtin_strcpy): Adjust.
21894 (check_strncat_sizes): Adjust.
21895 (expand_builtin_strncat): Adjust.
21896 (expand_builtin_strncpy): Adjust and simplify.
21897 (expand_builtin_memset): Adjust.
21898 (expand_builtin_bzero): Adjust.
21899 (expand_builtin_memcmp): Adjust.
21900 (expand_builtin): Handle memcmp.
21901 (maybe_emit_chk_warning): Check strncat just once.
21903 2017-05-04 Martin Sebor <msebor@redhat.com>
21905 PR preprocessor/79214
21906 PR middle-end/79222
21907 PR middle-end/79223
21908 * builtins.c (check_sizes): Add inlining context and issue
21909 warnings even when -Wno-system-headers is set.
21910 (check_strncat_sizes): Same.
21911 (expand_builtin_strncat): Same.
21912 (expand_builtin_memmove): New function.
21913 (expand_builtin_stpncpy): Same.
21914 (expand_builtin): Handle memmove and stpncpy.
21916 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21918 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21919 which is not used any more.
21921 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21923 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21925 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21927 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21928 (cortexa53_tunings): Likewise.
21929 (cortexa57_tunings): Likewise.
21930 (cortexa72_tunings): Likewise.
21931 (cortexa73_tunings): Likewise.
21933 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21935 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21936 Set loop alignment to 8.
21938 2017-05-04 Martin Sebor <msebor@redhat.com>
21940 PR translation/80280
21941 * builtins.c (expand_builtin_object_size): Add missing quoting to
21942 %D and like directives.
21943 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21944 (hsa_type_for_tree_type): Same.
21945 (verify_function_arguments): Same.
21946 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21947 * varasm.c (get_section): Same.
21950 2017-05-04 Martin Sebor <msebor@redhat.com>
21952 PR translation/80280
21953 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21955 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21957 * config/aarch64/aarch64.c (generic_addrcost_table):
21958 Change HI/TI mode setting.
21960 2017-05-04 Martin Jambor <mjambor@suse.cz>
21962 PR tree-optimization/80622
21963 * tree-sra.c (comes_initialized_p): New function.
21964 (build_accesses_from_assign): Only set write lazily when
21965 comes_initialized_p is false.
21966 (analyze_access_subtree): Use comes_initialized_p.
21967 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21968 instead of testing for PARM_DECL.
21970 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21972 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21973 constraint on operand 0 to allow more general addressing modes.
21974 Adjust output template.
21975 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21977 * config/aarch64/aarch64-protos.h
21978 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21979 * config/aarch64/constraints.md (Dp): New address constraint.
21980 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21983 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21985 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21986 update use of estimate_ipcp_clone_size_and_time.
21987 (estimate_local_effects): Update use of
21988 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21989 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21990 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21991 Return nonspecialized time.
21993 2017-05-04 Richard Biener <rguenther@suse.de>
21995 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21996 for the last VUSE which def dominates the PHI. Directly call
21998 (get_continuation_for_phi_1): Remove.
22000 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22002 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22003 to explain the use of truncating division. Cap the number of
22004 iterations to the maximum given by nb_iterations_upper_bound,
22007 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22009 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22010 * configure: Regenerate.
22011 * config.in: Regenerate.
22012 * config/i386/driver-mingw32.c: new file.
22013 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22014 * config.host: Link driver-mingw32.o on MinGW host.
22015 * doc/install.texi: Document new --enable-mingw-wildcard configure
22018 2017-05-04 Marek Polacek <polacek@redhat.com>
22020 PR tree-optimization/80612
22021 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22023 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22024 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22026 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22027 (movt splitter): Likewise.
22028 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22029 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22030 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22031 block for Thumb-1 with MOVT.
22032 (thumb2_legitimate_address_p): Move code block ...
22033 (can_avoid_literal_pool_for_label_p): ... into this new function.
22034 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22036 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22037 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22038 "M-profile targets with the MOVT instruction".
22040 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22042 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22043 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22044 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22046 2017-05-04 Martin Liska <mliska@suse.cz>
22048 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22049 variable cond_code.
22051 2017-05-04 Richard Biener <rguenther@suse.de>
22053 * tree.c (array_at_struct_end_p): Handle arrays at struct
22054 end with flexarrays more conservatively. Refactor and treat
22055 arrays of arrays or aggregates more strict. Fix
22056 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22057 * tree.c (array_at_struct_end_p): Adjust prototype.
22058 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22059 * gimple-fold.c (get_range_strlen): Likewise.
22060 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22062 2017-05-04 Richard Biener <rguenther@suse.de>
22064 PR tree-optimization/31130
22065 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22067 (supports_overflow_infinity): Likewise.
22068 (is_negative_overflow_infinity): Likewise.
22069 (is_positive_overflow_infinity): Likewise.
22070 (is_overflow_infinity): Likewise.
22071 (stmt_overflow_infinity): Likewise.
22072 (overflow_infinity_range_p): Likewise.
22073 (usable_range_p): Remove as always returning true.
22074 (make_overflow_infinity): Remove.
22075 (negative_overflow_infinity): Likewise.
22076 (positive_overflow_infinity): Likewise.
22077 (avoid_overflow_infinity): Likewise.
22078 (set_value_range): Adjust accordingly.
22079 (set_value_range_to_nonnegative): Likewise, remove now unused
22080 overflow_infinity arg.
22081 (vrp_operand_equal_p): Adjust.
22082 (update_value_range): Likewise.
22083 (range_int_cst_singleton_p): Likewise.
22084 (operand_less_p): Likewise.
22085 (compare_values_warnv): Likewise.
22086 (extract_range_for_var_from_comparison_expr): Likewise.
22087 (vrp_int_const_binop): Likewise.
22088 (zero_nonzero_bits_from_vr): Likewise.
22089 (extract_range_from_multiplicative_op_1): Likewise.
22090 (extract_range_from_binary_expr_1): Likewise.
22091 (extract_range_from_unary_expr): Likewise.
22092 (extract_range_from_comparison): Likewise.
22093 (extract_range_basic): Likewise.
22094 (adjust_range_with_scev): Likewise.
22095 (compare_ranges): Likewise.
22096 (compare_range_with_value): Likewise.
22097 (dump_value_range): Likewise.
22098 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22100 (simplify_cond_using_ranges): Adjust.
22102 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22104 * brig-builtins.def: Added a builtin for class_f64.
22105 * builtin-types.def: Added a builtin type needed by class_f64.
22107 2017-05-03 Jason Merrill <jason@redhat.com>
22109 * timevar.def: Add TV_CONSTEXPR.
22111 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22113 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22115 2017-05-03 Martin Jambor <mjambor@suse.cz>
22117 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22118 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22119 * ipa-cp.c (ipcp_propagate_stage): Do not call
22120 ipa_update_after_lto_read.
22121 * ipa-inline.c (ipa_inline): Likewise.
22123 2017-05-03 Martin Jambor <mjambor@suse.cz>
22125 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22126 tag. Added a default constructor and a destructor.
22127 (ipa_edge_args_sum_t): New class;
22128 (ipa_edge_args_sum): Declare.
22129 (ipa_edge_args_vector): Remove declaration.
22130 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22131 (ipa_free_edge_args_substructures): Remove declaration.
22132 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22133 (ipa_edge_args_info_available_for_edge_p): Likewise.
22134 * ipa-prop.c (ipa_edge_args_vector): Removed.
22135 (edge_removal_hook_holder): Likewise.
22136 (edge_duplication_hook_holder): Likewise.
22137 (ipa_edge_args_sum): New variable.
22138 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22139 ipa_edge_args_vector.
22140 (ipa_free_edge_args_substructures): Likewise.
22141 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22142 ipa_edge_args_vector.
22143 (ipa_edge_removal_hook): Turned into method
22144 ipa_edge_args_sum_t::remove.
22145 (ipa_edge_duplication_hook): Turned into method
22146 ipa_edge_args_sum_t::duplicate.
22147 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22148 registering edge hooks.
22149 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22150 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22151 ipa_edge_args_sum instead of ipa_edge_args_vector.
22152 * ipa-profile.c (ipa_profile): Likewise.
22154 2017-05-03 Martin Jambor <mjambor@suse.cz>
22156 * symbol-summary.h (function_summary): New method exists.
22157 (function_summary::symtab_removal): Deallocate through release.
22158 (call_summary): New class.
22159 (gt_ggc_mx): New overload.
22160 (gt_pch_nx): Likewise.
22161 (gt_pch_nx): Likewise.
22163 2017-05-03 Jeff Law <law@redhat.com>
22165 PR tree-optimization/78496
22166 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22167 from simplify_cond_using_ranges. Split off code to walk
22168 backwards through casts into ...
22169 (simplify_cond_using_ranges_2): New function.
22170 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22171 (execute_vrp): After identifying jump threads, call
22172 simplify_cond_using_ranges_2.
22174 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22177 * ipa-inline.h (inline_summary): Add ctor.
22178 (create_ggc): Do not use ggc_cleared_alloc.
22180 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22181 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22183 * gcc.c (handle_braces): Support escaping in switch matching
22185 * doc/invoke.texi (Spec Files): Document it.
22186 Remove superfluous @code markup in items.
22188 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22190 * diagnostic-show-locus.c (struct column_range): New struct.
22191 (get_affected_columns): New function.
22192 (get_printed_columns): New function.
22193 (struct correction): New struct.
22194 (correction::ensure_capacity): New function.
22195 (correction::ensure_terminated): New function.
22196 (struct line_corrections): New struct.
22197 (line_corrections::~line_corrections): New dtor.
22198 (line_corrections::add_hint): New function.
22199 (layout::print_trailing_fixits): Reimplement in terms of the new
22201 (selftest::test_overlapped_fixit_printing): New function.
22202 (selftest::diagnostic_show_locus_c_tests): Call it.
22204 2017-05-03 Nathan Sidwell <nathan@acm.org>
22206 Canonicalize canonical type hashing
22207 * tree.h (type_hash_canon_hash): Declare.
22208 * tree.c (type_hash_list, attribute_hash_list): Move into
22209 type_hash_canon_hash.
22210 (build_type_attribute_qual_variant): Break out hash code calc into
22211 type_hash_canon_hash.
22212 (type_hash_canon_hash): New. Generic type hash computation.
22213 (build_range_type_1, build_array_type_1, build_function_type,
22214 build_method_type_directly, build_offset_type, build_complex_type,
22215 make_vector_type): Call it.
22217 2017-05-03 Richard Biener <rguenther@suse.de>
22219 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22220 When all DRs have unknown misaligned do not always peel
22221 when there is a store but apply the same costing model as if
22222 there were only loads.
22224 2017-05-03 Richard Biener <rguenther@suse.de>
22227 PR tree-optimization/80492
22228 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22229 compare_base_decls returning dont-know properly.
22231 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22233 * config/arm/iterators.md (CCSI): New mode iterator.
22234 (arch): New mode attribute.
22235 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22236 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22237 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22238 code iterator for success result mode.
22239 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22240 the corresponding new insn generators.
22242 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22245 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22246 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22248 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22250 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22251 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22252 (DDR_A): Wrap DDR argument in brackets.
22253 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22254 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22255 (DDR_REVERSED_P): Likewise.
22257 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22259 PR tree-optimization/79472
22260 * tree-switch-conversion.c (struct switch_conv_info): Add
22261 contiguous_range and default_case_nonstandard fields.
22262 (collect_switch_conv_info): Compute contiguous_range and
22263 default_case_nonstandard fields, don't clear final_bb if
22264 contiguous_range and only the default case doesn't have the required
22266 (check_all_empty_except_final): Set default_case_nonstandard instead
22267 of failing if contiguous_range and the default case doesn't have empty
22269 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22270 and only the default case doesn't have the required constants. Skip
22272 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22273 if default_case_nonstandard.
22274 (build_constructors): Build constant 1 just once. Assert that default
22275 values aren't inserted in between cases if contiguous_range. Skip
22277 (build_arrays): Skip virtual phis.
22278 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22279 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22280 Handle virtual phis.
22281 (gen_inbound_check): Handle default_case_nonstandard case.
22282 (process_switch): Adjust check_final_bb caller. Call
22283 gather_default_values with the first non-default case instead of
22284 default case if default_case_nonstandard.
22286 2017-05-02 Nathan Sidwell <nathan@acm.org>
22288 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22289 check. Fix formatting.
22291 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22293 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22294 errors when comparing specialized and unspecialized times.
22296 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22298 * diagnostic-show-locus.c
22299 (layout::should_print_annotation_line_p): Make private.
22300 (layout::print_annotation_line): Make private.
22301 (layout::annotation_line_showed_range_p): Make private.
22302 (layout::show_ruler): Make private.
22303 (layout::print_source_line): Make private. Pass in line and
22304 line_width, rather than calling location_get_source_line. Drop
22306 (layout::print_leading_fixits): New method.
22307 (layout::print_any_fixits): Rename to...
22308 (layout::print_trailing_fixits): ...this, and make private.
22309 Don't print newline fixits.
22310 (diagnostic_show_locus): Move logic for printing one row into...
22311 (layout::print_line): ...this new function. Move the
22312 location_get_source_line call and error-handling from
22313 print_source_line to here. Call print_leading_fixits, and rename
22314 print_any_fixits to print_trailing_fixits.
22315 (selftest::test_fixit_insert_containing_newline): Update now that
22316 newlines are partially supported.
22317 (selftest::test_fixit_insert_containing_newline_2): New test.
22318 (selftest::test_fixit_replace_containing_newline): Update comments.
22319 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22320 * edit-context.c (class added_line): New class.
22321 (class edited_line): Describe newline handling in comment.
22322 (edited_line::actually_edited_p): New method.
22323 (edited_line::print_content): Delete redundant decl.
22324 (edited_line::m_predecessors): New field.
22325 (edited_file::print_content): Call edited_line::print_content.
22326 (edited_file::print_diff): Update to support newlines.
22327 (edited_file::print_diff_hunk): Likewise.
22328 (edited_file::print_run_of_changed_lines): New function.
22329 (edited_file::print_diff_line): Convert to...
22330 (print_diff_line): ...this.
22331 (edited_file::get_effective_line_count): New function.
22332 (edited_line::edited_line): Initialize new field m_predecessors.
22333 (edited_line::~edited_line): Clean up m_predecessors.
22334 (edited_line::apply_fixit): Handle newlines.
22335 (edited_line::get_effective_line_count): New function.
22336 (edited_line::print_content): New function.
22337 (edited_line::print_diff_lines): New function.
22338 (selftest::test_applying_fixits_insert_containing_newline): New
22340 (selftest::test_applying_fixits_replace_containing_newline): New
22342 (selftest::insert_line): New function.
22343 (selftest::test_applying_fixits_multiple_lines): Add example of
22345 (selftest::edit_context_c_tests): Call the new tests.
22347 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22349 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22350 parameter cand. Update dump information.
22351 (get_computation_cost): Update uses.
22353 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22355 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22356 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22357 (get_computation_at, rewrite_use_address): Update use of
22358 get_computation_aff.
22360 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22362 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22363 (get_computation): Delete.
22364 (get_computation_cost): Implement like get_computation_cost_at.
22365 Use get_computation_at.
22366 (get_computation_cost_at): Delete.
22367 (rewrite_use_nonlinear_expr): Use get_computation_at.
22368 (rewrite_use_compare, remove_unused_ivs): Ditto.
22370 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22372 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22374 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22376 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22377 (ivopts_global_cost_for_size): Rename parameter and update uses.
22378 (iv_ca_recount_cost): Update uses.
22379 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22380 candidates seperately in n_invs and n_cands.
22381 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22383 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22385 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22386 (find_inv_vars_cb): New.
22387 (find_depends): Renamed to ...
22388 (find_inv_vars): ... this.
22389 (add_candidate_1, force_var_cost): Call find_inv_vars.
22390 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22392 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22394 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22395 inv_vars. Add inv_exprs.
22396 (struct iv_cand): Rename depends_on to inv_vars.
22397 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22398 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22399 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22400 (dump_cand): Dump inv_vars.
22401 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22402 (record_invariant, find_depends, add_candidate_1): Ditto.
22403 (set_group_iv_cost, force_var_cost): Ditto.
22404 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22405 (get_computation_cost_at, get_computation_cost): Ditto.
22406 (determine_group_iv_cost_generic): Ditto.
22407 (determine_group_iv_cost_address): Ditto.
22408 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22409 (determine_group_iv_costs): Ditto.
22410 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22411 (iv_ca_set_remove_invariants): Renamed to ...
22412 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22413 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22414 (iv_ca_set_add_invariants): Renamed to ...
22415 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22416 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22417 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22418 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22419 (create_new_ivs): Remove useless dump.
22421 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22423 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22425 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22426 (iv_ca_set_no_cp, create_new_iv): Ditto.
22428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22430 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22432 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22434 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22435 function tree_check2.
22437 2017-05-02 Martin Liska <mliska@suse.cz>
22439 * doc/gcov.texi: Add missing preposition.
22440 * gcov.c (function_info::function_info): Properly fill up
22441 all member variables.
22443 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22445 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22447 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22449 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22451 2017-05-02 Martin Liska <mliska@suse.cz>
22454 * lto-streamer-in.c (lto_read_tree_1): Remove
22455 LTO_STREAMER_DEBUG.
22456 * lto-streamer.c (struct tree_hash_entry): Likewise.
22457 (struct tree_entry_hasher): Likewise.
22458 (tree_entry_hasher::hash): Likewise.
22459 (tree_entry_hasher::equal): Likewise.
22460 (lto_streamer_init): Likewise.
22461 (lto_orig_address_map): Likewise.
22462 (lto_orig_address_get): Likewise.
22463 (lto_orig_address_remove): Likewise.
22464 * lto-streamer.h: Likewise.
22465 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22466 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22468 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22470 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22471 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22472 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22473 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22474 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22475 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22476 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22477 (mm_maskz_sub_ss): New intrinsics.
22478 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22479 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22480 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22481 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22482 (__builtin_ia32_subss_mask_round): New builtins.
22483 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22484 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22485 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22487 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22488 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22490 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22493 2017-05-02 Martin Jambor <mjambor@suse.cz>
22495 PR tree-optimization/78687
22496 * tree-sra.c (access): New field parent.
22497 (process_subtree_disqualification): New function.
22498 (disqualify_candidate): Call it.
22499 (build_accesses_from_assign): Reset write flag if creating an
22501 (build_access_subtree): Fill in parent field and also prpagate
22502 down grp_write flag.
22503 (create_artificial_child_access): New parameter set_grp_write, set
22504 grp_write to its value.
22505 (propagate_subaccesses_across_link): Also propagate grp_write flag
22507 (propagate_all_subaccesses): Push the closest parent back to work
22508 queue if add_access_to_work_queue returned true.
22510 2017-05-02 Richard Biener <rguenther@suse.de>
22512 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22513 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22514 -fstrict-overflow documentation.
22515 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22516 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22517 flag_strict_overflow.
22518 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22519 * lto-opts.c (lto_write_options): Do not stream it.
22520 * lto-wrapper.c (merge_and_complain): Do not handle it.
22521 * opts.c (default_options_table): Do not set -fstrict-overflow.
22522 (finish_options): Likewise do not clear it when sanitizing.
22523 * simplify-rtx.c (simplify_const_relational_operation): Do not
22524 test flag_strict_overflow.
22526 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22528 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22529 using enabled attribute.
22530 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22531 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22532 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22533 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22534 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22535 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22536 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22537 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22538 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22539 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22541 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22543 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22545 2017-05-02 Richard Biener <rguenther@suse.de>
22547 PR tree-optimization/80591
22549 2017-04-10 Richard Biener <rguenther@suse.de>
22551 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22554 2017-05-02 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/80549
22557 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22558 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22559 headers if they do not have a preheader.
22561 2017-05-02 Martin Liska <mliska@suse.cz>
22564 * common.opt: Fix typo.
22565 * doc/invoke.texi: Likewise.
22567 2017-05-01 Jan Beulich <jbeulich@suse.com>
22569 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22570 swapping, add (x,x,m,x,n) alternative.
22572 2017-05-01 Nathan Sidwell <nathan@acm.org>
22574 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22575 unnecessary unadjusted_alignment check.
22577 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22580 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22582 * gimplify.c (gimplify_cilk_detach): New function.
22583 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22584 * tree-core.h: Document EXPR_CILK_SPAWN.
22585 * tree.h (EXPR_CILK_SPAWN): Define.
22587 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22589 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22590 to use new fixit_hint representation, using the "replace" logic.
22591 (get_line_span_for_fixit_hint): Likewise.
22592 (layout::print_any_fixits): Likewise.
22593 (selftest::test_one_liner_many_fixits): Rename to...
22594 (selftest::test_one_liner_many_fixits_1): ...this, and update
22595 comment and expected output to reflect that the multiple fix-it
22596 hints are now consolidated into one insertion.
22597 (selftest::test_one_liner_many_fixits_2): New test.
22598 (selftest::test_diagnostic_show_locus_one_liner): Update for
22600 (selftest::test_fixit_consolidation): Update for fix-it API
22602 * diagnostic.c (print_parseable_fixits): Likewise.
22603 * edit-context.c (edited_line::m_line_events): Convert from
22604 auto_vec <line_event *> to auto_vec <line_event>.
22605 (class line_event): Convert from abstract base class to a concrete
22606 class, taking over the role of replace_event.
22607 (class insert_event): Delete.
22608 (class replace_event): Rename to class line_event. Convert to
22610 (edit_context::add_fixits): Reimplement.
22611 (edit_context::apply_insert): Delete.
22612 (edit_context::apply_replace): Rename to...
22613 (edit_context::apply_fixit): ...this. Convert to half-open range.
22614 (edited_file::apply_insert): Delete.
22615 (edited_file::apply_replace): Rename to...
22616 (edited_file::apply_fixit): ...this.
22617 (edited_line::~edited_line): Drop deletion of events.
22618 (edited_line::apply_insert): Delete.
22619 (edited_line::apply_replace): Rename to...
22620 (edited_line::apply_fixit): ...this. Convert to half-open range.
22621 Update for change to type of m_line_events.
22622 * edit-context.h (edit_context::apply_insert): Delete.
22623 (edit_context::apply_replace): Rename to...
22624 (edit_context::apply_fixit): ...this.
22626 2017-05-01 Martin Sebor <msebor@redhat.com>
22628 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22631 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22634 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22635 __get_cpuid_max returns 0.
22636 (__get_cpuid_count): Ditto.
22638 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22640 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22641 replacement expression is another instance of one of its arguments.
22643 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22646 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22647 check for stack push/pop autoinc.
22648 * config/i386/i386.c (ix86_agi_dependent): Return false
22649 if the only reason why modified_in_p returned true is that
22650 addr is SP based and set_insn is a push or pop.
22652 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22654 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22657 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22660 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22661 (account_size_time): Use two predicates - exec_pred and
22663 (evaluate_conditions_for_known_args): Compute both clause and
22665 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22666 (inline_summary_t::duplicate): Update.
22667 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22669 (compute_inline_parameters): Likewise.
22670 (estimate_edge_size_and_time): Update caluclation of time.
22671 (estimate_node_size_and_time): Compute both time and nonspecialized
22673 (estimate_ipcp_clone_size_and_time): Update.
22674 (inline_merge_summary): Update.
22675 (do_estimate_edge_time): Update.
22676 (do_estimate_edge_size): Update.
22677 (do_estimate_edge_hints): Update.
22678 (inline_read_section, inline_write_summary): Stream both new predicates.
22679 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22681 (compute_inlined_call_time): Cleanup.
22682 (big_speedup_p): Update.
22683 (edge_badness): Update.
22684 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22685 (size_time_entry): Replace predicate by exec_predicate and
22686 nonconst_predicate.
22687 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22688 (estimate_edge_time): Return also nonspec_time.
22689 (reset_edge_growth_cache): Update.
22691 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22693 PR rtl-optimization/80491
22694 * ifcvt.c (noce_process_if_block): When looking for x setter
22695 with missing else_bb, don't check only the insn right before
22696 cond_earliest, but look for the last insn that x is modified in
22697 within the same bb.
22699 PR rtl-optimization/80491
22700 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22702 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22704 PR tree-optimization/80487
22705 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22707 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22709 PR tree-optimization/79697
22710 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22711 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22712 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22714 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22715 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22717 2017-04-28 Martin Sebor <msebor@redhat.com>
22719 PR tree-optimization/80523
22720 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22721 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22723 (maybe_warn, format_directive, parse_directive): Use new functions.
22724 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22726 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22728 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22730 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22732 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22733 target_header_dir): Set correctly.
22734 * configure: Regenerated.
22735 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22736 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22737 instead of SYSTEM_HEADER_DIR.
22739 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22741 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22742 (estimate_local_effects): Likewise.
22743 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22744 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22745 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22746 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22747 do_estimate_edge_time, estimate_edge_time): Likewise.
22748 * ipa-inline-analysis.c (estimate_node_size_and_time,
22749 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22750 (estimate_time_after_inlining): Remove.
22752 2017-04-28 Martin Liska <mliska@suse.cz>
22754 * doc/gcov.texi: Enhance documentation of gcov.
22756 2017-04-28 Martin Liska <mliska@suse.cz>
22758 * doc/gcov.texi: Sort options in alphabetic order.
22759 * doc/gcov-dump.texi: Likewise.
22760 * doc/gcov-tool.texi: Likewise.
22761 * gcov.c (print_usage): Likewise.
22762 * gcov-dump.c (print_usage): Likewise.
22763 * gcov-tool.c (print_merge_usage_message): Likewise.
22764 (print_rewrite_usage_message): Likewise.
22765 (print_overlap_usage_message): Likewise.
22767 2017-04-28 Martin Liska <mliska@suse.cz>
22769 PR gcov-profile/53915
22770 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22772 2017-04-28 Martin Liska <mliska@suse.cz>
22774 PR gcov-profile/79891
22775 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22776 is marked by compiler as living on a line.
22777 (get_cycles_count): Remove usage of the union.
22778 (output_intermediate_file): Likewise.
22779 (find_source): Fix GNU coding style.
22780 (accumulate_line_counts): Remove old non-all block mode.
22781 (output_lines): Remove usage of the union.
22782 * profile.c (output_location): Include all BBs, even if
22783 belonging to a same line (and file) as a previous BB.
22785 2017-04-28 Martin Liska <mliska@suse.cz>
22787 * gcov.c (process_args): Handle new argument 'w'.
22788 (read_graph_file): Assign ID to BBs.
22789 (output_branch_count): Display BB # if verbose flag is set.
22790 (output_lines): Likewise for arcs.
22791 (print_usage): Add '--verbose' option help.
22792 * doc/gcov.texi: Document --verbose (-w) option.
22794 2017-04-28 Martin Liska <mliska@suse.cz>
22796 * gcov.c (struct block_location_info): New struct.
22797 (process_file): Fill up the new structure.
22798 (read_graph_file): Replace usage of encoding by the newly added
22800 (add_line_counts): Likewise.
22801 (accumulate_line_counts): Remove usage of the union.
22802 (function_info::function_info): New function.
22803 (function_info::~function_info): Likewise.
22804 (process_file): Call delete instead of release_function.
22805 (release_function): Release the function.
22806 (release_structures): Call delete instead of release_function.
22807 (solve_flow_graph): Replace usage of num_blocks.
22808 (find_exception_blocks): Likewise.
22809 (output_lines): Fix GNU coding style.
22811 2017-04-28 Martin Liska <mliska@suse.cz>
22814 * coverage.c (coverage_remove_note_file): New function.
22815 * coverage.h: Declare the function.
22816 * toplev.c (finalize): Clean if an error has been seen.
22818 2017-04-28 Martin Liska <mliska@suse.cz>
22820 PR gcov-profile/80031
22821 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22822 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22823 * gcov.c (read_graph_file): Read just number of blocks.
22824 * profile.c (branch_prob): Do not stream 0 flags per a basic
22827 2017-04-28 Martin Liska <mliska@suse.cz>
22829 * gcov-dump.c (tag_*): Add new argument to declarations.
22830 (dump_gcov_file): Likewise.
22831 (tag_blocks): Add and use new argument depth.
22832 (tag_arcs): Likewise.
22833 (tag_lines): Likewise.
22834 (tag_counters): Likewise.
22835 (tag_summary): Likewise.
22836 (dump_working_sets): Use depth to do a proper indentation.
22838 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22841 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22842 * symtab.c (symtab_node::debug_symtab): Move definition here.
22844 2017-04-28 Richard Biener <rguenther@suse.de>
22846 * lto-streamer.h (LTO_major_version): Bump to 7.
22848 2017-04-28 Richard Biener <rguenther@suse.de>
22850 * tree-vrp.c (assert_info): New struct.
22851 (add_assert_info): New helper.
22852 (register_edge_assert_for_2): Refactor to add asserts to a vector
22854 (register_edge_assert_for_1): Likewise.
22855 (register_edge_assert_for): Likewise.
22856 (finish_register_edge_assert_for): New helper actually registering
22857 asserts where live on edge.
22858 (find_conditional_asserts): Adjust.
22859 (find_switch_asserts): Likewise.
22860 (evrp_dom_walker::try_find_new_range): Generalize.
22861 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22863 2017-04-27 Marek Polacek <polacek@redhat.com>
22866 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22867 arg10 and arg11 to itype.
22869 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22871 * doc/extend.texi (Object Size Checking): Improve grammar.
22873 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22876 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22877 that the logic for permitting reciprocal estimates matches that
22880 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22883 * tree.c (type_cache_hasher::equal): Only compare
22884 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22885 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22886 non-aggregate element types.
22887 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22888 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22890 2017-04-27 Richard Biener <rguenther@suse.de>
22892 PR middle-end/80533
22893 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22894 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22895 keeping a reference to a trailing array.
22897 2017-04-27 Richard Biener <rguenther@suse.de>
22899 PR middle-end/80539
22900 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22901 being in loop-closed SSA form conservatively.
22902 (chrec_fold_multiply_poly_poly): Likewise.
22904 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22906 PR middle-end/79665
22907 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22908 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22910 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22913 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22914 (aarch64_function_arg_alignment): Return unsigned int again, but still
22915 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22916 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22917 Don't emit -Wpsabi note.
22918 (aarch64_function_arg_boundary): Likewise.
22919 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22922 2017-04-26 Nathan Sidwell <nathan@acm.org>
22924 * tree.h (crc32_unsigned_n): Declare.
22925 (crc32_unsigned, crc32_unsigned): Make inline.
22926 * tree.c (crc32_unsigned_bits): Replace with ...
22927 (crc32_unsigned_n): ... this.
22928 (crc32_unsigned, crc32_byte): Remove.
22929 (crc32_string): Remove unnecessary braces.
22931 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22933 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22934 * ipa-inline-analysis.c (MAX_TIME): Remove.
22935 (account_size_time): Use sreal for time.
22936 (dump_inline_summary): Update.
22937 (estimate_function_body_sizes): Update.
22938 (estimate_edge_size_and_time): Update.
22939 (estimate_calls_size_and_time): Update.
22940 (estimate_node_size_and_time): Update.
22941 (inline_merge_summary): Update.
22942 (inline_update_overall_summary): Update.
22943 (estimate_time_after_inlining): Update.
22944 (inline_read_section): Update.
22945 (inline_write_summary): Update.
22946 * ipa-inline.c (compute_uninlined_call_time): Update.
22947 (compute_inlined_call_time): Update.
22948 (recursive_inlining): Update.
22949 (inline_small_functions): Update.
22950 (dump_overall_stats): Update.
22951 * ipa-inline.h: Include sreal.h.
22952 (size_time_entry): Turn time to sreal.
22953 (inline_summary): Turn self_time nad time to sreal.
22955 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22957 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22959 (sreal::stream_out, sreal::stream_in): New.
22960 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22962 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22964 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22967 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22970 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22972 (dimode_scalar_chain::compute_convert_gain): Ditto.
22973 (dimode_scalar_chain::make_vector_copies): Ditto.
22974 (dimode_scalar_chain::convert_reg): Ditto.
22975 (dimode_scalar_chain::convert_insn): Ditto.
22976 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22977 (VI248_AVX512BW_1): New mode iterator.
22978 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22979 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22982 2017-04-25 Martin Sebor <msebor@redhat.com>
22984 PR tree-optimization/80497
22985 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22986 constants are representable in HOST_WIDE_INT.
22987 (parse_directive): Ditto.
22989 2017-04-25 Martin Sebor <msebor@redhat.com>
22992 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22993 (new_zero_array): Adjust signature.
22994 (dom_info::dom_init): Used unsigned rather that size_t.
22995 (dom_info::dom_info): Same.
22997 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22998 Jakub Jelinek <jakub@redhat.com>
23001 * config/arm/arm.c: Include gimple.h.
23002 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23003 returns negative, increment ncrn only if it returned positive.
23004 (arm_needs_doubleword_align): Return int instead of bool,
23005 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23006 members, but if there is any such non-FIELD_DECL
23007 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23008 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23009 returns negative, increment nregs only if it returned positive.
23010 (arm_setup_incoming_varargs): Likewise.
23011 (arm_function_arg_boundary): Emit -Wpsabi note if
23012 arm_needs_doubleword_align returns negative, return
23013 DOUBLEWORD_ALIGNMENT only if it returned positive.
23015 2017-04-25 Marek Polacek <polacek@redhat.com>
23018 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23019 first argument to type.
23021 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23024 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23025 type checks to test for compatibility instead of equality.
23027 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23028 Jakub Jelinek <jakub@redhat.com>
23031 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23033 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23034 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23035 the alignment computation, but return their maximum in warn_alignment.
23036 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23037 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23039 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23040 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23043 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23045 * config/arc/simdext.md (dmpyh): Fix typo.
23047 2017-04-25 Richard Biener <rguenther@suse.de>
23049 PR tree-optimization/80492
23050 * alias.c (compare_base_decls): Handle registers with asm
23051 specification conservatively.
23052 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23053 compare_base_decls returning dont-know properly.
23055 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23057 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23058 (legitimate_offset_address_p): New function.
23059 (arc_legitimate_address_p): Use above function.
23061 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23063 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23067 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23068 ACCH registers whenever they are available.
23070 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23072 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23073 double regs fix when not used.
23075 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23077 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23079 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23080 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23082 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23084 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23085 when emitting short ADD instructions.
23087 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23089 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23091 (cmpsi_cc_c_insn): Likewise.
23092 (cbranchsi4_scratch): Compute proper instruction length using
23093 compact_hreg_operand.
23094 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23096 2017-04-25 Richard Biener <rguenther@suse.de>
23098 PR middle-end/80509
23099 * passes.c (pass_manager::pass_manager): Initialize
23100 m_name_to_pass_map.
23102 2017-04-25 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/79201
23105 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23107 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23110 * config/s390/vector.md: Split MEM->GPR vector moves for
23111 non-s_operand addresses.
23113 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23116 * config/s390/predicates.md (reload_const_wide_int_operand): New
23118 * config/s390/s390.md ("movti"): Remove d/P alternative.
23119 ("movti_bigconst"): New pattern definition.
23121 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23124 * s390-protos.h (s390_expand_cs_hqi): Removed.
23125 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23126 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23127 modes as well as CCZ1mode and CCZmode.
23128 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23129 signature of s390_emit_compare_and_swap.
23130 (s390_expand_cs_hqi): Likewise, make static.
23131 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23132 compare-and-swap, in some cases.
23133 (s390_expand_cs): Wrapper function.
23134 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23136 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23137 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23138 patterns for small and large integers. Forbid symref memory operands.
23139 Move expander to s390.c. Require cc register.
23140 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23141 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23142 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23143 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23144 symref memory operands. Remove CC mode and call s390_match_ccmode
23146 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23148 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23150 * config/s390/s390.md (define_peephole2): New peephole to help
23151 combining the load-and-test pattern with volatile memory.
23153 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23155 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23156 with CCZmode for TARGET_Z196.
23158 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23160 PR rtl-optimization/80501
23161 * combine.c (make_compound_operation_int): Set subreg_code to SET
23162 even for AND with mask of the sign bit of mode.
23164 PR rtl-optimization/80500
23165 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23166 sum's initial value.
23168 2017-04-25 Julian Brown <julian@codesourcery.com>
23169 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23171 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23173 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23175 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23177 2017-04-25 Julian Brown <julian@codesourcery.com>
23178 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23180 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23181 (thunderx2t99_sha): New Reservation.
23183 2017-04-25 Julian Brown <julian@codesourcery.com>
23184 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23186 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23187 type for 1-element load.
23189 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23191 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23193 2017-04-24 Martin Jambor <mjambor@suse.cz>
23195 PR tree-optimization/80293
23196 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23197 char arrays not totally scalarizable if it is false.
23198 (analyze_all_variable_accesses): Pass correct value in the new
23199 parameter. Add a statistics counter.
23201 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23203 PR middle-end/79931
23204 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23206 2017-04-24 Richard Biener <rguenther@suse.de>
23208 PR tree-optimization/80494
23209 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23210 out for complex types.
23212 2017-04-24 Richard Biener <rguenther@suse.de>
23214 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23215 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23216 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23217 (DFS): Adjust and never fail.
23218 (sccvn_dom_walker::fail): Remove.
23219 (sccvn_dom_walker::before_dom_children): Adjust.
23220 (run_scc_vn): Likewise and never fail.
23221 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23222 (pass_fre::execute): Likewise.
23224 2017-04-24 Richard Biener <rguenther@suse.de>
23226 PR tree-optimization/79725
23227 * tree-ssa-sink.c (statement_sink_location): Return whether
23228 failure reason was zero uses. Move that check later.
23229 (sink_code_in_bb): Deal with zero uses by removing the stmt
23232 2017-04-24 Richard Biener <rguenther@suse.de>
23235 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23236 pointer-based references.
23238 2017-04-24 Richard Biener <rguenther@suse.de>
23241 * pass_manager.h (pass_manager::operator new): Remove.
23242 (pass_manager::operator delete): Likewise.
23243 * passes.c (pass_manager::operator new): Remove.
23244 (pass_manager::operator delete): Likewise.
23245 (pass_manager::pass_manager): Zero individual pass members.
23247 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23250 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23251 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23252 Check "XEXP (src, 1)" operand here.
23253 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23254 Check "XEXP (src, 1)" operand here.
23255 (dimode_scalar_chain::make_vector_copies): Detect count register
23256 of a shift instruction. Zero extend count register from QImode
23257 to DImode to satisfy vector shift pattern count operand predicate.
23258 Substitute vector shift count operand with a DImode copy.
23259 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23262 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23264 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23265 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23266 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23267 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23268 (UNSPEC_NOREX_MEM): Remove definition.
23270 2017-04-21 Richard Biener <rguenther@suse.de>
23272 PR tree-optimization/79547
23273 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23274 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23275 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23276 without any constraints.
23278 2017-04-21 Richard Biener <rguenther@suse.de>
23280 PR tree-optimization/78847
23281 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23283 2017-04-21 Richard Biener <rguenther@suse.de>
23285 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23286 (build_distinct_type_copy): Likewise.
23287 (build_variant_type_copy): Likewise.
23288 * tree.c (build_qualified_type): Pass down mem-stat info.
23289 (build_distinct_type_copy): Likewise.
23290 (build_variant_type_copy): Likewise.
23292 2017-04-21 Richard Biener <rguenther@suse.de>
23294 PR tree-optimization/80237
23295 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23297 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23298 for a simplified result.
23300 2016-04-21 Richard Biener <rguenther@suse.de>
23302 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23303 sth as strict as a simple_iv but a chrec without symbols and an
23304 operand defined in the loop we are peeling (and not some subloop).
23305 (propagate_constants_for_unrolling): Propagate all constants.
23307 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23310 * config/i386/i386.c (print_reg): Remove assert for disalowed
23311 regno values, call output_operand_lossage instead.
23313 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23316 * config/i386/constraints.md (Yc): New register constraint.
23317 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23318 Use Yc constraint for alternative 2 of operand 0. Remove
23319 preferred_for_speed attribute.
23321 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23323 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23324 lastprivate clauses in SIMT case.
23326 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23328 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23330 2017-04-20 Richard Biener <rguenther@suse.de>
23332 PR tree-optimization/57796
23333 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23334 as N scalar stores.
23335 (vect_model_load_cost): Cost gathers as N scalar loads.
23337 2017-04-20 Richard Biener <rguenther@suse.de>
23339 * ggc-page.c (ggc_allocated_p): Rename to ...
23340 (safe_lookup_page_table_entry): ... this and return the lookup
23342 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23344 2017-04-20 Richard Biener <rguenther@suse.de>
23346 PR tree-optimization/80453
23347 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23348 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23349 from the conditions.
23350 (vn_phi_eq): Pass them down.
23351 (vn_phi_lookup): Record them.
23352 (vn_phi_insert): Likewise.
23354 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23356 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23357 uninitialized variable warning to avoid buffer overrun.
23359 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23362 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23363 is suppressed for '{ 0 }' in C.
23365 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23367 * BASE-VER: Set to 8.0.0.
23369 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23371 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23372 priority .init_array and .fini_array section with SECTION_NOTYPE
23375 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23377 PR middle-end/80423
23378 * tree.h (build_array_type): Add typeless_storage default argument.
23379 * tree.c (type_cache_hasher::equal): Also compare
23380 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23381 (build_array_type): Add typeless_storage argument, set
23382 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23384 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23385 (build_array_type): Likewise. Add typeless_storage argument.
23387 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23388 Jakub Jelinek <jakub@redhat.com>
23390 PR tree-optimization/80426
23391 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23392 operation on symbolic operands, also compute the overflow for the
23393 invariant part when the operation degenerates into a negation.
23395 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23398 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23399 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23402 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23404 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23407 * config/avr/avr.c (tree.h): Include it.
23408 (cgraph.h): Include it.
23409 (avr_encode_section_info): Don't warn for uninitialized progmem
23410 variable if it's just an alias.
23412 2017-04-19 Richard Biener <rguenther@suse.de>
23415 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23416 when needed by AutoPGO.
23418 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23421 * doc/lto.texi: Remove an extra 'that'.
23423 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23425 PR rtl-optimization/80429
23426 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23427 are only used in debug insns.
23429 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23430 Vladimir Makarov <vmakarov@redhat.com>
23432 * config/sparc/predicates.md (input_operand): Add comment. Return
23433 true for any memory operand when LRA is in progress.
23434 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23436 2017-04-18 Jeff Law <law@redhat.com>
23439 * mips.md ({return,simple_return}_internal): Do not overwrite
23442 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23444 PR tree-optimization/80443
23445 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23446 instead of adding 1, subtract -1 and similarly instead of subtracting
23449 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23451 PR rtl-optimization/80357
23452 * haifa-sched.c (tmp_bitmap): New variable.
23453 (model_recompute): Handle duplicate use records.
23454 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23455 (free_global_sched_pressure_data): Free it.
23457 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23460 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23461 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23462 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23463 instead of SYSTEM_HEADER_DIR.
23465 2017-04-18 Jeff Law <law@redhat.com>
23467 PR middle-end/80422
23468 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23469 predecessors after walking up the insn chain.
23471 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23474 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23475 sizetype type into debug info.
23477 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23480 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23481 unneeded test for TARGET_UPPER_REGS_SF.
23482 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23484 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23487 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23488 instead of gsi_after_labels.
23490 2017-04-18 Jeff Law <law@redhat.com>
23492 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23496 2017-04-13 Jeff Law <law@redhat.com>
23497 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23498 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23500 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23503 * config/avr/avr.c (intl.h): Include it.
23504 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23506 2017-04-18 Martin Liska <mliska@suse.cz>
23508 PR gcov-profile/78783
23509 * gcov-tool.c (gcov_output_files): Validate that destination
23510 file is either removed by the tool or by a user.
23512 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23513 Guy Benyei <guybe@mellanox.com>
23515 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23516 block, and do not negate it, the stored id is already negative.
23518 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23520 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23522 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23525 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23526 masks of options that should be turned off if the VSX vector
23527 options are turned off.
23528 (OTHER_P8_VECTOR_MASKS): Likewise.
23529 (OTHER_VSX_VECTOR_MASKS): Likewise.
23530 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23531 rs6000_disable_incompatible_switches to validate no type switches
23533 (rs6000_incompatible_switch): New function to disallow turning on
23534 other vector options if -mno-vsx, -mno-power8-vector, or
23535 -mno-power9-vector are specified.
23537 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23539 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23541 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23543 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23544 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23545 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23546 (ARG_POINTER_CFA_OFFSET): Likewise.
23548 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23550 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23551 conditions to take advantage of various optimizations.
23553 2017-04-13 Jeff Law <law@redhat.com>
23555 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23556 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23557 (zero_extendsidi2_dext): Likewise.
23559 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23562 * fold-const.c (fold_ternary_loc): Revert
23563 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23566 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23568 PR rtl-optimization/80343
23569 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23570 new scratch pseudo.
23572 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23575 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23576 to ubsan_encode_value.
23578 2017-04-13 Jeff Law <law@redhat.com>
23580 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23581 appearing in DEBUG_INSNs.
23583 2017-04-13 Martin Liska <mliska@suse.cz>
23585 PR gcov-profile/80413
23586 * gcov-io.c (gcov_write_string): Copy to buffer just when
23587 allocated size is greater than zero.
23589 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23592 * dwarf2out.c (decls_for_scope): Ignore declarations of
23593 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23595 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23598 * ipa-visibility.c (non_local_p): Fix typos.
23599 (localize_node): When localizing symbol in same comdat group,
23600 dissolve the group only when we know external symbols are going
23602 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23604 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23606 PR tree-optimization/79390
23607 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23608 order does not result in usable sequence, retry with reversed operand
23614 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23615 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23616 op0 instead of fold_convert_loc (loc, type, arg0).
23618 2017-04-12 Jeff Law <law@redhat.com>
23620 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23621 has a delay slot in the generated code.
23623 * config/cris/cris.md (cris_preferred_reload_class): Return
23624 GENNONACR_REGS rather than GENERAL_REGS.
23626 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23629 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23630 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23631 signedness of the result type.
23633 2017-04-12 Richard Biener <rguenther@suse.de>
23634 Jeff Law <law@redhat.com>
23636 PR tree-optimization/80359
23637 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23638 trim stores to TARGET_MEM_REFs.
23640 2017-04-12 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/79390
23643 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23644 threading case even more.
23646 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23649 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23650 for quad_address_p for TImode, instead of just not indexed_address.
23652 2017-04-12 Richard Biener <rguenther@suse.de>
23653 Bernd Edlinger <bernd.edlinger@hotmail.de>
23655 PR middle-end/79671
23656 * alias.c (component_uses_parent_alias_set_from): Handle
23657 TYPE_TYPELESS_STORAGE.
23658 (get_alias_set): Likewise.
23659 * tree-core.h (tree_type_common): Add typeless_storage flag.
23660 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23661 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23662 for types containing members with TYPE_TYPELESS_STORAGE.
23663 (place_field): Likewise.
23664 (layout_type): Likewise for ARRAY_TYPE.
23665 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23666 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23667 TYPE_TYPELESS_STORAGE.
23668 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23670 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23673 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23674 first argument to type.
23676 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23680 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23681 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23682 (rs6000_expand_binop_builtin): Likewise.
23683 (rs6000_expand_ternop_builtin): Likewise; also add missing
23684 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23685 vshasigma built-ins.
23686 * doc/extend.texi: Document that vec_xxpermdi's third argument
23687 must be a constant.
23689 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23691 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23692 Use shift_const cost parameter when calculating gain of STV shifts.
23694 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23696 PR rtl-optimization/70478
23697 * lra-constraints.c (process_alt_operands): Check memory for
23698 disfavoring memory insn operand.
23700 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23702 PR middle-end/80100
23703 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23704 left shift in unsigned HOST_WIDE_INT type.
23706 PR rtl-optimization/80385
23707 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23708 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23711 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23712 if they have any depend clauses.
23714 2017-04-11 Martin Liska <mliska@suse.cz>
23717 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23718 * ipa-split.c (split_function): Create a local comdat symbol
23719 if caller is in a comdat group.
23721 2017-04-11 Martin Liska <mliska@suse.cz>
23724 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23727 2017-04-11 Martin Sebor <msebor@redhat.com>
23729 PR middle-end/80364
23730 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23731 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23733 (directive::set_width, directive::set_precision, format_character):
23735 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23738 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23741 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23742 conflict, set target->arch_name instead of target->cpu_name.
23744 2017-04-11 Richard Biener <rguenther@suse.de>
23746 PR tree-optimization/80374
23747 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23748 build_zero_cst, remove fold_convertible_p check again.
23750 2017-04-11 Martin Liska <mliska@suse.cz>
23753 * ubsan.c (instrument_object_size): Do not instrument register
23756 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23759 * config/i386/i386-builtin-types.def
23760 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23761 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23762 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23763 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23764 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23765 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23766 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23767 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23768 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23769 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23770 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23771 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23772 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23773 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23774 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23775 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23776 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23777 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23778 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23779 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23780 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23781 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23782 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23783 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23784 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23785 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23786 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23787 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23788 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23789 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23790 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23791 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23792 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23793 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23794 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23795 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23796 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23797 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23798 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23799 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23800 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23801 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23802 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23803 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23804 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23805 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23807 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23808 flag to second_arg_count, handle 4 argument function type _COUNT
23809 aliases, handle second_arg_count on second argument rather than last.
23811 2017-04-10 Jeff Law <law@redhat.com>
23813 PR tree-optimization/80374
23814 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23815 record anything if we can not convert integer_zero_node to the
23818 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23821 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23822 Enhance special handling given to the TARGET_P9_MINMAX option in
23823 relation to certain other options.
23825 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23827 PR tree-optimization/80153
23828 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23829 remove POINTER_PLUS_EXPR's base part directly, rather than through
23832 2017-04-10 Richard Biener <rguenther@suse.de>
23833 Bin Cheng <bin.cheng@arm.com>
23835 PR tree-optimization/80153
23836 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23837 the first element of pointer type aff_tree. Build result expr in
23839 (add_elt_to_tree): Convert to type unconditionally. Remove other
23840 fold_convert calls.
23841 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23842 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23844 2017-04-10 Richard Biener <rguenther@suse.de>
23846 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23849 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23851 PR rtl-optimization/70478
23852 * lra-constraints.c (curr_small_class_check): New.
23853 (update_and_check_small_class_inputs): New.
23854 (process_alt_operands): Update curr_small_class_check. Disfavor
23855 alternative insn memory operands. Check available regs for small
23858 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23861 * config/mips/mips.opt (-mvirt): Update description.
23862 * doc/invoke.texi (-mvirt): Likewise.
23864 2017-04-10 Richard Biener <rguenther@suse.de>
23866 PR middle-end/80362
23867 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23868 looking for NEGATE_EXPR in -A / -B to A / B folding.
23870 2017-04-10 Martin Liska <mliska@suse.cz>
23872 PR gcov-profile/80224
23873 * gcov.c (print_usage): Fix usage string.
23874 (get_gcov_intermediate_filename): Remove.
23875 (output_gcov_file): Use both for normal and intermediate format.
23876 (generate_results): Do not initialize special file for
23877 intermediate format.
23879 2017-04-10 Richard Biener <rguenther@suse.de>
23881 PR tree-optimization/80304
23882 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23885 2017-04-10 Nathan Sidwell <nathan@acm.org>
23888 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23889 (rs6000_init_builtins): Use it.
23891 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23893 * config/arm/arm.md (<mrc>): Add mode to SET source.
23894 (<mrrc>): Likewise.
23896 2017-04-10 Richard Biener <rguenther@suse.de>
23898 PR middle-end/80344
23899 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23901 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23904 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23905 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23906 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23907 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23908 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23909 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23910 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23911 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23912 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23913 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23914 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23915 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23916 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23917 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23918 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23919 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23920 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23921 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23922 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23923 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23924 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23925 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23926 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23928 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23930 PR rtl-optimization/70478
23931 * lra-constraints.c: Reverse the last patch.
23933 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23935 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23936 Add comment for WCHAR_T.
23938 2017-04-08 Martin Liska <mliska@suse.cz>
23941 2017-04-07 Martin Liska <mliska@suse.cz>
23944 * ipa-split.c (split_function): Add function part to a same comdat
23947 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23950 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23952 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23954 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23955 (vec_pairing): ...to this.
23956 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23957 (rs6000_sched_init): Adjust for name change.
23958 (struct rs6000_sched_context): Likewise.
23959 (rs6000_init_sched_context): Likewise.
23960 (rs6000_set_sched_context): Likewise.
23962 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23968 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23970 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23971 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23972 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23974 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23976 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23978 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23980 PR rtl-optimization/70703
23981 * ira-color.c (update_conflict_hard_regno_costs): Use
23982 int64_t instead of HOST_WIDE_INT.
23984 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23986 PR rtl-optimization/70478
23987 * lra-constraints.c (process_alt_operands): Disfavor alternative
23988 insn memory operands.
23990 2017-04-07 Jeff Law <law@redhat.com>
23992 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23993 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23995 2017-04-07 Martin Liska <mliska@suse.cz>
23998 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23999 Show error message instead of an ICE.
24001 2017-04-07 Martin Liska <mliska@suse.cz>
24004 * ipa-split.c (split_function): Add function part to a same comdat
24007 2017-04-07 Richard Biener <rguenther@suse.de>
24009 PR middle-end/80341
24010 * tree.c (get_unwidened): Also handle ! for_type case for
24012 * convert.c (do_narrow): Split out from ...
24013 (convert_to_integer_1): ... here. Do not pass final truncation
24014 type to get_unwidened for TRUNC_DIV_EXPR.
24016 2017-04-07 Richard Biener <rguenther@suse.de>
24018 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24020 (aff_combination_const): Adjust.
24021 (aff_combination_scale): Likewise.
24022 (aff_combination_add_elt): Likewise.
24023 (aff_combination_add_cst): Likewise.
24024 (aff_combination_convert): Likewise.
24025 (add_elt_to_tree): Likewise. Remove unused argument.
24026 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24028 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24030 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24032 * config/arm/arm.c (arm_default_short_enums): Use
24033 ARM_DEFAULT_SHORT_ENUMS.
24034 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24036 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24039 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24040 members with redundant out-of-class redeclaration.
24042 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24045 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24046 * config/i386/i386.md (*zero_extendsidi2):
24047 Add (?*x,*x) and (?*v,*v) alternatives.
24049 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24052 * config/i386/i386.c (ix86_expand_builtin)
24053 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24054 mode from insn data. Convert operands to insn operand mode.
24055 Copy operands that don't satisfy insn predicate to a register.
24057 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24059 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24062 2017-04-06 Richard Biener <rguenther@suse.de>
24064 PR tree-optimization/80334
24065 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24066 preserve alignment of accesses.
24068 2017-04-06 Richard Biener <rguenther@suse.de>
24070 PR tree-optimization/80262
24071 * tree-sra.c (build_ref_for_offset): Preserve address-space
24073 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24074 Drop useless address-space information on MEM_REF offsets.
24076 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24078 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24080 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24082 PR rtl-optimization/70703
24083 * ira-color.c (update_conflict_hard_regno_costs): Use
24084 HOST_WIDE_INT instead of long.
24086 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24089 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24090 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24092 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24093 for x86_64 target. Handle -m3dnowa option.
24095 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24097 PR rtl-optimization/70703
24098 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24099 (update_conflict_hard_regno_costs): Use long instead of unsigned
24100 arithmetic for cost calculation.
24102 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24103 Bernd Edlinger <bernd.edlinger@hotmail.de>
24106 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24109 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24112 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24113 ptr_mode with Pmode throughout.
24114 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24115 into probe_stack_range and use DImode.
24117 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24120 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24121 call_eh_return is true.
24123 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24125 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24126 Initialize last_match_fntype_index.
24128 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24131 * tree-nvr.c: Include internal-fn.h.
24132 (pass_return_slot::execute): Ignore internal calls without
24135 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24136 Richard Biener <rguenther@suse.de>
24139 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24140 captures used multiple times, except for the last use.
24141 * generic-match-head.c: Include gimplify.h.
24143 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24145 PR tree-optimization/79390
24146 * target.h (struct noce_if_info): Declare.
24147 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24148 * target.def (noce_conversion_profitable_p): New target hook.
24149 * ifcvt.h (struct noce_if_info): New type, moved from ...
24150 * ifcvt.c (struct noce_if_info): ... here.
24151 (noce_conversion_profitable_p): Renamed to ...
24152 (default_noce_conversion_profitable_p): ... this. No longer
24154 (noce_try_store_flag_constants, noce_try_addcc,
24155 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24156 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24157 instead of noce_conversion_profitable_p.
24158 * config/i386/i386.c: Include ifcvt.h.
24159 (ix86_option_override_internal): Don't override
24160 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24161 (ix86_noce_conversion_profitable_p): New function.
24162 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24163 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24164 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24165 * doc/tm.texi: Regenerated.
24167 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24169 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24172 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24175 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24176 instructions for small multiply cores.
24178 2017-04-04 Jeff Law <law@redhat.com>
24180 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24182 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24183 that are not set by the loop over the elements.
24185 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24188 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24189 int mode, convert_modes it to mode as unsigned, otherwise use
24190 lowpart_subreg to mode rather than SImode.
24191 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24192 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24193 Use DImode instead of SImode for the shift count operand.
24194 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24197 2017-04-04 Richard Biener <rguenther@suse.de>
24199 PR middle-end/80281
24200 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24201 arithmetic done for the negate or the plus. Simplify.
24202 (A - (-B) -> A + B): Likewise.
24203 * fold-const.c (split_tree): Make sure to not negate pointers.
24205 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24207 PR rtl-optimization/60818
24208 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24209 a compare of comparisons with the thing compared if this results
24210 in a different machine mode.
24212 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24214 * alias.c (base_alias_check): Fix typo in comment.
24215 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24216 * cgraphunit.c (symbol_table::compile): Likewise.
24217 * collect2.c (maybe_run_lto_and_relink): Likewise.
24218 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24219 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24220 * config/avr/avr.c (avr_map_op_t): Likewise.
24221 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24222 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24223 * config/epiphany/epiphany.md (movcc): Likewise.
24224 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24225 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24227 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24228 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24229 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24230 * config/sh/sh.c (sh_rtx_costs): Likewise.
24231 * fold-const.c (fold_truth_andor): Likewise.
24232 * genautomata.c (collapse_flag): Likewise.
24233 * gengtype.h (struct type::u::s): Likewise.
24234 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24235 * input.c (FORMAT_AMOUNT): Likewise.
24236 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24237 (known_aggs_to_agg_replacement_list): Likewise.
24238 * ipa-inline-analysis.c: Likewise.
24239 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24240 * ipa-polymorphic-call.c
24241 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24242 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24243 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24245 * modulo-sched.c (apply_reg_moves): Likewise.
24246 * omp-expand.c (build_omp_regions_1): Likewise.
24247 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24248 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24249 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24250 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24251 * value-prof.c: Likewise.
24252 * var-tracking.c (val_reset): Likewise.
24254 2017-04-03 Richard Biener <rguenther@suse.de>
24256 PR tree-optimization/80275
24257 * fold-const.c (split_address_to_core_and_offset): Handle
24260 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24262 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24263 descriptors is at least equal to that of functions.
24265 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24267 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24269 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24272 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24273 (mov<IMOD4:mode>): New expander.
24274 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24276 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24278 PR rtl-optimization/79405
24279 * fwprop.c (propagations_left): New variable.
24280 (forward_propagate_into): Decrement it.
24281 (fwprop_init): Initialize it.
24282 (fw_prop): If the variable has reached zero, stop propagating.
24283 (fwprop_addr): Ditto.
24285 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24288 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24289 a FUNCTION_DECL, pass it as decl instead of origin to
24292 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24294 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24297 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24300 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24301 TARGET_VSX_SMALL_INTEGER.
24303 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24305 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24306 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24308 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24310 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24311 extraction from odd-numbered MSA register.
24313 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24315 PR middle-end/80173
24316 * expmed.c (store_bit_field_1): Don't attempt to create
24317 a word subreg out of hard registers wider than word if they
24318 have HARD_REGNO_NREGS of 1 for their mode.
24320 PR middle-end/80163
24321 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24322 conversions to integer types wider than word and pointer.
24325 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24326 (rtx_equal_for_cselib_p): Pass 0 to it.
24327 * cselib.c (cselib_hasher::equal): Likewise.
24328 (rtx_equal_for_cselib_1): Add depth argument. If depth
24329 is 128, don't look up VALUE locs and punt. Increment
24330 depth in recursive calls when walking VALUE locs.
24332 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24334 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24335 (make_gcov_file_name): Use the canonical path name for generating
24337 (read_line): Fix handling of files with ascii null bytes.
24339 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24341 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24342 to initialise a vector register instead
24343 of using a const_int.
24345 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24347 PR translation/80189
24348 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24349 diagnostic messages.
24351 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24354 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24355 (dfp_diex_<mode>): Update mode of operand 1.
24356 * doc/extend.texi (dxex, dxexq): Document change to return type.
24357 (diex, diexq): Document change to argument type.
24359 2017-03-30 Martin Jambor <mjambor@suse.cz>
24362 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24363 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24364 it reflects the signature changes performed at the callee side.
24365 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24366 to cgraph_build_function_type_skip_args.
24367 (build_function_decl_skip_args): Adjust call to the above function.
24369 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24372 * config/i386/sse.md
24373 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24374 register as dest whenever it is a MEM not rtx_equal_p to the
24375 corresponding dup operand, and when forcing into reg move the
24376 reg into the memory afterwards.
24377 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24378 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24379 for the force_reg mode.
24380 (avx512vl_vextractf128<mode>): Use register as dest either
24381 always when a MEM, or when it is a MEM not rtx_equal_p to the
24382 corresponding dup operand, or even not when it is a CONST_VECTOR
24383 depending on the mode and lo vs. hi.
24384 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24386 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24387 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24388 Likewise. Require that operands[2] is even.
24389 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24390 Remove extraneous parens. Require that operands[2] is a multiple
24392 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24393 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24394 disallow memory then.
24396 2017-03-30 Richard Biener <rguenther@suse.de>
24398 PR tree-optimization/77498
24399 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24400 to non-constants over backedges.
24402 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24404 PR rtl-optimization/80233
24405 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24406 as last_combined_insn. Do not test for BARRIER_P separately.
24408 2017-03-29 Andreas Schwab <schwab@suse.de>
24411 * calls.c (prepare_call_address): Convert funexp to Pmode before
24412 copying to temp reg.
24414 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24416 PR tree-optimization/80158
24417 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24418 Handle possible future case of more than one alternate
24420 (replace_rhs_if_not_dup): Likewise.
24421 (replace_one_candidate): Likewise.
24423 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24425 PR rtl-optimization/80193
24426 * ira.c (ira): Do not check allocation for LRA.
24428 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24430 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24431 (nvptx_output_simt_exit): Declare.
24432 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24433 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24434 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24435 (nvptx_declare_function_name): ...here. Emit declaration of local
24436 memory space buffer for omp_simt_enter insn.
24437 (nvptx_output_unisimt_switch): New.
24438 (nvptx_output_softstack_switch): New.
24439 (nvptx_output_simt_enter): New.
24440 (nvptx_output_simt_exit): New.
24441 * config/nvptx/nvptx.h (struct machine_function): New fields
24442 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24443 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24444 (UNSPECV_SIMT_EXIT): Ditto.
24445 (omp_simt_enter_insn): New insn.
24446 (omp_simt_enter): New expansion.
24447 (omp_simt_exit): New insn.
24448 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24450 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24451 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24452 (expand_GOMP_SIMT_EXIT): New.
24453 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24454 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24455 (GOMP_SIMT_EXIT): Ditto.
24456 * target-insns.def (omp_simt_enter): New insn.
24457 (omp_simt_exit): Ditto.
24458 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24460 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24461 (lower_rec_input_clauses): Likewise.
24462 (lower_lastprivate_clauses): Handle SIMT privatization.
24464 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24465 (ompdevlow_adjust_simt_enter): New.
24466 (find_simtpriv_var_op): New.
24467 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24468 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24470 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24471 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24472 (copy_decl_for_dup_finish): Ditto.
24474 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24476 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24479 * config/i386/i386.c (ix86_option_override_internal): Always
24480 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24482 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24484 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24486 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24488 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24489 mark new edge's irreducible flag accordign to it.
24490 (vect_do_peeling): Check loop preheader edge's irreducible flag
24491 and pass it to function slpeel_add_loop_guard.
24493 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24495 PR tree-optimization/80218
24496 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24497 Update block frequencies and counts.
24499 2017-03-28 Richard Biener <rguenther@suse.de>
24501 PR tree-optimization/78644
24502 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24503 of a simplification result we may not use it at all.
24505 2017-03-28 Richard Biener <rguenther@suse.de>
24508 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24509 without arguments, generate default definition of a SSA name.
24511 2017-03-28 Richard Biener <rguenther@suse.de>
24513 PR middle-end/80222
24514 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24515 TYPE_REF_CAN_ALIAS_ALL references.
24516 * fold-const.c (fold_indirect_ref_1): Likewise.
24518 2017-03-28 Martin Liska <mliska@suse.cz>
24521 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24522 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24524 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24525 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24527 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24528 (EXTRA_SPECS): Define.
24529 (SUBTARGET_EXTRA_SPECS): Likewise.
24530 (SUBTARGET_CPP_SPEC): Likewise.
24531 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24532 SUBTARGET_EXTRA_SPECS.
24533 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24535 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24537 * config/arc/simdext.md (vst64_insn): Update pattern.
24538 (vld32wh_insn): Likewise.
24539 (vld32wl_insn): Likewise.
24540 (vld64_insn): Likewise.
24541 (vld32_insn): Likewise.
24543 2017-03-28 Marek Polacek <polacek@redhat.com>
24546 * fold-const.c (fold_comparison): Use protected_set_expr_location
24547 instead of SET_EXPR_LOCATION.
24549 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24551 * tree.c (add_expr): Avoid name lookup warning.
24553 2017-03-27 Jeff Law <law@redhat.com>
24555 PR tree-optimization/80216
24556 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24557 function name. Limit recursion depth.
24558 (record_temporary_equivalences): Corresponding changes.
24560 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24562 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24565 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24568 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24570 * cfgcleanup.c (reg_note_cfa_p): New array.
24571 (insns_have_identical_cfa_notes): New function.
24572 (old_insns_match_p): Don't cross-jump in between /f
24573 and non-/f instructions. If both i1 and i2 are frame related,
24574 verify all CFA notes, their order and content.
24576 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24579 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24580 HImode and SImode with zero extend to DImode to one insn.
24581 (bswap<mode>2_extenddi): Likewise.
24582 (bswapsi2_extenddi): Likewise.
24583 (bswaphi2_extendsi): Likewise.
24584 (bswaphi2): Combine bswap HImode and SImode into one insn.
24585 Separate memory insns from swapping register.
24586 (bswapsi2): Likewise.
24587 (bswap<mode>2): Likewise.
24588 (bswaphi2_internal): Delete, no longer used.
24589 (bswapsi2_internal): Likewise.
24590 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24591 store, and gpr<-gpr swap insns.
24592 (bswap<mode>2_store): Likewise.
24593 (bswaphi2_reg): Register only splitter, combine with the splitter.
24594 (bswaphi2 splitter): Likewise.
24595 (bswapsi2_reg): Likewise.
24596 (bswapsi2 splitter): Likewise.
24597 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24598 the insns into load, store, and register/register insns.
24599 (bswapdi2_ldbrx): Likewise.
24600 (bswapdi2_load): Likewise.
24601 (bswapdi2_store): Likewise.
24602 (bswapdi2_reg): Likewise.
24604 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24606 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24607 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24609 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24612 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24615 special handling for target option conflicts between dform
24616 options (-mpower9-dform, -mpower9-dform-vector,
24617 -mpower9-dform-scalar) and -mno-direct-move.
24619 2017-03-27 Richard Biener <rguenther@suse.de>
24621 PR tree-optimization/80181
24622 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24624 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24626 * config/arc/predicates.md (move_double_src_operand): Replace the
24627 call to move_double_src_operand with a call to address_operand.
24629 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24631 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24632 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24633 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24635 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24637 * config/arc/predicates.md (long_immediate_loadstore_operand):
24638 Consider scaled addresses cases.
24640 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24642 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24643 restored when in interrupt.
24644 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24645 doesn't have delay slot.
24647 2017-03-27 Richard Biener <rguenther@suse.de>
24650 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24651 inlined thunk clones.
24653 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24656 * asan.c (instrument_derefs): Copy over last operand from
24657 original COMPONENT_REF to the new COMPONENT_REF with
24658 DECL_BIT_FIELD_REPRESENTATIVE.
24659 * ubsan.c (instrument_object_size): Likewise.
24661 2017-03-27 Richard Biener <rguenther@suse.de>
24663 PR tree-optimization/80170
24664 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24665 sure DR/SCEV didnt fold in constants we do not see when looking
24666 at the reference base alignment.
24668 2017-03-27 Richard Biener <rguenther@suse.de>
24670 PR middle-end/80171
24671 * gimple-fold.c (fold_ctor_reference): Properly guard against
24672 NULL return value from canonicalize_constructor_val.
24674 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24677 * config/i386/i386.c (ix86_expand_builtin)
24678 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24679 flags reg setting and flags reg using instructions.
24680 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24681 clobbering instructions to zero extend op2.
24683 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24685 * doc/install.texi (Configuration) <--with-aix-soname>:
24686 Update link to AIX ld.
24688 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24690 PR rtl-optimization/80160
24691 PR rtl-optimization/80159
24692 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24693 reg_alternate_class into account.
24695 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24698 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24699 to consider in curr_insn_transform.
24701 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24703 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24704 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24705 and emit_mode_inner.
24707 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24709 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24710 argument to the overloaded builtin variants. Use the new flag to
24711 deprecate certain builtin variants.
24712 * config/s390/s390-builtin-types.def: Add new builtin types.
24713 * config/s390/s390-builtins.h: Support new flags field for
24714 overloaded builtins.
24715 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24716 (s390_macro_to_expand): Enable vector float data type.
24717 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24718 builtins by incrementing the __VEC__ version number.
24719 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24721 (s390_resolve_overloaded_builtin): Emit error messages depending
24722 on the builtin flags.
24723 * config/s390/s390.c (s390_expand_builtin): Support additional
24724 flags argument. Change error message to match the messages
24725 emitted in s390-c.c.
24726 * config/s390/s390.md: New UNSPEC_* constants.
24727 (op_type): Add new instruction types.
24728 * config/s390/vecintrin.h: Add new builtins and test data class
24730 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24731 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24732 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24733 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24734 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24735 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24737 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24738 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24739 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24740 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24742 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24743 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24744 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24745 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24747 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24748 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24749 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24750 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24751 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24752 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24753 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24755 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24756 ("vec_scatter_element<V_HW_4:mode>_DI")
24757 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24758 ("vec_fpint<mode>", "vflls")
24759 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24760 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24761 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24762 ("*vec_cmphe<mode>_cc"): ... these.
24764 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24765 mode constant instead of magic value.
24767 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24769 * config/s390/s390.c (s390_expand_vec_compare): Support other
24770 vector floating point modes than just V2DF.
24771 (s390_expand_vcond): Likewise.
24772 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24773 (s390_cannot_change_mode_class): Prevent mode changes between TF
24774 and V1TF in vector registers.
24775 * config/s390/s390.md (DF, SF): New mode attributes.
24776 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24777 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24778 SFmode support for VRs.
24779 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24781 (VFT, VF_HW): New mode iterators.
24782 (vw, sdx): New mode attributes.
24783 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24784 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24785 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24786 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24787 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24788 also the new vector floating point modes. Renaming to ...
24790 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24791 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24792 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24793 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24794 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24795 ("vec_unordered<mode>"): ... these.
24797 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24798 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24799 ("*vec_extendv2df"): New insn definitions.
24801 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24803 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24804 ("mulditi3_2", "*muldi3_sign"): New patterns.
24805 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24806 rename the pattern definition.
24808 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24810 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24812 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24814 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24816 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24817 instruction if possible.
24818 * config/s390/vector.md (vec_halfnumelts): New mode
24820 ("*vec_vllezlf<mode>"): New pattern.
24822 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24824 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24825 ("popcountv4si2", "popcountv2di2"): Rename to ...
24826 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24827 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24829 ("popcount<mode>2_vxe"): New pattern.
24831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24833 * common/config/s390/s390-common.c (processor_flags_table): Add
24835 * config.gcc: Add arch12.
24836 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24837 Default to arch12 for unknown CPU model numbers.
24838 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24839 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24840 PROCESSOR_max sanity check.
24841 * config/s390/s390-opts.h (enum processor_type): Add
24843 * config/s390/s390.c (processor_table): Add arch12.
24844 (s390_expand_builtin): Add check for B_VXE flag.
24845 (s390_issue_rate): Add PROCESSOR_ARCH12.
24846 (s390_get_sched_attrmask): Likewise.
24847 (s390_get_unit_mask): Likewise.
24848 (s390_sched_score): Enable z13 scheduling for arch12.
24849 (s390_sched_reorder): Likewise.
24850 (s390_sched_variable_issue): Likewise.
24851 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24853 (s390_tune_attr): Use z13 scheduling also for arch12.
24854 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24855 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24856 (TARGET_VXE_P): New macros.
24857 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24858 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24859 * config/s390/s390.opt: Add arch12 as processor_type.
24861 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24863 * config/s390/s390.md
24864 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24865 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24866 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24868 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24869 Rename expanders to ...
24871 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24872 ("fixuns_truncdddi2_emu"): ... these.
24874 ("fixuns_trunc<mode>si2_emu"): New expander.
24876 ("*fixuns_truncdfdi2_z13"): Rename to ...
24877 ("*fixuns_truncdfdi2_vx"): ... this.
24879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24881 * config/s390/2964.md: Remove the single element vector compare
24882 instructions which are no longer used.
24883 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24885 (s390_canonicalize_comparison): Remove handling of DFmode
24887 (s390_expand_vec_compare_scalar): Remove function.
24888 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24889 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24891 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24893 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24895 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24897 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24898 will anyway by matched by mov<mode>_64dfp.
24900 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24902 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24903 vlef/vstef. Add missing operand to vleif.
24905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24907 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24908 pair for all vector types with 64 bit elements.
24909 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24910 * config/s390/vector.md (V_HW_64): ... here.
24911 (V_128_NOSINGLE): New mode iterator.
24912 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24913 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24914 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24915 ("*vec_load_pairv2di"): Change to ...
24916 ("*vec_load_pair<mode>"): ... this one.
24918 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24920 * config/s390/constraints.md: Add comments.
24921 (jKK): Reject element sizes > 8 bytes.
24922 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24924 * config/s390/s390.md: Add the s_operand checks formerly in
24925 s390_split_ok_p to various splitters where they are still
24927 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24928 for 128 bit vectors. Plus two splitters.
24930 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24932 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24935 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24938 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24939 error if the boundary argument is not constant.
24941 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24943 PR rtl-optimization/80112
24944 * loop-doloop.c (doloop_condition_get): Don't check condition
24945 if cmp isn't SET with IF_THEN_ELSE src.
24947 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24949 PR tree-optimization/80158
24950 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24951 replacing a candidate statement, also replace it for the
24952 candidate's alternate interpretation.
24953 (replace_rhs_if_not_dup): Likewise.
24954 (replace_one_candidate): Likewise.
24956 2017-03-24 Richard Biener <rguenther@suse.de>
24958 PR tree-optimization/80167
24959 * graphite-isl-ast-to-gimple.c
24960 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24962 (translate_isl_ast_to_gimple::get_rename): Likewise.
24964 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24966 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24967 handling of certain combinations of target options, including the
24968 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24969 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24971 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24974 * config/arm/arm.md (*load_multiple): Add reload_completed to
24975 matching condition.
24977 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24978 Richard Biener <rguenth@suse.de>
24980 PR tree-optimization/79908
24981 PR tree-optimization/80136
24982 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24983 been cast away, gimplify_and_add suffices.
24985 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24987 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24989 2017-03-23 Richard Biener <rguenther@suse.de>
24991 PR tree-optimization/80032
24992 * gimplify.c (gimple_push_cleanup): Forced unconditional
24993 cleanups still have to go to the conditional_cleanups
24996 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24998 PR tree-optimization/80072
24999 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25001 (next_operand_entry_id): Change type to unsigned int.
25002 (sort_by_operand_rank): Make sure to return the right return value
25003 even if unsigned fields are bigger than INT_MAX.
25004 (struct oecount): Change cnt and id type to unsigned int.
25005 (oecount_hasher::equal): Formatting fix.
25006 (oecount_cmp): Make sure to return the right return value
25007 even if unsigned fields are bigger than INT_MAX.
25008 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25011 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25012 TREE_READONLY on result if writing it more than once.
25015 * doc/invoke.texi (-fsanitize=thread): Document that with
25016 -fnon-call-exceptions atomics are not able to throw
25020 * tsan.c: Include tree-eh.h.
25021 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25022 maybe_clean_or_replace_eh_stmt where needed.
25023 (instrument_memory_accesses): Add cfg_changed argument.
25024 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25025 if it returned true.
25026 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25028 PR rtl-optimization/63191
25029 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25030 wrapper function, moved the whole old content into ...
25031 (ix86_delegitimize_address_1): ... this. New inline function.
25032 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25033 true as last argument instead of ix86_delegitimize_address.
25035 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25037 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25038 cortexa57_branch_cost.
25040 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25042 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25044 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25047 * doc/md.texi (Constraints): Document wA constraint.
25048 * config/rs6000/constraints.md (wA): New.
25049 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25050 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25051 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25052 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25054 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25057 * gimplify.c (is_oacc_declared): New function.
25058 (oacc_default_clause): Use it to set default flags for acc declared
25059 variables inside parallel regions.
25060 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25061 declared variables.
25062 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25063 declare attribute to any decl as necessary.
25065 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25068 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25069 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25070 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25071 (arm_arch_lpae): This.
25072 * config/arm/arm.c (arm_arch7ve): Rename into ...
25073 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25074 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25077 2017-03-22 Martin Liska <mliska@suse.cz>
25080 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25081 error message instead of an ICE.
25083 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25085 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25087 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25089 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25091 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25094 2017-03-21 Martin Sebor <msebor@redhat.com>
25096 * doc/extend.texi: Use "cannot" instead of "can't."
25097 * doc/hostconfig.texi: Same.
25098 * doc/install.texi: Same.
25099 * doc/invoke.texi: Same.
25100 * doc/loop.texi: Same.
25101 * doc/md.texi: Same.
25102 * doc/objc.texi: Same.
25103 * doc/rtl.texi: Same.
25104 * doc/tm.texi: Same.
25105 * doc/tm.texi.in: Same.
25106 * doc/trouble.texi: Same.
25108 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25111 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25112 (collect_checksum_attributes): Set it.
25113 (die_checksum_ordered): Use it.
25115 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25117 PR tree-optimization/79908
25118 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25119 change: For a VA_ARG whose LHS has been cast away, use
25120 force_gimple_operand to construct the side effects.
25122 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25124 PR translation/80001
25125 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25126 more amenable to translation.
25127 (oacc_loop_auto_partitions): Likewise.
25129 2017-03-21 Marek Polacek <polacek@redhat.com>
25130 Martin Sebor <msebor@redhat.com>
25132 PR tree-optimization/80109
25133 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25134 on INTEGRAL_TYPE_P.
25136 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25137 Segher Boessenkool <segher@kernel.crashing.org>
25140 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25141 check reg_used_between_p between insn and one of succ or succ2
25142 depending on if succ is artificial insn not inserted into insn
25145 2017-03-21 Martin Liska <mliska@suse.cz>
25147 PR gcov-profile/80081
25148 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25149 * doc/gcc.texi: Include gcov-dump stuff.
25150 * doc/gcov-dump.texi: New file.
25152 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25154 PR rtl-optimization/79150
25155 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25156 conditional jump, if the jump is the last insn of the loop.
25158 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25159 Richard Biener <rguenth@suse.de>
25161 PR tree-optimization/79908
25162 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25163 been cast away, use force_gimple_operand to construct the side
25166 2017-03-21 Martin Liska <mliska@suse.cz>
25168 PR libfortran/79956
25169 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25172 2017-03-21 Brad Spengler <spender@grsecurity.net>
25175 * plugin.c (htab_hash_plugin): New function.
25176 (add_new_plugin): Use it and adjust.
25177 (parse_plugin_arg_opt): Adjust.
25178 (init_one_plugin): Likewise.
25180 2017-03-21 Richard Biener <rguenther@suse.de>
25182 PR tree-optimization/80032
25183 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25184 if set force the cleanup to happen unconditionally.
25185 (gimplify_target_expr): Push inserted clobbers with force_uncond
25186 to avoid them being removed by control-dependent DCE.
25188 2017-03-21 Richard Biener <rguenther@suse.de>
25190 PR tree-optimization/80122
25191 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25192 va_arg_pack_len when the inlined call stmt requires pack
25194 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25196 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25199 * tsan.c (instrument_builtin_call): If the memory model argument
25200 is not a constant, assume it is valid.
25203 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25206 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25208 PR rtl-optimization/79910
25209 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25210 if its dest is used by an insn before I2 (other than the combined
25211 insns themselves, which are properly handled already).
25213 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25216 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25218 * combine.c (record_used_regs): New static function.
25219 (try_combine): Handle situations where there is an additional
25220 instruction between I2 and I3 which needs to have a LOG_LINK
25224 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25226 * combine.c (try_combine): Delete redundant i1 test. Call
25227 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25229 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25232 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25233 alternatives 13/14.
25235 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25237 PR tree-optimization/80054
25238 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25239 the optimization if a PHI or any of its arguments is not dominated
25240 by the candidate's basis. Use gphi* rather than gimple* as
25242 (replace_profitable_candidates): Clean up a gimple* variable that
25243 should be a gphi* variable.
25245 2017-03-20 Martin Sebor <msebor@redhat.com>
25248 * doc/extend.texi (attribute constructor): Document present limitation.
25250 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25253 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25254 __POWER9_VECTOR__ #ifdef control, change template definition to
25255 use Power9-specific built-in function.
25256 (vec_any_eq): Likewise.
25257 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25258 to control outcomes from this test.
25259 (vector_ae_<mode>p): For VEC_F modes, likewise.
25261 2017-03-20 Ian Lance Taylor <iant@google.com>
25263 * config/i386/i386.c (ix86_function_regparm): Save an extra
25264 register for -fsplit-stack with DECL_STATIC_CHAIN.
25266 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25269 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25270 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25272 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25274 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25276 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25279 2017-03-20 Marek Polacek <polacek@redhat.com>
25282 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25284 2017-03-20 Richard Biener <rguenther@suse.de>
25286 PR tree-optimization/80113
25287 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25288 allocate extra SSA name for PHI def.
25289 (add_close_phis_to_outer_loops): Likewise.
25290 (add_close_phis_to_merge_points): Likewise.
25291 (copy_loop_close_phi_args): Likewise.
25292 (copy_cond_phi_nodes): Likewise.
25294 2017-03-20 Martin Liska <mliska@suse.cz>
25296 PR middle-end/79753
25297 * tree-chkp.c (chkp_build_returned_bound): Do not build
25298 returned bounds for a LHS that's not a BOUNDED_P type.
25300 2017-03-20 Martin Liska <mliska@suse.cz>
25304 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25305 COMPLEX_CST and VECTOR_CST.
25307 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25310 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25311 target operand. A new splitter adds the clobber statement in case
25312 the target operand is dead anyway.
25314 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25316 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25317 to age-old versions of binutils and glibc.
25319 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25321 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25323 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25325 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25327 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25329 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25330 requirement for binutils 2.13.
25332 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25334 * combine.c (try_combine): Delete redundant i1 test. Call
25335 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25337 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25339 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25340 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25342 <riscv64-*-elf>: Re-arrange section
25343 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25344 <riscv32-*-linux>: Likewise.
25345 <riscv64-*-elf>: Likewise
25346 <riscv64-*-linux>: Likewise.
25348 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25351 * aarch64.opt(verbose-cost-dump): Fix typo.
25353 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25356 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25357 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25359 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25361 * reload.c (find_reloads): When reloading a nonoffsettable address,
25362 use RELOAD_OTHER for it and its address reloads.
25364 PR rtl-optimization/79910
25365 * combine.c (record_used_regs): New static function.
25366 (try_combine): Handle situations where there is an additional
25367 instruction between I2 and I3 which needs to have a LOG_LINK
25370 2017-03-17 Jeff Law <law@redhat.com>
25372 PR tree-optimization/71437
25373 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25374 conditional in the hash table first.
25375 (vrp_dom_walker::before_dom_children): Extract condition from
25376 ASSERT_EXPR. Record condition, its inverion and any implied
25377 conditions as well.
25379 2017-03-17 Marek Polacek <polacek@redhat.com>
25380 Markus Trippelsdorf <markus@trippelsdorf.de>
25382 PR tree-optimization/80079
25383 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25386 2017-03-17 Richard Biener <rguenther@suse.de>
25388 PR middle-end/80075
25389 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25390 Properly verify the LHS before the RHS possibly claims to be
25392 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25395 2017-03-17 Martin Jambor <mjambor@suse.cz>
25397 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25398 (List of -O2 options): Likewise.
25399 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25402 2017-03-17 Tom de Vries <tom@codesourcery.com>
25404 * gcov-dump.c (print_usage): Print bug_report_url.
25406 2017-03-17 Richard Biener <rguenther@suse.de>
25408 PR middle-end/80050
25409 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25410 (parser::peek): Likewise.
25412 2017-03-17 Richard Biener <rguenther@suse.de>
25414 PR tree-optimization/80048
25415 * sese.c (free_sese_info): Properly release rename_map and
25416 copied_bb_map elements.
25418 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25420 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25421 Add linked-list forward and backlinks. Insert on
25422 construction, remove on destruction.
25423 (class pass_store_merging): Add m_stores_head field.
25424 (pass_store_merging::terminate_and_process_all_chains):
25425 Iterate over m_stores_head list.
25426 (pass_store_merging::terminate_all_aliasing_chains):
25428 (pass_store_merging::execute): Check for debug stmts first.
25429 Push new chains onto the m_stores_head stack.
25431 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25434 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25435 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25436 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25438 2017-03-16 Jeff Law <law@redhat.com>
25440 PR tree-optimization/71437
25441 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25442 member function. Implementation moved into after_dom_children
25443 member function and into the threader's thread_outgoing_edges
25445 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25446 some code into new thread_outgoing_edges.
25447 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25448 definition. Simplify marker handling (do it here). Assume we always
25449 have the available expression and the const/copies tables.
25450 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25452 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25453 * tree-vrp.c (equiv_stack): No longer file scoped.
25454 (vrp_dom_walker): New class.
25455 (vrp_dom_walker::before_dom_children): New member function.
25456 (vrp_dom_walker::after_dom_children): Likewise.
25457 (identify_jump_threads): Setup domwalker. Use it rather than
25458 walking edges in a random order by hand. Simplify setup/finalization.
25459 (finalize_jump_threads): Remove.
25460 (vrp_finalize): Do not call identify_jump_threads here.
25461 (execute_vrp): Do it here instead and call thread_through_all_blocks
25464 PR tree-optimization/71437
25465 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25467 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25469 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25470 (dom_opt_dom_walker::thread_across_edge): Remove
25471 handle_dominating_asserts argument. All callers changed.
25472 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25473 changes. Remove calls to lhs_of_dominating_assert. Other
25474 uses of handle_dominating_asserts turn into unconditional code
25475 (simplify_control_stmt_condition_1): Likewise.
25476 (simplify_control_stmt_condition): Likewise.
25477 (thread_through_normal_block, thread_across_edge): Likewise.
25478 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25479 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25480 object if it is not an SSA_NAME.
25481 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25482 before calling into the VRP specific simplifiers.
25483 (identify_jump_threads): Remove handle_dominating_asserts
25486 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25489 * tree-diagnostic.c (default_tree_printer): No longer static.
25490 * tree-diagnostic.h (default_tree_printer): New prototype.
25492 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25494 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25495 Change ins into fmov.
25497 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25499 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25500 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25501 Use h_con constraint for operand 1.
25502 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25503 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25505 2017-03-15 Jeff Law <law@redhat.com>
25507 PR tree-optimization/71437
25508 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25509 (record_temporary_equivalences): Use it.
25511 PR tree-optimization/71437
25512 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25513 tree-ssa-scopedtables.
25514 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25515 (record_conditions, record_cond, vuse_eq): Likewise.
25516 (record_edge_info): Adjust to API tweak of record_conditions.
25517 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25518 (record_temporary_equivalences, optimize_stmt): Likewise.
25519 (eliminate_redundant_computations): Likewise.
25520 (record_equivalences_from_stmt): Likewise.
25521 * tree-ssa-scopedtables.c: Include options.h and params.h.
25522 (vuse_eq): New function, moved from tree-ssa-dom.c
25523 (build_and_record_new_cond): Likewise.
25524 (record_conditions): Likewise. Accept vector of conditions rather
25525 than edge_equivalence structure for first argument.
25526 for the first argument.
25527 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25528 from tree-ssa-dom.c.
25529 (avail_exprs_stack::record_cond): Likewise.
25530 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25531 from tree-ssa-dom.c.
25532 (avail_exprs_stack): Add new member functions lookup_avail_expr
25534 (record_conditions): Declare.
25536 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25539 * lra-constraints.c (process_alt_operands): Increase reject for
25540 reloading an input/output operand.
25542 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25545 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25546 insns to convert from signed/unsigned char/short to IEEE 128-bit
25548 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25550 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25553 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25554 subreg of inner mode for values already in registers.
25556 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25558 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25559 iteration reg is used after the loop.
25561 2017-03-14 Martin Sebor <msebor@redhat.com>
25563 PR tree-optimization/79800
25564 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25565 precision in negative-positive range.
25566 (format_floating): Call non-const overload with adjusted precision.
25568 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25571 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25574 2017-03-14 Martin Sebor <msebor@redhat.com>
25576 PR middle-end/80020
25577 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25578 * builtins.def (aligned_alloc): Use it.
25581 * Makefile.in (GTFILES): Add calls.c.
25582 * calls.c: Include "gt-calls.h".
25584 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25586 PR rtl-optimization/79728
25587 * regs.h (struct target_regs): New field
25588 x_contains_allocatable_regs_of_mode.
25589 (contains_allocatable_regs_of_mode): New macro.
25590 * reginfo.c (init_reg_sets_1): Initialize it, and change
25591 contains_reg_of_mode so it includes global regs as well.
25592 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25593 rather than contains_regs_of_mode.
25595 2017-03-14 Martin Liska <mliska@suse.cz>
25597 * doc/invoke.texi: Document options that can't be combined with
25598 -fcheck-pointer-bounds.
25600 2017-03-14 Martin Liska <mliska@suse.cz>
25602 PR middle-end/79831
25603 * doc/invoke.texi (-Wchkp): Document the option.
25605 2017-03-14 Martin Liska <mliska@suse.cz>
25607 * Makefile.in: Install gcov-dump.
25609 2017-03-14 Martin Liska <mliska@suse.cz>
25611 * multiple_target.c (expand_target_clones): Bail out for
25612 an invalid attribute.
25614 2017-03-14 Richard Biener <rguenther@suse.de>
25616 * alias.c (struct alias_set_entry): Pack properly.
25617 * cfgloop.h (struct loop): Likewise.
25618 * cse.c (struct set): Likewise.
25619 * ipa-utils.c (struct searchc_env): Likewise.
25620 * loop-invariant.c (struct invariant): Likewise.
25621 * lra-remat.c (struct cand): Likewise.
25622 * recog.c (struct change_t): Likewise.
25623 * rtl.h (struct address_info): Likewise.
25624 * symbol-summary.h (function_summary): Likewise.
25625 * tree-loop-distribution.c (struct partition): Likewise.
25626 * tree-object-size.c (struct object_size_info): Likewise.
25627 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25628 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25629 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25630 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25631 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25632 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25633 (struct _stmt_vec_info): Likewise.
25635 2017-03-14 Martin Liska <mliska@suse.cz>
25638 * multiple_target.c (create_dispatcher_calls): Check that
25639 a target can create a function dispatcher.
25641 2017-03-14 Martin Liska <mliska@suse.cz>
25644 * multiple_target.c (expand_target_clones): Drop local.local
25645 flag for default implementation.
25647 2017-03-14 Richard Biener <rguenther@suse.de>
25649 PR tree-optimization/80030
25650 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25652 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25654 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25655 gcc_fallthrough() instead of __attribute__((fallthrough));
25657 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25659 * doc/gcc.texi: Remove "up" link to (DIR).
25660 * doc/gccint.texi: Ditto.
25662 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25664 * doc/install.texi (Specific) <avr>: Remove reference to
25667 2017-03-13 Jeff Law <law@redhat.com>
25669 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25670 attribute rather than comments.
25672 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25673 match_scratch operand is highest.
25675 2017-03-13 Martin Liska <mliska@suse.cz>
25677 PR middle-end/78339
25678 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25679 is a CHKP clone, use original declaration.
25681 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25683 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25684 (arc_conditional_register_usage): Use a different allocation order
25685 when optimizing for size.
25686 * common/config/arc/arc-common.c (arc_option_optimization_table):
25687 Section anchors default on when optimizing for size.
25689 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25691 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25693 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25695 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25696 * config/arc/arc.md (cpu_facility): Add cd variant.
25697 (*movqi_insn): Add code density variant.
25698 (*movhi_insn): Likewise.
25699 (*movqi_insn): Likewise.
25700 (*addsi3_mixed): Likewise.
25701 (subsi3_insn): Likewise.
25703 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25705 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25707 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25709 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25710 expressions with MINUS and UNARY ops.
25712 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25715 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25717 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25718 between vec_select and vector argument.
25719 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25720 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25721 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25722 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25723 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25724 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25726 2017-03-13 Richard Biener <rguenther@suse.de>
25729 * params.def (vect-max-peeling-for-alignment): Fix typo.
25731 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25733 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25734 issue that only occurred with binutils below 2.18.
25736 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25738 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25739 refer to binutils 2.11/2.12 minimum.
25741 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25743 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25744 ftp.kernel.org and simplify binutils requirement.
25746 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25748 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25750 (Optimize Options): Ditto. Also remove redundancy.
25752 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25754 PR translation/79848
25755 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25757 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25758 to G_ to avoid double translation.
25760 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25762 PR translation/79923
25763 * auto-profile.c (get_combined_location): Convert leading
25764 character of diagnostics to lower case and remove trailing period.
25765 (read_profile): Likewise for various diagnostics.
25766 * config/arm/arm.c (arm_option_override): Remove trailing period
25767 from various diagnostics.
25768 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25769 (msp430_expand_delay_cycles): Likewise.
25771 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25774 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25775 full command-line argument, rather than just "str".
25776 (aarch64_validate_march): Likewise.
25777 (aarch64_validate_mtune): Likewise.
25779 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25781 PR rtl-optimization/78911
25782 * lra-assigns.c (must_not_spill_p): New function.
25783 (spill_for): Use it.
25785 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25787 PR tree-optimization/79981
25788 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25789 ATOMIC_COMPARE_EXCHANGE ifn result.
25790 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25791 IFN_ATOMIC_COMPARE_EXCHANGE.
25793 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25796 * opts.c (parse_sanitizer_options): Add missing question mark to
25797 "did you mean" message.
25799 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25801 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25803 (VMULEUH_UNS): Likewise.
25804 (VMULOUB_UNS): Likewise.
25805 (VMULOUH_UNS): Likewise.
25806 * config/rs6000/rs6000.c (builtin_function_type): Remove
25807 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25809 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25812 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25813 x with result of extra_parsing_for_operand_code_0.
25814 (function_reader::extra_parsing_for_operand_code_0): Convert
25815 return type from void to rtx, returning x. When reading
25816 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25817 larger size containing struct block_symbol.
25819 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25821 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25822 -mfloat128-hardware without -m64.
25824 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25827 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25828 entries to the case statement that marks unsigned arguments to
25829 overloaded functions.
25831 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25833 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25834 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25836 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25839 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25840 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25842 2017-03-10 Martin Liska <mliska@suse.cz>
25846 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25847 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25850 2017-03-10 Olivier Hainque <hainque@adacore.com>
25852 * tree-switch-conversion (array_value_type): Start by resetting
25853 candidate type to it's main variant.
25855 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25857 PR rtl-optimization/79909
25858 * combine.c (try_combine): Use simplify_replace_rtx on individual
25859 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25860 of the whole CALL_INSN_FUNCTION_USAGE.
25862 PR tree-optimization/79972
25863 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25864 get_range_info on SSA_NAMEs. Formatting fixes.
25866 2017-03-10 Richard Biener <rguenther@suse.de>
25867 Jakub Jelinek <jakub@redhat.com>
25869 PR tree-optimization/77975
25870 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25871 edge to be constant.
25872 (get_val_for): For constant x return it. Formatting fix.
25873 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25874 would use the same bases as the current one.
25876 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25878 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25879 instead of vec_select for V1TImode.
25880 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25882 (VSX_LE_128): Add V1TI to this mode iterator.
25883 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25884 (*vsx_le_perm_store_<mode>): Likewise.
25885 (pre-reload splitter for VSX stores): Likewise.
25886 (post-reload splitter for VSX stores): Likewise.
25887 (*vsx_xxpermdi2_le_<mode>): Likewise.
25888 (*vsx_lxvd2x2_le_<mode>): Likewise.
25889 (*vsx_stxvd2x2_le_<mode>): Likewise.
25891 2017-03-09 Michael Eager <eager@eagercon.com>
25893 Correct failures with --enable-checking=yes,rtl.
25895 * config/microblaze/microblaze.c (microblaze_expand_shift):
25896 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25897 test for const0_rtx.
25898 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25899 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25901 2017-03-09 Richard Biener <rguenther@suse.de>
25903 PR tree-optimization/79977
25904 * graphite-scop-detection.c (scop_detection::merge_sese):
25905 Handle the case of extra exits to blocks dominating the entry.
25907 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25909 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25912 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25914 PR rtl-optimization/79949
25915 * lra-constraints.c (process_alt_operands): Check memory when
25916 trying to predict a cycle. Print about the overall increase.
25918 2017-03-09 Richard Biener <rguenther@suse.de>
25920 PR middle-end/79971
25921 * gimple-expr.c (useless_type_conversion_p): Preserve
25922 TYPE_SATURATING for fixed-point types.
25924 2017-03-09 Richard Biener <rguenther@suse.de>
25927 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25928 alignment of BLKmode params.
25930 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25933 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25934 (VALL_NO_V2Q): Likewise.
25936 * config/aarch64/aarch64-simd.md
25937 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25939 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25940 VALL_NO_V2Q mode iterator.
25941 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25943 2017-03-09 Martin Liska <mliska@suse.cz>
25945 PR tree-optimization/79631
25946 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25947 tree_int_cst_sign_bit just for INTEGER constants.
25949 2017-03-09 Martin Liska <mliska@suse.cz>
25953 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25956 2017-03-09 Marek Polacek <polacek@redhat.com>
25959 * tree.c (inchash::add_expr): Handle TREE_VEC.
25961 2017-03-09 Martin Liska <mliska@suse.cz>
25964 (chkp_narrow_size_and_offset): New function.
25965 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25966 (void chkp_parse_bit_field_ref): New function.
25967 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25968 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25970 2017-03-09 Martin Liska <mliska@suse.cz>
25973 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25974 (chkp_find_bounds_1): Remove gcc_unreachable.
25976 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25979 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25980 BUILT_IN_SYNC*, determine the access type from the size suffix and
25981 always build a MEM_REF with that type. Handle forgotten
25982 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25985 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25986 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25987 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25988 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25989 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25990 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25991 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25992 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25993 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25994 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25995 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25996 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25997 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25998 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25999 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26000 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26001 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26002 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26003 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26004 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26005 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26006 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26007 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26008 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26009 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26010 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26011 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26012 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26013 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26014 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26015 definitions outside of __OPTIMIZE__ guarded section.
26018 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26019 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26020 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26021 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26024 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26026 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26027 ("vfenez<mode>"): Add missing constraints.
26029 2017-03-08 Martin Sebor <msebor@redhat.com>
26032 * config/nds32/nds32.c (nds32_option_override):
26033 Fix misspelled diagnostic.
26035 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26038 * gimplify.c (gimplify_omp_for): Replace index var in outer
26039 taskloop statement with an artificial variable and add
26040 OMP_CLAUSE_PRIVATE clause for it.
26042 2017-03-08 Richard Biener <rguenther@suse.de>
26044 PR tree-optimization/79955
26045 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26046 for accesses that are completely outside of the variable.
26048 2017-03-08 Andrew Haley <aph@redhat.com>
26050 PR tree-optimization/79943
26051 * tree-ssa-loop-split.c (compute_new_first_bound): When
26052 calculating the new upper bound, (END-BEG) should be added, not
26055 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26057 * config/avr/avr.md (setmemhi): Make sure match_dup
26058 operand number comes before match_scratch.
26060 2017-03-08 Richard Biener <rguenther@suse.de>
26062 PR tree-optimization/79920
26063 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26064 with ncopies == 1 to ...
26065 (vect_transform_slp_perm_load): ... here. Properly compute
26066 all element loads by iterating VF times over the group. Do
26067 not handle ncopies (computed in a broken way) in
26068 vect_create_mask_and_perm.
26070 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26073 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26074 is a uniform vector, use uniform_vector_p return value instead of
26075 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26077 2017-03-07 Marek Polacek <polacek@redhat.com>
26079 PR middle-end/79809
26080 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26081 (alloca_call_type): Likewise.
26083 2017-03-07 Martin Liska <mliska@suse.cz>
26085 * gcov.c (process_args): Put comment to correct location.
26087 2017-03-07 Martin Liska <mliska@suse.cz>
26089 PR middle-end/68270
26090 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26091 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26092 (chkp_narrow_bounds_for_field): Likewise.
26093 (chkp_parse_array_and_component_ref): Pass one more argument to
26096 2017-03-07 Richard Biener <rguenther@suse.de>
26098 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26101 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26103 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26104 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26106 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26109 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26110 to end of description.
26111 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26113 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26115 PR rtl-optimization/79901
26116 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26118 (*avx512f_<code><mode>3<mask_name>): ... this.
26119 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26120 iterator instead of VI8_AVX2_AVX512BW.
26122 PR rtl-optimization/79901
26123 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26124 min/max expander, expand it using expand_vec_cond_expr.
26127 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26130 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26133 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26134 to void * for PCH reasons.
26135 * dwarf2out.c (output_loc_operands, output_die): Cast
26136 v.val_vec.array to unsigned char *.
26138 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26141 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26144 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26146 PR rtl-optimization/79571
26147 * lra-constraints.c (process_alt_operands): Calculate static
26148 reject and subtract it from overall when only addresses will be
26151 2017-03-06 Julia Koval <julia.koval@intel.com>
26154 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26155 incoming stack boundary to 128 for 64-bit targets.
26157 2017-03-06 Richard Biener <rguenther@suse.de>
26159 PR tree-optimization/79894
26160 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26161 to NULL after folding it.
26163 2017-03-06 Richard Biener <rguenther@suse.de>
26165 PR tree-optimization/79824
26166 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26167 check disabling peeling for gaps.
26169 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26171 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26172 attributes): Document gettimeofday.
26174 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26176 * config/s390/s390.c (s390_option_override_internal): Set
26177 PARAM_MIN_VECT_LOOP_BOUND
26179 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26181 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26182 * config/s390/s390.md: Likewise.
26184 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26187 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26188 (<avx2_avx512>_perm<mode>): Rename to ...
26189 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26191 (<avx512>_perm<mode>_mask): Rename to ...
26192 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26194 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26195 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26196 instead of VI8F_256_512.
26197 (avx512f_perm<mode>): New define_expand.
26198 (avx512f_perm<mode>_mask): Likewise.
26199 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26200 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26202 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26204 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26205 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26207 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26209 2017-03-06 Martin Liska <mliska@suse.cz>
26212 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26213 when having a SSA NAME w/o VAR_DECL assigned to it.
26215 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26217 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26218 msa_dpsub_<su>_d): Fix MODE for vec_select.
26220 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26222 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26224 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26226 2017-03-06 Richard Biener <rguenther@suse.de>
26228 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26229 * plugin.c (register_plugin_info): Likewise.
26230 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26232 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26234 * config/i386/sse.md (sse_storehps, sse_storelps,
26235 avx_<castmode><avxsizesuffix>_<castmode>,
26236 avx512f_<castmode><avxsizesuffix>_<castmode>,
26237 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26238 in condition that at least one operand is not a MEM.
26240 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26242 PR middle-end/79805
26243 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26244 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26246 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26247 gimple_call_nothrow_p flag based on whether original builtin can throw.
26248 If it can, emit following stmts on the fallthrough edge.
26249 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26250 don't create new bb if inserting just debug stmts on the edge, try to
26251 insert them on the fallthru bb or just reset debug stmts.
26253 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26256 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26257 restore recog_data (including the operand rtxes inside it) around
26258 the call to get_insn_template.
26260 2017-03-03 Martin Sebor <msebor@redhat.com>
26262 PR tree-optimization/79699
26263 * context.c (context::~context): Free MPFR caches to avoid
26264 a memory leak on program exit.
26266 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26268 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26269 Use wide_int::ulow () instead of .elt (0).
26271 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26273 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26274 (*pushxf): Limit oF constraint to 32bit targets and add oC
26275 constraint for 64bit targets.
26276 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26277 (*pushdf): Change rmF constraint to rmC.
26279 2017-03-03 Martin Liska <mliska@suse.cz>
26281 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26282 Remove unused variable.
26284 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26287 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26288 is a memory operand, increase num_memory.
26289 (ix86_expand_args_builtin): Likewise.
26291 2017-03-03 Jan Hubicka <jh@suse.cz>
26294 * ipa-devirt.c (maybe_record_node): Properly handle
26295 __cxa_pure_virtual visibility.
26297 2017-03-03 Martin Liska <mliska@suse.cz>
26299 PR tree-optimization/79803
26300 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26302 (pass_loop_prefetch::execute): Disabled optimization if an
26303 assumption about L1 cache size is not met.
26305 2017-03-03 Martin Liska <mliska@suse.cz>
26307 PR rtl-optimization/79574
26308 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26309 (hash_scan_set): Likewise.
26310 (dump_hash_table): Likewise.
26311 (hoist_code): Likewise.
26313 2017-03-03 Richard Biener <rguenther@suse.de>
26315 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26316 in place of uhigh ().
26317 (fixed_convert_from_real): Likewise.
26319 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26322 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26324 2017-03-03 Richard Biener <rguenther@suse.de>
26326 PR middle-end/79818
26327 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26328 TYPE_OVERFLOW_UNDEFINED check.
26330 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26332 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26334 (vector_ae_<mode>_p): Likewise.
26335 (vector_nez_<mode>_p): Likewise.
26336 (vector_ne_v2di_p): Likewise.
26337 (vector_ae_v2di_p): Likewise.
26338 (vector_ne_<mode>_p): Likewise.
26339 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26341 (vsx_tsqrt<mode>2_fe): Likewise.
26343 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26346 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26348 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26350 PR rtl-optimization/79780
26351 * cprop.c (one_cprop_pass): When second and further conditional trap
26352 in a single basic block is turned into an unconditional trap, turn it
26353 into a deleted note to avoid RTL verification failures.
26355 2017-03-02 Richard Biener <rguenther@suse.de>
26357 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26359 2017-03-02 Richard Biener <rguenther@suse.de>
26361 PR tree-optimization/79345
26363 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26364 param and abort the walk, returning -1 if it is hit.
26365 (walk_aliased_vdefs): Take a limit param and pass it on.
26366 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26367 defaulting to 0 and return a signed int.
26368 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26369 (check_defs): New helper.
26370 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26371 about uninitialized memory.
26372 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26373 bogus uninitialized warning.
26374 (fixed_convert_from_real): Likewise.
26376 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26378 PR tree-optimization/66768
26379 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26380 iv_use if base object can't be determined.
26382 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26384 PR tree-optimization/79345
26385 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26386 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26387 (get_pattern_stats): Initialize it.
26388 * genemit.c (gen_expand): Verify match_scratch numbers come after
26389 match_operand/match_dup numbers.
26390 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26391 match_scratch numbers.
26392 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26394 * config/s390/s390.md (trunctdsd2): Likewise.
26396 2017-03-02 Richard Biener <rguenther@suse.de>
26398 * wide-int.h (wide_int_storage::operator=): Implement in terms
26401 2017-03-02 Richard Biener <rguenther@suse.de>
26403 PR tree-optimization/79777
26404 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26405 the to insert expression to sth existing.
26407 2017-03-01 Martin Sebor <msebor@redhat.com>
26409 PR middle-end/79692
26410 * gimple-ssa-sprintf.c
26411 (directive::known_width_and_precision): New function.
26412 (format_integer): Use it.
26413 (get_mpfr_format_length): Consider the full range of precision
26414 when computing %g output with the # flag. Set the likely byte
26415 count to 3 rather than 1 when precision is indeterminate.
26416 (format_floating): Correct the lower bound of precision.
26418 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26420 * doc/invoke.texi: Document default code model for 64-bit Linux.
26422 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26425 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26426 udiv rather than div since input pattern is unsigned.
26428 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26430 * config/i386/i386.c (print_reg): Warn for values of
26431 unsupported size in integer register.
26433 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26436 * config/rs6000/predicates.md (current_file_function_operand): Do
26437 not allow self calls to be local if the function is replaceable.
26439 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26442 * config/rs6000/altivec.h (vec_ctz and others): Change the
26443 preprocessor macro that controls conditional compilation from
26444 _ARCH_PWR9 to __POWER9_VECTOR__.
26445 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26446 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26447 control (instead of _ARCH_PWR9 control) so that template
26448 definition uses power9-specific function.
26449 (vec_any_eq): Likewise.
26450 (vec_all_ne): Change macro definition to use a power9-specific
26451 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26452 _ARCH_PWR9 control).
26453 (vec_any_eq) Likewise.
26454 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26455 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26456 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26457 support for xvcmpnedp instruction.
26458 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26459 macro expansion so that Power9 implementation of vec_all_ne does
26460 not use the AltiVec predicate framework.
26461 (VCMPNEH_P): Likewise.
26462 (VCMPNEW_P): Likewise.
26463 (VCMPNED_P): Likewise.
26464 (VCMPNEFP_P): Likewise.
26465 (VCMPNEDP_P): Likewise.
26466 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26467 implementation of vec_any_eq to not use AltiVec predicate
26469 (VCMPAEH_P): Likewise.
26470 (VCMPAEW_P): Likewise.
26471 (VCMPAED_P): Likewise.
26472 (VCMPAEFP_P): Likewise.
26473 (VCMPAEDP_P): Likewise.
26474 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26475 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26476 not use the AltiVec predicate framework.
26477 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26478 of vec_any_eq to not use AltiVec predicate framework.
26479 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26480 support for predefined __POWER9_VECTOR__ macro to indicate that
26481 Power9 instruction selection is enabled.
26482 (altivec_overloaded_builtins): Remove extraneous
26483 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26484 function argument types RS6000_BTI_bool_V16QI and
26485 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26486 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26487 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26488 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26489 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26490 Power9 for implementations of vec_cmpne. Change the signature for
26491 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26492 (representing vec_all_ne) to remove the previously described first
26493 argument of type RS6000_BTI_INTSI, as this was an artifact of
26494 reliance on the AltiVec predicate framework, which is no longer
26495 used in the implementation of these functions. Add
26496 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26497 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26498 since, unlike the AltiVec predicate framework implementation, we
26499 do not share function descriptors between vec_alle and vec_anyeq.
26500 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26501 set of modes that receive special treatment even when
26502 TARGET_P9_VECTOR is true. The special treatment emits code that
26503 does not depend on Power9 instructions.
26504 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26505 define_expand to not rely on AltiVec predicate framework.
26506 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26508 (vector_ne_v2di_p): Change this define_expand to not rely on
26509 AltiVec predicate framework.
26510 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26512 (vector_ne_<mode>_p): Change this define_expand to not rely on
26513 AltiVec predicate framework.
26514 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26516 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26517 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26518 define_insn pattern.
26519 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26520 define_insn pattern because the xvcmpne<VSs>. instruction is not
26522 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26523 instruction is not supported.
26525 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26527 * config/nvptx/nvptx.c: Include intl.h.
26529 2017-03-01 Martin Jambor <mjambor@suse.cz>
26532 * ipa-prop.h (ipa_bits): Removed field known.
26533 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26534 to pointers. Adjusted their comments to warn about their sharing.
26535 (ipcp_transformation_summary): Change bits to a vector of pointers.
26536 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26537 (ipa_get_ipa_bits_for_value): Declare.
26538 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26539 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26540 (ipa_bits_hash_table): Likewise.
26541 (ipa_vr_ggc_hash_traits): Likewise.
26542 (ipa_vr_hash_table): Likewise.
26543 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26544 being pointers and vr_known being removed.
26545 (ipa_set_jf_unknown): Likewise.
26546 (ipa_get_ipa_bits_for_value): New function.
26547 (ipa_set_jfunc_bits): Likewise.
26548 (ipa_get_value_range): New overloaded functions.
26549 (ipa_set_jfunc_vr): Likewise.
26550 (ipa_compute_jump_functions_for_edge): Use the above functions to
26551 construct bits and vr parts of jump functions.
26552 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26553 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26555 (ipcp_grow_transformations_if_necessary): Also allocate
26556 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26558 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26559 them. Fix too long lines.
26560 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26561 vr_known being removed.
26562 (ipa_read_jump_function): Use new setter functions to construct bits
26563 and vr parts of jump functions or set them to NULL.
26564 (write_ipcp_transformation_info): Adjust for bits being pointers.
26565 (read_ipcp_transformation_info): Likewise.
26566 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26568 Include gt-ipa-prop.h.
26569 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26571 (ipcp_store_bits_results): Likewise.
26572 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26573 Do not write to existing jump functions but use a temporary instead.
26575 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26578 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26579 attempt to use its first operand as BIT_FIELD_REF base.
26581 2017-03-01 Richard Biener <rguenther@suse.de>
26583 PR middle-end/79721
26584 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26585 interpolating formula in wrapping arithmetic.
26586 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26588 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26590 PR tree-optimization/79734
26591 * tree-vect-generic.c (expand_vector_condition): Optimize
26592 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26593 Handle VEC_COND_EXPR where comparison has different inner width from
26594 type's inner width.
26596 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26598 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26599 markup, and similar issues. Remove @opindex entries for things
26600 that aren't options. Add missing -mmpy-option entries.
26602 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26604 PR tree-optimization/79737
26605 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26606 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26607 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26608 instead of byte_size. Formatting fix.
26609 (shift_bytes_in_array_right): Formatting fix.
26611 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26614 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26615 condition on optimize for the leaf function test.
26617 2017-02-28 Martin Liska <mliska@suse.cz>
26620 * read-rtl-function.c (function_reader::handle_unknown_directive):
26621 Bail out when one uses -flto.
26623 2017-02-28 Martin Liska <mliska@suse.cz>
26625 * common.opt: Replace space with tabular for options of <number>
26627 * config/i386/i386.opt: Show <number> value for
26628 -mlarge-data-threshold.
26629 * opts.c (print_filtered_help): Do not display number in hexadecimal
26632 2017-02-28 Martin Liska <mliska@suse.cz>
26634 * common.opt: Fix --help=option -Q for options which are of
26637 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26639 * config/i386/i386.c (print_reg): Error out for values
26640 of 8-bit size in invalid integer register.
26642 2017-02-28 Martin Sebor <msebor@redhat.com>
26644 PR tree-optimization/79691
26645 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26647 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26650 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26651 gcc_unreachable with output_operand_lossage.
26653 2017-02-28 Richard Biener <rguenther@suse.de>
26655 PR tree-optimization/79740
26656 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26658 (visit_nary_op): Insert the nary into the hashtable if we
26659 pattern-matched sth.
26660 * tree-ssa-pre.c (eliminate_insert): Robustify.
26662 2017-02-28 Richard Biener <rguenther@suse.de>
26664 PR middle-end/79731
26665 * fold-const.c (decode_field_reference): Reject out-of-bound
26668 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26670 * config/i386/i386.c: Include intl.h.
26671 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26672 instead of just cond ? "..." : "...".
26673 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26674 * coverage.c (read_counts_file): Likewise.
26675 * omp-offload.c: Include intl.h.
26676 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26677 of just cond ? "..." : "...".
26678 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26679 of just cond ? "..." : "...".
26681 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26684 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26686 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26687 'tune for' CPU name.
26688 * config/arm/arm-cpu-data.h: Regenerated.
26690 2017-02-28 Richard Biener <rguenther@suse.de>
26692 PR tree-optimization/79732
26693 * tree-inline.c (expand_call_inline): Do not shadow var.
26695 2017-02-28 Richard Biener <rguenther@suse.de>
26697 PR tree-optimization/79723
26698 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26699 address-space properly.
26701 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26703 * doc/optinfo.texi (Optimization groups): Fix option used for
26705 * doc/invoke.texi (-fopt-info): Document "omp".
26706 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26707 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26708 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26709 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26710 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26712 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26714 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26715 OPTGROUP_OMP with "omp".
26717 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26720 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26721 for arithmetic shift of unsigned V2DI.
26723 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26725 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26726 arc/linux.h headers.
26727 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26728 (LINK_SPEC): Likewise.
26729 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26730 (EXTRA_SPECS): Likewise.
26731 (STARTFILE_SPEC): Likewise.
26732 (ENDFILE_SPEC): Likewise.
26733 (LIB_SPEC): Likewise.
26734 (TARGET_SDATA_DEFAULT): Likewise.
26735 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26736 (MULTILIB_DEFAULTS): Likewise.
26737 (DWARF2_UNWIND_INFO): Likewise.
26738 * config/arc/big.h: New file.
26739 * config/arc/elf.h: Likewise.
26740 * config/arc/linux.h: Likewise.
26741 * config/arc/t-uClibc: Remove.
26743 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26745 PR tree-optimization/77536
26746 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26747 (tree_transform_and_unroll_loop): Use above function to compute the
26748 estimated niter of unrolled loop and use it when scaling profile.
26749 Also use count info rather than frequency if it's non-zero.
26750 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26751 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26752 (vect_transform_loop): Call above function.
26754 2017-02-27 Richard Biener <rguenther@suse.de>
26756 PR tree-optimization/45397
26757 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26758 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26759 (visit_nary_op): Add pattern matching for CSEing sign-changed
26760 or truncated operations with wider ones.
26762 2017-02-27 Richard Biener <rguenther@suse.de>
26764 PR tree-optimization/79690
26765 * tree-vect-stmts.c (vectorizable_store): Use vector type
26766 built from the DR with address-space.
26768 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26770 * doc/invoke.texi (Optimize Options): Refine the description
26771 of asan-use-after-return.
26773 2017-02-25 Alan Modra <amodra@gmail.com>
26775 PR rtl-optimization/79584
26776 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26777 base, not ad->base_term, the reg within base. Remove assertion
26778 that ad->base == ad->base_term. Replace gen_int_mode using
26779 bogus mode with const0_rtx.
26781 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26783 PR middle-end/79396
26784 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26785 FMA_EXPR like tcc_binary or tcc_unary.
26787 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26790 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26792 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26793 (output_loc_operands): Handle DW_OP_call_ref and
26794 DW_OP_GNU_variable_value.
26795 (struct variable_value_struct): New type.
26796 (struct variable_value_hasher): Likewise.
26797 (variable_value_hash): New variable.
26798 (string_types): Remove.
26799 (copy_loc_descr): New function.
26800 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26801 (prepend_loc_descr_to_each): New function.
26802 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26803 instead of add_loc_descr_to_each if the first argument is single
26804 location list and the second has multiple.
26805 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26806 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26807 when looking for variable value which doesn't have other location info.
26808 (loc_list_from_tree): Formatting fix.
26809 (gen_array_type_die): Simplify DW_AT_string_length handling.
26810 (adjust_string_types): Remove.
26811 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26812 string_types. Call resolve_variable_values.
26813 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26814 (resolve_addr_in_expr): Likewise. Add A argument.
26815 (copy_deref_exprloc): Remove deref argument. Adjust for the
26816 original expression being DW_OP_GNU_variable_value with optionally
26817 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26818 optionally after it.
26819 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26820 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26821 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26822 (variable_value_hasher::hash, variable_value_hasher::equal): New
26824 (resolve_variable_value_in_expr, resolve_variable_value,
26825 resolve_variable_values, note_variable_value_in_expr,
26826 note_variable_value): New functions.
26827 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26830 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26833 * opts.h (handle_generated_option): Add GENERATED_P argument.
26834 * opts-common.c (handle_option): Adjust function comment.
26835 (handle_generated_option): Add GENERATED_P argument, pass it to
26837 (control_warning_option): Pass false to handle_generated_option
26839 * opts.c (maybe_default_option): Pass true to handle_generated_option
26841 * optc-gen.awk: Likewise.
26843 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26845 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26846 a REG, look at the REG it is a SUBREG of.
26847 (splitter for cmpeqsi_t): Ditto.
26849 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26851 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26852 the special USEs with the pattern of the insn, not the insn itself.
26854 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26857 * doc/invoke.texi: Document -mload-store-pairs.
26859 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26860 Sandra Loosemore <sandra@codesourcery.com>
26862 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26863 argument isn't a CONST_INT.
26864 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26865 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26866 (nios2_validate_compare): Bypass alternate compare logic if *op2
26867 is not a CONST_INT.
26868 (ldstwm_operation_p): Return false if first_base is not a REG or
26869 if first_offset is not a CONST_INT.
26871 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26873 * config/cris/cris.md: Use correct operand in a define_peephole2.
26875 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26877 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26879 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26881 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26882 this_insn if it is an INSN or JUMP_INSN.
26883 (force_offsettable): Look at base, not at addr.
26884 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26885 on things that aren't necessarily CONST_INTs.
26887 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26889 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26890 -mfpmath=sse is the default also for x86-32 targets with SSE2
26891 instruction set when @option{-ffast-math} is enabled
26893 2017-02-24 Jeff Law <law@redhat.com>
26895 PR rtl-optimizatoin/79286
26896 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26899 2017-02-24 Richard Biener <rguenther@suse.de>
26901 PR tree-optimization/79389
26902 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26905 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26907 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26908 function comment to reflect reality.
26909 (loop_exits_before_overflow): Fix typo in function description.
26911 2017-02-24 Richard Biener <rguenther@suse.de>
26913 PR tree-optimization/79389
26914 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26915 properly that a threading opportunity exists. Detect conditional
26916 copy/constant propagation opportunities.
26918 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26920 * config/visium/visium.md (type): Add trap.
26921 (b): New mode attribute.
26922 (*btst): Rename into...
26923 (*btst<mode>): ...this and adjust.
26924 (*cbranchsi4_btst_insn): Rename into...
26925 (*cbranch<mode>4_btst_insn): ...this and adjust.
26926 (trap): New define_insn.
26928 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26930 PR tree-optimization/79389
26931 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26932 (noce_reversed_cond_code): New function.
26933 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26934 reversed_comparison_code. Formatting fix.
26935 (noce_try_store_flag): Test rev_cond != NULL in addition to
26936 reversed_comparison_code.
26937 (noce_try_store_flag_constants): Likewise.
26938 (noce_try_store_flag_mask): Likewise.
26939 (noce_try_addcc): Use rev_cond if non-NULL instead of
26940 reversed_comparison_code.
26941 (noce_try_cmove_arith): Likewise. Formatting fixes.
26942 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26943 (noce_find_if_block): Initialize rev_cond.
26944 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26945 instead of false as last argument never attempt to reverse it
26948 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26950 PR tree-optimization/79663
26951 * tree-predcom.c (combine_chains): Process refs in reverse order
26952 only for ZERO length chains, and add explaining comment.
26954 2017-02-23 Jeff Law <law@redhat.com>
26956 PR tree-optimization/79578
26957 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26958 in call to operand_equal_p.
26960 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26963 * config/i386/cpuid.h: Fix another undefined behavior.
26965 2017-02-23 Richard Biener <rguenther@suse.de>
26967 PR tree-optimization/79683
26968 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26969 vector types for data-refs.
26971 2017-02-23 Martin Liska <mliska@suse.cz>
26973 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26975 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26977 PR middle-end/79665
26978 * internal-fn.c (get_range_pos_neg): Moved to ...
26979 * tree.c (get_range_pos_neg): ... here. No longer static.
26980 * tree.h (get_range_pos_neg): New prototype.
26981 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26982 are known to be in between 0 and signed maximum inclusive, try to
26983 expand both unsigned and signed divmod and use the cheaper one from
26986 2017-02-22 Jeff Law <law@redhat.com>
26988 PR tree-optimization/79578
26989 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26990 to compare base operands.
26992 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26995 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26996 gpc_reg_operand instead of fpr_reg_operand.
26998 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27000 * config/mips/mips.c (mips_return_in_memory): Force FP
27001 vector types to be returned in memory for o32 ABI.
27003 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27005 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27006 instead of DW_TAG_member for static data member declarations and don't
27007 set no_linkage_name for static inline data members.
27008 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27011 2017-02-22 Martin Liska <mliska@suse.cz>
27013 * doc/invoke.texi: Replace inequality signs with square brackets
27016 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27018 PR tree-optimization/68644
27019 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27021 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27024 * lra-constraints.c (simplify_operand_subreg): Handle
27025 WORD_REGISTER_OPERATIONS targets.
27027 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27030 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27031 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27032 elimination by swapping fld*.
27034 2017-02-22 Richard Biener <rguenther@suse.de>
27036 PR tree-optimization/79673
27037 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27038 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27039 irrelevant address-space qualifiers and avoiding a
27040 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27042 2017-02-22 Richard Biener <rguenther@suse.de>
27044 PR tree-optimization/79666
27045 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27046 to not symbolically negate if that may introduce undefined
27049 2017-02-22 Martin Liska <mliska@suse.cz>
27052 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27053 * data-streamer-out.c (streamer_write_gcov_count_stream):
27055 * value-prof.c (stream_out_histogram_value): Make assert more
27056 precise based on type of counter.
27058 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27061 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27062 Use nonimmediate_operand instead of memory_operand for operand 1.
27063 (float-extend standard_x87sse_constant_load splitter): Ditto.
27065 2017-02-21 Jeff Law <law@redhat.com>
27067 PR tree-optimization/79621
27068 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27069 blocks with edges to themselves.
27071 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27074 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27075 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27076 Use gimple_call_builtin_p.
27079 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27080 on temporarily removed DEBUG_INSNs.
27082 PR tree-optimization/79649
27083 * tree-loop-distribution.c (classify_partition): Give up on
27084 non-generic address space loads/stores.
27086 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27088 * doc/loop.texi (Loop manipulation): Remove nonexistent
27089 tree_ssa_loop_version from the documentation.
27090 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27092 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27095 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27096 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27097 * config/rs6000/rs6000.c: Include except.h.
27098 (rs6000_expand_split_stack_prologue): Call
27099 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27101 2017-02-21 Martin Jambor <mjambor@suse.cz>
27104 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27105 have been analyzed.
27107 2017-02-21 Martin Jambor <mjambor@suse.cz>
27109 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27110 for backward compatibility only.
27111 * doc/invoke.texi (Option Summary): Remove all references to
27112 -fipa-cp-alignment.
27114 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27118 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27120 * lra-constraints.c (curr_insn_transform): Handle
27121 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27123 2017-02-21 Martin Liska <mliska@suse.cz>
27125 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27127 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27129 PR translation/79638
27130 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27132 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27135 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27136 (arm_function_ok_for_sibcall): Return false for an indirect call by
27137 descriptor if all the argument registers are used.
27138 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27139 alignment of the function.
27141 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27143 PR tree-optimization/61441
27144 * simplify-rtx.c (simplify_const_unary_operation): For
27145 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27146 the sNaN unmodified.
27148 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27150 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27151 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27152 instead of SYSTEM_HEADER_DIR.
27154 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27155 Martin Liška <mliska@suse.cz>
27157 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27158 Fix typos and grammar, use active voice, and clarify.
27160 2017-02-20 Marek Polacek <polacek@redhat.com>
27162 PR middle-end/79537
27163 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27166 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27168 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27171 * config/i386/i386.c (ix86_expand_builtin): Handle
27172 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27173 ix86_builtins_isa[fcode].isa as a requirement of those
27174 flags and any other flag in the bitmask.
27175 (ix86_init_mmx_sse_builtins): Use 0 instead of
27176 ~OPTION_MASK_ISA_64BIT as mask.
27177 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27178 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27179 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27180 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27182 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27185 * lra-constraints.c (split_reg): Check requested split mode
27186 is supported by the register.
27188 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27190 * lra-constraints.c (simplify_operand_subreg): Remove early
27193 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27196 * lra-constraints.c (curr_insn_transform): Tighten condition
27197 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27199 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27202 * lra-constraints.c (curr_insn_transform): Handle
27203 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27205 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27208 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27210 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27212 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27215 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27218 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27220 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27223 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27225 * final.c (last_columnnum, override_columnnum): New variables.
27226 (final_start_function): Set last_columnnum, pass it to begin_prologue
27227 hook and pass 0 to dwarf2out_begin_prologue.
27228 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27229 to source_line debug hook.
27230 (notice_source_line): Compute last_columnnum and for debug_column_info
27231 return true on column changes.
27232 * debug.h (struct gcc_debug_hooks): Add column argument to
27233 source_line and begin_prologue hooks.
27234 (debug_nothing_int_charstar_int_bool): Remove prototype.
27235 (debug_nothing_int_int_charstar,
27236 debug_nothing_int_int_charstar_int_bool): New prototypes.
27237 (dwarf2out_begin_prologue): Add column argument.
27238 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27239 begin_prologue hooks.
27240 (debug_nothing_int_charstar_int_bool): Remove.
27241 (debug_nothing_int_int_charstar,
27242 debug_nothing_int_int_charstar_int_bool): New functions.
27243 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27244 through to dwarf2out_source_line.
27245 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27246 (dwarf2out_source_line): Add column argument, emit it if requested.
27247 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27249 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27250 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27251 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27252 through to dwarf2out_begin_prologue.
27253 (vmsdbgout_source_line): Add column argument, pass it through to
27254 dwarf2out_source_line.
27255 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27256 dbxout_source_line caller.
27257 (dbxout_source_line): Add column argument.
27259 * common.opt (gno-column-info, gcolumn-info): New options.
27260 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27261 (check_die): Also test for multiple DW_AT_decl_column attributes.
27262 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27263 DW_AT_decl_column if requested.
27264 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27266 (gen_variable_die): Likewise.
27267 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27268 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27271 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27272 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27273 (ix86_handle_option): Handle OPT_m3dnowa.
27274 * doc/invoke.texi (-m3dnowa): Document.
27275 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27276 -m3dnowa instead of -m3dnow -march=athlon.
27279 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27280 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27282 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27285 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27286 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27287 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27288 generator for vsx_xxpermdi_<mode>_be.
27289 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27290 force big-endian semantics.
27291 (vsx_xxpermdi_<mode>_be): New define_expand with same
27292 implementation as previous version of vsx_xxpermdi_<mode>.
27294 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27296 PR tree-optimization/79327
27297 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27298 variable, its initialization and use.
27300 2017-02-17 Julia Koval <julia.koval@intel.com>
27302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27303 (OPTION_MASK_ISA_PKU_UNSET): New.
27304 (ix86_handle_option): Handle -mrdpid.
27305 * config/i386/cpuid.h (bit_RDPID): New.
27306 * config/i386/driver-i386.c (host_detect_local_cpu):
27307 Detect RDPID feature.
27308 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27309 * config/i386/i386-c.c (ix86_target_macros_internal):
27311 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27312 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27313 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27314 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27315 * config/i386/i386.md (define_insn "rdpid"): New.
27316 * config/i386/i386.opt Add -mrdpid.
27317 * config/i386/immintrin.h (_rdpid_u32): New.
27319 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27321 PR rtl-optimization/79541
27322 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27323 instead of transforming it into USE.
27325 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27327 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27328 If HONOR_SNANS (SFmode) force the input to a register.
27329 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27330 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27331 an frsp or similar insn.
27333 2017-02-17 Martin Liska <mliska@suse.cz>
27335 PR rtl-optimization/79577
27336 * params.def (selsched-max-sched-times): Increase minimum to 1.
27338 2017-02-17 Martin Liska <mliska@suse.cz>
27340 PR rtl-optimization/79574
27341 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27343 2017-02-17 Martin Liska <mliska@suse.cz>
27345 PR tree-optimization/79529
27346 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27347 ssa_defined_default_def_p to handle cases which are implicitly
27349 * tree-ssa.c (ssa_defined_default_def_p): New function.
27350 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27351 which are implicitly defined.
27352 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27354 2017-02-17 Richard Biener <rguenther@suse.de>
27356 PR middle-end/79576
27357 * params.def (max-ssa-name-query-depth): Limit to 10.
27359 2017-02-17 Richard Biener <rguenther@suse.de>
27361 PR tree-optimization/79552
27362 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27365 2017-02-17 Richard Biener <rguenther@suse.de>
27368 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27370 2017-02-17 Marek Polacek <polacek@redhat.com>
27372 PR middle-end/79536
27373 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27374 (fold_negate_expr): New wrapper.
27376 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27378 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27379 Correct terminology and de-emphasize pre-standard behavior.
27381 2017-02-16 Alan Modra <amodra@gmail.com>
27383 PR rtl-optimization/79286
27384 * ira.c (def_dominates_uses): New function.
27385 (update_equiv_regs): Don't create an equivalence for insns that
27386 may trap where the register def does not dominate the use.
27388 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27390 PR rtl-optimization/78127
27391 * lra.c (lra): Call lra_eliminate before finish the loop after
27394 2017-02-16 Richard Biener <rguenther@suse.de>
27396 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27398 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27399 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27400 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27401 (isl_val_int_from_wi): New function.
27402 (extract_affine_gmp): Rename to ...
27403 (extract_affine_wi): ... this, take a widest_int.
27404 (extract_affine_int): Just wrap extract_affine_wi.
27405 (add_param_constraints): Use isl_val_int_from_wi.
27406 (add_loop_constraints): Likewise, and extract_affine_wi.
27408 2017-02-15 Jeff Law <law@redhat.com>
27410 PR middle-end/79521
27411 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27412 ira_init_register_move_cost_if_necessary.
27414 2017-02-15 Martin Sebor <msebor@redhat.com>
27416 PR middle-end/32003
27417 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27418 removed in a prior commit.
27420 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27422 PR tree-optimization/79347
27423 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27424 counters during peeling.
27426 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27428 * Makefile.in (site.exp): Remove "set ISLVER".
27430 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27433 * real.c (real_from_integer): Call real_convert even for decimal.
27435 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27438 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27440 2017-02-14 Andrew Pinski <apinski@cavium.com>
27442 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27443 cores and change the partno/implementer to be correct.
27444 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27445 the 'B" as the implementer.
27446 * config/aarch64/aarch64-tune.md: Regenerate.
27448 2017-02-14 Carl Love <cel@us.ibm.com>
27450 * config/rs6000/rs6000.c: Add case statement entry to make the
27451 xvcvuxdsp built-in argument unsigned.
27452 * config/rs6000/vsx.md: Fix the source and return operand types so they
27453 match the instruction definitions from the ISA document. Fix typo
27454 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27457 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27460 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27461 member early_clobber_alts.
27462 * lra-lives.c (reg_early_clobber_p): New.
27463 (process_bb_lives): Use it.
27464 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27465 (debug_operand_data): Initialize early_clobber_alts.
27466 (setup_operand_alternative): Set up early_clobber_alts.
27467 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27468 alternatives to new_insn_reg.
27469 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27471 (lra_update_insn_regno_info): Pass the new arg.
27473 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27475 PR middle-end/79505
27476 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27477 (new_oacc_loop_raw): Don't clear already cleared fields.
27480 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27481 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27482 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27484 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27487 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27489 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27492 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27493 the extra instruction to the right place to store 128-bit constant
27496 2017-02-14 Martin Sebor <msebor@redhat.com>
27498 PR middle-end/79448
27499 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27500 warning for strings of unknown length.
27502 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27504 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27506 2017-02-14 Jeff Law <law@redhat.com>
27509 * ira-costs.c (scan_one_insn): Initialize register move costs
27510 for pseudos seen in USE/CLOBBER insns.
27512 PR tree-optimization/79095
27513 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27514 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27515 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27516 if the operands are known to be not equal, then the resulting range
27518 (intersect_ranges): If the new range is ~[0,0] and the old range is
27519 wide, then prefer ~[0,0].
27520 * tree-vrp.c (overflow_comparison_p_1): New function.
27521 (overflow_comparison_p): New function.
27522 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27523 if NAME is used in an overflow test.
27524 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27525 overflow check that can be expressed as an equality test, then adjust
27526 ops to be that equality test.
27528 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27530 * config/s390/s390-builtin-types.def: Remove flags argument.
27531 * config/s390/s390.c (s390_init_builtins): Likewise.
27533 2017-02-14 Martin Liska <mliska@suse.cz>
27535 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27536 vector. Fix trailing white spaces.
27538 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27540 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27543 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27545 PR rtl-optimization/68664
27546 * config/arm/arm.c (arm_sched_can_speculate_insn):
27547 New function. Declare prototype.
27548 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27550 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27552 PR rtl-optimization/68664
27553 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27555 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27557 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27559 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27560 max skip bytes for function, loop and jump.
27562 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27564 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27565 ABS_EXPR for gimple dump.
27567 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27570 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27572 PR tree-optimization/79408
27573 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27574 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27575 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27576 also if rhs1 is INTEGER_CST.
27578 2017-02-14 Richard Biener <rguenther@suse.de>
27580 PR middle-end/79432
27581 * tree-into-ssa.c (insert_phi_nodes): When the function can
27582 have abnormal edges rewrite SSA names with broken use-def
27583 dominance out of SSA and register them for PHI insertion.
27585 2017-02-13 Martin Sebor <msebor@redhat.com>
27587 PR middle-end/79496
27588 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27589 clearing info.nowrite flag when snprintf size argument is a range.
27591 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27593 * cprop.c (cprop_jump): Add missing space in string literal.
27594 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27595 (get_constraint_for_component_ref): Likewise.
27596 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27597 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27598 * lra-constraints.c (process_alt_operands): Likewise.
27599 * ipa-inline.c (inline_small_functions): Likewise.
27600 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27601 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27602 * trans-mem.c (diagnose_tm_1_op): Likewise.
27603 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27604 (grid_parallel_clauses_gridifiable): Likewise.
27606 * config/nvptx/mkoffload.c (process): Add space in between
27609 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27610 "MOD4_SSE_REGS" and "ALL_REGS".
27612 * spellcheck.c (test_data): Add , in between "foo" and "food".
27614 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27617 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27618 boundary crossing check and subsequent code generation agree.
27620 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27622 * config/aarch64/aarch64.c (has_memory_op): Delete.
27623 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27626 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27628 PR rtl-optimization/79388
27629 PR rtl-optimization/79450
27630 * combine.c (distribute_notes): When removing TEM_INSN for which
27631 corresponding dest has last value recorded, invalidate that last
27634 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27636 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27637 of explicit '@'. Add missing assembly comment marker on branch costs
27640 2017-02-13 Nathan Sidwell <nathan@acm.org>
27642 * gengtype-lex.l (<in_struct>): Add '/'.
27644 2017-02-13 Martin Liska <mliska@suse.cz>
27647 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27649 2017-02-13 Richard Biener <rguenther@suse.de>
27651 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27653 * configure: Re-generate.
27654 * config.in: Likewise.
27655 * graphite-dependences.c: Simplify as if
27656 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27657 * graphite-isl-ast-to-gimple.c: Likewise.
27658 * graphite-optimize-isl.c: Likewise.
27659 * graphite-poly.c: Likewise.
27660 * graphite-sese-to-poly.c: Likewise.
27661 * graphite.h: Likewise.
27662 * toplev.c: Include isl/version.h and use isl_version () for
27663 printing the ISL version.
27664 * doc/install.texi: Update ISL requirement.
27666 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27668 * doc/standards.texi (Standards): Update reference to
27671 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27673 * doc/extend.texi (Named Address Spaces): sourceware.org now
27675 * doc/install.texi (Binaries): Ditto.
27678 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27680 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27681 standard terminology "stringize"/"stringizing" throughout.
27682 * doc/cppinternals.texi: Likewise.
27684 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27686 * doc/extend.texi: Fix some spelling mistakes and typos.
27687 * doc/invoke.texi: Likewise.
27689 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27692 * params.def (inline-min-speedup) Change from 10 to 8.
27694 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27696 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27699 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27702 * ipa-inline-analysis.c (get_minimal_bb): New function.
27703 (record_modified): Use it.
27704 (remap_edge_change_prob): Handle also ancestor functions.
27706 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27708 * doc/contrib.texi (Contributors): Remove broken link into
27709 the Mauve CVS repository.
27711 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27713 PR middle-end/79454
27714 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27715 result computation whenever lhs doesn't have vector mode, not
27716 just when it has BLKmode.
27718 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27720 * doc/makefile.texi (profiledbootstrap): Refer to the
27721 installation instructions only in textual form.
27723 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27726 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27728 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27730 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27731 (Specific): Update mingw-w64 reference.
27733 (Specific): Remove broken link to Renesas RX processor.
27735 2017-02-10 Richard Biener <rguenther@suse.de>
27737 * toplev.c (process_options): Do not mention obsolete graphite
27738 options when printing sorry message about missing graphite support.
27739 Mention -floop-nest-optimize.
27741 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27743 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27744 (vtst_p16): Likewise.
27745 (vtstq_p8): Likewise.
27746 (vtstq_p16): Likewise.
27748 (vtstq_p64): Likewise.
27749 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27750 (vset_lane_p64): New.
27751 (vsetq_lane_p64): New.
27753 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27755 PR tree-optimization/79411
27756 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27757 stmt operands are SSA_NAMEs used in abnormal phis.
27758 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27761 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27764 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27767 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27769 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27771 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27773 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27777 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27778 not arbitrary TREE_CONSTANT.
27781 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27782 "omp declare target link" attribute unless is_global_var.
27783 * omp-offload.c (find_link_var_op): Likewise.
27785 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27786 Chung-Lin Tang <cltang@codesourcery.com>
27788 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27790 (gimplify_adjust_omp_clauses): Don't delete TILE.
27791 (gimplify_omp_for): Deal with TILE.
27792 * internal-fn.c (expand_GOACC_TILE): New function.
27793 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27795 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27796 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27798 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27799 avoid DIV for outermost collapse var.
27800 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27801 Remove out of date comments, fix whitespace.
27802 * omp-general.c (omp_extract_for_data): Deal with tiling.
27803 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27804 adjust OLF_DIM_BASE value.
27805 (struct omp_for_data): Add tiling field.
27806 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27807 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27808 for auto loops. Remove default auto determining, moved to
27809 oacc_loop_fixed_partitions.
27810 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27811 stmts, add e_mask field.
27812 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27813 (oacc_thread_numbers): Use oacc_dim_call.
27814 (oacc_xform_tile): New.
27815 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27816 (finish_oacc_loop): Adjust for ifns vector.
27817 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27818 add case for GOACC_TILE fns.
27819 (oacc_loop_xform_loop): Delete.
27820 (oacc_loop_process): Iterate over call list directly, and add
27821 handling for GOACC_TILE fns.
27822 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27824 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27825 vector partitioning to outer loops. Assign 2 partitions to loops
27826 when available. Add TILE handling.
27827 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27828 (execite_oacc_device_lower): Process GOACC_TILE fns,
27829 ignore unknown specs.
27830 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27831 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27832 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27834 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27836 * configure.ac (ACX_BUGURL): Update.
27837 * configure: Regenerate.
27839 2017-02-09 Richard Biener <rguenther@suse.de>
27841 PR tree-optimization/69823
27842 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27843 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27845 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27847 * config/arc/arc-c.def: Add __NPS400__ definition.
27848 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27849 (TARGET_NPS400): Define.
27851 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27853 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27855 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27856 pointer, arch_info.
27857 (arc_cpu_types): Fill the arch_info field with a pointer into the
27858 arc_arch_types table.
27859 (arc_selected_cpu): Declare.
27860 * config/arc/arc.c (arc_selected_cpu): Make global.
27861 (arc_selected_arch): Delete.
27862 (arc_base_cpu): Delete.
27863 (arc_override_options): Remove references to deleted variables,
27864 update access to arch information.
27865 (ARC_OPT): Update access to arch information.
27866 (ARC_OPTX): Likewise.
27867 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27868 (TARGET_ARC600): Update access to arch information.
27869 (TARGET_ARC601): Likewise.
27870 (TARGET_ARC700): Likewise.
27871 (TARGET_EM): Likewise.
27872 (TARGET_HS): Likewise.
27873 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27876 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27879 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27880 condition/operands for integer GE/LE/GEU/LEU operations.
27882 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27884 PR translation/79397
27885 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27888 2017-02-08 Martin Jambor <mjambor@suse.cz>
27891 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27892 whether allocation happened.
27893 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27894 nothing was allocated.
27896 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27898 PR tree-optimization/79408
27899 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27900 constant, but SSA_NAME with a known integer range, use the minimum
27901 of that range instead of op1 to determine if modulo can be replaced
27902 with its first operand.
27904 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27906 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27908 2017-02-08 Richard Biener <rguenther@suse.de>
27910 PR tree-optimization/71824
27911 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27912 Check all loops contained in the merged region.
27914 2017-02-07 Andrew Pinski <apinski@cavium.com>
27916 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27918 2017-02-07 Andrew Pinski <apinski@cavium.com>
27920 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27921 (thunderxt88): Likewise.
27922 (thunderxt81): Disable LSE and change v8.1 to v8.
27923 (thunderxt83): Likewise.
27925 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27926 Richard Biener <rguenther@suse.de>
27928 PR middle-end/79399
27929 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27930 type from int to size_t.
27931 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27933 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27935 PR rtl-optimization/79386
27936 * cprop.c (bypass_conditional_jumps): Initialize
27937 bypass_last_basic_block already before splitting bbs after
27938 unconditional traps...
27939 (bypass_conditional_jumps): ... rather than here.
27942 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27943 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27944 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27945 fix -masm=intel patterns.
27947 2017-02-07 Richard Biener <rguenther@suse.de>
27949 PR tree-optimization/79256
27950 PR middle-end/79278
27951 * builtins.c (get_object_alignment_2): Use min_align_of_type
27952 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27953 and ADJUST_FIELD_ALIGN.
27955 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27957 * doc/tm.texi: Regenerate.
27958 * stor-layout.c (layout_decl): Adjust.
27959 (update_alignment_for_field): Likewise.
27960 (place_field): Likewise.
27961 (min_align_of_type): Likewise.
27962 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27963 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27964 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27965 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27966 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27967 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27968 * config/i386/i386.c (x86_field_alignment): Likewise.
27969 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27970 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27971 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27972 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27973 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27974 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27978 2017-01-30 Richard Biener <rguenther@suse.de>
27980 PR tree-optimization/79256
27981 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27982 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27985 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27987 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27988 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27989 builtins to SImode and emit a zero-extend, if necessary.
27991 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27993 * docs/invoke.texi (RISC-V Options): Alphabetize.
27995 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27997 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28000 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28002 * config/riscv/riscv.c: New file.
28003 * common/config/riscv/riscv-common.c: Likewise.
28004 * config.gcc: Likewise.
28005 * config/riscv/constraints.md: Likewise.
28006 * config/riscv/elf.h: Likewise.
28007 * config/riscv/generic.md: Likewise.
28008 * config/riscv/linux.h: Likewise.
28009 * config/riscv/multilib-generator: Likewise.
28010 * config/riscv/peephole.md: Likewise.
28011 * config/riscv/pic.md: Likewise.
28012 * config/riscv/predicates.md: Likewise.
28013 * config/riscv/riscv-builtins.c: Likewise.
28014 * config/riscv/riscv-c.c: Likewise.
28015 * config/riscv/riscv-ftypes.def: Likewise.
28016 * config/riscv/riscv-modes.def: Likewise.
28017 * config/riscv/riscv-opts.h: Likewise.
28018 * config/riscv/riscv-protos.h: Likewise.
28019 * config/riscv/riscv.h: Likewise.
28020 * config/riscv/riscv.md: Likewise.
28021 * config/riscv/riscv.opt: Likewise.
28022 * config/riscv/sync.md: Likewise.
28023 * config/riscv/t-elf-multilib: Likewise.
28024 * config/riscv/t-linux: Likewise.
28025 * config/riscv/t-linux-multilib: Likewise.
28026 * config/riscv/t-riscv: Likewise.
28027 * configure.ac: Likewise.
28028 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28029 Waterman as RISC-V maintainers.
28030 * doc/install.texi: Add RISC-V entries.
28031 * doc/invoke.texi: Add RISC-V options section.
28032 * doc/md.texi: Add RISC-V constraints section.
28033 * configure: Regenerated.
28035 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28038 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28039 false values to be constant vectors with all 0 or all 1 bits set.
28040 (vcondu<mode><mode>): Likewise.
28041 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28043 (fpmask_comparison_operator): Update comment.
28044 (vecint_comparison_operator): New predicate.
28045 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28046 vector conditionals when the true and false values are constant
28047 vectors with all 0 bits or all 1 bits set.
28049 2017-02-06 Martin Sebor <msebor@redhat.com>
28051 PR tree-optimization/79376
28052 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28054 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28056 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28057 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28058 to simplify split condition.
28060 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28062 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28063 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28066 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28068 PR rtl-optimization/68664
28069 * target.def (can_speculate_insn): New hook.
28070 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28071 * doc/tm.texi: Regenerate.
28072 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28073 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28074 (rs6000_sched_can_speculate_insn): New function.
28076 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28078 PR tree-optimization/79284
28079 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28080 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28081 vectorizable_mask_load_store, vectorizable_operation,
28082 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28083 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28084 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28085 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28086 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28087 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28088 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28089 is_gimple_assign (stmt). Replace another such test with
28090 is_gimple_assign (stmt).
28092 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28095 * config/avr/avr.c (rtl-iter.h): Include it.
28096 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28097 (avr_legitimate_combined_insn): ...and implementation.
28099 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28101 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28102 * config/s390/s390.c (s390_const_operand_ok)
28103 (s390_canonicalize_comparison, s390_extract_part)
28104 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28105 (s390_contiguous_bitmask_p, s390_rtx_costs)
28106 (legitimize_pic_address): Likewise.
28107 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28108 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28109 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28110 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28111 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28113 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28115 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28116 REGNO($0) == REGNO($1).
28118 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28120 * config/s390/linux.h(SIZE_TYPE): Add comment.
28122 2017-02-06 Julian Brown <julian@codesourcery.com>
28123 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28124 Virendra Pathak <virendra.pathak@broadcom.com>
28126 * config/aarch64/aarch64-cores.def: Change the scheduler
28128 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28129 * config/aarch64/thunderx2t99.md: New file.
28131 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28133 * doc/standards.texi (Go Language): Update link to language
28136 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28138 * tree-eh.c (lower_resx): Sanitize profile.
28139 (cleanup_empty_eh_move_lp): Likewise.
28141 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28144 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28146 * cfgloopmanip.h (loop_version): Update prototype.
28147 * modulo-sched.c (sms_schedule): Update call of loop_version.
28148 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28149 * tree-parloops.c (gen_parallel_loop): Likewise.
28150 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28151 * tree-ssa-loop-split.c (split_loop): Likewise.
28152 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28153 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28155 2017-02-05 Martin Liska <mliska@suse.cz>
28158 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28160 (print_operand_address): Initialize a struct to zero.
28162 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28164 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28165 garbage collector only in textual form.
28167 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28169 * doc/extend.texi (x86 specific memory model extensions for
28170 transactional memory): Simplify a phrase.
28172 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28175 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28176 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28177 (atomic_storedi_1): Likewise.
28179 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28181 PR tree-optimization/79338
28182 * tree-parloops.c (gather_scalar_reductions): Don't call
28183 vect_analyze_loop_form for loop->inner before destroying loop's
28186 2017-02-03 Martin Sebor <msebor@redhat.com>
28188 PR tree-optimization/79327
28189 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28190 when precision has resulted in leading zeros.
28191 (format_integer): Adjust the likely counter to assume an unknown
28192 argument that may be zero is non-zero.
28194 2017-02-03 Jason Merrill <jason@redhat.com>
28197 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28198 avoid copying non-taken branch.
28200 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28202 PR tree-optimization/79340
28203 * tree-vect-loop.c (vectorizable_reduction): Release
28204 vec_defs elements after safe_splicing them into other vectors.
28207 PR tree-optimization/79327
28208 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28209 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28211 (format_integer): Use wide_int_to_tree instead of build_int_cst
28212 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28213 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28214 of shortest and longest sequence.
28216 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28218 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28219 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28221 2017-02-03 Walter Lee <walt@tilera.com>
28224 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28225 after initial stackframe link reg save.
28226 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28228 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28231 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28232 wu for stxssp alternative.
28234 2017-02-03 Martin Sebor <msebor@redhat.com>
28236 PR tree-optimization/79352
28237 * gimple-fold.c (get_range_strlen): Add argument.
28238 (get_range_strlen): Change return type to bool.
28239 (get_maxval_strlen): Pass in a dummy argument.
28240 * gimple-fold.h (get_range_strlen): Change return type to bool.
28241 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28242 * tree.h (array_at_struct_end_p): Add argument.
28243 * tree.c (array_at_struct_end_p): Handle it.
28245 2017-02-03 Martin Liska <mliska@suse.cz>
28248 * multiple_target.c (create_dispatcher_calls): Redirect edge
28249 from a caller of a dispatcher.
28250 (expand_target_clones): Make the clones local.
28251 (ipa_target_clone): Do both target clones and resolvers.
28252 (ipa_dispatcher_calls): Remove the pass.
28253 (pass_dispatcher_calls::gate): Likewise.
28254 (make_pass_dispatcher_calls): Likewise.
28255 * passes.def (pass_target_clone): Put as very first IPA early
28258 2017-02-03 Martin Liska <mliska@suse.cz>
28260 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28261 in case of a function with ifunc attribute.
28263 2017-02-03 Martin Liska <mliska@suse.cz>
28265 * cgraph.c (cgraph_node::dump): Dump function version info.
28266 * symtab.c (symtab_node::dump_base): Add missing new line.
28268 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28270 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28271 (ifcombine_ifandif): Use it.
28273 2017-02-03 Martin Liska <mliska@suse.cz>
28275 * doc/invoke.texi: Document default value for
28276 use-after-scope-direct-emission-threshold.
28278 2017-02-03 Martin Liska <mliska@suse.cz>
28280 PR tree-optimization/79339
28281 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28282 (format_floating): Likewise.
28284 2017-02-03 Martin Liska <mliska@suse.cz>
28287 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28289 (ipa_node_params_t::remove): Likewise.
28290 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28291 initialization from removed ipa_node_params_t::insert.
28292 (ipa_node_params::~ipa_node_params): Move from removed
28293 ipa_node_params_t::release.
28294 * symbol-summary.h (symbol_summary::m_released): New member.
28295 Do not release a summary twice. Do not allow to call finalizer
28296 for types of a summary that live in GGC memory.
28298 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28300 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28303 2017-02-02 Martin Sebor <msebor@redhat.com>
28305 PR middle-end/79275
28306 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28307 (format_string): Tighten up the range of output for non-constant
28308 strings and correct the expected range for wide non-constant strings.
28310 2017-02-02 Martin Sebor <msebor@redhat.com>
28312 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28314 PR middle-end/32003
28315 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28317 (-fdump-tree-@var): Add to index and document how to come up
28318 with pass-specific option and dump file names.
28319 (-fdump-passes): Clarify where to look for output.
28321 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28323 PR middle-end/77445
28324 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28325 statistics of the analyzed path; allow threading for speed when
28326 any of BBs along the path are optimized for speed.
28328 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28330 PR middle-end/78468
28331 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28332 settings of the virtual registers.
28335 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28337 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28338 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28341 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28343 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28344 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28346 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28348 * config/s390/s390.md: Add missing comments with the expanded
28350 * config/s390/vector.md: Likewise.
28351 * config/s390/vx-builtins.md: Likewise.
28353 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28356 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28357 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28358 conditions on a single line.
28360 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28362 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28363 __S390_VX__ to __VX__.
28365 2017-02-01 Andrew Pinski <apinski@cavium.com>
28367 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28368 stmt_info to record_stmt_cost.
28369 (vect_get_known_peeling_cost): Pass stmt_info if known to
28371 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28372 cpu_vector_cost field into
28373 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28374 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28375 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28376 splitting of scalar_stmt_cost and vec_stmt_cost.
28377 (thunderx_vector_cost): Likewise.
28378 (cortexa57_vector_cost): LIkewise.
28379 (exynosm1_vector_cost): Likewise.
28380 (xgene1_vector_cost): Likewise.
28381 (thunderx2t99_vector_cost): Improve after the splitting of the two
28383 (aarch64_builtin_vectorization_cost): Update for the splitting of
28384 scalar_stmt_cost and vec_stmt_cost.
28386 2017-02-01 Torvald Riegel <triegel@redhat.com>
28387 Richard Henderson <rth@redhat.com>
28389 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28390 conditional on existance of a fast atomic load.
28391 * optabs-query.c (can_atomic_load_p): New function.
28392 * optabs-query.h (can_atomic_load_p): Declare it.
28393 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28394 no fast atomic load is available for the particular size of access.
28395 (expand_atomic_compare_and_swap): Likewise.
28396 (expand_atomic_load): Likewise.
28397 (expand_atomic_store): Likewise.
28398 (expand_atomic_fetch_op): Likewise.
28399 * testsuite/lib/target-supports.exp
28400 (check_effective_target_sync_int_128): Remove x86 because it provides
28401 no fast atomic load.
28402 (check_effective_target_sync_int_128_runtime): Likewise.
28404 2017-02-01 Richard Biener <rguenther@suse.de>
28406 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28407 (graphite_transform_loops): Provide opt-info for optimized nests.
28408 * tree-parloop.c (parallelize_loops): Provide opt-info for
28409 parallelized loops.
28411 2017-02-01 Richard Biener <rguenther@suse.de>
28413 PR middle-end/79315
28414 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28415 was not set before.
28417 2017-02-01 Richard Biener <rguenther@suse.de>
28419 PR tree-optimization/71824
28420 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28421 Verify the loops are valid in the merged SESE region.
28422 (scop_detection::can_represent_loop_1): Check analyzing the
28423 evolution of the number of iterations in the region succeeds.
28425 2017-01-31 Ian Lance Taylor <iant@golang.org>
28427 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28428 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28430 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28432 PR preprocessor/79210
28433 * input.c (get_substring_ranges_for_loc): Replace line_width
28434 assertion with error-handling.
28436 2017-01-31 Richard Biener <rguenther@suse.de>
28438 PR tree-optimization/77318
28439 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28440 (create_pw_aff_from_tree): Take loop parameter.
28441 (add_condition_to_pbb): Pass loop of the condition to
28442 create_pw_aff_from_tree.
28444 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28446 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28447 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28449 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28453 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28455 (convert_int_to_float128): Likewise.
28456 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28457 (convert_int_to_float128): Likewise.
28458 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28459 (UNSPEC_IEEE128_CONVERT): Likewise.
28460 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28461 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28462 Use local variables for IBM extended format.
28463 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28464 (fix_trunc<mode>si2_fprs): Likewise.
28465 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28466 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28467 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28468 to know that we can now have integers of all sizes in vector
28470 (fix<uns>_<mode>di2_hw): Likewise.
28471 (float<uns>_<mode>si2_hw): Likewise.
28472 (fix_<mode>si2_hw): Likewise.
28473 (fixuns_<mode>si2_hw): Likewise.
28474 (float<uns>_<mode>di2_hw): Likewise.
28475 (float_<mode>di2_hw): Likewise.
28476 (float_<mode>si2_hw): Likewise.
28477 (floatuns_<mode>di2_hw): Likewise.
28478 (floatuns_<mode>si2_hw): Likewise.
28479 (xscvqp<su>wz_<mode>): Delete, no longer used.
28480 (xscvqp<su>dz_<mode>): Likewise.
28481 (xscv<su>dqp_<mode>): Likewise.
28482 (ieee128_mfvsrd_64bit): Likewise.
28483 (ieee128_mfvsrd_32bit): Likewise.
28484 (ieee128_mfvsrwz): Likewise.
28485 (ieee128_mtvsrw): Likewise.
28486 (ieee128_mtvsrd_64bit): Likewise.
28487 (ieee128_mtvsrd_32bit): Likewise.
28489 2017-01-31 Martin Liska <mliska@suse.cz>
28492 * ipa-prop.c (ipa_free_all_node_params): Call release method
28493 instead of ~sumbol_summary to not to trigger double times
28496 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28498 PR tree-optimization/71691
28499 * bitmap.h (class auto_bitmap): New.
28500 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28501 is_maybe_undefined instead of ssa_undefined_value_p.
28503 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28505 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28506 __S390_ARCH_LEVEL__ to __ARCH__.
28508 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28510 PR tree-optimization/79267
28511 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28512 if should_remove_lhs_p is true.
28514 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28517 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28518 (add_alignment_attribute): New.
28519 (base_type_die): Add alignment attribute.
28520 (subrange_type_die): Likewise.
28521 (modified_type_die): Likewise.
28522 (gen_array_type_die): Likewise.
28523 (gen_descr_array_type_die: Likewise.
28524 (gen_enumeration_type_die): Likewise.
28525 (gen_subprogram_die): Likewise.
28526 (gen_variable_die): Likewise.
28527 (gen_field_die): Likewise.
28528 (gen_ptr_to_mbr_type_die): Likewise.
28529 (gen_struct_or_union_type_die): Likewise.
28530 (gen_subroutine_type_die): Likewise.
28531 (gen_typedef_die): Likewise.
28532 (base_type_cmp): Compare alignment attribute.
28534 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28537 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28538 (setb_unsigned) New pattern for setb with CCUNS.
28539 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28540 subfc./subfe sequence to avoid overflow problems. Generate a
28541 shorter sequence with cmpld/setb for power9.
28542 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28543 for generating subfc. instruction.
28544 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28545 now uses this instruction.
28547 2017-01-30 Ian Lance Taylor <iant@google.com>
28550 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28551 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28553 2017-01-30 Martin Sebor <msebor@redhat.com>
28555 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28556 Move constant to the right of a relational operator.
28557 (get_mpfr_format_length, format_character, format_string): Ditto.
28558 (should_warn_p, maybe_warn): Same.
28560 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28562 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28565 * asan.c (get_translation_unit_decl): Remove function.
28566 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28568 2017-01-30 Martin Liska <mliska@suse.cz>
28570 PR gcov-profile/79259
28571 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28572 -fprofile-generate.
28574 2017-01-30 Martin Liska <mliska@suse.cz>
28577 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28578 Initialize variables with NULL value.
28580 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28583 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28585 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28587 2017-01-30 Richard Biener <rguenther@suse.de>
28589 PR tree-optimization/79276
28590 * tree-vrp.c (process_assert_insertions): Properly adjust common
28591 when removing a duplicate.
28593 * gcc.dg/torture/pr79276.c: New testcase.
28595 2017-01-30 Richard Biener <rguenther@suse.de>
28597 PR tree-optimization/79256
28598 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28599 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28601 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28603 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28606 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28607 ("*r<noxa>sbg_<mode>_sll_bitmask")
28608 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28609 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28610 Use contiguous_bitmask_nowrap_operand.
28612 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28615 * config/rs6000/altivec.h (vec_xl): Revise #define.
28616 (vec_xst): Likewise.
28618 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28620 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28622 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28624 PR rtl-optimization/79194
28625 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28626 traps before call to bypass_conditional_jumps.
28628 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28630 PR tree-optimization/71374
28631 * lra-constraints.c (check_conflict_input_operands): New.
28632 (match_reload): Use it.
28634 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28637 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28638 account to calculate conflict_set.
28640 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28642 PR rtl-optimization/78559
28643 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28644 other_insn in combine.
28646 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28648 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28649 uint16_type_node for BT_UINT16.
28651 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28653 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28654 "RTL Tests" to menu.
28655 (GIMPLE Tests): New node.
28656 (RTL Tests): New node.
28658 2017-01-27 Richard Biener <rguenther@suse.de>
28660 PR tree-optimization/79245
28661 * tree-loop-distribution.c (distribute_loop): Apply cost
28662 modeling also to detected patterns.
28664 2017-01-27 Richard Biener <rguenther@suse.de>
28666 PR tree-optimization/71433
28667 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28668 (compare_assert_loc): New function.
28669 (process_assert_insertions): Sort and optimize assert locations
28670 to remove duplicates and push down identical assertions on
28671 edges to their destination block.
28673 2017-01-27 Richard Biener <rguenther@suse.de>
28675 PR tree-optimization/79244
28676 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28677 out SSA names even if abnormal.
28679 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28681 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28682 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28683 instead of MPFR_RNDN.
28685 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28688 * arm.c (arm_option_override): Don't call build_target_option_node
28689 until after doing all option overrides.
28690 (arm_valid_target_attribute_tree): Likewise.
28692 2017-01-27 Martin Liska <mliska@suse.cz>
28694 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28695 for {cd}tors and C++ {cd}tors.
28697 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28699 * config/s390/s390.md ("*setmem_long_and")
28700 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28702 2017-01-26 Martin Sebor <msebor@redhat.com>
28704 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28707 2017-01-26 Martin Sebor <msebor@redhat.com>
28709 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28710 HAVE_DFmode before using XFmode or DFmode.
28711 (parse_directive): Avoid using the z length modifier to avoid
28712 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28714 PR middle-end/78703
28715 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28716 to accept adjustment as an array.
28717 (get_int_range): New function.
28718 (struct directive): Make width and prec arrays.
28719 (directive::set_width, directive::set_precision): Call get_int_range.
28720 (format_integer, format_floating): Handle width and precision ranges.
28721 (format_string, parse_directive): Same.
28723 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28726 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28727 comdat_type_p set, just clone them, but keep the children in the
28731 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28732 which have direct callers with -fvar-tracking-assignments enabled
28734 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28735 inside of type units.
28737 2017-01-26 Martin Sebor <msebor@redhat.com>
28739 PR middle-end/78703
28740 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28742 (struct format_result): Replace number_chars, number_chars_min,
28743 and number_chars_max with a single member of struct result_range.
28745 (format_result::operator+=): Adjust.
28746 (struct fmtresult): Remove bounded. Handle likely and unlikely
28748 (fmtresult::adjust_for_width_or_precision): New function.
28749 (fmtresult:type_max_digits): New function.
28750 (bytes_remaining): Handle likely and unlikely counters.
28751 (min_bytes_remaining): Remove.
28752 (format_percent): Simplify.
28753 (format_integer, format_floating): Set likely and unlikely counters.
28754 (get_string_length, format_character, format_string): Same.
28755 (format_plain, should_warn_p): New function.
28756 (maybe_warn): Call should_warn_p. Update diagnostic messages
28757 and handle those for all directives, including plain strings.
28758 (format_directive): Handle likely and unlikely counters.
28759 Remove unnecessary quoting from diagnostics. Add an informational
28761 (add_bytes): Remove.
28762 (pass_sprintf_length::compute_format_length): Simplify.
28763 (try_substitute_return_value): Handle likely and unlikely counters.
28765 2017-01-26 Carl Love <cel@us.ibm.com>
28767 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28768 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28770 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28773 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28774 endianess for subregs into account.
28775 * lra-constraints.c (lra_constraints): Do risky transformations
28776 always on the first iteration.
28777 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28778 last_call_used_reg_set.
28779 (process_bb_lives): Define and use last_call_used_reg_set.
28780 * lra.c (lra): Always continue after lra_constraints on the first
28783 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28785 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28787 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28789 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28791 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28792 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28793 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28794 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28795 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28796 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28797 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28798 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28799 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28801 2017-01-26 Marek Polacek <polacek@redhat.com>
28804 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28805 for the third operand.
28807 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28809 PR middle-end/79236
28810 * omp-low.c (struct omp_context): Add simt_stmt field.
28811 (scan_omp_for): Return omp_context *.
28812 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28813 context to the _simt_ SIMD stmt.
28814 (lower_omp_for): For combined SIMD with sibling _simt_
28815 SIMD, make sure to use the same decls in _looptemp_
28816 clauses as in the sibling.
28818 2017-01-26 David Sherwood <david.sherwood@arm.com>
28820 PR middle-end/79212
28821 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28824 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28827 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28828 emit fld b; fld a; if possible.
28830 * brig-builtins.def: Update copyright years.
28831 * config/arm/arm_acle_builtins.def: Update copyright years.
28833 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28836 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28837 constraint instead of o for the stxsd instruction.
28839 2017-01-25 Carl Love <cel@us.ibm.com>
28841 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28842 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28844 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28846 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28848 2017-01-25 Richard Biener <rguenther@suse.de>
28850 PR tree-optimization/69264
28851 * target.def (vector_alignment_reachable): Improve documentation.
28852 * doc/tm.texi: Regenerate.
28853 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28855 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28856 earlier changes with respect to TYPE_USER_ALIGN.
28857 (vector_alignment_reachable_p): Likewise. Improve dumping.
28859 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28862 * config/arm/arm.md (xordi3): Force constant operand into a register
28865 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28867 * doc/invoke.texi (-fstore-merging): Correct default optimization
28868 levels at which it is enabled.
28869 (-O): Move -fstore-merging from list to...
28872 2017-01-25 Richard Biener <rguenther@suse.de>
28875 * omp-expand.c: Include debug.h.
28876 (expand_omp_taskreg): Make sure to generate early debug before
28877 outlining anything from a function.
28878 (expand_omp_target): Likewise.
28879 (grid_expand_target_grid_body): Likewise.
28881 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28884 * asan.c (get_translation_unit_decl): New function.
28885 (asan_add_global): Extract modules file name from globals
28886 TRANSLATION_UNIT_DECL name.
28888 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28891 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28892 for long calls with APCS frame and VFP.
28894 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28896 * cfg.c (original_copy_tables_initialized_p): New function.
28897 * cfg.h (original_copy_tables_initialized_p): New decl.
28898 * cfgrtl.c (relink_block_chain): Guard the call to
28899 free_original_copy_tables with a call to
28900 original_copy_tables_initialized_p.
28901 * cgraph.h (symtab_node::native_rtl_p): New decl.
28902 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28903 (symtab_node::needed_p): Don't assert for early assembly output
28904 for __RTL functions.
28905 (cgraph_node::finalize_function): Set "force_output" for __RTL
28907 (cgraph_node::analyze): Bail out early for __RTL functions.
28908 (analyze_functions): Update assertion to support __RTL functions.
28909 (cgraph_node::expand): Bail out early for __RTL functions.
28910 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28912 * function.h (struct function): Update comment for field
28914 * gimple-expr.c: Include "tree-pass.h".
28915 (gimple_has_body_p): Return false for __RTL functions.
28916 * Makefile.in (OBJS): Add run-rtl-passes.o.
28917 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28919 (gcc::pass_manager::get_clean_slate): New accessor.
28920 * passes.c: Include "insn-addr.h".
28921 (should_skip_pass_p): Add logging. Update logic for running
28922 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28923 property-provider override so it is only done for gimple passes.
28925 (skip_pass): New function.
28926 (execute_one_pass): Call skip_pass when skipping passes.
28927 * read-md.c (md_reader::read_char): Support filtering
28928 the input to a subset of line numbers.
28929 (md_reader::md_reader): Initialize fields
28930 m_first_line and m_last_line.
28931 (md_reader::read_file_fragment): New function.
28932 * read-md.h (md_reader::read_file_fragment): New decl.
28933 (md_reader::m_first_line): New field.
28934 (md_reader::m_last_line): New field.
28935 * read-rtl-function.c (function_reader::create_function): Only
28936 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28937 curr_properties. Set DECL_INITIAL to a dummy block.
28938 (read_rtl_function_body_from_file_range): New function.
28939 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28941 * run-rtl-passes.c: New file.
28942 * run-rtl-passes.h: New file.
28944 2017-01-24 Jeff Law <law@redhat.com>
28946 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28949 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28951 PR tree-optimization/79159
28952 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28953 (record_nonwrapping_iv): Improve boundary using above function if no
28954 value range information.
28956 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28957 Martin Jambor <mjambor@suse.cz>
28959 * brig-builtins.def: New file.
28960 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28961 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28962 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28963 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28964 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28965 * builtin-types.def (BT_INT8): New.
28966 (BT_INT16): Likewise.
28967 (BT_UINT8): Likewise.
28968 (BT_UINT16): Likewise.
28969 (BT_FN_ULONG): Likewise.
28970 (BT_FN_UINT_INT): Likewise.
28971 (BT_FN_UINT_ULONG): Likewise.
28972 (BT_FN_UINT_LONG): Likewise.
28973 (BT_FN_UINT_PTR): Likewise.
28974 (BT_FN_ULONG_PTR): Likewise.
28975 (BT_FN_INT8_FLOAT): Likewise.
28976 (BT_FN_INT16_FLOAT): Likewise.
28977 (BT_FN_UINT32_FLOAT): Likewise.
28978 (BT_FN_UINT16_FLOAT): Likewise.
28979 (BT_FN_UINT8_FLOAT): Likewise.
28980 (BT_FN_UINT64_FLOAT): Likewise.
28981 (BT_FN_UINT16_UINT32): Likewise.
28982 (BT_FN_UINT32_UINT16): Likewise.
28983 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28984 (BT_FN_INT_PTR_INT): Likewise.
28985 (BT_FN_UINT_PTR_UINT): Likewise.
28986 (BT_FN_LONG_PTR_LONG): Likewise.
28987 (BT_FN_ULONG_PTR_ULONG): Likewise.
28988 (BT_FN_VOID_UINT64_UINT64): Likewise.
28989 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28990 (BT_FN_INT8_INT8_INT8): Likewise.
28991 (BT_FN_INT16_INT16_INT16): Likewise.
28992 (BT_FN_INT_INT_INT): Likewise.
28993 (BT_FN_UINT_FLOAT_UINT): Likewise.
28994 (BT_FN_FLOAT_UINT_UINT): Likewise.
28995 (BT_FN_ULONG_UINT_UINT): Likewise.
28996 (BT_FN_ULONG_UINT_PTR): Likewise.
28997 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28998 (BT_FN_UINT_UINT_UINT): Likewise.
28999 (BT_FN_VOID_UINT_PTR): Likewise.
29000 (BT_FN_UINT_UINT_PTR: Likewise.
29001 (BT_FN_UINT32_UINT64_PTR): Likewise.
29002 (BT_FN_INT_INT_UINT_UINT): Likewise.
29003 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29004 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29005 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29006 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29007 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29008 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29009 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29010 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29011 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29012 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29013 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29014 * doc/frontends.texi: List BRIG FE.
29015 * doc/install.texi (Testing): Add BRIG tesring requirements.
29016 * doc/invoke.texi (Overall Options): Mention BRIG.
29017 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29019 2017-01-24 Richard Biener <rguenther@suse.de>
29021 PR translation/79208
29022 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29024 2017-01-24 Martin Jambor <mjambor@suse.cz>
29027 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29028 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29029 and known_contexts.
29031 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29033 PR middle-end/79123
29034 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29035 casts from signed to unsigned really don't have a range.
29037 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29039 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29040 GMP_RNDx for compatiblity.
29042 2017-01-24 Martin Liska <mliska@suse.cz>
29045 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29046 that would prevent us to call alloca with -1 as argument.
29048 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29050 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29051 Avoid -Wformat-security warning.
29053 2017-01-23 Andrew Pinski <apinski@cavium.com>
29055 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29058 2017-01-23 Martin Sebor <msebor@redhat.com>
29060 PR middle-end/78703
29061 * gimple-ssa-sprintf.c (warn_level): New global.
29062 (format_integer): Use it here and throughout the rest of the file.
29063 Use the same switch to compute sign as base.
29064 (maybe_warn): New function.
29065 (format_directive): Factor out warnings into maybe_warn.
29066 Add debugging output. Use warn_level.
29067 (add_bytes): Use warn_level.
29068 (pass_sprintf_length::compute_format_length): Add debugging output.
29069 (try_substitute_return_value): Same.
29070 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29072 PR middle-end/78703
29073 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29074 (struct fmtresult, format_integer, format_floating): Adjust.
29075 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29076 (get_string_length, format_string,format_directive): Same.
29077 (pass_sprintf_length::compute_format_length): Same.
29078 (try_substitute_return_value): Simplify slightly.
29080 PR middle-end/78703
29081 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29082 (fmtresult::operator+=): Outlined.
29083 (struct fmtresult): Add ctors.
29084 (struct conversion_spec): Rename...
29085 (struct directive): ...to this. Add and remove data members.
29086 (directive::set_width, directive::set_precision): New functions.
29087 (format_percent): Use fmtresult ctor.
29088 (get_width_and_precision): Remove.
29089 (format_integer): Make naming changes. Avoid computing width and
29091 (format_floating): Same. Adjust indentation.
29092 (format_character, format_none): New functions.
29093 (format_string): Moved character handling to format_character.
29094 (format_directive): Remove arguments, change return type.
29095 (parse_directive): New function.
29096 (pass_sprintf_length::compute_format_length): Move directive
29097 parsing to parse_directive.
29099 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29101 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29102 (assign_assembler_name_if_needed): ... this.
29103 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29104 (assign_assembler_name_if_needed): ... this.
29105 (free_lang_data_in_cgraph): Adjust callers.
29106 * cgraphunit.c (cgraph_node::analyze): Likewise.
29107 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29110 2017-01-23 Richard Biener <rguenther@suse.de>
29112 PR tree-optimization/79088
29113 PR tree-optimization/79188
29114 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29115 resetting loop bounds after last path deletion. Reset loop
29116 bounds of the target loop, make code match the comments.
29117 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29118 Make sure loops need no fixups.
29120 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29122 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29123 exponent support with double type for first argument.
29124 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29125 type returned by __builtin_vec_extract_sig,
29126 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29127 functions from "vector int" to "vector unsigned int" or from
29128 "vector long long int" to "vector unsigned long long int".
29129 Changed type returned by __builtin_vec_extract_exp,
29130 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29131 functions from "vector int" to "vector unsigned int" or from
29132 "vector long long int" to "vector unsigned long long int".
29133 Changed return type of __builtin_vec_test_data_class,
29134 __builtin_vec_test_data_class_sp, and
29135 __builtin_vec_test_data_class_dp from "vector int" to
29136 "vector bool int" or from "vector long long int" to "vector bool
29137 long long int" and changed second argument type from "unsigned
29138 int" to "int". Added new overloaded function forms "vector float
29139 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29140 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29141 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29142 double, vector unsigned long long int)" and "vector double
29143 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29144 long int)". Changed return type of
29145 __builtin_scalar_test_data_class and
29146 __builtin_scalar_test_data_class_sp and
29147 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29148 int" and changed second argument from "unsigned int" to "int".
29149 Changed type returned by __builtin_scalar_test_neg,
29150 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29151 from "int" to "bool int". Added new overloaded function form
29152 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29153 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29154 exponent double-precision with floating point first argument.
29155 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29156 documentation of scalar_test_data_class, scalar_test_neg,
29157 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29158 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29159 vec_test_data_class built-in functions to reflect refinements in
29160 their type signatures.
29162 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29164 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29166 (aarch64_elf_asm_destructor): Likewise.
29168 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29170 PR rtl-optimization/78634
29171 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29172 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29173 * ifcvt.c (noce_try_cmove): Add missing cost check.
29175 PR rtl-optimization/71724
29176 * combine.c (if_then_else_cond): Look for situations where it is
29177 beneficial to undo the work of one of the recursive calls.
29179 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29181 PR tree-optimization/70754
29182 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29183 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29184 combined stmt before it if not NULL.
29185 (combine_chains): Process refs reversely and compute dominance point
29188 2017-01-23 Martin Liska <mliska@suse.cz>
29190 PR tree-optimization/79196
29191 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29192 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29194 (strlen_optimize_stmt): Call the renamed function.
29196 2017-01-23 Michael Matz <matz@suse.de>
29198 PR tree-optimization/78384
29199 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29201 2017-01-23 Richard Biener <rguenther@suse.de>
29203 PR tree-optimization/79186
29204 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29205 both incoming edges before moving an assert.
29207 2017-01-23 Martin Jambor <mjambor@suse.cz>
29209 * ipa-prop.c (load_from_param_1): Removed.
29210 (load_from_unmodified_param): Bits from load_from_param_1 put back
29212 (load_from_param): Removed.
29213 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29214 with stmt. Reverted back to use of load_from_unmodified_param.
29216 2017-01-23 Martin Jambor <mjambor@suse.cz>
29219 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29220 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29221 field a pointer to garbage collected vector, mark lattices and
29222 ipcp_orig_node with GTY((skip)).
29223 (ipa_get_param_count): Adjust to descriptors being a pointer.
29224 (ipa_get_param): Likewise.
29225 (ipa_get_type): Likewise.
29226 (ipa_get_param_move_cost): Likewise.
29227 (ipa_set_param_used): Likewise.
29228 (ipa_get_controlled_uses): Likewise.
29229 (ipa_set_controlled_uses): Likewise.
29230 (ipa_is_param_used): Likewise.
29231 (ipa_node_params_t): Move into garbage collector. New methods insert
29233 (ipa_node_params_sum): Annotate wth GTY(()).
29234 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29236 (ipa_load_from_parm_agg): Adjust declaration.
29237 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29238 * ipa-profile.c (ipa_profile): Likewise.
29239 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29240 (ipa_populate_param_decls): Make descriptors parameter garbage
29242 (ipa_dump_param): Adjust to descriptors being a pointer.
29243 (ipa_alloc_node_params): Likewise.
29244 (ipa_initialize_node_params): Likewise.
29245 (load_from_param_1): Make descriptors parameter garbage collected.
29246 (load_from_unmodified_param): Likewise.
29247 (load_from_param): Likewise.
29248 (ipa_load_from_parm_agg): Likewise.
29249 (ipa_node_params::~ipa_node_params): Removed.
29250 (ipa_free_all_node_params): Remove call to delete operator.
29251 (ipa_node_params_t::insert): New.
29252 (ipa_node_params_t::remove): Likewise.
29253 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29254 copy known_csts and known_contexts vectors.
29255 (ipa_read_node_info): Adjust to descriptors being a pointer.
29256 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29258 (ipcp_transform_function): Make descriptors variable garbage
29261 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29263 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29264 * config/i386/avx512dqintrin.h: Ditto.
29265 * config/i386/avx512fintrin.h: Ditto.
29266 * config/i386/i386.c: Handle new builtins.
29267 * config/i386/i386-builtin.def: Add new builtins.
29268 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29269 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29271 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29272 Martin Liska <mliska@suse.cz>
29274 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29275 * asan.c (asan_expand_poison_ifn): Support stores and use
29276 appropriate ASAN report function.
29277 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29278 * internal-fn.def (ASAN_POISON_USE): Declare.
29279 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29280 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29281 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29282 ASAN_POISON calls w/o LHS.
29283 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29284 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29285 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29286 * gimplify.c (asan_poison_variables): Add attribute
29287 use_after_scope_memory to variables that really needs to live
29289 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29290 having the attribute.
29292 2017-01-23 Martin Liska <mliska@suse.cz>
29294 * asan.c (create_asan_shadow_var): New function.
29295 (asan_expand_poison_ifn): Likewise.
29296 * asan.h (asan_expand_poison_ifn): New declaration.
29297 * internal-fn.c (expand_ASAN_POISON): Likewise.
29298 * internal-fn.def (ASAN_POISON): New builtin.
29299 * sanopt.c (pass_sanopt::execute): Expand
29300 asan_expand_poison_ifn.
29301 * tree-inline.c (copy_decl_for_dup_finish): Make function
29303 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29304 * tree-ssa.c (is_asan_mark_p): New function.
29305 (execute_update_addresses_taken): Rewrite local variables
29306 (identified just by use-after-scope as addressable) into SSA.
29308 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29310 * doc/install.texi (Specific): opensource.apple.com uses https
29311 now. Remove trailing slash.
29313 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29315 * README.Portability: Remove note on an Irix compatibility issue.
29317 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29319 * gcov.c (INCLUDE_ALGORITHM): Define.
29320 (INCLUDE_VECTOR): Define.
29321 No longer include <vector> and <algorithm> directly.
29323 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29325 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29327 * doc/invoke.texi (Code Gen Options): Ditto.
29329 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29332 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29334 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29336 rtl-optimization/79125
29337 * cprop.c (local_cprop_pass): Handle cases where we make an
29338 unconditional trap.
29340 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29344 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29345 read from, for big endian.
29347 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29349 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29350 register pauth builtins for LP64 only.
29352 2017-01-20 Marek Polacek <polacek@redhat.com>
29355 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29358 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29360 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29362 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29363 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29364 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29366 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29369 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29370 in big-endian mode when they are not a single duplicated value.
29372 2017-01-20 Richard Biener <rguenther@suse.de>
29374 * BASE-VER: Bump to 7.0.1.
29376 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29378 * omp-low.c (omplow_simd_context): New struct. Use it...
29379 (lower_rec_simd_input_clauses): ...here and...
29380 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29381 references to idx, lane, max_vf, is_simt.
29383 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29385 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29388 2017-01-20 Martin Jambor <mjambor@suse.cz>
29390 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29391 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29393 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29394 (GTFILES): Rename hsa.c to hsa-common.c.
29395 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29396 * hsa-dump.c: Likewise.
29397 * hsa-gen.c: Likewise.
29398 * hsa-regalloc.c: Likewise.
29399 * ipa-hsa.c: Likewise.
29400 * omp-expand.c: Likewise.
29401 * omp-low.c: Likewise.
29402 * toplev.c: Likewise.
29404 2017-01-20 Marek Polacek <polacek@redhat.com>
29407 * doc/invoke.texi: Document -Wduplicated-branches.
29408 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29409 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29410 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29411 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29412 return 0 only when not OEP_LEXICOGRAPHIC.
29413 (fold_build_cleanup_point_expr): Use the expression
29414 location when building CLEANUP_POINT_EXPR.
29415 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29416 * tree.c (add_expr): Handle error_mark_node.
29418 2017-01-20 Martin Liska <mliska@suse.cz>
29421 * tree-profile.c (init_ic_make_global_vars): Do not call
29423 (gimple_init_gcov_profiler): Likewise.
29425 2017-01-20 Martin Liska <mliska@suse.cz>
29428 * cgraph.h (maybe_create_reference): Remove argument and
29430 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29432 * ipa-cp.c (create_specialized_node): Likewise.
29433 * symtab.c (symtab_node::maybe_create_reference): Handle
29434 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29436 2017-01-20 Martin Liska <mliska@suse.cz>
29438 * read-rtl-function.c (function_reader::create_function): Use
29439 build_decl instread of build_decl_stat.
29441 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29443 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29444 * config/i386/avx512dqintrin.h: Ditto.
29445 * config/i386/avx512fintrin.h: Ditto.
29446 * config/i386/i386-builtin-types.def: Add new types.
29447 * config/i386/i386.c: Handle new types.
29448 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29449 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29450 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29451 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29452 (__builtin_ia32_kshiftridi): New.
29453 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29455 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29459 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29460 define to rs6000_init_stack_protect_guard.
29461 (rs6000_init_stack_protect_guard): New function.
29463 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29464 Yunqiang Su <yunqiang.su@imgtec.com>
29466 * config.gcc (supported_defaults): Add madd4.
29467 (with_madd4): Add validation.
29468 (all_defaults): Add madd4.
29469 * config/mips/mips.opt (mmadd4): New option.
29470 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29472 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29474 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29475 (ISA_HAS_FUSED_MADD4): Likewise.
29476 * doc/invoke.texi (-mmadd4): Document the new option.
29477 * doc/install.texi (--with-madd4): Document the new option.
29479 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29481 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29482 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29483 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29484 (aarch64_init_pauth_hint_builtins): New.
29485 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29486 (aarch64_expand_builtin): Expand new builtins.
29488 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29490 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29491 * combine-stack-adj.c (no_unhandled_cfa): Handle
29492 REG_CFA_TOGGLE_RA_MANGLE.
29493 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29494 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29495 info for return address signing.
29496 (aarch64_expand_epilogue): Likewise.
29498 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29500 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29501 * config/aarch64/aarch64-protos.h
29502 (aarch64_return_address_signing_enabled): New declaration.
29503 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29505 (aarch64_expand_prologue): Sign return address before it's pushed onto
29507 (aarch64_expand_epilogue): Authenticate return address fetched from
29509 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29510 (aarch64_attributes): New function attributes for "sign-return-address".
29511 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29512 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29513 ("*do_return"): Generate combined instructions according to key index.
29514 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29515 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29517 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29518 * config/aarch64/aarch64.opt (msign-return-address=): New.
29519 * doc/extend.texi (AArch64 Function Attributes): Documents
29520 "sign-return-address=".
29521 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29523 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29525 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29526 overall option summary.
29528 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29530 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29531 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29532 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29533 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29535 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29537 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29538 -mpower9-minmax by default for -mcpu=power9.
29539 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29540 128-bit floating point.
29542 2017-01-20 Alan Modra <amodra@gmail.com>
29544 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29545 optimizing for size.
29547 2017-01-20 Alan Modra <amodra@gmail.com>
29550 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29551 for strcmp and strncmp from corresponding builtin decl.
29553 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29555 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29556 instead of i386/rtems-64.h.
29557 * config/i386/rtems-64.h: Remove.
29559 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29563 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29565 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29567 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29569 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29570 Change int to HOST_WIDE_INT.
29571 * config/aarch64/aarch64-protos.h
29572 (aarch64_simd_gen_const_vector_dup): Likewise.
29573 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29575 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29577 * langhooks-def.h (lhd_type_for_size): New decl.
29578 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29579 * langhooks.c (lhd_type_for_size): New function, taken from
29582 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29584 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29585 define_bypass for CR latency.
29586 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29587 (power9-alu2): Add define_bypass for CR latency.
29589 (power9-mul): Update insn latency.
29590 (power9-mul-compare): Update insn latency, bypass latency and remove
29593 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29595 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29597 * config/aarch64/aarch64.md
29598 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29599 aarch64_nopcrelative_literal_loads.
29600 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29602 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29604 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29605 TARGET_LOONGSON_3A.
29606 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29608 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29611 * config.gcc (supported_defaults): Add lxc1-sxc1.
29612 (with_lxc1_sxc1): Add validation.
29613 (all_defaults): Add lxc1-sxc1.
29614 * config/mips/mips.opt (mlxc1-sxc1): New option.
29615 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29617 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29618 __mips_no_lxc1_sxc1.
29619 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29620 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29621 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29623 2017-01-19 Richard Biener <rguenther@suse.de>
29625 PR tree-optimization/72488
29626 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29627 sure to restore SSA info.
29628 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29630 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29632 PR rtl-optimization/79121
29633 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29634 of the inner type when shifting an extended value.
29636 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29639 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29640 interposable aliases.
29642 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29645 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29646 Use the evmergelohi instruction.
29647 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29648 (mov_sitf_e500_subreg8_2_be): Likewise.
29649 (mov_sitf_e500_subreg12_2_le): Likewise.
29650 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29651 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29652 (mov_sitf_e500_subreg8_2_le): Likewise.
29653 (mov_sitf_e500_subreg12_2_be): Likewise.
29655 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29657 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29658 attribute from vecsimple to vecperm.
29659 (altivec_vbpermq2): Likewise.
29661 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29664 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29666 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29667 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29668 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29669 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29670 case where N arg is SIZE_MAX.
29671 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29672 (cmpstrsi): Add pattern.
29674 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29677 __builtin_vec_revb builtins.
29678 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29679 built-in functions to support generation of the ISA 3.0 XXBR<x>
29680 vector byte reverse instructions.
29681 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29682 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29683 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29684 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29685 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29686 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29687 (P9V_BUILTIN_VEC_REVB): Likewise.
29688 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29689 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29690 (p9_xxbrq_v16qi): Likewise.
29691 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29692 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29693 (p9_xxbrh_v8hi): Likewise.
29694 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29695 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29696 vec_revb built-in functions.
29698 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29700 PR rtl-optimization/78952
29701 * config/i386/i386.md (any_extract): New code iterator.
29702 (*insvqi_2): Use any_extract for source operand.
29703 (*insvqi_3): Use any_shiftrt for source operand.
29705 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29707 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29709 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29711 2017-01-18 Matthias Klose <doko@ubuntu.com>
29713 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29715 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29717 * config/rs6000/altivec.h (vec_bperm): Change #define.
29718 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29719 (altivec_vbpermq2): New define_insn.
29720 (altivec_vbpermd): Likewise.
29721 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29722 function interface.
29723 (VBPERMD): Likewise.
29724 (VBPERM): New polymorphic function interface.
29725 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29726 Add entries for P9V_BUILTIN_VEC_VBPERM.
29727 * doc/extend.texi: Add interfaces for vec_bperm.
29729 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29731 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29732 first letter of error messages.
29733 (s390_resolve_overloaded_builtin): Likewise.
29734 * config/s390/s390.c (s390_expand_builtin): Likewise.
29735 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29736 (s390_valid_target_attribute_inner_p): Likewise.
29737 * config/s390/s390.md ("tabort"): Likewise.
29739 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29741 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29742 (ISA_AVOID_DIV_HILO): New macro.
29743 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29744 (ISA_HAS_DDIV): Likewise.
29746 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29748 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29750 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29752 * doc/invoke.texi (fabi-version): Spelling fix.
29754 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29757 * doc/invoke.texi (fabi-version): Mention mangling fix for
29760 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29763 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29765 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29768 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29769 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29771 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29773 (stack_protect_test): Ditto.
29774 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29775 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29777 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29778 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29779 -mstack-protector-guard-offset=.
29780 (RS/6000 and PowerPC Options): Ditto.
29782 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29784 * config/i386/i386.h (MASK_CLASS_P): New define.
29785 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29786 there are no registers from different register sets also when
29787 mask registers are used. Update function comment.
29788 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29789 to (*k/*r) and (*k/*km) alternatives.
29791 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29793 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29794 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29795 (EH_RETURN_HANDLER_RTX): New define.
29796 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29797 Force frame pointer in EH return functions.
29798 (aarch64_expand_epilogue): Add barrier for eh_return.
29799 (aarch64_final_eh_return_addr): Remove.
29800 (aarch64_eh_return_handler_rtx): New function.
29801 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29803 (aarch64_eh_return_handler_rtx): New prototype.
29805 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29807 * config/rs6000/altivec.h (vec_rlmi): New #define.
29808 (vec_vrlnm): Likewise.
29809 (vec_rlnm): Likewise.
29810 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29811 (UNSPEC_VRLNM): Likewise.
29812 (VIlong): New mode iterator.
29813 (altivec_vrl<VI_char>mi): New define_insn.
29814 (altivec_vrl<VI_char>nm): Likewise.
29815 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29817 (VRLDNM): Likewise.
29818 (RLNM): New polymorphic function entry.
29819 (VRLWMI): New monomorphic function entry.
29820 (VRLDMI): Likewise.
29821 (RLMI): New polymorphic function entry.
29822 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29823 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29824 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29827 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29830 * dwarf2out.c (field_byte_offset): Restore the
29831 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29832 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29833 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29836 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29839 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29841 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29844 * dwarf2out.c (add_data_member_location_attribute): For constant
29845 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29846 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29847 DW_AT_byte_size attributes.
29849 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29851 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29852 after forcing to constant memory when the code model is medium.
29854 2017-01-17 Julia Koval <julia.koval@intel.com>
29857 * config/i386/avx512fintrin.h
29858 (_mm512_i32gather_ps): Change __addr type to void const*.
29859 (_mm512_mask_i32gather_ps): Ditto.
29860 (_mm512_i32gather_pd): Ditto.
29861 (_mm512_mask_i32gather_pd): Ditto.
29862 (_mm512_i64gather_ps): Ditto.
29863 (_mm512_mask_i64gather_ps): Ditto.
29864 (_mm512_i64gather_pd): Ditto.
29865 (_mm512_mask_i64gather_pd): Ditto.
29866 (_mm512_i32gather_epi32): Ditto.
29867 (_mm512_mask_i32gather_epi32): Ditto.
29868 (_mm512_i32gather_epi64): Ditto.
29869 (_mm512_mask_i32gather_epi64): Ditto.
29870 (_mm512_i64gather_epi32): Ditto.
29871 (_mm512_mask_i64gather_epi32): Ditto.
29872 (_mm512_i64gather_epi64): Ditto.
29873 (_mm512_mask_i64gather_epi64): Ditto.
29874 (_mm512_i32scatter_ps): Change __addr type to void*.
29875 (_mm512_mask_i32scatter_ps): Ditto.
29876 (_mm512_i32scatter_pd): Ditto.
29877 (_mm512_mask_i32scatter_pd): Ditto.
29878 (_mm512_i64scatter_ps): Ditto.
29879 (_mm512_mask_i64scatter_ps): Ditto.
29880 (_mm512_i64scatter_pd): Ditto.
29881 (_mm512_mask_i64scatter_pd): Ditto.
29882 (_mm512_i32scatter_epi32): Ditto.
29883 (_mm512_mask_i32scatter_epi32): Ditto.
29884 (_mm512_i32scatter_epi64): Ditto.
29885 (_mm512_mask_i32scatter_epi64): Ditto.
29886 (_mm512_i64scatter_epi32): Ditto.
29887 (_mm512_mask_i64scatter_epi32): Ditto.
29888 (_mm512_i64scatter_epi64): Ditto.
29889 (_mm512_mask_i64scatter_epi64): Ditto.
29890 * config/i386/avx512pfintrin.h
29891 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29892 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29893 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29894 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29895 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29896 (_mm512_prefetch_i32scatter_ps): Ditto.
29897 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29898 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29899 (_mm512_prefetch_i64scatter_pd): Ditto.
29900 (_mm512_prefetch_i64scatter_ps): Ditto.
29901 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29902 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29903 * config/i386/avx512vlintrin.h
29904 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29905 (_mm_mmask_i32gather_ps): Ditto.
29906 (_mm256_mmask_i32gather_pd): Ditto.
29907 (_mm_mmask_i32gather_pd): Ditto.
29908 (_mm256_mmask_i64gather_ps): Ditto.
29909 (_mm_mmask_i64gather_ps): Ditto.
29910 (_mm256_mmask_i64gather_pd): Ditto.
29911 (_mm_mmask_i64gather_pd): Ditto.
29912 (_mm256_mmask_i32gather_epi32): Ditto.
29913 (_mm_mmask_i32gather_epi32): Ditto.
29914 (_mm256_mmask_i32gather_epi64): Ditto.
29915 (_mm_mmask_i32gather_epi64): Ditto.
29916 (_mm256_mmask_i64gather_epi32): Ditto.
29917 (_mm_mmask_i64gather_epi32): Ditto.
29918 (_mm256_mmask_i64gather_epi64): Ditto.
29919 (_mm_mmask_i64gather_epi64): Ditto.
29920 (_mm256_i32scatter_ps): Change __addr type to void*.
29921 (_mm256_mask_i32scatter_ps): Ditto.
29922 (_mm_i32scatter_ps): Ditto.
29923 (_mm_mask_i32scatter_ps): Ditto.
29924 (_mm256_i32scatter_pd): Ditto.
29925 (_mm256_mask_i32scatter_pd): Ditto.
29926 (_mm_i32scatter_pd): Ditto.
29927 (_mm_mask_i32scatter_pd): Ditto.
29928 (_mm256_i64scatter_ps): Ditto.
29929 (_mm256_mask_i64scatter_ps): Ditto.
29930 (_mm_i64scatter_ps): Ditto.
29931 (_mm_mask_i64scatter_ps): Ditto.
29932 (_mm256_i64scatter_pd): Ditto.
29933 (_mm256_mask_i64scatter_pd): Ditto.
29934 (_mm_i64scatter_pd): Ditto.
29935 (_mm_mask_i64scatter_pd): Ditto.
29936 (_mm256_i32scatter_epi32): Ditto.
29937 (_mm256_mask_i32scatter_epi32): Ditto.
29938 (_mm_i32scatter_epi32): Ditto.
29939 (_mm_mask_i32scatter_epi32): Ditto.
29940 (_mm256_i32scatter_epi64): Ditto.
29941 (_mm256_mask_i32scatter_epi64): Ditto.
29942 (_mm_i32scatter_epi64): Ditto.
29943 (_mm_mask_i32scatter_epi64): Ditto.
29944 (_mm256_i64scatter_epi32): Ditto.
29945 (_mm256_mask_i64scatter_epi32): Ditto.
29946 (_mm_i64scatter_epi32): Ditto.
29947 (_mm_mask_i64scatter_epi32): Ditto.
29948 (_mm256_i64scatter_epi64): Ditto.
29949 (_mm256_mask_i64scatter_epi64): Ditto.
29950 (_mm_i64scatter_epi64): Ditto.
29951 (_mm_mask_i64scatter_epi64): Ditto.
29952 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29953 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29954 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29955 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29956 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29957 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29958 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29959 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29960 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29961 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29962 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29963 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29964 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29965 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29966 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29967 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29968 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29969 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29970 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29971 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29972 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29973 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29974 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29975 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29976 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29977 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29978 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29979 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29980 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29981 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29982 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29983 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29984 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29985 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29986 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29987 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29988 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29989 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29990 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29991 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29992 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29993 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29994 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29995 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29996 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29997 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29998 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29999 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30000 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30001 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30002 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30003 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30004 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30005 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30006 definitions accordingly.
30008 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30009 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30012 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30015 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30018 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30019 hard regs for inner regno.
30021 2017-01-17 Martin Liska <mliska@suse.cz>
30024 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30025 assumption and add comment.
30027 2017-01-17 Nathan Sidwell <nathan@acm.org>
30029 * ipa-visibility.c (localize_node): New function, broken out of ...
30030 (function_and_variable_visibility): ... here. Call it.
30032 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30034 PR middle-end/77445
30035 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30036 correctly set frequency of oudgoing edge.
30037 (duplicate_thread_path): Fix profile updating.
30039 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30042 * configure.ac: Add GCC_BASE_VER.
30043 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30044 version from BASE-VER file.
30045 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30046 (gcc.o): Depend on $(BASEVER).
30047 * common.opt (dumpfullversion): New option.
30048 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30049 * doc/invoke.texi: Document -dumpfullversion.
30050 * doc/install.texi: Document --with-gcc-major-version-only.
30051 * configure: Regenerated.
30053 2017-01-17 Richard Biener <rguenther@suse.de>
30055 PR tree-optimization/71433
30056 * tree-vrp.c (register_new_assert_for): Merge same asserts
30057 on all incoming edges.
30058 (process_assert_insertions_for): Handle insertions at the
30061 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30063 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30064 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30066 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30069 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30072 2017-01-17 Alan Modra <amodra@gmail.com>
30075 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30076 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30077 symbolic stack limit when pic.
30079 2017-01-16 Martin Sebor <msebor@redhat.com>
30081 PR tree-optimization/78608
30082 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30084 2017-01-16 Jeff Law <law@redhat.com>
30087 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30088 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30089 for several include directories that may be relative to sysroot.
30090 * config/i386/x-mingw32 (gplus_includedir): Define.
30091 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30092 (native_system_includedir): Likewise.
30093 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30094 override if TARGET_SYSTEM_ROOT is defined.
30095 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30097 PR tree-optimization/79090
30098 PR tree-optimization/33562
30099 PR tree-optimization/61912
30100 PR tree-optimization/77485
30101 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30102 and computed trims into the dump file.
30104 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30106 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30108 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30111 * gimplify.c (gimplify_init_constructor): If want_value and
30112 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30116 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30117 sequence. Formatting fixes.
30118 (doloop_optimize): Formatting fixes.
30121 * gcc.c (debug_level_greater_than_spec_func): New function.
30122 (static_spec_functions): Add debug-level-gt spec function.
30123 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30125 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30126 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30127 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30128 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30129 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30130 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30132 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30134 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30135 QImode fixups to general and mask registers only.
30137 2017-01-16 Carl Love <cel@us.ibm.com>
30139 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30140 for built-in functions
30141 vector signed char vec_nabs (vector signed char)
30142 vector signed short vec_nabs (vector signed short)
30143 vector signed int vec_nabs (vector signed int)
30144 vector signed long long vec_nabs (vector signed long long)
30145 vector float vec_nabs (vector float)
30146 vector double vec_nabs (vector double)
30147 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30149 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30150 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30151 * doc/extend.texi: Update the documentation file for the new built-in
30154 2017-01-16 Martin Sebor <msebor@redhat.com>
30156 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30159 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30161 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30162 UNSPEC_VSX__XXSPLTD to require special splat handling.
30164 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30167 * system.h: Poison strndup.
30169 2017-01-16 Alan Modra <amodra@gmail.com>
30172 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30175 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30177 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30179 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30181 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30182 call recog here. Assert that INSN_CODE (insn) is non-negative.
30184 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30187 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30189 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30190 in the currently scheduled RTL fragment.
30192 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30194 PR rtl-optimization/78751
30195 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30198 2017-01-14 Jeff Law <law@redhat.com>
30200 PR tree-optimization/79090
30201 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30202 variable length stores.
30203 (compute_trims): Delete dead assignment to *trim_tail.
30204 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30207 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30209 PR rtl-optimization/78626
30210 PR rtl-optimization/78727
30211 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30212 of a block, and split such blocks after everything else is finished.
30214 2017-01-14 Alan Modra <amodra@gmail.com>
30217 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30218 target legitimate_combined_insn.
30219 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30220 (rs6000_legitimate_combined_insn): New function.
30221 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30223 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30224 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30225 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30227 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30229 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30231 2017-01-13 Jeff Law <law@redhat.com>
30233 PR tree-optimization/33562
30234 PR tree-optimization/61912
30235 PR tree-optimization/77485
30236 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30238 (delete_dead_assignment): Likewise.
30239 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30240 statement to delete_dead_call and delete_dead_assignment.
30242 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30245 * substring-locations.c (format_warning_va): Strengthen case 1 so
30246 that both endpoints of the substring must be within the format
30247 range for just the substring to be printed.
30249 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30251 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30252 * config/i386/i386.c (ix86_target_string): Add missing options
30253 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30254 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30255 flags_other and ix86_target_other to flags2_other. Display unknown
30257 (ix86_valid_target_attribute_inner_p): Add missing options and
30258 reorder options by implied ISAs, as in ix86_target_string.
30260 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30262 * hash-table.h (hash_table::too_empty_p): New function.
30263 (hash_table::expand): Use it.
30264 (hash_table::traverse): Likewise.
30265 (hash_table::empty_slot): Use sizeof (value_type) instead of
30266 sizeof (PTR) to convert bytes to elements. Shrink the table
30267 if the current size is excessive for the current number of
30270 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30272 * ira-costs.c (record_reg_classes): Break from the inner loop
30273 early once alt_fail is known to be true. Update outer loop
30274 handling accordingly.
30276 2017-01-13 Jeff Law <law@redhat.com>
30278 * tree-ssa-dse.c (decrement_count): New function.
30279 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30280 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30281 when we know the partially dead statement is a mem* function.
30283 PR tree-optimization/61912
30284 PR tree-optimization/77485
30285 * tree-ssa-dse.c: Include expr.h.
30286 (maybe_trim_constructor_store): New function.
30287 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30289 PR tree-optimization/33562
30290 PR tree-optimization/61912
30291 PR tree-optimization/77485
30292 * doc/invoke.texi: Document new dse-max-object-size param.
30293 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30294 * tree-ssa-dse.c: Include params.h.
30295 (dse_store_status): New enum.
30296 (initialize_ao_ref_for_dse): New, partially extracted from
30298 (valid_ao_ref_for_dse, normalize_ref): New.
30299 (setup_live_bytes_from_ref, compute_trims): Likewise.
30300 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30301 (maybe_trim_partially_dead_store): Likewise.
30302 (maybe_trim_complex_store): Likewise.
30303 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30304 Track what bytes live from the original store. Return tri-state
30305 for dead, partially dead or live.
30306 (dse_dom_walker): Add constructor, destructor and new private members.
30307 (delete_dead_call, delete_dead_assignment): New extracted from
30309 (dse_optimize_stmt): Make a member of dse_dom_walker.
30310 Use initialize_ao_ref_for_dse.
30312 PR tree-optimization/33562
30313 PR tree-optimization/61912
30314 PR tree-optimization/77485
30315 * sbitmap.h (bitmap_count_bits): Prototype.
30316 (bitmap_clear_range, bitmap_set_range): Likewise.
30317 * sbitmap.c (bitmap_clear_range): New function.
30318 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30320 2017-01-13 Martin Liska <mliska@suse.cz>
30323 * function.c (set_cfun): Add new argument force.
30324 * function.h (set_cfun): Likewise.
30325 * ipa-inline-transform.c (inline_call): Use the function when
30326 strict alising from is dropped for function we inline to.
30328 2017-01-13 Richard Biener <rguenther@suse.de>
30330 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30331 for dumping GIMPLE INTEGER_CSTs.
30333 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30335 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30336 to 201112L since C++17.
30338 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30341 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30342 if -fsanitize=kernel-address is present.
30344 2017-01-13 Richard Biener <rguenther@suse.de>
30346 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30347 as _Literal ( type ) number in case usual suffixes do not
30348 preserve all information.
30350 2017-01-13 Richard Biener <rguenther@suse.de>
30352 PR tree-optimization/77283
30353 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30354 and ssa-iterators.h.
30355 (is_feasible_trace): Implement a cost model based on joiner
30358 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30361 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30362 char or short to __float128/_Float128 directly.
30364 2017-01-12 Martin Sebor <msebor@redhat.com>
30366 to -Wformat-overflow.
30367 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30368 (min_bytes_remaining): Same.
30369 (get_string_length): Same.
30370 (format_string): Same.
30371 (format_directive): Same.
30373 (pass_sprintf_length::handle_gimple_call): Same.
30375 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30377 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30378 info.nowrite calls with no lhs that can't throw. Return bool
30379 whether gsi_remove has been called or not.
30380 (pass_sprintf_length::handle_gimple_call): Return bool whether
30381 try_substitute_return_value called gsi_remove. Formatting fix.
30382 (pass_sprintf_length::execute): Don't use gsi_remove if
30383 handle_gimple_call returned true.
30386 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30387 be removed due to side-effects, don't remove following barrier nor
30388 turn the successor edge into fallthru edge.
30390 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30393 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30394 element-reversing loads and stores as not swappable.
30396 2017-01-12 Nathan Sidwell <nathan@acm.org>
30397 Nicolai Stange <nicstange@gmail.com>
30399 * combine.c (try_combine): Don't ignore result of overlap checking
30400 loop. Combine overlap & asm check into single loop.
30402 2017-01-12 Richard Biener <rguenther@suse.de>
30404 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30405 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30407 2017-01-12 Richard Biener <rguenther@suse.de>
30409 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30410 and TS_TARGET_OPTION directly derive from TS_BASE.
30411 * tree-core.h (tree_optimization_option): Derive from tree_base.
30412 (tree_target_option): Likewise.
30414 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30416 * config/i386/i386.c (memory_address_length): Increase len
30417 only when rip_relative_addr_p returns false.
30419 2017-01-11 Julia Koval <julia.koval@intel.com>
30421 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30422 (OPTION_MASK_ISA_SGX_SET): New.
30423 (ix86_handle_option): Handle OPT_msgx.
30424 * config.gcc: Added sgxintrin.h.
30425 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30426 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30427 * config/i386/i386.c (ix86_target_string): Add -msgx.
30429 (ix86_option_override_internal): Handle new options.
30430 (ix86_valid_target_attribute_inner_p): Add sgx.
30431 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30432 * config/i386/i386.opt: Add msgx.
30433 * config/i386/sgxintrin.h: New file.
30434 * config/i386/x86intrin.h: Add sgxintrin.h.
30436 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30439 * fold-const.c (maybe_nonzero_address): Return 1 for function
30441 (tree_single_nonzero_warnv_p): Don't handle function local objects
30445 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30448 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30451 * opts.c: Include "spellcheck.h"
30452 (struct string_fragment): New struct.
30453 (struct edit_distance_traits<const string_fragment &>): New
30455 (get_closest_sanitizer_option): New function.
30456 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30458 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30460 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30462 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30463 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30464 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30465 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30466 for initial die_offset if dwarf_split_debug_info.
30467 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30468 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30470 (output_skeleton_debug_sections): Formatting fix. Use
30471 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30472 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30474 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30476 * config/arm/cortex-a53.md: Add bypasses for
30477 cortex_a53_r2f_cvt.
30478 (cortex_a53_r2f): Only use for transfers.
30479 (cortex_a53_f2r): Likewise.
30480 (cortex_a53_r2f_cvt): Add reservation for conversions.
30481 (cortex_a53_f2r_cvt): Likewise.
30483 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30485 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30486 to all inlined functions, change static to extern.
30488 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30491 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30493 (arm_assemble_integer): Likewise.
30495 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30497 * config.gcc: Use new awk script to check CPU, FPU and architecture
30498 parameters for --with-... options.
30499 * config/arm/parsecpu.awk: New file
30500 * config/arm/arm-cpus.in: New file.
30501 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30503 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30505 * config/arm/t-arm: Update dependency rules.
30506 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30507 of processing .def files.
30508 * config/arm/genopt.sh: Deleted.
30509 * config/arm/gentune.sh: Deleted.
30510 * config/arm/arm-cores.def: Deleted.
30511 * config/arm/arm-arches.def: Deleted.
30512 * config/arm/arm-fpus.def: Deleted.
30513 * config/arm/arm-tune.md: Regenerated.
30514 * config/arm/arm-tables.opt: Regenerated.
30515 * config/arm/arm-cpu.h: New generated file.
30516 * config/arm/arm-cpu-data.h: New generated file.
30517 * config/arm/arm-cpu-cdata.h: New generated file.
30519 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30522 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30524 (input_varpool_node): Unpack dynamically_initialized bit.
30526 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30528 PR rtl-optimization/79032
30529 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30530 the alignment of the adjusted memory reference against that of MODE,
30531 instead of the alignment of the original memory reference.
30533 2017-01-11 Martin Jambor <mjambor@suse.cz>
30535 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30537 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30538 decorated functions.
30540 2017-01-11 Richard Biener <rguenther@suse.de>
30542 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30543 set range/nonnull info for PHI results. Do not set it on
30544 stmts marked for removal.
30546 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30548 * expr.c (store_field): In the bitfield case, fetch the return value
30549 from the registers before applying a single big-endian adjustment.
30550 Always do a final load for a BLKmode value not larger than a word.
30552 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30555 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30556 that we correctly handle column numbers greater than
30557 LINE_MAP_MAX_COLUMN_NUMBER.
30559 2017-01-10 Martin Sebor <msebor@redhat.com>
30561 PR middle-end/78245
30562 * gimple-ssa-sprintf.c (get_destination_size): Call
30563 {init,fini}object_sizes.
30564 * tree-object-size.c (addr_object_size): Adjust.
30565 (pass_through_call): Adjust.
30566 (pass_object_sizes::execute): Adjust.
30567 * tree-object-size.h (fini_object_sizes): Declare.
30569 2017-01-10 Martin Sebor <msebor@redhat.com>
30571 PR tree-optimization/78775
30572 * builtins.c (get_size_range): Move...
30573 * calls.c: ...to here.
30574 (alloc_max_size): Accept zero argument.
30575 (operand_signed_p): Remove.
30576 (maybe_warn_alloc_args_overflow): Call get_size_range.
30577 * calls.h (get_size_range): Declare.
30579 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30581 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30582 from TI's devices.csv file as of September 2016.
30583 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30585 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30587 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30588 * doc/invoke.texi: Likewise.
30589 * doc/md.texi: Likewise.
30590 * doc/objc.texi: Likewise.
30592 2017-01-10 Joshua Conner <joshconner@google.com>
30594 * config/arm/fuchsia-elf.h: New file.
30595 * config/fuchsia.h: New file.
30596 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30597 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30599 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30601 2016-01-10 Richard Biener <rguenther@suse.de>
30603 PR tree-optimization/79034
30604 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30605 Propagate out degenerate PHIs in the joiner.
30607 2017-01-10 Martin Liska <mliska@suse.cz>
30609 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30610 (sort_congruence_classes_by_decl_uid): Likewise.
30611 (sort_congruence_class_groups_by_decl_uid): Likewise.
30612 (sem_item_optimizer::merge_classes): Sort class, groups in these
30613 classes and members in the groups by DECL_UID of declarations.
30614 This would make merge operations stable.
30616 2017-01-10 Martin Liska <mliska@suse.cz>
30618 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30619 usage of m_classes_vec.
30620 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30621 (sem_item_optimizer::get_group_by_hash): Likewise.
30622 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30623 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30624 (sem_item_optimizer::verify_classes): Likewise.
30625 (sem_item_optimizer::process_cong_reduction): Likewise.
30626 (sem_item_optimizer::dump_cong_classes): Likewise.
30627 (sem_item_optimizer::merge_classes): Likewise.
30628 * ipa-icf.h (congruence_class_hash): Rename from
30629 congruence_class_group_hash. Remove declaration of m_classes_vec.
30631 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30633 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30634 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30635 * config.gcc: Add avx512vpopcntdqintrin.h.
30636 * config/i386/avx512vpopcntdqintrin.h: New.
30637 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30638 * config/i386/i386-builtin-types.def: Add new types.
30639 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30640 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30641 __builtin_ia32_vpopcountq_v8di_mask): New.
30642 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30643 __AVX512VPOPCNTDQ__.
30644 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30645 (PTA_AVX512VPOPCNTDQ): Define.
30646 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30647 TARGET_AVX512VPOPCNTDQ_P): Define.
30648 * config/i386/i386.opt: Add mavx512vpopcntdq.
30649 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30650 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30652 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30654 PR middle-end/77484
30655 * predict.def (PRED_CALL): Set to 67.
30657 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30659 * expr.c (store_field): In the bitfield case, if the value comes from
30660 a function call and is of an aggregate type returned in registers, do
30661 not modify the field mode; extract the value in all cases if the mode
30662 is BLKmode and the size is not larger than a word.
30664 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30667 * config/i386/cpuid.h: Fix undefined behavior.
30669 2017-01-04 Jeff Law <law@redhat.com>
30671 PR tree-optimization/79007
30672 PR tree-optimization/67955
30673 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30674 conservative for pt.null when flag_non_call_exceptions is on.
30676 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30678 PR translation/79019
30679 PR translation/79020
30680 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30681 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30682 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30684 * config/avr/avr.opt (maccumulate-args): Likewise.
30685 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30686 * common.opt (freport-bug): Likewise.
30687 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30688 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30689 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30690 translatable string.
30691 * config/i386/i386.c (function_value_32): Likewise.
30692 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30693 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30695 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30696 * common/config/msp430/msp430-common.c (msp430_handle_option):
30698 * symtab.c (symtab_node::verify_base): Likewise.
30699 * opts.c (set_debug_level): Likewise.
30700 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30701 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30702 missing whitespace to translatable strings.
30703 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30704 * config/sh/superh.h: Likewise.
30705 * config/i386/xopintrin.h: Likewise.
30706 * config/i386/znver1.md: Likewise.
30707 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30708 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30709 * double-int.h (struct double_int): Likewise.
30710 * double-int.c (div_and_round_double): Likewise.
30711 * wide-int.cc: Likewise.
30712 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30713 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30714 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30715 (crossjumps_occurred): ... this.
30716 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30719 PR tree-optimization/78899
30720 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30721 returning bool return struct loop *, NULL for failure and the new
30723 (versionable_outer_loop_p): Don't version outer loop if it has
30724 dont_vectorized bit set.
30725 (tree_if_conversion): When versioning outer loop, ensure
30726 tree_if_conversion is performed also on the inner loop of the
30727 non-vectorizable outer loop copy.
30728 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30729 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30730 prevent vectorization of it.
30731 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30732 the outer loop vectorization of the non-scalar version is attempted
30733 before vectorization of the inner loop in scalar version. If
30734 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30735 vectorization of its inner loop.
30736 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30737 has 2 inner loops, rename also on edges from bb whose single pred
30738 is outer_loop->header. Fix typo in function comment.
30740 2017-01-09 Martin Sebor <msebor@redhat.com>
30743 * asan.c (asan_emit_stack_protection): Increase local buffer size
30744 to avoid snprintf truncation warning.
30746 2017-01-09 Andrew Pinski <apinski@cavium.com>
30748 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30749 to reference thunderx2t99 for the tuning structure
30750 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30752 (thunderx2t99_extra_costs): This.
30753 * config/aarch64/aarch64-tune.md: Regenerate.
30754 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30755 (vulcan_addrcost_table): This.
30756 (vulcan_regmove_cost): Rename to ...
30757 (thunderx2t99_regmove_cost): This.
30758 (vulcan_vector_cost): Rename to ...
30759 (thunderx2t99_vector_cost): this.
30760 (vulcan_branch_cost): Rename to ...
30761 (thunderx2t99_branch_cost): This.
30762 (vulcan_tunings): Rename to ...
30763 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30764 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30766 2017-01-09 Martin Jambor <mjambor@suse.cz>
30770 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30771 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30772 (propagate_vr_accross_jump_function): Use the above function for all
30773 value range computations for pass-through jump functions and type
30774 converasion from explicit value range values.
30775 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30776 parameters from TYPE_ARG_TYPES.
30777 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30778 (ipa_write_node_info): Stream type of the actual argument.
30779 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30781 2017-01-09 Martin Liska <mliska@suse.cz>
30784 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30785 (lookup_compiler): Do not show error message with have_E.
30787 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30789 PR tree-optimization/78938
30790 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30791 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30792 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30793 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30796 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30798 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30801 2017-01-09 Richard Biener <rguenther@suse.de>
30803 PR tree-optimization/78997
30804 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30805 name condition properly.
30807 2017-01-09 Richard Biener <rguenther@suse.de>
30810 * dwarf2out.c (is_cxx): New overload with context.
30811 (is_naming_typedef_decl): Use it.
30813 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30815 * invoke.texi (Option Summary): Correct spacing in option lists
30816 and add line breaks to fix over-long lines.
30818 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30820 PR middle-end/17660
30822 * extend.texi (Common Variable Attributes): Add xref to GCC
30823 Internals manual to explain mode attribute keywords.
30825 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30828 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30829 and Preprocessor Options.
30830 (Options for Linking): Document -pthread here....
30831 (RS/6000 and PowerPC Options): ...not here.
30832 (Solaris 2 Options): ...or here.
30833 * doc/cppopts.texi: Document -pthread.
30835 2017-01-08 Martin Sebor <msebor@redhat.com>
30837 PR middle-end/77708
30838 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30839 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30840 New member functions.
30841 (format_directive): Used them.
30843 (pass_sprintf_length::handle_gimple_call): Same.
30844 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30845 to avoid truncation for any argument.
30846 (extract_affine_mul): Same.
30847 * tree.c (get_file_function_name): Same.
30849 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30851 PR middle-end/77484
30852 * predict.def (PRED_INDIR_CALL): Set to 86.
30854 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30856 PR preprocessor/54124
30857 * doc/cppopts.texi: Reformat -d subtable to list the full name
30858 of the options. Add cross-reference to the docs for the general
30859 compiler -d options.
30860 * doc/invoke.texi (Developer Options): Add cross-reference to the
30861 preprocessor-specific -d option documentation.
30863 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30865 PR preprocessor/13498
30866 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30867 redudant material, and reflect new command-line options.
30868 (System Headers): Likewise.
30870 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30872 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30873 -isystem, and -idirafter. Copy-edit.
30874 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30875 default for -ftrack-macro-expansion. Delete obsolete and
30876 badly-formatted implementation details about -fdebug-cpp output.
30877 * doc/cppwarnopts.texi: Copy-edit.
30879 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30882 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30883 that the transition from a max line width >= 1<<10 to narrower
30884 lines works correctly.
30886 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30888 * doc/options.texi (PerFunction): New.
30889 * opt-functions.awk (switch_flags): Map both Optimization and
30890 PerFunction to CL_OPTIMIZATION.
30891 * opth-gen.awk: Test for PerFunction flag along with
30893 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30894 it only when the latter is present. Skip those that don't in
30895 the hash function generator.
30896 * common.opt (fvar-tracking): Mark as PerFunction instead of
30898 (fvar-tracking-assignments): Likewise.
30899 (fvar-tracking-assignments-toggle): Likewise.
30900 (fvar-tracking-uninit): Likewise.
30902 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30904 PR translation/79018
30905 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30908 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30911 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30912 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30913 TARGET_LONG_JUMP_TABLE_OFFSETS.
30914 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30915 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30916 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30917 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30918 * config/m68k/m68k.md (tablejump expander): Likewise.
30919 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30920 TARGET_LONG_JUMP_TABLE_OFFSETS.
30921 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30922 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30924 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30925 David Holsgrove <david.holsgrove@xilinx.com>
30927 * common/config/microblaze/microblaze-common.c
30928 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30929 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30931 * config/microblaze/microblaze.c (microblaze_must_save_register)
30932 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30934 (microblaze_eh_return): New function.
30935 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30936 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30937 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30938 * config/microblaze/microblaze.md (eh_return): New pattern.
30940 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30942 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30943 GCC_DIAGNOSTIC_STRINGIFY): Define.
30945 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30947 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30949 * config/arm/arm.md (<mcrr>): New.
30951 * config/arm/arm.c (arm_arch5te): New.
30952 (arm_option_override): Set arm_arch5te.
30953 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30955 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30956 (arm_mcrr_qualifiers): ... this. New.
30957 (MRRC_QUALIFIERS): Define to...
30958 (arm_mrrc_qualifiers): ... this. New.
30959 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30961 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30962 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30963 (MRRCI, mrrc, MRRC): New.
30964 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30965 VUNSPEC_MRRC2): New.
30967 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30969 * config/arm/arm.md (<mcr>): New.
30971 * config/arm/arm.c (arm_coproc_builtin_available): Add
30972 support for mcr, mrc, mcr2 and mrc2.
30973 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30974 (arm_mcr_qualifiers): ... this. New.
30975 (MRC_QUALIFIERS): Define to ...
30976 (arm_mrc_qualifiers): ... this. New.
30977 (MCR_QUALIFIERS): Define to ...
30978 (arm_mcr_qualifiers): ... this. New.
30979 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30981 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30982 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30983 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30984 VUNSPEC_MRC2): New.
30986 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30988 * config/arm/arm.md (*ldc): New.
30992 * config/arm/arm.c (arm_coproc_builtin_available): Add
30993 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30994 (arm_coproc_ldc_stc_legitimate_address): New.
30995 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30996 'qualifier_const_pointer'.
30997 (LDC_QUALIFIERS): Define to...
30998 (arm_ldc_qualifiers): ... this. New.
30999 (STC_QUALIFIERS): Define to...
31000 (arm_stc_qualifiers): ... this. New.
31001 * config/arm/arm-protos.h
31002 (arm_coproc_ldc_stc_legitimate_address): New.
31003 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31004 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31005 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31006 stc2, stcl, stc2l): New.
31007 * config/arm/constraints.md (Uz): New.
31008 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31009 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31010 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31011 VUNSPEC_STC2L): New.
31013 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31015 * config/arm/arm.md (<cdp>): New.
31016 * config/arm/arm.c (neon_const_bounds): Rename this ...
31017 (arm_const_bounds): ... this.
31018 (arm_coproc_builtin_available): New.
31019 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31020 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31021 (CDP_QUALIFIERS): Define to...
31022 (arm_cdp_qualifiers): ... this. New.
31024 (arm_expand_builtin_args): Add case for 6 arguments.
31025 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31026 (arm_const_bounds): ... this.
31027 (arm_coproc_builtin_available): New.
31028 * config/arm/arm_acle.h (__arm_cdp): New.
31030 * config/arm/arm_acle_builtins.def (cdp): New.
31032 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31033 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31034 'arm_const_bounds'.
31035 * config/arm/types.md (coproc): New.
31036 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31037 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31038 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31039 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31041 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31043 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31044 (UBINOP_QUALIFIERS): New.
31046 (acle_builtin_data): New. Change comment.
31047 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31048 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31049 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31050 arm_acle_builtins.def.
31051 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31052 (arm_init_acle_builtins): New.
31053 (CRC32_BUILTIN): Remove.
31054 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31055 crc32cb, crc32ch and crc32cw.
31056 (arm_init_crc32_builtins): Remove.
31057 (arm_init_builtins): Use arm_init_acle_builtins rather
31058 than arm_init_crc32_builtins.
31059 (arm_expand_acle_builtin): New.
31060 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31061 * config/arm/arm_acle_builtins.def: New.
31063 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31065 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31066 (arm_builtin_datum): ... this.
31067 (arm_init_neon_builtin): Rename to ...
31068 (arm_init_builtin): ... this. Add a new parameters PREFIX
31069 and USE_SIG_IN_NAME.
31070 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31071 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31072 'arm_builtin_datum'.
31073 (arm_init_vfp_builtins): Likewise.
31074 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31075 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31076 (arm_expand_neon_args): Rename to ...
31077 (arm_expand_builtin_args): ... this. Rename builtin_arg
31078 enum values and differentiate between ARG_BUILTIN_MEMORY
31079 and ARG_BUILTIN_NEON_MEMORY.
31080 (arm_expand_neon_builtin_1): Rename to ...
31081 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31082 values, arm_expand_builtin_args and add bool parameter NEON.
31083 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31084 (arm_expand_vfp_builtin): Likewise.
31085 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31087 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31089 PR middle-end/77484
31090 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31091 * predict.c (tree_estimate_probability_bb): Reverse direction of
31092 polymorphic call predictor.
31094 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31096 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31097 (determine_pass_name_match): ...this new function and...
31098 (should_skip_pass_p): ...this new function.
31100 2017-01-06 Nathan Sidwell <nathan@acm.org>
31102 * ipa-visibility.c (function_and_variable_visibility): Reformat
31103 comments and long lines. Remove extrneous if.
31104 * symtab.c (symtab_node::make_decl_local): Fix code format.
31105 (symtab_node::set_section_for_node): Fix comment typo.
31107 2017-01-06 Martin Liska <mliska@suse.cz>
31110 * lra-constraints.c: Rename invariant to lra_invariant.
31111 * predict.c (set_even_probabilities): Initialize e to NULL.
31113 2017-01-05 Martin Sebor <msebor@redhat.com>
31115 PR tree-optimization/78910
31116 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31117 (format_integer): Correct off-by-one error in the handling
31118 of precision with negative numbers in signed conversions..
31120 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31122 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31124 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31126 PR tree-optimization/71016
31127 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31128 factor_out_conditional_conversion. Formatting fix.
31129 (factor_out_conditional_conversion): Add cond_stmt argument.
31130 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31131 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31134 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31136 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31137 read-rtl-function.o, and selftest-rtl.o.
31138 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31139 (selftest::aarch64_test_loading_full_dump): New function.
31140 (selftest::aarch64_run_selftests): New function.
31141 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31142 selftest::aarch64_run_selftests.
31143 * config/i386/i386.c
31144 (selftest::ix86_test_loading_dump_fragment_1): New function.
31145 (selftest::ix86_test_loading_call_insn): New function.
31146 (selftest::ix86_test_loading_full_dump): New function.
31147 (selftest::ix86_test_loading_unspec): New function.
31148 (selftest::ix86_run_selftests): Call the new functions.
31149 * emit-rtl.c (maybe_set_max_label_num): New function.
31150 * emit-rtl.h (maybe_set_max_label_num): New decl.
31151 * function.c (instantiate_decls): Guard call to
31152 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31153 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31155 * gensupport.c (gen_reader::gen_reader): Pass "false"
31156 for new "compact" param of rtx_reader.
31157 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31158 rather than an empty string for NULL strings.
31159 * read-md.c: Potentially include config.h rather than bconfig.h.
31160 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31161 (have_error): New global, copied from errors.c.
31162 (md_reader::read_name): Rename to...
31163 (md_reader::read_name_1): ...this, adding "out_loc" param,
31164 and converting "missing name or number" to returning false, rather
31166 (md_reader::read_name): Reimplement in terms of read_name_1.
31167 (md_reader::read_name_or_nil): New function.
31168 (md_reader::read_string): Handle "(nil)" by returning NULL.
31169 (md_reader::md_reader): Add new param "compact".
31170 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31171 (md_reader::read_file): New method.
31172 * read-md.h (md_reader::md_reader): Add new param "compact".
31173 (md_reader::read_file): New method.
31174 (md_reader::is_compact): New accessor.
31175 (md_reader::read_name): Convert return type from void to file_location.
31176 (md_reader::read_name_or_nil): New decl.
31177 (md_reader::read_name_1): New decl.
31178 (md_reader::m_compact): New field.
31179 (noop_reader::noop_reader): Pass "false" for new "compact" param
31181 (rtx_reader::rtx_reader): Add new "compact" param.
31182 (rtx_reader::read_rtx_operand): Make virtual and convert return
31183 type from void to rtx.
31184 (rtx_reader::read_until): New decl.
31185 (rtx_reader::handle_any_trailing_information): New virtual function.
31186 (rtx_reader::postprocess): New virtual function.
31187 (rtx_reader::finalize_string): New virtual function.
31188 (rtx_reader::m_in_call_function_usage): New field.
31189 (rtx_reader::m_reuse_rtx_by_id): New field.
31190 * read-rtl-function.c: New file.
31191 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31192 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31193 (selftest::verify_three_block_rtl_cfg): New decl.
31194 * read-rtl-function.h: New file.
31195 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31196 For host, include function.h, memmodel.h, and emit-rtl.h.
31197 (one_time_initialization): New function.
31198 (struct compact_insn_name): New struct.
31199 (compact_insn_names): New array.
31200 (find_code): Handle insn codes in compact dumps.
31201 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31202 (bind_subst_iter_and_attr): Likewise.
31203 (add_condition_to_string): Likewise.
31204 (add_condition_to_rtx): Likewise.
31205 (apply_attribute_uses): Likewise.
31206 (add_current_iterators): Likewise.
31207 (apply_iterators): Likewise.
31208 (initialize_iterators): Guard usage of apply_subst_iterator with
31209 #ifdef GENERATOR_FILE.
31210 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31211 (md_reader::read_mapping): Likewise.
31212 (add_define_attr_for_define_subst): Likewise.
31213 (add_define_subst_attr): Likewise.
31214 (read_subst_mapping): Likewise.
31215 (check_code_iterator): Likewise.
31216 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31218 (one_time_initialization): New function.
31219 (rtx_reader::read_until): New method.
31220 (read_flags): New function.
31221 (parse_reg_note_name): New function.
31222 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31223 Handle reuse_rtx ids.
31224 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31225 Add parsing support for RTL dumps, mirroring the special-cases in
31226 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31227 values, and calling handle_any_trailing_information.
31228 (rtx_reader::read_rtx_operand): Convert return type from void
31229 to rtx, returning return_rtx. Handle case 'e'. Call
31230 finalize_string on XSTR and XTMPL fields.
31231 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31232 "(nil)" values were omitted. Call the postprocess vfunc on the
31234 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31235 class ctor. Initialize m_in_call_function_usage. Call
31236 one_time_initialization.
31237 * rtl-tests.c (selftest::test_uncond_jump): Call
31238 set_new_first_and_last_insn.
31239 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31240 * selftest-rtl.c: New file.
31241 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31242 (selftest::get_insn_by_uid): New decl.
31243 * selftest-run-tests.c (selftest::run_tests): Call
31244 read_rtl_function_c_tests.
31245 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31246 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31249 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31251 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31252 operands in a special way. Assert that pos+len <= mode precision.
31254 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31256 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31257 3 argument Alias with unlimited for the negative form.
31258 (fno-vect-cost-model): Removed.
31260 2017-01-05 Martin Liska <mliska@suse.cz>
31262 * hsa-gen.c (gen_hsa_divmod): New function.
31263 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31265 2017-01-05 Martin Liska <mliska@suse.cz>
31268 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31271 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31273 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31274 small constant length operands.
31276 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31278 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31279 between loop iterations.
31281 2017-01-05 Martin Liska <mliska@suse.cz>
31284 * gimplify.c (gimplify_decl_expr): Compare to
31285 asan_poisoned_variables instread of checking flags.
31286 (gimplify_target_expr): Likewise.
31287 (gimplify_expr): Likewise.
31288 (gimplify_function_tree): Conditionally initialize
31289 asan_poisoned_variables.
31291 2017-01-04 Jeff Law <law@redhat.com>
31293 PR tree-optimizatin/78812
31294 * rtl.h (contains_mem_rtx_p): Prototype.
31295 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31296 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31297 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31298 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31299 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31301 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31303 * input.c (assert_char_at_range): Default-initialize actual_range.
31305 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31307 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31308 to match the caller.
31310 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31312 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31313 insns after final jump in test to emit dummy move.
31315 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31317 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31318 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31320 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31322 * multiple_target.c (create_dispatcher_calls): Init e_next.
31323 * tree-ssa-loop-split.c (split_loop): Init border.
31324 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31327 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31332 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31333 (altivec_register_operand): Do not return true if the operand
31334 contains a SUBREG mixing SImode and SFmode.
31335 (vsx_register_operand): Likewise.
31336 (vsx_reg_sfsubreg_ok): New predicate.
31337 (vfloat_operand): Do not return true if the operand contains a
31338 SUBREG mixing SImode and SFmode.
31339 (vint_operand): Likewise.
31340 (vlogical_operand): Likewise.
31341 (gpc_reg_operand): Likewise.
31342 (int_reg_operand): Likewise.
31343 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31344 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31345 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31347 (rs6000_emit_move_si_sf_subreg): New helper function.
31348 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31349 fixup SUBREGs involving SImode and SFmode.
31350 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31351 numbers for the new peephole2 optimization.
31352 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31353 the GLIBC math library that do AND/IOR/XOR operations on single
31354 precision floating point.
31355 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31356 target macros to say whether we need to avoid SUBREGs mixing
31358 (TARGET_ALLOW_SF_SUBREG): Likewise.
31359 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31360 (UNSPEC_SI_FROM_SF): Likewise.
31361 (iorxor): Change spacing.
31362 (and_ior_xor): New iterator for AND, IOR, and XOR.
31363 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31364 (movdi_from_sf_zero_ext): Likewise.
31365 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31366 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31367 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31368 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31369 (fms<mode>4): Likewise.
31370 (fnma<mode>4): Likewise.
31371 (fnms<mode>4): Likewise.
31372 (nfma<mode>4): Likewise.
31373 (nfms<mode>4): Likewise.
31375 2017-01-04 Marek Polacek <polacek@redhat.com>
31378 * doc/invoke.texi: Document -Wpointer-compare.
31380 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31382 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31385 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31386 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31389 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31392 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31393 documentation of the powerpc_popcntb_ok attribute.
31394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31395 code to issue warning messages if a requested CPU configuration is
31396 not supported by the binary (assembler and loader) toolchain.
31397 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31398 made to define a built-in function that has been disabled.
31399 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31400 made to define a built-in function that has been disabled.
31401 (altivec_init_builtins): Add comment explaining why definition
31402 of the DST built-in functions is not preceded by an assertion
31403 check. Add assertions to prevent ICE if attempts are made to
31404 define an altivec predicate or an abs* built-in function that has
31406 (htm_init_builtins): Add comment explaining why definition of the
31407 htm built-in functions is not preceded by an assertion check.
31409 2017-01-04 Jeff Law <law@redhat.com>
31411 PR tree-optimizatin/67955
31412 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31413 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31414 the points-to solution does not include pt_null. Use DECL_PT_UID
31417 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31419 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31420 Use gen_int_mode instead of gen_lopwart for const_int operands.
31422 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31424 PR tree-optimization/71563
31425 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31426 out of range value - has low bits known to be zero.
31428 2017-01-04 Alan Modra <amodra@gmail.com>
31430 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31431 * configure: Regenerate.
31432 * config.in: Regenerate.
31434 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31437 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31438 a substring of the message, but strcmp with the whole message. Ifdef
31439 ENABLE_NLS, translate the message first using dgettext.
31441 2017-01-03 Jeff Law <law@redhat.com>
31443 PR tree-optimizatin/78856
31444 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31445 (mark_threaded_blocks): Remove code to truncate thread paths that
31446 cross multiple loop headers. Instead invalidate the cached loop
31447 iteration information and handle case of a thread path walking
31448 into an irreducible region.
31450 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31453 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31454 assertions. Add support for doing the signbit if the IEEE 128-bit
31455 floating point value is in a GPR.
31456 * config/rs6000/rs6000.md (Fsignbit): Delete.
31457 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31458 Update the length attribute if the value is in a GPR.
31459 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31460 the sign or zero extension instruction, since the value is always 0/1.
31461 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31464 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31465 extracting SImode to a GPR register so that we can generate a
31466 store, limit the vector to be in a traditional Altivec register
31467 for the vextuwrx instruction.
31469 2017-01-03 Ian Lance Taylor <iant@google.com>
31471 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31473 2017-01-03 Martin Sebor <msebor@redhat.com>
31475 PR tree-optimization/78696
31476 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31477 precision. Use MPFR for %f for greater fidelity. Correct handling
31479 (pass_sprintf_length::compute_format_length): Set width and precision
31480 specified by asrerisk to void_node for vararg functions.
31481 (try_substitute_return_value): Adjust dump output.
31483 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31485 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31487 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31489 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31490 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31491 -mlra/-mno-lra was passed to the compiler.
31493 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31495 PR rtl-optimization/65618
31496 * emit-rtl.c (try_split): Move initialization of "before" and
31497 "after" to just before the call to emit_insn_after_setloc.
31499 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31501 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31503 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31505 * dwarf2out.c (gen_enumeration_type_die): When
31506 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31508 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31510 PR tree-optimization/78965
31511 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31512 Change first argument from const call_info & to call_info &. For %n
31513 set info.nowrite to false.
31515 PR middle-end/78901
31516 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31517 possibly throwing calls.
31519 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31520 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31521 and fns handling, rather than in a separate case SSA_NAME.
31523 2017-01-02 Jeff Law <law@redhat.com>
31525 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31526 fixes for first_period and second_period variables.
31528 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31531 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31532 (*insvqi_1): New insn pattern.
31533 (*insvqi_1_mem_rex64): Ditto.
31534 (*insvqi_2): Ditto.
31535 (*insvqi_3): Rename from *insvqi.
31537 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31539 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31541 * doc/cfg.texi (Edges): Remove reference to Java.
31542 (Maintaining the CFG): Ditto.
31544 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31546 PR middle-end/77674
31547 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31548 transparent aliases.
31550 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31552 PR middle-end/77484
31553 * predict.def (PRED_CALL): Update hitrate.
31554 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31555 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31556 into direct/indirect/polymorphic variants.
31558 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31560 Update copyright years.
31562 * gcc.c (process_command): Update copyright notice dates.
31563 * gcov-dump.c (print_version): Ditto.
31564 * gcov.c (print_version): Ditto.
31565 * gcov-tool.c (print_version): Ditto.
31566 * gengtype.c (create_file): Ditto.
31567 * doc/cpp.texi: Bump @copying's copyright year.
31568 * doc/cppinternals.texi: Ditto.
31569 * doc/gcc.texi: Ditto.
31570 * doc/gccint.texi: Ditto.
31571 * doc/gcov.texi: Ditto.
31572 * doc/install.texi: Ditto.
31573 * doc/invoke.texi: Ditto.
31575 Copyright (C) 2017 Free Software Foundation, Inc.
31577 Copying and distribution of this file, with or without modification,
31578 are permitted in any medium without royalty provided the copyright
31579 notice and this notice are preserved.