1 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * machmode.h (MACRO_MODE): New macro.
6 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
7 * targhooks.c (default_libcall_value, default_secondary_reload)
8 (default_memory_move_cost, default_register_move_cost)
9 (default_class_max_nregs): Likewise.
11 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
12 Alan Hayward <alan.hayward@arm.com>
13 David Sherwood <david.sherwood@arm.com>
15 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
16 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
18 * genautomata.c (main): Likewise.
19 * genconditions.c (write_header): Likewise.
20 * genemit.c (main): Likewise.
21 * genextract.c (print_header): Likewise.
22 * genopinit.c (main): Likewise.
23 * genoutput.c (output_prologue): Likewise.
24 * genpeep.c (main): Likewise.
25 * genpreds.c (write_insn_preds_c): Likewise.
26 * genrecog.c (writer_header): Likewise.
27 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
28 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
29 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
30 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
31 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
32 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
33 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
34 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
35 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
36 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
37 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
38 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
39 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
40 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
41 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
42 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
43 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
44 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
45 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
46 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
47 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
48 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
49 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
50 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
51 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
52 * config/darwin.c (IN_TARGET_CODE): Likewise.
53 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
54 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
55 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
56 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
57 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
58 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
59 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
60 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
61 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
62 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
63 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
64 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
65 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
66 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
67 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
68 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
69 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
70 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
71 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
72 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
73 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
74 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
75 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
76 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
77 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
78 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
79 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
80 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
81 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
82 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
83 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
84 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
85 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
86 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
87 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
88 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
89 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
90 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
91 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
92 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
93 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
94 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
95 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
96 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
97 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
98 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
99 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
100 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
101 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
102 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
103 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
104 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
105 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
106 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
107 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
108 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
109 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
110 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
111 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
112 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
113 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
114 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
115 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
116 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
117 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
118 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
119 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
120 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
121 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
122 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
123 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
124 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
125 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
126 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
127 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
128 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
129 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
130 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
131 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
132 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
133 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
134 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
135 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
136 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
137 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
138 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
139 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
140 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
141 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
142 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
143 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
144 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
145 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
146 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
147 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
148 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
149 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
150 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
151 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
152 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
153 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
154 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
155 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
156 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
157 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
158 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
160 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
161 Alan Hayward <alan.hayward@arm.com>
162 David Sherwood <david.sherwood@arm.com>
164 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
167 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
168 Alan Hayward <alan.hayward@arm.com>
169 David Sherwood <david.sherwood@arm.com>
171 * doc/generic.texi (VEC_SERIES_EXPR): Document.
172 * doc/md.texi (vec_series@var{m}): Document.
173 * tree.def (VEC_SERIES_EXPR): New tree code.
174 * tree.h (build_vec_series): Declare.
175 * tree.c (build_vec_series): New function.
176 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
177 * tree-pretty-print.c (dump_generic_node): Likewise.
178 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
179 * tree-inline.c (estimate_operator_cost): Likewise.
180 * expr.c (expand_expr_real_2): Likewise.
181 * optabs-tree.c (optab_for_tree_code): Likewise.
182 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
183 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
184 * expmed.c (make_tree): Handle VEC_SERIES.
185 * optabs.def (vec_series_optab): New optab.
186 * optabs.h (expand_vec_series_expr): Declare.
187 * optabs.c (expand_vec_series_expr): New function.
188 * tree-vect-generic.c (expand_vector_operations_1): Check that
189 the operands also have vector type.
191 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hawyard@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
196 (VEC_COND_EXPR): Add missing @tindex.
197 * doc/md.texi (vec_duplicate@var{m}): Document.
198 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
199 * tree.c (build_vector_from_val): Add stubbed-out handling of
200 variable-length vectors, using VEC_DUPLICATE_EXPR.
201 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
202 * cfgexpand.c (expand_debug_expr): Likewise.
203 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
204 * tree-inline.c (estimate_operator_cost): Likewise.
205 * tree-pretty-print.c (dump_generic_node): Likewise.
206 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
207 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
208 (test_vec_duplicate_folding): New function.
209 (fold_const_c_tests): Call it.
210 * optabs.def (vec_duplicate_optab): New optab.
211 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
212 * optabs.h (expand_vector_broadcast): Declare.
213 * optabs.c (expand_vector_broadcast): Make non-static. Try using
215 * expr.c (store_constructor): Try using vec_duplicate_optab for
217 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
219 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
222 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
223 div/mod latencies a bit.
225 2017-12-15 Jeff Law <law@redhat.com>
227 PR tree-optimization/36550
228 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
229 (mark_threaded_blocks): Rewrite code to avoid block copying when
230 optimizing for size. Don't pessimize blocks which will be
231 copied, but all the statements will be dead.
233 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
235 PR tree-optimization/81165
236 * tree-ssa-threadupdate.c (uses_in_bb): New.
237 (estimate_threading_killed_stmts): New.
238 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
239 * tree-ssa-threadedge.c
240 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
243 2017-12-15 Jeff Law <law@redhat.com>
245 PR tree-optimization/83410
246 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
247 threads when parallelizing loops.
249 2017-12-15 Jakub Jelinek <jakub@redhat.com>
251 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
253 * config/alpha/alpha.c (vms_attribute_table): Swap
254 affects_type_identity and handler fields, adjust comments.
255 * config/mips/mips.c (mips_attribute_table): Likewise.
256 * config/visium/visium.c (visium_attribute_table): Likewise.
257 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
258 * config/microblaze/microblaze.c (microblaze_attribute_table):
260 * config/spu/spu.c (spu_attribute_table): Likewise.
261 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
262 * config/arc/arc.c (arc_attribute_table): Likewise.
263 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
264 * config/v850/v850.c (v850_handle_interrupt_attribute,
265 v850_handle_data_area_attribute): Formatting fixes.
266 (v850_attribute_table): Swap affects_type_identity and handler
267 fields, adjust comments.
268 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
269 * config/arm/arm.c (arm_attribute_table): Likewise.
270 * config/avr/avr.c (avr_attribute_table): Likewise.
271 * config/s390/s390.c (s390_attribute_table): Likewise.
272 * config/sh/sh.c (sh_attribute_table): Likewise.
273 * config/i386/i386.c (ix86_handle_cconv_attribute,
274 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
275 (ix86_attribute_table): Swap affects_type_identity and handler
276 fields, adjust comments.
277 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
278 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
279 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
280 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
281 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
282 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
283 * config/rx/rx.c (rx_attribute_table): Likewise.
284 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
285 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
286 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
287 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
288 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
289 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
290 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
291 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
292 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
293 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
294 * doc/plugins.texi (user_attr): Likewise. Add NULL for
296 * attribs.c (empty_attribute_table): Swap affects_type_identity and
298 (register_scoped_attributes, decl_attributes): Formatting fixes.
300 PR tree-optimization/83269
301 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
302 subtraction in arg0's type if type is signed and arg0 is unsigned.
306 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
307 convert? on @0 instead of convert. Check type of @1, not @0.
308 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
309 convert? on @0 instead of convert. Check type of @1, not @0.
310 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
311 only optimize if either both @1 and @2 types are narrower
312 precision, or both are wider or equal precision, and in the former
313 case only if both have undefined overflow.
315 2017-12-15 Richard Biener <rguenther@suse.de>
318 * internal-fn.def (IFN_NOP): Add.
319 * internal-fn.c (expand_NOP): Do nothing.
320 * lto-streamer-in.c (input_function): Instead of removing
321 sanitizer calls replace them with IFN_NOP calls.
323 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
324 Alan Hayward <alan.hayward@arm.com>
325 David Sherwood <david.sherwood@arm.com>
327 * dse.c (store_info, read_info_type): Replace begin and end with
329 (print_range): New function.
330 (set_all_positions_unneeded, any_positions_needed_p)
331 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
333 (record_store): Likewise. Optimize the case in which all positions
335 (get_stored_val): Replace read_begin and read_end with read_offset
337 (replace_read): Update call accordingly.
339 2017-12-15 Bin Cheng <bin.cheng@arm.com>
341 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
342 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
343 (loop_cand::loop_cand): Initialize above members.
344 (loop_cand::supported_operations): Delete.
345 (loop_cand::can_interchange_p): Inline above function.
346 (loop_cand::classify_simple_reduction): Record number of constant
347 initialized simple reductions.
348 (should_interchange_loops): New parameters. Check stmt cost of loops
350 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
351 Update call to should_interchange_loops.
352 (should_interchange_loop_nest): Update call to
353 should_interchange_loops.
355 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
358 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
359 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
360 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
362 2017-12-15 Julia Koval <julia.koval@intel.com>
364 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
365 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
366 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
367 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
368 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
370 2017-12-15 Julia Koval <julia.koval@intel.com>
372 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
373 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
374 * config/i386/sse.md (vaesenc_<mode>): New pattern.
375 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
376 _mm_aesenc_epi128): New intrinsics.
378 2017-12-15 Julia Koval <julia.koval@intel.com>
380 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
381 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
382 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
383 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
384 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
386 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
388 * gimple-ssa-strength-reduction.c (analyze_increments):
389 Distinguish replacement costs for constant strides from those for
392 2017-12-14 Jakub Jelinek <jakub@redhat.com>
394 * var-tracking.c (variable_tracking_main_1): Formatting fix.
396 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
398 * doc/invoke.texi: Document -Wcast-function-type.
399 * recog.h (stored_funcptr): Change signature.
400 * tree-dump.c (dump_node): Avoid warning.
401 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
403 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
406 * gimple-fold.c (get_range_strlen): Add the handling of non-member
409 2017-12-14 David Malcolm <dmalcolm@redhat.com>
411 PR tree-optimization/83312
412 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
413 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
414 "val" param, and to cope with arbitrary basic blocks.
415 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
416 handle NULL_TREE for "val", dropping "bb" param.
417 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
418 drop "bb" param. Handle NULL_TREE for "val".
419 (find_case_label_for_value): Make "switch_stmt" param const.
420 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
422 (vrp_prop::check_all_array_refs): Reimplement as...
423 (check_array_bounds_dom_walker::before_dom_children): ...this new
424 vfunc. Replace linear search through BB block list, excluding
425 those with non-executable in-edges via dominator walk.
427 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
429 * config/arm/arm.opt (mverbose-cost-dump): New option.
430 * config/arm/arm.c (arm_rtx_costs): Use it.
432 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
435 * reload1.c (emit_input_reload_insns): Skip debug markers.
437 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
439 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
443 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
444 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
445 * config/sh/sh.c: Skip debug insns besides notes.
446 * config/sh/sh.md: Likewise.
447 * config/sh/sh_treg_combine.cc: Likewise.
448 * config/sh/sync.md: Likewise.
450 2017-12-14 Tom de Vries <tom@codesourcery.com>
452 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
453 Add item for weak_undefined.
455 2017-12-14 Richard Biener <rguenther@suse.de>
457 PR tree-optimization/67842
458 * sese.h (bb_in_region): Remove #if 0'ed code.
460 2017-12-14 Richard Biener <rguenther@suse.de>
462 PR tree-optimization/83326
463 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
464 may_be_zero parameter and handle it by not marking the first
465 peeled copy as not exiting the loop.
466 (try_peel_loop): Likewise.
467 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
468 to handle the case of constant or zero iterations and perform
469 loop header copying on-the-fly.
471 2017-12-14 Richard Biener <rguenther@suse.de>
473 PR tree-optimization/83418
474 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
475 Instead of asserting we don't get unfolded comparisons deal with
478 2017-12-14 Jakub Jelinek <jakub@redhat.com>
481 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
482 basic blocks. Assert debug bind insns don't appear outside of bbs,
483 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
486 PR tree-optimization/83198
487 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
488 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
489 value if arg is a REAL_CST with incompatible type.
491 2017-12-14 Sudakshina Das <sudi.das@arm.com>
492 Bin Cheng <bin.cheng@arm.com>
495 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
497 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
500 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
502 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
503 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
505 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * poly-int.h: New file.
510 * poly-int-types.h: Likewise.
511 * coretypes.h: Include them.
512 (POLY_INT_CONVERSION): Define.
513 * target.def (estimated_poly_value): New hook.
514 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
515 * doc/tm.texi: Regenerate.
516 * doc/poly-int.texi: New file.
517 * doc/gccint.texi: Include it.
518 * doc/rtl.texi: Describe restrictions on subreg modes.
519 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
520 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
521 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
522 * targhooks.h (default_estimated_poly_value): Declare.
523 * targhooks.c (default_estimated_poly_value): New function.
524 * target.h (estimated_poly_value): Likewise.
525 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
526 (wi::unary_traits): Delete.
527 (wi::binary_traits::signed_shift_result_type): Define for
528 offset_int << HOST_WIDE_INT, etc.
529 (generic_wide_int::operator <<=): Define for all types and use
530 wi::lshift instead of <<.
531 (wi::hwi_with_prec): Add a default constructor.
532 (wi::ints_for): New class.
533 (operator <<): Define for all wide-int types.
534 (operator /): New function.
535 (operator %): Likewise.
536 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
537 (ASSERT_MAYBE_NE_AT): New macros.
539 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
540 Dominik Vogt <vogt@linux.vnet.ibm.com>
543 * emit-rtl.c (init_emit): Remove ??? comment.
544 * explow.c (get_dynamic_stack_size): Take known alignment of stack
545 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
546 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
547 alignment of 3 virtual registers to BITS_PER_WORD.
549 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
551 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
553 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
554 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
556 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
559 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
561 2017-12-13 Jakub Jelinek <jakub@redhat.com>
563 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
564 with the exception of debug begin stmt markers appear before
568 * final.c (rest_of_handle_final): Call variable_tracking_main only
569 if !flag_var_tracking.
571 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
572 Jakub Jelinek <jakub@redhat.com>
576 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
577 labels debug stmts that can only appear after labels.
579 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
581 PR rtl-optimization/82398
582 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
583 EXPR_USEFULNESS in priority comparison.
585 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
587 PR rtl-optimization/83393
588 * combine.c (move_deaths): If reg_stat points to a too new insn in
589 last_death, do not use it: find the proper insn instead.
591 2017-12-12 Jeff Law <law@redhat.com>
593 PR tree-optimization/83298
594 PR tree-optimization/83362
595 PR tree-optimization/83383
596 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
597 push_value_range a public interface. Add new argument to
598 record_ranges_from_stmt.
599 * gimple-ssa-evrp-analyze.c
600 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
601 Update comments. Handle recording temporary equivalences.
602 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
603 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
604 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
605 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
606 gimple-ssa-evrp-analyze.h.
607 (record_temporary_equivalences_from_phis): Add new argument. When
608 the PHI arg is an SSA_NAME, set the result's range to the range
610 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
612 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
613 Pass it down to children as needed.
614 (thread_outgoing_edges): Likewise.
615 (thread_across_edge): Likewise. Push/pop range state as needed.
616 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
618 2017-12-12 Julia Koval <julia.koval@intel.com>
620 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
621 (PTA_CANNONLAKE): Remove PTA_CLWB.
623 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
626 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
627 (vcondv2div2df): Likewise.
628 (vconduv2dfv2di): Likewise.
630 2017-12-12 Jakub Jelinek <jakub@redhat.com>
632 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
633 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
634 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
635 ATTR_MATHFN_FPROUNDING_ERRNO.
637 2017-12-12 Richard Biener <rguenther@suse.de>
639 PR tree-optimization/83385
640 * graphite-scop-detection.c (get_order, order): Remove.
641 (bb_to_rpo): New global.
643 (build_scops): Sort pbbs in RPO order.
645 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
647 * combine.c (simplify_set): Do not transform subregs to zero_extends
648 if the destination mode is a vector mode.
650 2017-12-12 Jakub Jelinek <jakub@redhat.com>
652 PR tree-optimization/83359
653 * tree-cfg.h (fold_loop_internal_call): Declare.
654 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
655 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
656 (find_loop_dist_alias): New function.
657 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
658 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
661 PR tree-optimization/80631
662 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
663 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
664 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
665 hardcoding zero as the value if COND_EXPR is never true. For
666 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
667 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
668 hardcoding MAX_EXPR as the reduction operation.
669 (is_nonwrapping_integer_induction): Allow negative step.
670 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
671 vect_create_epilog_for_reduction, if no value is suitable, don't
672 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
674 2017-12-12 Richard Biener <rguenther@suse.de>
676 PR tree-optimization/81889
677 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
678 range info from the non-wrapping IV instead of just the range
681 2017-12-12 Julia Koval <julia.koval@intel.com>
683 * config.gcc: Add vaesintrin.h.
684 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
686 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
687 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
689 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
690 * config/i386/immintrin.h: Include vaesintrin.h.
691 * config/i386/sse.md (vaesdec_<mode>): New pattern.
692 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
693 _mm_aesdec_epi128): New intrinsics.
695 2017-12-12 Julia Koval <julia.koval@intel.com>
697 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
698 OPTION_MASK_ISA_VAES_UNSET): New.
699 (ix86_handle_option): Handle -mvaes.
700 * config/i386/cpuid.h: Define bit_VAES.
701 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
702 * config/i386/i386-c.c (__VAES__): New.
703 * config/i386/i386.c (ix86_target_string): Add -mvaes.
704 (ix86_valid_target_attribute_inner_p): Ditto.
705 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
706 * config/i386/i386.opt: Add -mvaes.
707 * doc/invoke.texi: Ditto.
709 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
711 * debug.h (gcc_debug_hooks): Add inline_entry.
712 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
713 * debug.c (do_nothing_debug_hooks): Likewise.
714 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
715 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
716 (dwarf2_lineno_debug_hooks): Likewise.
718 * common.opt (gstatement-frontiers): New, setting
719 debug_nonbind_markers_p.
720 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
721 * toplev.c (process_options): Autodetect value for debug statement
723 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
724 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
726 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
727 markers. Integrate source bind into debug stmt expand loop.
728 (pass_expand::execute): Check debug marker limit. Avoid deep
729 TER and expand debug locations for debug bind insns only.
730 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
732 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
733 * final.c (reemit_insn_block_notes): Take current block from
734 nonbind markers. Declare note where it's first set.
735 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
736 begin stmt markers if enabled.
737 (notice_source_line): Handle nonbind markers. Fail if their
738 location is unknown or that of builtins.
739 (rest_of_handle_final): Convert begin stmt markers to notes if
740 var-tracking didn't run.
741 (rest_of_clean_state): Skip begin stmt markers.
742 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
744 * function.c (allocate_struct_function): Set begin_stmt_markers.
745 * function.h (struct function): Add debug_marker_count counter
746 and debug_nonbind_markers flag.
747 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
748 * gimple-low.c (lower_function_body): Adjust
749 debug_nonbind_markers.
750 (lower_stmt): Drop or skip gimple debug stmts.
751 (lower_try_catch): Skip debug stmts.
752 * gimple.c (gimple_build_debug_begin_stmt): New.
753 (gimple_copy): Increment debug_marker_count if copying one.
754 * gimple.h (gimple_build_debug_begin_stmt): Declare.
755 * gimplify.c (rexpr_location): New.
756 (rexpr_has_location): New.
757 (warn_switch_unreachable_r): Handle gimple debug stmts.
758 (shortcut_cond_r): Call expr_location.
760 (find_goto_label): New.
761 (shortcut_cond_expr): Call expr_has_location, expr_location, and
763 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
765 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
766 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
767 (LANG_HOOKS_INITIALIZER): ... this.
768 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
769 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
771 (update_ebb_live_info): Skip debug insn markers.
772 * lra.c (debug_insn_static_data): Rename to...
773 (debug_bind_static_data): ... this.
774 (debug_marker_static_data): New.
775 (lra_set_insn_recog_data): Select one of the above depending
777 (lra_update_isn_regno_info): Don't assume debug insns have
779 (push_insns): Skip debug insns.
780 * lto-streamer-in.c (input_function): Drop debug stmts
781 depending on active options. Adjust debug_nonbind_markers.
782 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
783 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
784 begin stmt marker notes.
785 (print_insn): Likewise.
786 * recog.c (extract_insn): Recognize rtl for debug markers.
787 * rtl.def (DEBUG_MARKER): New.
788 * tree-inline.c: Include params.h.
789 (remap_gimple_stmt): Handle nonbind markers.
790 (maybe_move_debug_stmts_to_successors): Likewise.
791 (copy_debug_stmt): Likewise.
792 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
793 markers regardless of no side effects.
794 (tsi_link_before): Don't update container's side effects when adding
796 (tsi_link_after): Likewise.
797 (expr_first): Skip begin stmt markers.
798 (expr_last): Likewise.
799 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
800 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
801 Disregard nonbind markers.
802 * tree.c (make_node_stat): Don't set side effects for begin stmt
804 (build1_stat): Likewise.
805 * tree.def (DEBUG_BEGIN_STMT): New.
806 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
807 * var-tracking.c (delete_debug_insns): Renamed to...
808 (delete_vta_debug_insns): ... this.
809 (reemit_marker_as_note): New.
810 (vt_initialize): Reemit markers.
811 (delete_vta_debug_insns): Likewise.
812 (vt_debug_insns_local): Reemit or delete markers.
813 (variable_tracking_main_1): Likewise.
814 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
815 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
816 (gimple_debug_nonbind_marker_p): New.
817 (gimple_build_debug_bind): Adjust.
818 (gimple_build_debug_begin_stmt): New.
819 * doc/invoke.texi (max-debug-marker-count): New param.
820 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
821 (debug_parameter_ref, debug_marker): New.
822 (NOTE_INSN_BEGIN_STMT): New.
823 (DEBUG_INSN): Describe begin stmt markers.
825 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
826 without debug insns, we wouldn't, but clean up debug insns
827 after a control flow insn nevertheless.
829 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
830 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
831 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
832 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
833 (rtl_tidy_fallthru_edge): Likewise.
834 (rtl_verify_fallthru): Likewise.
835 (rtl_verify_bb_layout): Likewise.
836 (skip_insns_after_block): Likewise.
837 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
838 * dwarf2out.c: Include print-rtl.h.
839 (dwarf2out_next_real_insn): New.
840 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
841 Dump debug binds in asm comments.
842 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
843 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
844 callers to use gsi_start_nondebug_bb instead.
845 (gsi_after_labels): Skip gimple debug stmts.
846 (gsi_start_nondebug): New.
847 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
848 (proper_loop_form_for_interchange): Adjust.
849 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
850 * gimple.h (gimple_seq_last_nondebug_stmt): New.
851 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
852 (collect_fallthrough_labels): Likewise.
853 (should_warn_for_implicit_fallthrough): Likewise.
854 (warn_implicit_fallthrough_r): Likewise.
855 (expand_FALLTHROUGH_r): Likewise.
856 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
857 (graphite_copy_stmts_from_block): Skip nonbind markers.
858 * haifa-sched.c (sched_extend_bb): Skip debug insns.
859 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
860 * jump.c (clean_barriers): Skip debug insns.
861 * omp-expand.c (expand_parallel_call): Skip debug insns.
862 (expand_task_call): Likewise.
863 (remove_exit_barrier): Likewise.
864 (expand_omp_taskreg): Likewise.
865 (expand_omp_for_init_counts): Likewise.
866 (expand_omp_for_generic): Likewise.
867 (expand_omp_for_static_nochunk): Likewise.
868 (expand_omp_for_static_chunk): Likewise.
869 (expand_omp_simd): Likewise.
870 (expand_omp_taskloop_for_outer): Likewise.
871 (expand_omp_taskloop_for_inner): Likewise.
872 (expand_oacc_for): Likewise.
873 (expand_omp_sections): Likewise.
874 (expand_omp_single): Likewise.
875 (expand_omp_synch): Likewise.
876 (expand_omp_atomic_load): Likewise.
877 (expand_omp_atomic_store): Likewise.
878 (expand_omp_atomic_fetch_op): Likewise.
879 (expand_omp_atomic_pipeline): Likewise.
880 (expand_omp_atomic_mutex): Likewise.
881 (expand_omp_target): Likewise.
882 (grid_expand_omp_for_loop): Likewise.
883 (grid_expand_target_grid_body): Likewise.
884 (build_omp_regions_1): Likewise.
885 * omp-low.c (check_combined_parallel): Skip debug stmts.
886 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
887 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
888 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
890 * tree-cfg.c (make_blobs_1): Skip debug stmts.
891 (make_edges): Likewise.
892 (cleanup_dead_labels): Likewise.
893 (gimple_can_merge_blocks_p): Likewise.
894 (stmt_starts_bb_p): Likewise.
895 (gimple_block_label): Likewise.
896 (gimple_redirect_edge_and_branch): Likewise.
897 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
899 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
901 * tree-pretty-print (print_declaration): Omit initializer in slim
903 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
905 (eliminate_unnecessary_stmts): Stabilize block removal order.
906 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
907 * var-tracking.c (get_first_insn): New.
908 (vt_emit_notes): Call it.
909 (vt_initialize): Walk any insns before the first BB.
910 (delete_debug_insns): Likewise.
912 * gimple.h (enum gimple_debug_subcode): Add
913 GIMPLE_DEBUG_BEGIN_STMT.
914 (gimple_debug_begin_stmt_p): New.
915 (gimple_debug_nonbind_marker_p): New.
916 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
917 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
918 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
919 * insn-notes.def (BEGIN_STMT): New.
920 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
921 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
922 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
923 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
924 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
925 (INSN_DEBUG_MARKER_KIND): New.
926 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
927 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
928 (INSN_VAR_LOCATION_PTR): New.
929 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
931 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
932 deps only if debug bind insns are enabled.
933 (pass_expand::execute): Avoid deep TER and expand
934 debug locations for debug bind insns only.
935 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
936 debug stmts special handling down to debug bind stmts.
937 * combine.c (try_combine): Narrow debug insns special handling
938 down to debug bind insns.
939 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
940 Narrow debug insns preexisting special handling down to debug
942 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
943 handling down to debug bind insns.
944 * function.c (instantiate_virtual_regs): Skip debug markers,
945 adjust handling of debug binds.
946 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
947 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
948 * haifa-sched.c (schedule_insn): Narrow special handling of debug
949 insns to debug bind insns.
950 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
951 special handling of debug stmts to debug bind stmts.
952 * ipa-split.c (split_function): Likewise.
953 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
954 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
956 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
957 * regrename.c (build_def_use): Likewise.
958 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
959 (pass_cprop_hardreg): Narrow special casing of debug insns to
961 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
962 * reload1.c (reload): Likewise.
963 * sese.c (sese_insert_phis_for_liveouts): Narrow special
964 casing of debug stmts to debug bind stmts.
965 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
966 * ssa-iterators.h (num_imm_uses): Likewise.
967 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
968 debug stmts to debug bind stmts.
969 * tree-inline.c (tree_function_versioning): Narrow special casing
970 of debug stmts to debug bind stmts.
971 * tree-loop-distribution.c (generate_loops_for_partition):
972 Narrow special casing of debug stmts to debug bind stmts.
973 * tree-sra.c (analyze_access_subtree): Narrow special casing
974 of debug stmts to debug bind stmts.
975 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
976 stmts to debug bind stmts.
977 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
978 casing of debug stmts to debug bind stmts.
979 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
980 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
981 casing of debug stmts to debug bind stmts.
982 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
984 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
985 debug stmts to debug bind stmts.
986 (gimple_replace_ssa_lhs): Likewise.
987 (insert_debug_temp_for_var_def): Likewise.
988 (insert_debug_temps_for_defs): Likewise.
989 (reset_debug_uses): Likewise.
990 * tree-ssanames.c (release_ssa_name_fn): Likewise.
991 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
992 (adjust_debug_stmts): Likewise.
993 (adjust_phi_and_debug_stmts): Likewise.
994 (vect_do_peeling): Likewise.
995 * tree-vect-loop.c (vect_transform_loop): Likewise.
996 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
997 * var-tracking.c (adjust_mems): Narrow special casing of debug
998 insns to debug bind insns.
999 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1000 (compute_bb_dataflow, vt_find_locations): Likewise.
1001 (vt_expand_loc, emit_notes_for_changes): Likewise.
1002 (vt_init_cfa_base): Likewise.
1003 (vt_emit_notes): Likewise.
1004 (vt_initialize): Likewise.
1005 (vt_finalize): Likewise.
1007 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1008 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1009 (next_nonnote_nondebug_insn_bb): New.
1010 (prev_nonnote_nondebug_insn_bb): New.
1011 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1012 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1013 (prev_nonnote_nondebug_insn_bb): Declare.
1014 (next_nonnote_nondebug_insn_bb): Declare.
1015 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1016 * cfgrtl.c (get_last_bb_insn): Likewise.
1017 * lra.c (push_insns): Likewise.
1019 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1022 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1024 * selftest.c (selftest::test_runner::test_runner): New ctor.
1025 (selftest::test_runner::~test_runner): New dtor.
1026 * selftest.h (class selftest::test_runner): New class.
1028 2017-12-11 Carl Love <cel@us.ibm.com>
1030 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1031 vec_extract_fp32_from_shortl]): Add #defines.
1032 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1033 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1034 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1035 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1036 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1037 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1038 * doc/extend.texi: Add documentation for the added builtins.
1040 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1042 PR rtl-optimization/80693
1043 PR rtl-optimization/81019
1044 PR rtl-optimization/81020
1045 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1046 are not mentioned in i3. Place the REG_UNUSED note on i2,
1047 possibly modified to REG_DEAD, if it did not originate in i3.
1049 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1051 * recog.c (store_data_bypass_p_1): New function.
1052 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1053 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1055 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1057 PR rtl-optimization/83361
1058 * ifcvt.c (if_convert): Call fixup_partitions.
1060 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1063 early folding of splat_u{8,16,32}.
1065 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1067 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1068 output_operand_lossage first argument with capital letter.
1069 (aarch64_override_options): Don't start error and sorry first argument
1070 with capital letter.
1072 2017-12-11 Andi Kleen <ak@linux.intel.com>
1074 PR gcov-profile/83355
1075 * auto-profile.c (string_table::get_index_by_decl): Don't
1076 recurse when abstract origin points to itself.
1078 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1080 PR tree-optimization/83320
1081 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1082 (prune_datarefs_not_in_loop): Ditto.
1084 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1086 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1088 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1090 PR tree-optimization/83337
1091 * gimple-loop-interchange.cc (compute_access_stride): Handle
1092 bitfield DRs properly.
1094 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1096 PR tree-optimization/83338
1097 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1098 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1099 vector of unsigned integers to vector of signed integers.
1101 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1103 PR rtl-optimization/83317
1104 * lra-constraints.c (process_address_1): Add insn code check.
1106 2017-12-08 Michael Matz <matz@suse.de>
1108 Fix PR tree-optimization/83323
1109 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1110 head-controlled loops and loop BBs.
1111 * common.opt (funroll-and-jam): Remove, instead ...
1112 (floop-unroll-and-jam): ... reuse this option.
1113 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1114 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1115 (-floop-unroll-and-jam): ... this option.
1117 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1119 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1120 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1121 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1123 2017-12-08 Richard Biener <rguenther@suse.de>
1126 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1127 handle accesses outside of zero-sized vars.
1129 2017-12-08 Martin Jambor <mjambor@suse.cz>
1131 PR tree-optimization/83141
1132 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1133 test for MEM_REFs implicitely changing types with padding. Remove
1135 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1137 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1139 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1141 * config/avr/avr.c (avr_attribute_table): Likewise.
1142 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1143 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1144 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1145 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1146 of struct attribute_spec.
1147 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1149 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1152 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1154 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1157 * profile-count.c (profile_count::from_gcov_type): Move from
1158 profile-count.h; handle overflow.
1159 * profile-count.h (profile_count::from_gcov_type): Move offline.
1161 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1163 PR rtl-optimization/83304
1164 * combine.c (move_deaths): If we do not know where a register died,
1167 2017-12-08 Richard Biener <rguenther@suse.de>
1169 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1170 Provide -fopt-info-loop feedback when we interchange in a nest.
1172 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1174 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1175 for armv6 ARM CPU IDs.
1177 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1179 * common/config/arm/arm-common.c: Include <algorithm>.
1180 (INCLUDE_VECTOR): Define.
1181 (compare_opt_names): New function.
1182 (arm_rewrite_selected_arch): Only strip out extensions that can be
1183 expressed through -mfpu. Sort the remaining extensions
1186 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1188 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1189 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1190 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1191 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1193 2017-06-08 Tristan Gingold <gindold@adacore.com>
1196 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1197 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1199 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1201 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1202 of struct attribute_spec.
1204 2017-12-08 Julia Koval <julia.koval@intel.com>
1206 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1207 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1208 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1209 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1210 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1211 _mm_maskz_dpwssds_epi32): Ditto.
1213 2017-12-08 Richard Biener <rguenther@suse.de>
1215 PR tree-optimization/81303
1216 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1217 conditions try to create a comparison vector type matching
1218 the data vector type.
1219 (vectorizable_condition): Adjust.
1220 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1221 Leave invariant conditions alone in case we can vectorize those.
1223 2017-12-08 Julia Koval <julia.koval@intel.com>
1225 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1226 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1227 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1228 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1229 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1231 2017-12-08 Julia Koval <julia.koval@intel.com>
1233 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1234 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1235 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1236 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1237 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1238 _mm_maskz_dpbusds_epi32): New intrinsics.
1240 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1242 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1244 (ld<bh>io_signed): Likewise.
1245 (st<bhw>io): Likewise.
1246 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1247 SMALL_INT12 constant integer operand.
1249 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1251 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1253 (nios2_large_constant_allowed): ...this. Adjust uses.
1254 (nios2_plus_symbolic_constant_p): Rename to...
1255 (nios2_plus_large_constant_p): ...this. Adjust uses.
1256 (nios2_legitimate_address_p): Correct CONST_INT handling.
1257 (nios2_symbolic_memory_operand_p): Rename to...
1258 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1259 (nios2_large_constant_p): Check for large constant integers too.
1260 (nios2_split_large_constant): Handle constant integers.
1261 (nios2_split_symbolic_memory_operand): Rename to...
1262 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1263 (nios2_legitimize_constant_address): Handle constant integers.
1264 (r0rel_constant_p): Handle small constant integers.
1265 (nios2_print_operand_address): Handle r0-relative integer addresses.
1266 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1267 * config/nios2/nios2.md: Adjust for renamed functions.
1269 2017-12-07 Andrew Waterman <andrew@sifive.com>
1271 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1272 (TARGET_HAVE_SRODATA_SECTION): New define.
1273 (riscv_select_section): New function.
1275 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1276 Alexander Monakov <amonakov@ispras.ru>
1277 Jakub Jelinek <jakub@redhat.com>
1280 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1282 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1284 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1285 a bool success value. Don't call output_operand_lossage here.
1286 (aarch64_print_ldpstp_address): Return a bool success value.
1287 (aarch64_print_operand_address): Call output_addr_const if
1288 aarch64_print_address_internal fails.
1289 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1290 'y'; call output_operand_lossage instead. Call output_operand_lossage
1291 if aarch64_print_ldpstp_address fails.
1293 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1295 * tree-vector-builder.h
1296 (tree_vector_builder::binary_encoded_nelts): Declare.
1297 * tree-vector-builder.c
1298 (tree_vector_builder::binary_encoded_nelts): New function.
1299 * fold-const.c (negate_expr_p): Likewise.
1300 (operand_equal_p, fold_checksum_tree): Likewise.
1301 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1302 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1303 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1304 (uniform_vector_p): Likewise.
1305 * varasm.c (const_hash_1, compare_constant): Likewise.
1306 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1307 (valid_lattice_transition): Operate directly on the VECTOR_CST
1309 * ipa-icf.c: Include tree-vector-builder.h.
1310 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1311 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1313 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1315 * tree.c (build_vector): Delete.
1316 * tree.h (build_vector): Make static and move into the self-testing
1319 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1321 * vector-builder.h (vector_builder::derived): New const overload.
1322 (vector_builder::elt): New function.
1323 * tree-vector-builder.h (tree_vector_builder::type): New function.
1324 (tree_vector_builder::apply_step): Declare.
1325 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1327 * gimple-fold.h (tree_vector_builder): Declare.
1328 (gimple_build_vector): Take a tree_vector_builder instead of a
1329 type and vector of elements.
1330 * gimple-fold.c (gimple_build_vector): Likewise.
1331 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1333 (get_initial_defs_for_reduction): Likewise.
1334 (vectorizable_induction): Likewise.
1336 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1338 * tree-vector-builder.h
1339 (tree_vector_builder::new_binary_operation): Declare.
1340 * tree-vector-builder.c
1341 (tree_vector_builder::new_binary_operation): New function.
1342 * fold-const.c (fold_relational_const): Use it.
1343 (const_binop): Likewise. Check that both input vectors have
1344 the same number of elements, thus excluding things like WIDEN_SUM.
1345 Check whether it is possible to operate directly on the encodings
1348 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1350 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1351 new_unary_operation, operating only on the encoded elements.
1352 (const_unop): Likewise.
1353 (exact_inverse): Likewise.
1354 (distributes_over_addition_p): New function.
1355 (const_binop): Use tree_vector_builder and new_unary_operation
1356 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1357 on the encoded elements unless the encoding is strided and the
1358 operation does not distribute over addition.
1359 (fold_convert_const): Use tree_vector_builder and
1360 new_unary_operation. Operate only on the encoded elements
1361 for truncating integer conversions, or for non-stepped encodings.
1363 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1365 * config/sparc/sparc.c: Include tree-vector-builder.h.
1366 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1367 * expmed.c: Include tree-vector-builder.h.
1368 (make_tree): Use tree_vector_builder instead of build_vector.
1369 * fold-const.c: Include tree-vector-builder.h.
1370 (const_binop): Use tree_vector_builder instead of build_vector.
1371 (const_unop): Likewise.
1372 (native_interpret_vector): Likewise.
1373 (fold_vec_perm): Likewise.
1374 (fold_ternary_loc): Likewise.
1375 * gimple-fold.c: Include tree-vector-builder.h.
1376 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1378 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1379 (simplify_vector_constructor): Use tree_vector_builder instead
1381 * tree-vect-generic.c: Include tree-vector-builder.h.
1382 (add_rshift): Use tree_vector_builder instead of build_vector.
1383 (expand_vector_divmod): Likewise.
1384 (optimize_vector_constructor): Likewise.
1385 * tree-vect-loop.c: Include tree-vector-builder.h.
1386 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1387 of build_vector. Explicitly use a stepped encoding for
1389 * tree-vect-slp.c: Include tree-vector-builder.h.
1390 (vect_get_constant_vectors): Use tree_vector_builder instead
1392 (vect_transform_slp_perm_load): Likewise.
1393 (vect_schedule_slp_instance): Likewise.
1394 * tree-vect-stmts.c: Include tree-vector-builder.h.
1395 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1396 (vect_gen_perm_mask_any): Likewise.
1397 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1398 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1400 (build_vector_from_val): Likewise. Explicitly use a duplicate
1403 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1405 * doc/generic.texi (VECTOR_CST): Describe new representation of
1407 * vector-builder.h: New file.
1408 * tree-vector-builder.h: Likewise.
1409 * tree-vector-builder.c: Likewise.
1410 * Makefile.in (OBJS): Add tree-vector-builder.o.
1411 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1412 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1413 (tree_vector): Change the number of elements to
1414 vector_cst_encoded_nelts.
1415 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1416 (VECTOR_CST_ELTS): Delete.
1417 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1418 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1419 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1420 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1421 (VECTOR_CST_ENCODED_ELT): Likewise.
1422 (vector_cst_encoded_nelts): New function.
1423 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1424 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1425 (vector_cst_int_elt, vector_cst_elt): Declare.
1426 * tree.c: Include tree-vector-builder.h.
1427 (tree_code_size): Abort if passed VECTOR_CST.
1428 (tree_size): Update for new VECTOR_CST layout.
1429 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1430 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1431 (build_vector): Use tree_vector_builder.
1432 (vector_cst_int_elt, vector_cst_elt): New functions.
1433 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1434 encoded elements and then create the vector in the canonical form.
1435 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1436 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1437 (tree_c_tests): Call test_vector_cst_patterns.
1438 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1440 (hash_tree): Likewise.
1441 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1442 (streamer_write_tree_header): Likewise.
1443 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1444 (streamer_alloc_tree): Likewise. Update call to make_vector.
1445 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1447 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1449 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1450 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1451 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1453 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1454 Richard Biener <rguenther@suse.de>
1456 PR tree-optimization/81303
1457 * Makefile.in (gimple-loop-interchange.o): New object file.
1458 * common.opt (floop-interchange): Reuse the option from graphite.
1459 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1460 -floop-interchange and mention it for -O3.
1461 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1462 * gimple-loop-interchange.cc: New file.
1463 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1464 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1465 * passes.def (pass_linterchange): New pass.
1466 * timevar.def (TV_LINTERCHANGE): New time var.
1467 * tree-pass.h (make_pass_linterchange): New declaration.
1468 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1469 interchange. Record IV before/after increment in new parameters.
1470 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1471 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1473 (check_reduction_path): ...New function here.
1474 * tree-vectorizer.h (check_reduction_path): New declaration.
1476 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1479 PR rtl-optimization/80818
1480 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1481 always early clobbered.
1482 * lra-lives.c (process_bb_lives): Check input hard regs for early
1483 clobbered non-operand hard reg.
1485 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1488 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1489 types_compatible_p, just that TYPE_MODE is the same.
1491 2017-12-07 Martin Sebor <msebor@redhat.com>
1494 * attribs.c (empty_attribute_table): Initialize new member of
1495 struct attribute_spec.
1496 (decl_attributes): Add argument. Handle mutually exclusive
1497 combinations of attributes.
1498 (selftests::test_attribute_exclusions): New function.
1499 (selftests::attribute_c_tests): Ditto.
1500 * attribs.h (decl_attributes): Add default argument.
1501 * selftest.h (attribute_c_tests): Declare.
1502 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1503 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1505 * doc/extend.texi (Common Function Attributes): Update const and pure.
1506 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1507 of struct attribute_spec.
1508 * config/arc/arc.c (arc_attribute_table): Same.
1509 * config/arm/arm.c (arm_attribute_table): Same.
1510 * config/avr/avr.c ( avr_attribute_table): Same.
1511 * config/bfin/bfin.c (bfin_attribute_table): Same.
1512 * config/cr16/cr16.c (cr16_attribute_table): Same.
1513 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1514 * config/h8300/h8300.c (h8300_attribute_table): Same.
1515 * config/i386/i386.c (ix86_attribute_table): Same.
1516 * config/ia64/ia64.c (ia64_attribute_table): Same.
1517 * config/m32c/m32c.c (m32c_attribute_table): Same.
1518 * config/m32r/m32r.c (m32r_attribute_table): Same.
1519 * config/m68k/m68k.c (m68k_attribute_table): Same.
1520 * config/mcore/mcore.c (mcore_attribute_table): Same.
1521 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1522 * config/mips/mips.c (mips_attribute_table): Same.
1523 * config/msp430/msp430.c (msp430_attribute_table): Same.
1524 * config/nds32/nds32.c (nds32_attribute_table): Same.
1525 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1526 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1527 * config/rl78/rl78.c (rl78__attribute_table): Same.
1528 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1529 * onfig/rx/rx.c (rx_attribute_table): Same.
1530 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1531 * config/sh/sh.c (sh_attribute_table): Same.
1532 * config/sparc/sparc.c (sparc_attribute_table): Same.
1533 * config/spu/spu.c (spu_attribute_table): Same.
1534 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1535 * config/v850/v850.c (v850_attribute_table): Same.
1536 * config/visium/visium.c (visium_attribute_table): Same.
1538 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1541 * config/arm/arm.c (INCLUDE_STRING): Define.
1542 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1543 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1546 2017-12-07 Michael Matz <matz@suse.de>
1548 Add unroll and jam pass
1550 * gimple-loop-jam.c: New file.
1551 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1552 * common.opt (funroll-and-jam): New option.
1553 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1554 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1555 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1556 * passes.def: Add pass_loop_jam.
1557 * timevar.def (TV_LOOP_JAM): Add.
1558 * tree-pass.h (make_pass_loop_jam): Declare.
1559 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1560 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1561 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1562 to flow_loop_tree_node_add.
1563 (duplicate_subloops, copy_loops_to): Append to sibling list.
1564 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1565 * doc/invoke.texi (-funroll-and-jam): Document new option.
1566 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1568 2017-12-07 Richard Biener <rguenther@suse.de>
1570 PR tree-optimization/83296
1571 PR tree-optimization/67769
1572 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1573 flow sensitive info in an unrelated BB.
1574 (value_replacement): Use reset_flow_sensitive_info.
1575 (minmax_replacement): Reset flow sensitive info on the def
1576 we move. Do not reset flow sensitive info in the whole BB
1577 we move the stmt to.
1578 (abs_replacement): Likewise.
1580 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1583 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1584 rs6000_cpu to the given -mcpu=, or to the default processor.
1586 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1588 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1589 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1590 instead of rs6000_cpu_attr.
1591 (rs6000_adjust_cost): Ditto.
1592 (is_microcoded_insn): Ditto.
1593 (rs6000_adjust_priority): Ditto.
1594 (rs6000_issue_rate): Ditto.
1595 (rs6000_use_sched_lookahead): Ditto.
1596 (rs6000_use_sched_lookahead_guard): Ditto.
1597 (rs6000_sched_reorder): Ditto.
1598 (force_new_group): Ditto.
1599 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1600 (group_ending_nop): Ditto.
1602 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1604 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1606 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1607 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1609 (rs6000_reassociation_width): Ditto.
1610 (rs6000_emit_epilogue): Ditto.
1611 (rs6000_adjust_cost): Ditto.
1612 (is_microcoded_insn): Ditto.
1613 (is_cracked_insn): Ditto.
1614 (rs6000_adjust_priority): Ditto.
1615 (rs6000_sched_reorder): Ditto.
1616 (rs6000_sched_reorder2): Ditto.
1617 (insn_must_be_first_in_group): Ditto.
1618 (insn_must_be_last_in_group): Ditto.
1619 (rs6000_register_move_cost): Ditto.
1620 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1623 2017-12-07 Julia Koval <julia.koval@intel.com>
1625 * config.gcc: Add vaesintrin.h.
1626 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1628 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1629 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1631 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1632 * config/i386/immintrin.h: Include vaesintrin.h.
1633 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1634 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1635 _mm_aesdec_epi128): New intrinsics.
1637 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1639 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1640 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1641 c-family/c-spellcheck.cc.
1642 (best_macro_match::best_macro_match): Likewise.
1644 (struct edit_distance_traits<cpp_hashnode *>): Move to
1645 c-family/c-spellcheck.h.
1646 (class best_macro_match): Likewise.
1648 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1650 PR tree-optimization/83293
1651 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1652 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1653 might insert into empty bb.
1656 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1657 simplify for plus with :c added, and pointer_plus without that.
1658 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1659 with undefined overflow and the conversion is not widening,
1660 perform negation in utype and only convert to type afterwards.
1661 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1662 simplify for plus with :c added, and pointer_plus without that.
1663 If type is integral with undefined overflow and the conversion is
1664 not widening, perform minus in utype and only convert to type
1665 afterwards. Move the last pointer_diff_expr simplify into the
1668 2017-12-06 Martin Sebor <msebor@redhat.com>
1670 PR tree-optimization/82646
1671 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1672 strncpy, not maxlen.
1674 2017-12-06 Martin Sebor <msebor@redhat.com>
1676 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1679 PR tree-optimization/83075
1680 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1681 strncat/strncpy don't change length of source string.
1683 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1686 2017-11-29 Martin Aberg <maberg@gaisler.com>
1688 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1689 to prevent b2bst errata sequence.
1690 (sqrtdf2_fix): Likewise.
1692 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1694 PR tree-optimization/81945
1695 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1696 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1697 to dest_cfun has orig_loop_num set, either remap it to the new
1698 loop number if the loop got moved too, or clear it.
1700 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1702 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1704 (thunderx2t99_nothing): New insn reservation.
1705 (thunderx2t99_mrs): New insn reservation.
1706 (thunderx2t99_multiple): New insn reservation.
1707 (thunderx2t99_alu_basi): Add bfx to reservation.
1708 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1710 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1713 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1715 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1717 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1718 tree-ssa-pre.c::remove_dead_inserted_code.
1719 * tree-ssa-dce.h: New file.
1720 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1721 (remove_dead_inserted_code): Move and rename to function
1722 tree-ssa-dce.c::simple_dce_from_worklist.
1723 (pass_pre::execute): Update use.
1725 2017-12-05 Richard Biener <rguenther@suse.de>
1727 PR tree-optimization/83277
1728 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1729 to code-gen liveout vars.
1731 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1733 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1734 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1735 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1738 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1740 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1742 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1743 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1746 2017-12-05 Richard Biener <rguenther@suse.de>
1748 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1749 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1750 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1751 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1753 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1755 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1757 2017-12-05 Martin Liska <mliska@suse.cz>
1758 Jakub Jelinek <jakub@redhat.com>
1760 * doc/invoke.texi: Document the options.
1761 * flag-types.h (enum sanitize_code): Add
1762 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1763 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1764 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1765 * opts.c: Define new sanitizer options.
1766 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1767 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1769 2017-12-05 Julia Koval <julia.koval@intel.com>
1771 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1772 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1773 (ix86_handle_option): Handle -mavx512vnni.
1774 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1775 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1776 * config/i386/i386-c (__AVX512VNNI__): New.
1777 * config/i386/i386.c (ix86_target_string): Handle new option.
1778 (ix86_valid_target_attribute_inner_p): Handle new option.
1779 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1780 * config/i386/i386.opt (mavx512vnni): New option.
1782 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1785 * athlon.md: Disable for generic.
1786 * haswell.md: Enable for generic.
1787 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1788 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1790 (ix86_adjust_cost): Move generic to haswell path.
1792 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1794 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1795 instead of MEM_P in a couple more places. Fix formatting issues.
1797 2017-12-04 Jim Wilson <jimw@sifive.com>
1799 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1800 instead of GP_REG_LAST-1.
1801 (riscv_adjust_libcall_cfi_prologue): Likewise.
1802 (riscv_adjust_libcall_cri_epilogue): Likewise.
1803 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1806 2017-12-04 Luis Machado <luis.machado@linaro.org>
1808 * ipa-pure-const.c (check_decl): Add missing newline.
1809 (state_from_flags): Likewise.
1811 2017-12-04 Jeff Law <law@redhat.com>
1813 PR tree-optimizatin/78496
1814 * gimple-ssa-evrp-analyze.h
1815 (evrp_range_analyzer::get_vr_values): Simplify.
1816 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1817 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1818 and gimple-ssa-evrp-analyze.h.
1819 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1820 (simplify_stmt_for_jump_threading): Copy a blob of code from
1821 tree-vrp.c to use ranges to simplify statements.
1822 (dom_opt_dom_walker::before_dom_children): Call
1823 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1824 (dom_opt_dom_walker::after_dom_children): Similarly for
1825 evrp_range_analyzer::leave.
1826 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1829 * gimple-ssa-evrp-analyze.c
1830 (evrp_range_analyzer::extract_range_from_stmt): Always use
1831 vr_values::update_value_range so preexisting range info is
1832 medged with new range info, even if the new range is VR_VARYING.
1834 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1836 * combine.c: Adjust comment.
1837 (use_crosses_set_p): Delete.
1838 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1839 (try_combine): Ditto.
1841 2017-12-04 Richard Biener <rguenther@suse.de>
1843 PR tree-optimization/83255
1844 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1845 Re-add zero-iteration check.
1847 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1849 PR rtl-optimization/83245
1850 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1851 hard registers as earlyclobber, also if not in an asm.
1853 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1857 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1861 rs6000_cpu based on cpu_index, not tune_index.
1863 2017-12-04 Richard Biener <rguenther@suse.de>
1865 PR tree-optimization/83238
1866 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1867 code match comment, rejecting invalid SESE regions.
1869 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1871 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1872 require base operand is a REG_POINTER prior to reload on targets
1873 with non-equivalent space registers.
1875 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1877 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1878 (update_specialized_profile): Fix updating of counts.
1879 (perhaps_add_new_callers): Likewise.
1881 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1884 * x86-tune.def: Remove obsolette FIXMEs.
1885 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1886 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1887 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1889 (X86_TUNE_PAD_RETURNS): Disable for generic.
1891 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1893 PR tree-optimization/83170
1894 PR tree-optimization/83241
1895 * gimple-ssa-store-merging.c
1896 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1897 gimple_vuse (ins_stmt) in case it has changed.
1898 (imm_store_chain_info::output_merged_store): Likewise.
1900 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1904 * tree-cfg.c (pass_warn_function_return::execute): Handle
1905 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1906 with BUILTINS_LOCATION.
1910 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1911 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1912 instead of DECL_MODE.
1914 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1915 last argument to rtx pointer.
1916 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1917 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1918 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1919 depending on the chosen ISAs.
1920 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1921 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1922 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1924 * config/i386/sse.md (mov<mode>_internal): Likewise.
1925 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1927 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1929 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1931 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1933 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1934 parameters from prototype.
1935 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1936 parameters. Don't print a comment.
1937 (emit_fusion_gpr_load): Adjust.
1938 (emit_fusion_load_store): Adjust.
1939 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1940 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1941 comment on the second line.
1943 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1946 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1947 rs6000_cpu based on cpu_index, not tune_index.
1949 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1951 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1952 which_alternative instead of which_alternative + 1.
1953 (output_asm_insn): Print an extra tab if the template is short.
1955 2017-12-01 Jim Wilson <jimw@sifive.com>
1957 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1959 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1960 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1962 * doc/tm.texi: Regenerate.
1964 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1967 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1968 whether we can allocate pseudos before trying to fix an address.
1969 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1970 memory address is indexed or indirect.
1971 (floatuns_<mode>si2_hw2): Likewise.
1973 2017-12-01 Jason Merrill <jason@redhat.com>
1975 * Makefile.in (TAGS): Add c-family/*.cc.
1977 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1979 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1980 (call_value_insn): Likewise.
1981 (sibcall_insn): Likewise.
1982 (sibcall_value_insn): Likewise.
1983 (movsi_aarch64): Likewise.
1984 (movdi_aarch64): Likewise.
1985 (add_losym_): Likewise.
1986 (ldr_got_small_): Likewise.
1987 (ldr_got_small_sidi): Likewise.
1988 (ldr_got_small_28k_): Likewise.
1989 (ldr_got_small_28k_sidi): Likewise.
1990 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1991 Move output_addr_const to symbolic case. Add error check.
1993 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1995 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1996 * config/s390/s390.md: Disable MVC merging peephole if it would
1997 disable operand forwarding.
1998 (new peephole2): Split MVCs if it would turn them into up to 2
2001 2017-12-01 Richard Biener <rguenther@suse.de>
2003 PR tree-optimization/83232
2004 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2005 detection of same access. Instead of breaking the group here
2006 do not consider the duplicate. Add comment explaining real fix.
2008 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2010 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2012 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2014 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2016 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2018 * function.h (struct function): Remove cilk_frame_decl,
2019 is_cilk_function and calls_cilk_spawn fields.
2020 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2022 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2023 cilk_elemental field.
2024 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2025 * target.def: Adjust comment.
2026 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2027 Don't test cilk_elemental.
2029 PR tree-optimization/83233
2030 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2031 bswap_stat name for the struct.
2034 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2035 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2036 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2037 SWITCH_BREAK_LABEL_P set on the label.
2038 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2039 added for default case if it was missing and not all cases covered.
2040 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2041 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2043 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2044 to avoid -Wimplicit-fallthrough warning.
2045 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2046 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2048 PR tree-optimization/83221
2049 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2051 (init_reassoc): Formatting fix.
2054 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2055 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2059 * tree-cfg.c: Include asan.h.
2060 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2062 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2064 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2066 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2068 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2069 vpcmpeqd instruction.
2071 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2073 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2074 registers type for 128bit mode.
2076 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2078 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2080 * gengtype-state.c (read_state_pair): Likewise.
2081 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2082 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2083 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2084 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2085 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2086 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2087 * builtins.c (fold_builtin_3): Likewise.
2088 * graphite-scop-detection.c
2089 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2090 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2093 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2096 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2097 parameter. Return true if ignore_decl_rtl_set_p is true and other
2098 conditions are satisfied.
2099 * asan.h (asan_protect_global): Add new parameter.
2100 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2101 to asan_protect_global calls.
2103 2017-11-30 Jim Wilson <jimw@sifive.com>
2105 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2106 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2107 -mstrict-align, add info on default value. Delete redundant lines for
2108 -mabi. Add missing -mexplicit-relocs docs.
2110 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2112 * config/arc/arc.md (trap): New pattern.
2114 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2116 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2117 instruction to end into a delay slot.
2118 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2119 can be placed into a delay slot against reg_note.
2121 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2123 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2124 labels number of usages.
2126 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2128 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2130 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2132 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2135 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2136 multiplication by power of 2 constant into two shifts + comparison.
2138 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2141 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2143 2017-11-30 Richard Biener <rguenther@suse.de>
2145 PR tree-optimization/83202
2146 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2147 allow_peel argument and guard peeling.
2148 (canonicalize_loop_induction_variables): Likewise.
2149 (canonicalize_induction_variables): Pass false.
2150 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2151 peeling from cunrolli.
2153 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2155 * combine.c (try_combine): Print a message to dump file whenever
2156 I0, I1, or I2 cannot be combined into I3.
2158 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2160 PR rtl-optimization/83156
2161 PR rtl-optimization/82621
2162 * combine.c (try_combine): Don't split an I2 if one of the dests is
2163 set again before I3. Allow unused dests.
2165 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2167 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2169 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2171 PR rtl-optimization/80818
2172 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2173 recursively. Use insn code for clobber.
2174 (lra_set_insn_recog_data): Pass the new arg to
2175 collect_non_operand_hard_regs.
2176 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2178 (lra_update_insn_regno_info): Pass insn to
2179 add_regs_to_insn_regno_info.
2181 2017-11-29 Jim Wilson <jimw@sifive.com>
2182 Andrew Waterman <andrew@sifive.com>
2184 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2185 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2186 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2187 use SINGLE_SHIFT_COST.
2188 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2189 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2191 2017-11-29 Julia Koval <julia.koval@intel.com>
2193 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2194 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2195 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2196 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2197 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2198 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2199 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2200 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2201 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2202 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2203 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2204 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2205 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2206 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2207 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2208 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2209 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2210 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2211 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2212 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2213 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2214 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2215 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2216 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2217 __builtin_ia32_vpshldv_v2di_mask,
2218 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2219 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2220 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2222 2017-11-29 Julia Koval <julia.koval@intel.com>
2224 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2225 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2226 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2227 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2228 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2229 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2230 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2231 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2232 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2233 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2234 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2235 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2236 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2237 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2238 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2239 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2240 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2241 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2242 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2243 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2244 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2246 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2248 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2249 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2252 2017-11-29 Martin Aberg <maberg@gaisler.com>
2254 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2255 to prevent b2bst errata sequence.
2256 (sqrtdf2_fix): Likewise.
2258 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2260 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2261 (div_sqrt_insn_p): New function.
2262 (sparc_do_work_around_errata): Insert NOP instructions to
2263 prevent sequences that could trigger the TN-0013 errata for
2264 certain LEON3 processors.
2265 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2266 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2267 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2268 (in_branch_delay): Prevent div and sqrt in delay slot if
2270 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2272 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2274 * config/sparc/sparc.c (atomic_insn_p): New function.
2275 (sparc_do_work_around_errata): Insert NOP instructions to
2276 prevent sequences that could trigger the TN-0010 errata for
2278 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2279 instruction referable in atomic_insns_p.
2281 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2283 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2284 (atomic_compare_and_swap_leon3_1): Likewise.
2287 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2289 * config/sparc/sparc.c (fpop_insn_p): New function.
2290 (sparc_do_work_around_errata): Insert NOP instructions to
2291 prevent sequences that could trigger the TN-0012 errata for
2293 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2294 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2295 (in_branch_annul_delay): Prevent floating-point instructions
2296 in delay slot of annulled integer branch.
2298 2017-11-29 Richard Biener <rguenther@suse.de>
2300 PR tree-optimization/83202
2301 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2303 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2304 nodes vectorized to the same stmts multiple times.
2305 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2306 (vect_analyze_slp_instance): Adjust.
2307 (scalar_stmts_to_slp_tree_map_t): New typedef.
2308 (vect_schedule_slp_instance): Add a map recording the SLP node
2309 representing the vectorized stmts for a set of scalar stmts.
2310 Avoid code-generating redundancies.
2311 (vect_schedule_slp): Allocate map and pass it down.
2313 2017-11-29 Nathan Sidwell <nathan@acm.org>
2316 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2319 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2321 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2322 cases for printing LDP/STP memory addresses.
2323 (aarch64_print_address_internal): Renamed from
2324 aarch64_print_operand_address, added parameter, add Pmode check.
2325 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2326 (aarch64_print_operand_address): Indirect to
2327 aarch64_print_address_internal.
2328 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2329 'y' operand output specifier.
2331 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2334 * tree.c (build_simple_mem_ref_loc): Handle
2335 get_addr_base_and_unit_offset returning a MEM_REF.
2338 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2341 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2344 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2346 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2348 (layout::move_to_column): Likewise.
2350 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2353 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2354 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2356 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2357 Martin Jambor <mjambor@suse.cz>
2360 * tree.h (expr_type_first_operand_type_p): Declare
2361 * tree.c (expr_type_first_operand_type_p): New function.
2362 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2363 (ipa_value_from_jfunc): Adjust declaration.
2364 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2365 Use it as result type for arithmetics, unless it is NULL in which case
2366 be more conservative.
2367 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2368 ipa_get_jf_pass_through_result.
2369 (propagate_vals_across_pass_through): Likewise.
2370 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2371 is to propagate_vals_across_pass_through.
2372 (propagate_constants_across_call): Pass PARM_TYPE to
2373 propagate_scalar_across_jump_function.
2374 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2375 ipa_value_from_jfunc.
2376 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2377 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2378 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2379 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2380 target_type, pass it to ipa_value_from_jfunc.
2381 (update_indirect_edges_after_inlining): Pass parameter type to
2382 try_make_edge_direct_simple_call.
2384 2017-11-28 Jeff Law <law@redhat.com>
2386 * gimple-ssa-evrp-analyze.c
2387 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2388 refine ranges if scev_initialized_p returns true.
2389 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2391 2017-11-28 Julia Koval <julia.koval@intel.com>
2393 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2394 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2395 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2396 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2397 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2398 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2399 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2400 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2401 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2402 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2403 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2404 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2405 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2406 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2407 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2408 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2409 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2410 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2411 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2412 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2413 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2414 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2416 2017-11-28 Julia Koval <julia.koval@intel.com>
2418 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2419 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2420 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2421 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2422 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2423 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2424 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2425 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2426 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2427 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2428 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2429 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2430 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2431 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2432 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2433 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2434 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2435 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2436 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2437 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2438 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2439 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2440 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2441 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2442 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2443 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2444 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2445 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2446 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2447 __builtin_ia32_vpshld_v2di_mask): New builtins.
2448 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2449 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2451 2017-11-28 Richard Biener <rguenther@suse.de>
2453 PR tree-optimization/80776
2454 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2456 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2458 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2459 If the incoming edge is an effective fallthru because the other
2460 edge only reaches a __builtin_unreachable () then record ranges
2461 derived from the controlling condition in SSA info.
2462 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2463 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2465 2017-11-28 Olivier Hainque <hainque@adacore.com>
2467 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2468 on mingw build hosts.
2470 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2473 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2475 2017-11-28 Richard Biener <rguenther@suse.de>
2478 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2479 copies generated from memcpy use a character array as reference
2482 2017-11-28 Julia Koval <julia.koval@intel.com>
2483 Sebastian Peryt <sebastian.peryt@intel.com>
2485 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2486 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2487 cilk-common.o, cilk.h, cilk-common.c): Remove.
2489 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2490 * builtins.c (is_builtin_name): Remove cilkplus condition.
2491 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2492 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2493 cilk-builtins.def, cilkplus.def): Remove.
2494 * cif-code.def (CILK_SPAWN): Remove.
2495 * cilk-builtins.def: Delete.
2496 * cilk-common.c: Ditto.
2498 * cilkplus.def: Ditto.
2499 * config/darwin.h (fcilkplus): Delete.
2500 * cppbuiltin.c: Ditto.
2501 * doc/extend.texi: Remove cilkplus doc.
2502 * doc/generic.texi: Ditto.
2503 * doc/invoke.texi: Ditto.
2504 * doc/passes.texi: Ditto.
2505 * gcc.c (fcilkplus): Remove.
2506 * gengtype.c (cilk.h): Remove.
2507 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2509 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2511 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2513 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2514 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2515 cilkplus conditions.
2516 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2517 inline_read_section): Ditto.
2518 * ipa-inline-analysis.c (cilk.h): Remove.
2519 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2520 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2521 append_linker_options): Remove condition for fcilkplus.
2522 * lto/lto-lang.c (cilk.h): Remove.
2523 (lto_init): Remove condition for fcilkplus.
2524 * omp-expand.c (expand_cilk_for_call): Delete.
2525 (expand_omp_taskreg, expand_omp_for_static_chunk,
2526 expand_omp_for): Remove cilkplus
2528 (expand_cilk_for): Delete.
2529 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2530 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2531 execute_lower_omp, diagnose_sb_0): Ditto.
2532 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2533 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2534 * tree-nested.c: Ditto.
2535 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2536 (dump_generic_node): Ditto.
2537 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2538 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2540 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2542 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2544 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2545 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2546 for vector mode and !TARGET_SIMD.
2548 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2550 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2552 * tree.h (SWITCH_LABELS): Remove.
2553 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2554 assert SWITCH_BODY is non-NULL.
2555 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2557 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2559 PR tree-optimization/80788
2560 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2561 has TREE_OVERFLOW set, call drop_tree_overflow.
2563 2017-11-28 Richard Biener <rguenther@suse.de>
2565 PR tree-optimization/83158
2566 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2568 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2571 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2572 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2574 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2576 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2577 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2579 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2582 * varasm.c (compare_constant): Take the mode of the constants into
2583 account when comparing floating point constants.
2585 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2587 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2588 from explicit instantiation of debug_helper.
2589 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2591 2017-11-27 Richard Biener <rguenther@suse.de>
2593 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2596 2017-11-27 Richard Biener <rguenther@suse.de>
2598 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2599 (type_hash_canon): Also clear int_cst_hash_table entry for
2601 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2603 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2605 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2606 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2607 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2609 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2611 * hash-map.h (gt_cleare_cache): Avoid UB.
2613 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2615 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2616 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2617 (decide_unroll_runtime_iterations): Likewise.
2618 (decide_unroll_stupid): Likewise.
2620 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2623 * config/i386/i386.md: Add a loop with incssp.
2625 2017-11-27 Martin Jambor <mjambor@suse.cz>
2627 PR tree-optimization/81248
2628 * tree-sra.c (splice_param_accesses): Remove size check.
2629 (decide_one_param_reduction): Fix size check.
2630 * gimple-pretty-print.c (dump_profile): Silence warning.
2631 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2633 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2636 * dbxout.c (lastlineno): New variable.
2637 (dbx_debug_hooks): Use dbxout_switch_text_section as
2638 switch_text_section debug hook.
2639 (dbxout_function_end): Switch to current_function_section
2640 rather than function_section. If crtl->has_bb_partition,
2641 output just one N_FUN, depending on in_cold_section_p.
2642 (dbxout_source_line): Remember last lineno in lastlineno.
2643 (dbxout_switch_text_section): New function.
2644 (dbxout_function_decl): Adjust dbxout_block caller.
2645 (dbx_block_with_cold_children): New function.
2646 (dbxout_block): Return true if any LBRAC/RBRAC have been
2647 emitted. Use dbx_block_with_cold_children at depth == 0
2648 in second partition. Add PARENT_BLOCKNUM argument, pass
2649 it optionally adjusted to children. Output LBRAC/RBRAC
2650 around recursive call only if the block is in the current
2651 partition, if not and anything was output, emit empty
2653 * final.c (final_scan_insn): Compute cold_function_name
2654 before calling switch_text_section debug hook. Call
2655 that hook even if dwarf2out_do_frame if not emitting
2659 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2660 TREE_READONLY decls.
2662 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2664 PR rtl-optimization/82488
2665 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2667 2017-11-26 Julia Koval <julia.koval@intel.com>
2669 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2671 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2674 2017-11-26 Julia Koval <julia.koval@intel.com>
2676 * config/i386/driver-i386.c (host_detect_local_cpu):
2677 Detect skylake-avx512.
2679 2017-11-26 Julia Koval <julia.koval@intel.com>
2681 * config.gcc: Add -march=cannonlake.
2682 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2683 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2684 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2685 (PTA_CANNONLAKE): New.
2686 (processor_target_table): Add cannonlake.
2687 (ix86_option_override_internal): Ditto.
2688 (fold_builtin_cpu): Ditto.
2689 (get_builtin_code_for_version): Handle cannonlake.
2690 (M_INTEL_COREI7_CANNONLAKE): New.
2691 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2692 * doc/invoke.texi: Add -march=cannonlake.
2694 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2696 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2697 (try_init_one_plugin): Alternative implementation for MinGW.
2698 * Makefile.in (plugin_implib): New.
2699 (gengtype-lex.c): Fix broken AIX workaround.
2700 * configure: Regenerate.
2701 * doc/plugins.texi: Document support for MinGW.
2703 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2705 PR rtl-optimization/81553
2706 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2707 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2708 is a shift where C1 has different mode than the whole shift, use C1's
2709 mode for MULT rather than the shift's mode.
2712 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2713 builtins not enabled in the currently selected ISA.
2715 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2717 PR tree-optimization/71026
2718 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2719 (insert_reciprocals): Change to insert reciprocals before a division
2720 by a square and to insert the square of a reciprocal.
2721 (execute_cse_reciprocals_1): Change to consider division by a square.
2722 (register_division_in): Add importance parameter.
2724 2017-11-24 Richard Biener <rguenther@suse.de>
2726 PR tree-optimization/82402
2727 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2728 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2730 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2732 * match.pd (0-ptr): New transformation.
2734 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2737 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2738 when skipping checking.
2740 2017-11-24 Richard Biener <rguenther@suse.de>
2742 PR tree-optimization/83128
2743 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2744 (vn_reference_lookup_3): Likewise.
2746 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2749 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2750 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2753 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2754 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2755 check for the latter with gimple_call_builtin_p. Do not handle
2756 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2758 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2760 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2763 2017-11-23 Julia Koval <julia.koval@intel.com>
2765 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2766 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2767 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2768 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2769 _mm512_maskz_expandloadu_epi16): New intrinsics.
2770 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2771 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2772 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2773 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2774 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2775 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2776 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2777 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2778 _mm256_maskz_expandloadu_epi8): New intrinsics.
2779 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2780 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2781 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2782 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2783 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2784 * config/i386/sse.md (VI248_VLBW): New iterator.
2785 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2787 2017-11-23 Julia Koval <julia.koval@intel.com>
2789 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2790 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2791 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2792 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2793 _mm512_mask_compressstoreu_epi16): New.
2794 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2795 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2796 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2797 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2798 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2799 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2800 _mm256_mask_compressstoreu_epi8): New.
2801 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2802 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2803 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2804 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2805 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2806 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2807 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2808 __builtin_ia32_compresshi128_mask,
2809 __builtin_ia32_compressstoreuqi512_mask,
2810 __builtin_ia32_compressstoreuhi512_mask,
2811 __builtin_ia32_compressstoreuqi256_mask,
2812 __builtin_ia32_compressstoreuqi128_mask,
2813 __builtin_ia32_compressstoreuhi256_mask,
2814 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2815 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2817 (ix86_expand_special_args_builtin): Handle new types.
2818 (s4fma_expand): Handle new builtin array.
2819 * config/i386/immintrin.h: Include new headers.
2820 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2821 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2823 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2826 fixes. Declare temp and g variables at the top in order to avoid
2827 {} in most of the cases.
2829 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2831 * match.pd (ptr-0): New transformation.
2833 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2835 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2836 (struct addr_mode_cost_table): New.
2837 (struct tune_params): Add field addr_mode_costs.
2838 * config/arm/arm.c (generic_addr_mode_costs): New.
2839 (arm_slowmul_tune): Initialise addr_mode_costs field.
2840 (arm_fastmul_tune): Likewise.
2841 (arm_strongarm_tune): Likewise.
2842 (arm_xscale_tune): Likewise.
2843 (arm_9e_tune): Likewise.
2844 (arm_marvell_pj4_tune): Likewise.
2845 (arm_v6t2_tune): Likewise.
2846 (arm_cortex_tune): Likewise.
2847 (arm_cortex_a8_tune): Likewise.
2848 (arm_cortex_a7_tune): Likewise.
2849 (arm_cortex_a15_tune): Likewise.
2850 (arm_cortex_a35_tune): Likewise.
2851 (arm_cortex_a53_tune): Likewise.
2852 (arm_cortex_a57_tune): Likewise.
2853 (arm_exynosm1_tune): Likewise.
2854 (arm_xgene1_tune): Likewise.
2855 (arm_cortex_a5_tune): Likewise.
2856 (arm_cortex_a9_tune): Likewise.
2857 (arm_cortex_a12_tune): Likewise.
2858 (arm_cortex_a73_tune): Likewise.
2859 (arm_v7m_tune): Likewise.
2860 (arm_cortex_m7_tune): Likewise.
2861 (arm_v6m_tune): Likewise.
2862 (arm_fa726te_tune): Likewise.
2863 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2865 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2867 * config/arm/arm.c (arm_mem_costs): New function.
2868 (arm_rtx_costs_internal): Use arm_mem_costs.
2870 2017-11-23 Mark Wielaard <mark@klomp.org>
2872 * dwarf2out.c (init_sections_and_labels): Use generation to create
2873 unique ranges_section_label and ranges_base_label. Return generation.
2874 (output_rnglists): Add generation argument. Use generation to create
2875 unique ranges labels.
2876 (dwarf2out_finish): Get generation from init_sections_and_labels
2877 and pass generation to output_rnglists.
2879 2017-11-23 Mike Stump <mikestump@comcast.net>
2880 Eric Botcazou <ebotcazou@adacore.com>
2882 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2883 * cfgloop.h (struct loop): Add unroll field.
2884 * function.h (struct function): Add has_unroll bitfield.
2885 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2886 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2887 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2888 (pass_rtl_unroll_loops::gate): Likewise.
2889 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2890 for which loop->unroll==1.
2891 (decide_unroll_constant_iterations): Use note for consistency and
2892 take loop->unroll into account. Return early if loop->unroll is set.
2893 Fix thinko in existing test.
2894 (decide_unroll_runtime_iterations): Use note for consistency and
2895 take loop->unroll into account.
2896 (decide_unroll_stupid): Likewise.
2897 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2898 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2899 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2901 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2902 (print_loop): Print loop->unroll if set.
2903 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2904 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2905 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2907 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2908 loop->unroll is set and smaller than the trip count. Otherwise bypass
2909 entirely the heuristics if loop->unroll is set. Remove dead note.
2910 Fix off-by-one bug in other note.
2911 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2912 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2914 (tree_unroll_loops_completely): Make static.
2915 (pass_complete_unroll::execute): Use correct type for variable.
2916 (pass_complete_unrolli::execute): Fix formatting.
2917 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2919 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2921 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2922 enable when TARGET_PREFER_AVX128 is set.
2924 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2926 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2928 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2929 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2931 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2933 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2934 frequencies. Fix estimation of aggregate parameters.
2936 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2938 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2941 2017-11-23 Tom de Vries <tom@codesourcery.com>
2943 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2945 2017-11-23 Tom de Vries <tom@codesourcery.com>
2947 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2950 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2953 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2954 sibcall_value_pcrel_fdpic): Use local variable instead of
2956 (calli_tbr_rel): Add missing operand 2.
2957 (call_valuei_tbr_rel): Add missing operand 3.
2959 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2962 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2963 bitpos/bitsize covering the whole destination, use store_expr only if
2964 the complex mode is the same. Otherwise, use expand_normal and if
2965 it returns CONCAT, subreg each part separately instead of trying to
2966 subreg the whole result.
2968 2017-11-23 Richard Biener <rguenther@suse.de>
2970 PR tree-optimization/23094
2971 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2972 come here from walking over backedges in the first iteration.
2973 (vn_reference_lookup_3): Skip clobbers that store the same value.
2975 2017-11-23 Richard Biener <rguenther@suse.de>
2977 PR tree-optimization/81403
2978 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2979 a block we need a leader relative to.
2980 (phi_translate_1): For nary processing require a leader from
2981 get_representative_for given we run expression simplification
2982 using match-and-simplify. Remove previous fix.
2984 2017-11-22 Jeff Law <law@redhat.com>
2986 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2987 Use new method allocate_value_range rather than accessing the
2988 vrp_value_range_pool data member directly.
2989 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2990 to use extract_range_from_stmt method to avoid need for
2991 extract_range_from_assignment method.
2992 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2993 method rather than setting values_propgated data member directly.
2994 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2995 and values propagated data members and extract_range_from_assignment
2996 method. Reorder private data members to conform to standards.
2997 Add new methods set_lattice_propagation_complete and
2998 allocate_value_range.
3000 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3002 PR rtl-optimization/83030
3003 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3004 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3005 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3006 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3007 CROSSING_JUMP_P flag is already set before setting it.
3008 * cfgrtl.c (fixup_partition_crossing): Likewise.
3009 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3012 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3014 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3015 Handle the case where both arguments are using gen_const_vec_series.
3017 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3020 * pretty-print.c (pp_format): Move quoting implementation to
3021 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3022 to pass address of "quote" local.
3023 (pp_begin_quote): New function.
3024 (pp_end_quote): New function.
3025 * pretty-print.h (printer_fn): Convert penultimate param from bool
3027 (pp_begin_quote): New decl.
3028 (pp_end_quote): New decl.
3029 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3030 param from bool to bool *.
3031 * tree-diagnostic.h (default_tree_printer): Likewise.
3033 2017-11-22 Jeff Law <law@redhat.com>
3035 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3036 Initialize vr_values.
3037 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3038 vr_values via vr_values class instance rather than delegators.
3039 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3040 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3041 (evrp_range_analyzer::push_value_range): Likewise.
3042 (evrp_range_analyzer::pop_value_range): Likewise.
3043 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3044 most delegators. Those remaining are exposed as public interfaces.
3045 Make vr_values a pointer and private.
3046 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3048 (evrp_range_analyzer::get_vr_value): New method.
3049 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3050 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3051 (class evrp_dom_walker): Attach evrp_folder class, initialize
3052 it in the ctor. Remove temporary delegators.
3053 (evrp_dom_walker::before_dom_children): Call methods in attached
3054 evrp_range_analyzer class via class instance pointer. Use
3055 free value_range_constant_singleton to remove need for
3056 op_with_constant_singleton_value delegator method. Do not
3057 create a vrp_prop class instance for every call! Narrow
3058 scope of a couple variables.
3059 (evrp_dom_walker::cleanup): Call methods in attached
3060 evrp_range_analyzer class via class instance pointer.
3061 * vr-values.h (class vr_values): Privatize many methods and
3064 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3066 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3067 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3068 * expr.c (expand_expr_real_2): Likewise.
3069 * fold-const.c (const_unop): Likewise.
3070 * optabs-tree.c (optab_for_tree_code): Likewise.
3071 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3072 * tree-inline.c (estimate_operator_cost): Likewise.
3073 * tree-pretty-print.c (dump_generic_node): Likewise.
3074 (op_code_prio): Likewise.
3075 (op_symbol_code): Likewise.
3076 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3077 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3078 * internal-fn.c (direct_internal_fn_optab): New function.
3079 (direct_internal_fn_array, direct_internal_fn_supported_p
3080 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3081 * fold-const-call.c (fold_const_reduction): New function.
3082 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3084 * tree-vect-loop.c: Include internal-fn.h.
3085 (reduction_code_for_scalar_code): Rename to...
3086 (reduction_fn_for_scalar_code): ...this and return an internal
3088 (vect_model_reduction_cost): Take an internal_fn rather than
3090 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3092 (vectorizable_reduction): Use internal functions rather than tree
3093 codes for the reduction operation. Update calls to the functions
3095 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3096 Use calls to internal functions rather than REDUC tree codes.
3097 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3099 2017-11-22 Olivier Hainque <hainque@adacore.com>
3101 * config/vxworks.c (vxworks_override_options): Pick default
3102 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3103 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3104 DWARF_GNAT_ENCODINGS_DEFAULT.
3105 * config/vxworksae.h: Likewise.
3107 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3109 PR tree-optimization/83104
3110 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3113 2017-11-22 Marek Polacek <polacek@redhat.com>
3114 H.J. Lu <hongjiu.lu@intel.com>
3115 Jason Merrill <jason@redhat.com>
3120 * calls.c (initialize_argument_information): Call
3121 warn_parameter_passing_abi target hook.
3122 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3123 argument onto stack.
3124 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3125 * common.opt: Update -fabi-version description.
3126 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3127 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3129 (ix86_is_empty_record): New function.
3130 (ix86_warn_parameter_passing_abi): New function.
3131 (TARGET_EMPTY_RECORD_P): Redefine.
3132 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3133 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3134 * doc/tm.texi: Regenerated.
3135 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3136 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3137 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3138 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3139 instead of int_size_in_bytes.
3140 * expr.c (copy_blkmode_to_reg): Likewise.
3141 * function.c (aggregate_value_p): Return 0 for empty types.
3142 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3144 (locate_and_pad_parm): Call arg size_in_bytes instead
3146 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3147 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3148 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3150 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3151 (std_gimplify_va_arg_expr): Skip empty records. Call
3152 arg_size_in_bytes instead size_in_bytes.
3153 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3154 * tree-core.h (tree_type_common): Add empty_flag.
3155 (tree_decl_common): Update comments.
3156 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3158 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3159 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3161 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3162 * tree.c (default_is_empty_type): New function.
3163 (default_is_empty_record): New function.
3164 (arg_int_size_in_bytes): New function.
3165 (arg_size_in_bytes): New function.
3166 (get_ultimate_context): New function.
3167 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3168 TRANSLATION_UNIT_WARN_EMPTY_P.
3169 (default_is_empty_record, arg_int_size_in_bytes,
3170 arg_size_in_bytes, get_ultimate_context): Declare.
3172 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3174 * config/arm/arm.c (cmse_clear_registers): New function.
3175 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3176 code by call to cmse_clear_registers.
3177 (cmse_nonsecure_entry_clear_before_return): Likewise.
3179 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3181 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3182 (vdot_s32, vdotq_s32): New.
3183 (vdot_lane_u32, vdotq_lane_u32): New.
3184 (vdot_lane_s32, vdotq_lane_s32): New.
3187 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3190 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3191 for unsigned values with fewer HWIs than the precision.
3192 (test_overflow): New function.
3193 (wide_int_cc_tests): Call it.
3195 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3197 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3198 has the same mode class as Pmode.
3200 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3202 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3203 padding_bits_to_clear_ptr.
3204 (cmse_nonsecure_entry_clear_before_return): Likewise.
3206 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3208 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3209 auto_sbitap instead of integer bitfield to control register needing
3212 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3214 PR tree-optimization/83044
3215 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3216 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3217 that need the upper bound. Subtract offset from
3218 get_addr_base_and_unit_offset only if positive and subtract it
3219 before division by eltsize rather than after it.
3222 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3223 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3224 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3228 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3229 expand_binop with *mul_widen_optab, make sure at least one of the
3230 operands doesn't have VOIDmode.
3233 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3235 PR rtl-optimization/82044
3236 PR tree-optimization/82042
3237 * dse.c (record_store): Check for overflow.
3238 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3239 add_wild_read instead of clear_rhs_from_active_local_stores on
3240 overflow and log it into dump_file.
3242 2017-11-22 Richard Biener <rguenther@suse.de>
3244 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3245 fake edges to exit when looking for a place to insert.
3246 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3248 (insert_into_preds_of_block): Commit edge insertion immediately,
3249 assert that doesn't require new BBs.
3250 (fini_pre): Release expressions.
3251 (pass_pre::execute): Shuffle things around a bit, if the fn
3252 is too large do not compute AVAIL either as this is really the
3255 2017-11-22 Richard Biener <rguenther@suse.de>
3257 PR tree-optimization/83089
3258 * tree-if-conv.c (pass_if_conversion::execute): If anything
3259 changed reset SCEV and free the number of iteration estimates.
3261 2017-11-21 Martin Sebor <msebor@redhat.com>
3263 PR tree-optimization/82945
3264 * calls.h (warn_nonstring_bound): Remove unused function.
3266 2017-11-21 Martin Sebor <msebor@redhat.com>
3268 PR tree-optimization/82945
3269 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3270 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3271 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3273 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3274 * calls.h (get_attr_nonstring_decl): Declare new function.
3275 * doc/extend.texi (attribute nonstring): Update.
3276 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3277 get_attr_nonstring_decl and handle it.
3278 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3279 detection of nul-termination.
3280 (strlen_to_stridx): Change to a pointer.
3281 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3282 (pass_strlen::execute): Same.
3284 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3286 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3287 for the new option -mprefer-vector-width=[none|128|256|512].
3288 * config/i386/i386.c (ix86_target_string): remove old style options
3289 -mprefer-avx256 and make -mprefer-avx128 as alias.
3290 (ix86_option_override_internal): Apply defaults for the
3291 -mprefer-vector-width=[128|256] option.
3292 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3293 Implement macros to work with -mprefer-vector-width=.
3294 * config/i386/i386.opt: Implemented option
3295 -mprefer-vector-width=[none|128|256|512].
3296 * doc/invoke.texi: Documentation for
3297 -mprefer-vector-width=[none|128|256|512].
3299 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3301 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3303 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3305 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3307 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3309 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3311 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3312 * expr.c (expand_expr_real_2): Likewise.
3313 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3314 fold_binary_loc): Likewise.
3315 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3316 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3317 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3318 MINUS_EXPR transformations.
3319 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3320 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3321 * tree-inline.c (estimate_operator_cost): Likewise.
3322 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3323 op_symbol_code): Likewise.
3324 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3325 * vr-values.c (extract_range_from_binary_expr): Likewise.
3326 * varasm.c (initializer_constant_valid_p_1): Likewise.
3327 * tree.def: New tree code POINTER_DIFF_EXPR.
3329 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3331 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3332 integer suffix to movbe mnemonic.
3333 (*bswaphi2_movbe): Ditto.
3334 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3336 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3339 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3340 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3341 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3343 2017-11-21 Martin Liska <mliska@suse.cz>
3345 * tree-inline.c (expand_call_inline): Remove not needed
3348 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3349 Jakub Jelinek <jakub@redhat.com>
3352 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3353 Remove static keyword from f variable.
3355 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3357 PR tree-optimization/83086
3358 * gimple-ssa-store-merging.c
3359 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3360 rather than n.base_addr.
3362 2017-11-21 Martin Liska <mliska@suse.cz>
3364 PR rtl-optimization/82044
3365 PR tree-optimization/82042
3366 * dse.c (check_mem_read_rtx): Check for overflow.
3368 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3370 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3373 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3375 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3376 and its corresponding call arg location note.
3378 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3379 Andrew Burgess <andrew.burgess@embecosm.com>
3381 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3383 (arc_return_slot_offset): Likewise.
3384 (arc_eh_return_address_location): New declaration.
3385 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3386 (MUST_SAVE_REGISTER): Add exception handler case.
3387 (MUST_SAVE_RETURN_ADDR): Likewise.
3388 (arc_frame_pointer_required): Likewise.
3389 (arc_frame_pointer_needed): New function.
3390 (arc_compute_frame_size): Changed.
3391 (arc_expand_prologue): Likewise.
3392 (arc_expand_epilogue): Likewise.
3393 (arc_initial_elimination_offset): Likewise.
3394 (arc_return_slot_offset): Delete.
3395 (arc_eh_return_address_location): New function.
3396 (arc_builtin_setjmp_frame_value): Likewise.
3397 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3398 (EH_RETURN_STACKADJ_RTX): Define.
3399 (EH_RETURN_HANDLER_RTX): Likewise.
3400 * config/arc/arc.md (eh_return): Delete.
3402 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3404 * print-tree.h (debug_vec_tree): Remove prototype.
3405 * gdbinit.in (pvt): Remove macro.
3407 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3409 PR tree-optimization/83047
3410 * gimple-ssa-store-merging.c
3411 (imm_store_chain_info::output_merged_store): If the loads with the
3412 same vuse are in different basic blocks, for load_gsi pick a load
3413 location that is dominated by the other loads.
3416 * config/i386/i386.c (ix86_memmodel_check): Start
3417 -Winvalid-memory-model diagnostics with lowercase letter.
3420 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3421 set in_cold_section_p to the partition containing loc_list->first.
3422 When seeing loc_list->last_before_switch node, update secname and
3423 perform range_across_switch second partition handling only after that.
3426 * run-rtl-passes.c: Include debug.h.
3427 (run_rtl_passes): Call debug_hooks->assembly_start.
3428 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3432 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3433 OPTAB_DIRECT in calls to expand_simple_binop.
3435 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3438 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3440 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3443 * spellcheck.h (best_match::blithely_get_best_candidate): New
3446 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3448 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3449 with lower case letter.
3451 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3453 * config/i386/i386.md (bswaphi2): New expander.
3454 (*bswaphi2_movbe): New insn pattern.
3455 (bswaphi -> rorhi pepehole2): New peephole pattern.
3457 2017-11-20 Jeff Law <law@redhat.com>
3459 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3460 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3461 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3462 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3463 gimple-ssa-evrp-analyze.h.
3465 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3466 set BB_VISITED here.
3467 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3469 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3471 * tree-predcom.c: Add general comment on Store-Store chains.
3472 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3474 (get_chain_last_ref_at): Rename into...
3475 (get_chain_last_write_at): ...this.
3476 (get_chain_last_write_before_load): New function.
3477 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3478 CT_STORE_STORE when write reference is added.
3479 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3480 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3481 (initialize_root_vars_store_elim_1): Ditto.
3482 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3483 definition is created.
3484 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3485 chain by replacing it with dominant stored value.
3487 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3489 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3491 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3493 * vr-values.c (extract_range_from_binary_expr): Use a full range
3496 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3498 * config/arm/arm.md (R4_REGNUM): Define constant.
3499 (nonsecure_call_internal): Remove r4 clobber.
3500 (nonsecure_call_value_internal): Likewise.
3501 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3502 clobber and resequence match_operands.
3503 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3504 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3505 (nonsecure_call_value_reg_thumb2): Likewise.
3507 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3509 PR tree-optimization/78821
3510 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3511 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3512 to the address of the base rather than the base itself.
3513 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3514 (find_bswap_or_nop_finalize): New function.
3515 (find_bswap_or_nop): Use it.
3516 (bswap_replace): Return a tree rather than bool, change first
3517 argument from gimple * to gimple_stmt_iterator, allow inserting
3518 into an empty sequence, allow ins_stmt to be NULL - then emit
3519 all stmts into gsi. Fix up MEM_REF address gimplification.
3520 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3521 (struct store_immediate_info): Add N and INS_STMT non-static
3523 (store_immediate_info::store_immediate_info): Initialize them
3524 from newly added ctor args.
3525 (merged_store_group::apply_stores): Formatting fixes. Sort by
3527 (stmts_may_clobber_ref_p): For stores call also
3528 refs_anti_dependent_p.
3529 (gather_bswap_load_refs): New function.
3530 (imm_store_chain_info::try_coalesce_bswap): New method.
3531 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3532 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3533 (imm_store_chain_info::output_merged_store): Fail if number of
3534 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3535 and NOP_EXPR rhs_code.
3536 (pass_store_merging::process_store): Compute n and ins_stmt, if
3537 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3538 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3540 (pass_store_merging::execute): Calculate dominators.
3542 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3543 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3544 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3545 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3546 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3547 class pass_optimize_bswap, bswap_replace,
3548 pass_optimize_bswap::execute): Moved to ...
3549 * gimple-ssa-store-merging.c: ... this file.
3550 Include optabs-tree.h.
3551 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3552 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3553 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3554 anonymous namespace, remove static keywords.
3555 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3556 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3558 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3561 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3563 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3565 * vec.h (debug_helper): New function.
3566 (DEFINE_DEBUG_VEC): New macro.
3567 * hash-set.h (debug_helper): New function.
3568 (DEFINE_DEBUG_HASH_SET): New macro.
3569 * cfg.c (debug_slim (edge)): New function.
3570 Call DEFINE_DEBUG_VEC for edges.
3571 Call DEFINE_DEBUG_HASH_SET for edges.
3572 * cfghooks.c (debug_slim (basic_block)): New function.
3573 Call DEFINE_DEBUG_VEC for basic blocks.
3574 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3575 * print-tree.c (debug_slim): New function to handle trees.
3576 Call DEFINE_DEBUG_VEC for trees.
3577 Call DEFINE_DEBUG_HASH_SET for trees.
3578 (debug (vec<tree, va_gc>) &): Remove.
3579 (debug (<vec<tree, va_gc>) *): Remove.
3580 * print-rtl.c (debug_slim): New function to handle const_rtx.
3581 Call DEFINE_DEBUG_VEC for rtx_def.
3582 Call DEFINE_DEBUG_VEC for rtx_insn.
3583 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3584 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3585 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3586 (debug (vec<rtx_insn *> *ptr): Remove.
3587 (debug_insn_vector): Remove.
3588 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3590 2017-11-20 Tom de Vries <tom@codesourcery.com>
3592 PR rtl-optimization/82020
3593 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3594 IF_THEN_ELSE condition.
3596 2017-11-19 Jeff Law <law@redhat.com>
3598 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3599 of degenerates resulting from ignoring an edge.
3601 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3604 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3606 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3609 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3610 for uninitialied counts.
3612 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3615 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3617 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3620 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3622 2017-11-19 Tom de Vries <tom@codesourcery.com>
3624 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3627 2017-11-19 Tom de Vries <tom@codesourcery.com>
3629 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3632 2017-11-19 Tom de Vries <tom@codesourcery.com>
3634 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3637 2017-11-19 Tom de Vries <tom@codesourcery.com>
3639 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3641 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3643 2017-11-19 Tom de Vries <tom@codesourcery.com>
3645 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3647 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3648 semicolon after MCORE_EXPORT_NAME call.
3650 2017-11-19 Tom de Vries <tom@codesourcery.com>
3653 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3654 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3656 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3658 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3659 (cgraph_edge::clone): Cleanup updating of profile.
3660 * ipa-cp.c (update_profiling_info): Likewise.
3661 * ipa-inline-transform.c (inline_transform): Likewise.
3662 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3663 * ipa-split.c (execute_split_functions): Do not split when function
3665 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3666 * profile-count.c (profile_count::dump): Add global0.
3667 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3669 (profile_count::to_sreal_scale): Likewise.
3670 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3671 (profile_count::combine_with_ipa_count): New function.
3672 * profile-count.h (profile_guessed_global0adjusted): New.
3673 (profile_count::adjusted_zero): New.
3674 (profile_count::global0adjusted): New.
3675 (profile_count::combine_with_ipa_count): New.
3676 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3677 correct profile of return block of split functions.
3678 (copy_cfg_body): Remove unused profile_count.
3679 (copy_body): Likewise.
3680 (expand_call_inline): Update.
3681 (tree_function_versioning): Update.
3683 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3685 * hash-set.h (hash_set::empty): New.
3686 * tree-ssa-threadbackward.h: Delete.
3687 * tree-ssa-threadbackward.c (class thread_jumps): New.
3688 Move max_threaded_paths into class.
3689 (fsm_find_thread_path): Remove arguments that are now in class.
3690 (profitable_jump_thread_path): Rename to...
3691 (thread_jumps::profitable_jump_thread_path): ...this.
3692 (convert_and_register_jump_thread_path): Rename to...
3693 (thread_jumps::convert_and_register_current_path): ...this.
3694 (check_subpath_and_update_thread_path): Rename to...
3695 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3696 (register_jump_thread_path_if_profitable): Rename to...
3697 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3698 (handle_phi): Rename to...
3699 (thread_jumps::handle_phi): ...this.
3700 (handle_assignment): Rename to...
3701 (thread_jumps::handle_assignment): ...this.
3702 (fsm_find_control_statement_thread_paths): Rename to...
3703 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3704 (find_jump_threads_backwards): Rename to...
3705 (thread_jumps::find_jump_threads_backwards): ...this.
3706 Initialize path local data.
3707 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3708 from within thread_jumps class.
3709 (pass_early_thread_jumps::execute): Same.
3711 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3713 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3715 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3717 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3718 to canonical location.
3720 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3723 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3725 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3727 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3729 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3730 rcrt1.o%s/grcrt1.o%s for -static-pie.
3732 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3734 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3735 ix86_shift_rotate_cost): Break out from ...
3736 (ix86_rtx_costs): ... here.
3737 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3740 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3742 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3743 when entry block was promoted unlikely.
3744 (estimate_bb_frequencies): Increase frequency scale.
3745 * profile-count.h (profile_count): Export precision info.
3747 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3749 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3750 disturbin profile of entry block.
3752 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3754 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3757 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3759 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3760 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3762 2017-11-17 Jeff Law <law@redhat.com>
3764 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3765 from evrp_dom_walker class. Various methods moved into new class.
3766 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3767 (evrp_range_analyzer::enter): New method.
3768 (evrp_range_analyzer::leave): New method.
3769 (evrp_dom_walker): Remove delegators no longer needed by this class.
3770 Replace vr_values data member with evrp_range_analyzer
3772 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3773 method extracted from evrp_dom_walker::before_dom_children.
3774 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3775 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3777 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3778 Add private copy constructor and move assignment operators.
3779 Privatize methods and class data where trivially possible.
3780 (evrp_dom_walker::cleanup): New function, extracted from
3781 execute_early_vrp. Simplify access to class data.
3783 * vr-values.h (get_output_for_vrp): Prototype.
3784 * vr-values.c (get_output_for_vrp): New function extracted from
3785 vrp_visit_assignment_or_call and extract_range_from_stmt.
3786 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3788 2017-11-17 Luis Machado <luis.machado@linaro.org>
3790 * config/aarch64/aarch64.c
3791 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3792 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3793 tune_params::AUTOPREFETCHER_WEAK.
3795 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3798 * config/arm/arm.c (arm_valid_target_attribute_rec):
3799 Parse "arch=" and "+<ext>".
3800 (arm_valid_target_attribute_tree): Re-init global options.
3801 (arm_option_override): Make non-static.
3802 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3803 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3804 (__ARM_FEATURE_CRC32): Support undef.
3805 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3806 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3808 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3810 * gdbinit.in (break-on-diagnostic): New command.
3812 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3814 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3815 return to indirect jump for EH return if control-flow
3816 protection is enabled. Change explicit 'false' argument in
3817 pro_epilogue_adjust_stack with a value of
3819 * config/i386/i386.md (simple_return_indirect_internal):
3820 Remove SImode restriction to support 64-bit.
3822 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3824 * combine.c (added_notes_insn): New.
3825 (try_combine): Handle added_notes_insn like added_links_insn.
3826 Rewrite return value code.
3827 (distribute_notes): Set added_notes_insn to the earliest insn we added
3830 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3832 PR rtl-optimization/82621
3833 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3834 dest of one of those SETs is unused.
3836 2017-11-17 Richard Biener <rguenther@suse.de>
3839 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3840 * tree-pretty-print.c (dump_generic_node): Handle
3841 annot_expr_parallel_kind.
3842 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3843 * gimplify.c (gimple_boolify): Likewise.
3845 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3847 * config.gcc (extra_headers): Add cet.h for x86 targets.
3848 * config/i386/cet.h: New file.
3849 * doc/install.texi: Add --enable-cet/--disable-cet.
3851 2017-11-17 Richard Biener <rguenther@suse.de>
3853 PR tree-optimization/83017
3854 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3855 (gen_parallel_loop): Properly count iterations.
3856 (parallelize_loops): Handle loop->can_be_parallel independent
3857 of flag_loop_parallelize_all. Make static profitability test match
3859 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3860 * invoke.texi (parloops-min-per-thread): Document.
3862 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3864 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3865 upstreaming review comments.
3867 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3869 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3870 with fast unaligned access.
3871 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3873 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3875 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3876 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3878 2017-11-17 Richard Biener <rguenther@suse.de>
3880 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3881 folding of references.
3883 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3886 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3887 of replacing call to strncmp with corresponding call to strcmp when
3890 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3892 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3893 option prefer-avx256 for skylake-avx512 configuration.
3894 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3895 (get_builtin_code_for_version): Ditto.
3897 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3898 Monk Chiang <sh.chiang04@gmail.com>
3900 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3901 (FIXED_REGISTERS): Reserve more register numbers.
3902 (CALL_USED_REGISTERS): Likewise.
3903 (REG_ALLOC_ORDER): Likewise.
3904 (REG_CLASS_CONTENTS): Likewise.
3905 (REGISTER_NAMES): Likewise.
3907 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3908 Kito Cheng <kito.cheng@gmail.com>
3910 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3912 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3914 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3916 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3917 parameter list for vec_splats.
3919 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3921 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3923 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3925 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3928 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3929 with the required version of automake 1.11.6.
3931 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3933 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3935 (DU_C2_3_power9): Correct reservation combinations.
3936 (FP_DIV_power9, VEC_DIV_power9): New.
3937 (power9-alu): Split out rotate/shift...
3938 (power9-rot): ...to here, correct dispatch resource.
3939 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3941 (power9-fp): Correct latency.
3942 (power9-sdiv): Add div/sqrt resource.
3943 (power9-ddiv): Correct latency, add div/sqrt resource.
3944 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3945 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3947 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3949 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3951 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3952 switch statement mapping KF built-ins to TF built-ins if we don't
3953 have the proper ISA 3.0 assembler support.
3955 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3957 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3958 (lower_emutls_function_body): Do not compute it.
3960 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3962 * ipa-split.c (split_bb_info): Turn time to sreal.
3963 (split_point): Likewise.
3964 (dump_split_point): Likewise.
3965 (fine_split_points): Likewise.
3966 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3968 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3970 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3973 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3975 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3977 (expensive_function_p): Remove useless assert.
3978 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3980 2017-11-16 Martin Sebor <msebor@redhat.com>
3982 PR tree-optimization/82588
3983 PR tree-optimization/82583
3984 * tree-vrp.c (check_array_ref): Handle flexible array members,
3985 string literals, and inner indices.
3986 (search_for_addr_array): Add detail to diagnostics.
3988 2017-11-16 Nathan Sidwell <nathan@acm.org>
3992 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3993 (SET_DECL_ASSEMBLER_NAME): Forward to
3994 overwrite_decl_assembler_name.
3995 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3996 (overwrite_decl_assembler_name): Declare.
3997 * tree.c (overwrite_decl_assembler_name): New.
3998 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3999 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4000 (LANG_HOOKS_INITIALIZER): Add it.
4001 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4002 * langhooks.c (lhd_set_decl_assembler_name): Use
4003 SET_DECL_ASSEMBLER_NAME.
4004 (lhd_overwrite_decl_assembler_name): Default implementation.
4006 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4007 Jackson Woodruff <jackson.woodruff@arm.com>
4009 PR tree-optimization/71026
4010 * match.pd: Canonicalize constant multiplies in division.
4012 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4014 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4015 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4016 Remove OPT_fomit_frame_pointer entry.
4017 * common/config/arc/arc-common.c: Likewise.
4018 * common/config/arm/arm-common.c: Likewise.
4019 * common/config/avr/avr-common.c: Likewise.
4020 * common/config/c6x/c6x-common.c: Likewise.
4021 * common/config/cr16/cr16-common.c: Likewise.
4022 * common/config/cris/cris-common.c: Likewise.
4023 * common/config/epiphany/epiphany-common.c: Likewise.
4024 * common/config/fr30/fr30-common.c: Likewise.
4025 * common/config/frv/frv-common.c: Likewise.
4026 * common/config/ia64/ia64-common.c: Likewise.
4027 * common/config/iq2000/iq2000-common.c: Likewise.
4028 * common/config/lm32/lm32-common.c: Likewise.
4029 * common/config/m32r/m32r-common.c: Likewise.
4030 * common/config/mcore/mcore-common.c: Likewise.
4031 * common/config/microblaze/microblaze-common.c: Likewise.
4032 * common/config/mips/mips-common.c: Likewise.
4033 * common/config/mmix/mmix-common.c: Likewise.
4034 * common/config/mn10300/mn10300-common.c: Likewise.
4035 * common/config/nios2/nios2-common.c: Likewise.
4036 * common/config/pa/pa-common.c: Likewise.
4037 * common/config/pdp11/pdp11-common.c: Likewise.
4038 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4039 * common/config/riscv/riscv-common.c: Likewise.
4040 * common/config/rs6000/rs6000-common.c: Likewise.
4041 * common/config/rx/rx-common.c: Likewise.
4042 * common/config/s390/s390-common.c: Likewise.
4043 * common/config/sh/sh-common.c: Likewise.
4044 * common/config/sparc/sparc-common.c: Likewise.
4045 * common/config/tilegx/tilegx-common.c: Likewise.
4046 * common/config/tilepro/tilepro-common.c: Likewise.
4047 * common/config/v850/v850-common.c: Likewise.
4048 * common/config/visium/visium-common.c: Likewise.
4049 * common/config/xstormy16/xstormy16-common.c: Likewise.
4050 * common/config/xtensa/xtensa-common.c: Likewise.
4051 * invoke.texi (-fomit-frame-pointer): Update documentation.
4053 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4055 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4057 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4059 * cfg.c (scale_bbs_frequencies_int,
4060 cale_bbs_frequencies_gcov_type): Remove.
4061 * cfg.h (scale_bbs_frequencies_int,
4062 cale_bbs_frequencies_gcov_type): Remove.
4064 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4066 * tree-ssa-loop-manip.c
4067 (scale_dominated_blocks_in_loop): Update to profile counts.
4068 (tree_transform_and_unroll_loop): Likewise.
4070 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4072 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4073 scale_bbs_frequencies_int.
4075 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4077 * final.c (compute_alignments): Use counts rather than frequencies.
4079 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4081 * cfgloopanal.c: Include sreal.h
4082 (average_num_loop_insns): Use counts and sreal for accounting.
4084 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4086 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4089 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4090 Kito Cheng <kito.cheng@gmail.com>
4092 * config/nds32/constraints.md: Provide more constraints.
4093 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4094 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4095 support constraints usage.
4097 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4099 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4101 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4102 Kito Cheng <kito.cheng@gmail.com>
4104 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4105 * config/nds32/nds32.opt: Refine the layout.
4106 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4107 TARGET_EXT_STRING): Support new options.
4108 * config/nds32/nds32.h: Likewise.
4109 * config/nds32/nds32.md: Likewise.
4110 * config/nds32/nds32-predicates.c: Likewise.
4111 * config/nds32/constraints.md: Likewise.
4112 * common/config/nds32/nds32-common.c: Likewise.
4114 2017-11-16 Julia Koval <julia.koval@intel.com>
4117 * config/i386/gfniintrin.h: Add sse check.
4118 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4120 2017-11-16 Julia Koval <julia.koval@intel.com>
4122 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4123 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4124 (ix86_handle_option): Handle -mavx512vbmi2.
4125 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4126 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4127 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4128 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4129 (ix86_valid_target_attribute_inner_p): Ditto.
4130 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4131 * config/i386/i386.opt (mavx512vbmi2): New option.
4132 * doc/invoke.texi: Add new option.
4134 2017-11-16 Julia Koval <julia.koval@intel.com>
4136 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4137 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4138 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4139 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4140 _mm512_gf2p8mul_epi8): New intrinsics.
4141 * config/i386/i386-builtin-types.def
4142 (V64QI_FTYPE_V64QI_V64QI): New type.
4143 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4144 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4145 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4146 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4147 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4148 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4150 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4152 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4153 explicitly as a stream of bytes.
4155 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4157 * config/rs6000/altivec.h (vec_xst_be): New #define.
4158 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4159 and externalize from *altivec_vperm_<mode>_internal.
4160 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4162 (XL_BE_V8HI): Likewise.
4163 (XL_BE_V4SI): Likewise.
4164 (XL_BE_V4SI): Likewise.
4165 (XL_BE_V2DI): Likewise.
4166 (XL_BE_V4SF): Likewise.
4167 (XL_BE_V2DF): Likewise.
4168 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4169 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4170 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4171 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4172 VSX_BUILTIN_VEC_XST_BE.
4173 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4174 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4176 (altivec_init_builtins): Replace conditional calls to def_builtin
4177 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4178 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4179 with unconditional calls. Remove calls to def_builtin for
4180 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4181 __builtin_vec_xst_be.
4182 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4183 to define_expand, and add alternate RTL generation for P8.
4184 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4185 vsx_ld_elemrev_v8hi.
4186 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4187 add alternate RTL generation for P8.
4188 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4189 vsx_ld_elemrev_v16qi.
4190 (vsx_st_elemrev_v8hi): Convert define_insn
4191 to define_expand, and add alternate RTL generation for P8.
4192 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4193 vsx_st_elemrev_v8hi.
4194 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4195 add alternate RTL generation for P8.
4196 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4197 vsx_st_elemrev_v16qi.
4199 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4202 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4203 TARGET_AVX512ER check.
4204 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4205 neither -mzeroupper nor -mno-zeroupper is used and
4206 TARGET_EMIT_VZEROUPPER is set.
4207 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4208 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4210 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4212 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4213 folding of vector compares.
4214 (fold_build_vec_cmp): New helper function.
4215 (fold_compare_helper): New helper function.
4216 (builtin_function_type): Add compare builtins to the list of functions
4217 having unsigned arguments. Cosmetic updates to comment indentation.
4218 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4219 the not+eq combination.
4221 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4223 PR tree-optimization/82726
4224 PR tree-optimization/70754
4225 * tree-predcom.c (order_drefs_by_pos): New function.
4226 (combine_chains): Move code setting has_max_use_after to...
4227 (try_combine_chains): ...here. New parameter. Sort combined chains
4228 according to position information.
4229 (tree_predictive_commoning_loop): Update call to above function.
4230 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4232 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4234 PR tree-optimization/82726
4236 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4238 PR tree-optimization/70754
4239 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4240 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4241 combined stmt before it if not NULL.
4242 (combine_chains): Process refs reversely and compute dominance point
4246 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4248 PR tree-optimization/79663
4249 * tree-predcom.c (combine_chains): Process refs in reverse order
4250 only for ZERO length chains, and add explaining comment.
4252 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4254 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4255 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4256 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4258 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4260 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4262 2017-11-15 Martin Liska <mliska@suse.cz>
4264 * tree-cfg.c (pass_warn_function_return::execute):
4265 Compare warn_return_type for greater than zero.
4267 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4271 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4272 to return true on Xeon and not on Xeon Phi.
4273 (ix86_check_avx256_register): Changed to ...
4274 (ix86_check_avx_upper_register): ... this. Add extra check for
4275 VALID_AVX512F_REG_OR_XI_MODE.
4276 (ix86_avx_u128_mode_needed): Changed
4277 ix86_check_avx256_register to ix86_check_avx_upper_register.
4278 (ix86_check_avx256_stores): Changed to ...
4279 (ix86_check_avx_upper_stores): ... this. Changed
4280 ix86_check_avx256_register to ix86_check_avx_upper_register.
4281 (ix86_avx_u128_mode_after): Changed
4282 avx_reg256_found to avx_upper_reg_found. Changed
4283 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4284 (ix86_avx_u128_mode_entry): Changed
4285 ix86_check_avx256_register to ix86_check_avx_upper_register.
4286 (ix86_avx_u128_mode_exit): Ditto.
4287 * config/i386/i386.h: (host_detect_local_cpu): New define.
4289 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4291 * config/arm/xgene1.md (xgene1): Split into automatons
4292 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4293 (xgene1_f_load): Adjust reservations and/or types.
4294 (xgene1_f_store): Likewise.
4295 (xgene1_load_pair): Likewise.
4296 (xgene1_store_pair): Likewise.
4297 (xgene1_fp_load1): Likewise.
4298 (xgene1_load1): Likewise.
4299 (xgene1_store1): Likewise.
4300 (xgene1_move): Likewise.
4301 (xgene1_alu): Likewise.
4302 (xgene1_simd): Likewise.
4303 (xgene1_bfm): Likewise.
4304 (xgene1_neon_load1): Likewise.
4305 (xgene1_neon_store1): Likewise.
4306 (xgene1_neon_logic): Likewise.
4307 (xgene1_neon_st1): Likewise.
4308 (xgene1_neon_ld1r): Likewise.
4309 (xgene1_alu_cond): Added.
4310 (xgene1_shift_reg): Likwise.
4311 (xgene1_bfx): Likewise.
4312 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4314 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4317 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4319 (can_widen_mult_without_libcall): New function.
4320 (expand_mul_overflow): If only checking unsigned mul overflow,
4321 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4322 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4323 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4324 (expand_DIVMOD): Formatting fix.
4325 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4326 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4327 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4329 PR tree-optimization/82977
4330 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4331 constructed temporary to strlen_to_stridx.put.
4333 2017-11-15 Martin Liska <mliska@suse.cz>
4335 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4336 * configure: Regenerate.
4338 2017-11-15 Martin Liska <mliska@suse.cz>
4341 * config/sh/sh-mem.cc: Use proper probability for
4344 2017-11-14 Jeff Law <law@redhat.com>
4346 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4347 the red zone for stack_clash_protection_final_dynamic_probe targets
4348 when the total dynamic stack size is zero bytes.
4350 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4351 blocks is post order.
4353 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4355 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4356 * final.c (rest_of_clean_state): Set it for the
4357 -fcompare-debug dump.
4358 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4359 class when TDF_COMPARE_DEBUG is set.
4361 * dwarf2out.c (gen_producer_string): Discard
4364 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4367 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4368 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4369 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4370 (__TGMATH_REAL_2_3): Remove macros.
4371 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4372 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4374 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4376 (remquo): Define using __TGMATH_REAL_3.
4378 2017-11-14 Jeff Law <law@redhat.com>
4380 * vr-values.c: New file with contents extracted from tree-vrp.c.
4381 * Makefile.in (OBJS): Add vr-values.o
4382 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4383 (set_value_range, set_and_canonicalize_value_range): Likewise.
4384 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4385 (value_range_constant_singleton, symbolic_range_p): Likewise.
4386 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4387 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4388 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4389 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4390 (find_case_label_range, find_case_label_index): Likewise.
4391 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4392 (range_int_cst_singleton_p, value_inside_range): Likewise.
4393 (get_single_symbol): Likewise.
4394 (switch_update): Move structure definition here.
4395 (to_remove_edges, to_update_switch_stmts): Provide externs.
4396 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4397 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4398 (vrp_val_is_min, set_value_range): Likewise.
4399 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4400 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4401 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4402 (range_is_nonnull, range_int_cst_p): Likewwise.
4403 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4404 (get_single_symbol, operand_less_p): Likewise
4405 (compare_values_warnv, compare_values): Likewise.
4406 (value_inside_range, value_range_constant_singleton): Likewise.
4407 (zero_nonzero_bitgs_from_vr): Likewise.
4408 (extract_range_from_binary_expr_1): Likewise.
4409 (overflow_comparison_p): Likewise.
4410 (to_remove_edges, to_update_switch_stmts): Likewise.
4411 (find_case_label-index, find_case_label_range): Likewise.
4412 (switch_update, set_value_range_to_nonnegative): Remove.
4413 (set_value_range_to_truthvalue): Likewise.
4414 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4415 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4416 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4417 (find_case_label_ranges, test_for_singularity): Likewise.
4418 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4419 (x_vr_values): Move to its remaining use site.
4421 2017-11-10 Jeff Law <law@redhat.com>
4423 * vr-values.h (VR_INITIALIZER): Move #define here.
4424 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4425 * Makefile.in (OBJS): Add tree-evrp.o
4426 * tree-vrp.h (assert_info): Move structure definition here.
4427 (set_value_range_to_varying): Prototype.
4428 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4429 (infer_value_range, register_edge_assert_for): Likewise.
4430 (stmt_interesting_for_vrp): Likewise.
4431 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4432 (set_value_range_to_varying): No longer static.
4433 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4434 (infer_value_range, register_edge_assert_for): Likewise.
4436 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4438 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4439 generate the XXBRD instruction.
4441 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4442 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4443 as being compatible if -mabi=ieeelongdouble.
4444 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4445 to setup float128 built-ins with hardware support.
4446 (BU_FLOAT128_HW_2): Likewise.
4447 (BU_FLOAT128_HW_3): Likewise.
4448 (BU_FLOAT128_HW_VSX_1): Likewise.
4449 (BU_FLOAT128_HW_VSX_2): Likewise.
4450 (scalar_extract_expq): Change float128 built-in functions to
4451 accommodate having both KFmode and TFmode functions. Use the
4452 KFmode variant as the default.
4453 (scalar_extract_sigq): Likewise.
4454 (scalar_test_neg_qp): Likewise.
4455 (scalar_insert_exp_q): Likewise.
4456 (scalar_insert_exp_qp): Likewise.
4457 (scalar_test_data_class_qp): Likewise.
4458 (sqrtf128_round_to_odd): Delete processing the round to odd
4459 built-in functions as special built-in functions, and define them
4460 as float128 built-ins. Use the KFmode variant as the default.
4461 (truncf128_round_to_odd): Likewise.
4462 (addf128_round_to_odd): Likewise.
4463 (subf128_round_to_odd): Likewise.
4464 (mulf128_round_to_odd): Likewise.
4465 (divf128_round_to_odd): Likewise.
4466 (fmaf128_round_to_odd): Likewise.
4467 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4468 support for KFmode and TFmode xststdcqp calls.
4469 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4470 point, switch the built-in handlers for the get/set float128
4471 exponent, get float128 mantissa, float128 test built-ins, and the
4472 float128 round to odd built-in functions. Eliminate creating the
4473 float128 round to odd built-in functions as special built-ins.
4474 (rs6000_init_builtins): Eliminate special creation of the float128
4475 round to odd built-in functions.
4476 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4477 function insns to support both TFmode and KFmode variants.
4478 (xsxsigqp_<mode>): Likewise.
4479 (xsiexpqpf_<mode>): Likewise.
4480 (xsiexpqp_<mode>): Likewise.
4481 (xststdcqp_<mode>): Likewise.
4482 (xststdcnegqp_<mode>): Likewise.
4483 (xststdcqp_<mode>): Likewise.
4485 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4487 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4488 unused path_in_freq_ptr parameter.
4489 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4491 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4493 * ipa-inline.c (edge_badness): Dump sreal frequency.
4494 (compute_inlined_call_time): Match natural implementaiton ...
4495 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4496 forgotten division by CGRAPH_FREQ_BASE.
4498 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4500 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4501 Solaris 11. Update comment.
4502 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4504 * config/sol2.h (STARTFILE_SPEC): Likewise.
4505 * configure: Regenerate.
4507 2017-11-14 Carl Love <cel@us.ibm.com>
4509 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4510 le_ and be_ prefixes to swap* variables. Remove
4511 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4514 2017-11-14 Jason Merrill <jason@redhat.com>
4516 Support GTY((cache)) on hash_map.
4517 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4518 (ggc_cache_remove): Override it instead of ggc_mx.
4519 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4520 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4521 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4522 (simple_cache_map_traits): Override maybe_mx.
4523 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4524 (hash_map): Friend gt_cleare_cache.
4525 (gt_cleare_cache): New.
4526 * tree.h (tree_cache_traits): New hash_map traits class.
4527 (tree_cache_map): New typedef.
4529 2017-11-14 Richard Biener <rguenther@suse.de>
4531 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4532 paramter and handling.
4533 (cleanup_control_flow_bb): Likewise.
4534 (cleanup_control_flow_pre): New helper performing a DFS walk
4535 to call cleanup_control_flow_bb in PRE order.
4536 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4537 via cleanup_control_flow_pre.
4539 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4541 * config/aarch64/aarch64-simd.md
4542 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4543 (*aarch64_simd_bsl<mode>_alt): Likewise.
4544 (aarch64_simd_bsldi_internal): New.
4545 (aarch64_simd_bsldi_alt): Likewise.
4547 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4549 * tracer.c (better_p): Do not compare frequencies.
4550 * reg-stack.c (better_edge): Likewise.
4551 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4554 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4556 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4557 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4558 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4559 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4560 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4561 * predict.c (maybe_hot_frequency_p): Inline to ...
4562 (maybe_hot_count_p): ... here; rewrite to counts.
4563 (counts_to_freqs): Rename to ...
4564 (update_max_bb_count): ... this one.
4565 (expensive_function_p): Use counts.
4566 (estimate_bb_frequencies): Update.
4567 (rebuild_frequencies): Update.
4568 * predict.h (counts_to_freqs): Rename to ...
4569 (update_max_bb_count): ... this one.
4570 * profile.c (compute_branch_probabilities): Add debug info
4571 * tree-inline.c (expand_call_inline): Update debug info.
4572 (optimize_inline_calls): Use update_max_bb_count..
4573 (tree_function_versioning): Use update_max_bb_count..
4574 * value-prof.c (gimple_value_profile_transformations):
4575 Do not use update_max_bb_count.
4577 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4579 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4580 always use frequencies.
4582 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4584 * bb-reorder.c: Remove frequencies from comments.
4585 (better_edge_p): Use profile counts.
4586 (find_traces): Dump profile counts.
4587 (rotate_loop): Use profile counts.
4588 (find_traces_1_round): Likewise.
4589 (connect_better_edge_p): Use counts instead of probabilities for
4591 (copy_bb_p): Drop early check for non-0 frequency.
4592 (sanitize_hot_paths): Update comments.
4594 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4596 * ipa-split.c (struct split_point): Add count.
4597 (consider_split): Do not compute incoming frequency; compute incoming
4598 count and store it to split_point.
4599 (split_function): Set count of the call to split part correctly.
4601 2017-11-13 Carl Love <cel@us.ibm.com>
4603 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4605 2017-11-13 Tom Tromey <tom@tromey.com>
4607 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4609 2017-11-13 Carl Love <cel@us.ibm.com>
4611 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4612 Add support for builtins:
4613 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4614 vector {un,}signed {char,int,short},
4615 vector {un,}signed {char,int,short}) arguments.
4616 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4617 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4618 Add BU_P9V_AV_2 expansions for the builtins.
4619 * config/rs6000/altivec.h (vec_first_match_index,
4620 vec_first_mismatch_index, vec_first_match_or_eos_index,
4621 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4622 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4623 new extern declaration.
4624 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4625 * config/rs6000/vsx.md (first_match_index_<mode>,
4626 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4627 first_mismatch_or_eos_index_<mode>): Add define expand.
4628 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4629 * doc/extend.texi: Update the built-in documenation file for the new
4632 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4634 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4635 into the min/max operations for _Float<N> and _Float<N>X types.
4637 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4640 * dwarf2out.c (do_eh_frame): New static variable.
4641 (dwarf2out_begin_prologue): Set it.
4642 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4644 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4646 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4648 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4650 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4653 * debug.h (dwarf2out_do_eh_frame): Declare.
4654 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4655 (dwarf2out_do_frame): Use it.
4656 (dwarf2out_do_cfi_asm): Likewise.
4657 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4658 (dwarf2out_assembly_start): Likewise.
4659 (dwarf2out_begin_prologue): Fix comment.
4660 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4661 if the target needs either debug or unwind DWARF2 info.
4662 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4663 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4664 -ftrapping-math, -ftrapv and -fwrapv.
4666 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4668 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4669 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4670 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4671 (estimate_edge_size_and_time): Likewise.
4672 (ipa_merge_fn_summary_after_inlining): Likewise.
4673 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4674 (compute_uninlined_call_time): Use sreal_frequency.
4675 (compute_inlined_call_time): Likewise.
4676 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4677 * profile-count.c: Include sreal.h.
4678 (profile_count::to_sreal_scale): New.
4679 * profile-count.h: Forward declare sreal.
4680 (profile_count::to_sreal_scale): Declare.
4682 2017-11-13 Nathan Sidwell <nathan@acm.org>
4684 * diagnostic.c (maybe_line_and_column): New.
4685 (diagnostic_get_location_text): Use it.
4686 (diagnostic_report_current_module): Likewise.
4687 (test_diagnostic_get_location_text): Add tests.
4689 2017-11-13 Luis Machado <luis.machado@linaro.org>
4691 * doc/md.texi (Specifying processor pipeline description): Fix
4692 incorrect latency for the div instruction example.
4694 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4696 PR tree-optimization/78821
4697 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4698 that bit_not_p is the same.
4699 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4700 (split_group): Count precisely bit_not_p bits in each statement.
4701 (invert_op): New function.
4702 (imm_store_chain_info::output_merged_store): Use invert_op to
4703 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4704 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4706 2017-11-13 Martin Liska <mliska@suse.cz>
4708 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4709 (struct source_info): Likewise.
4710 (add_branch_counts): Likewise.
4711 (add_line_counts): Likewise.
4712 (function_summary): Likewise.
4713 (output_intermediate_line): Likewise.
4714 (generate_results): Likewise.
4716 2017-11-13 Martin Liska <mliska@suse.cz>
4718 * gcov.c (struct block_info): Remove typedef for block_t.
4719 (struct line_info): Likewise.
4720 (line_info::has_block): Likewise.
4721 (EXIT_BLOCK): Likewise.
4722 (unblock): Likewise.
4723 (circuit): Likewise.
4724 (get_cycles_count): Likewise.
4725 (process_file): Likewise.
4726 (read_graph_file): Likewise.
4727 (solve_flow_graph): Likewise.
4728 (find_exception_blocks): Likewise.
4729 (add_line_counts): Likewise.
4730 (accumulate_line_info): Likewise.
4731 (output_line_details): Likewise.
4733 2017-11-13 Martin Liska <mliska@suse.cz>
4735 * gcov.c (struct arc_info): Remove typedef for arc_t.
4736 (struct line_info): Likewise.
4737 (add_branch_counts): Likewise.
4738 (output_branch_count): Likewise.
4739 (function_info::~function_info): Likewise.
4740 (circuit): Likewise.
4741 (output_intermediate_line): Likewise.
4742 (read_graph_file): Likewise.
4743 (solve_flow_graph): Likewise.
4744 (find_exception_blocks): Likewise.
4745 (add_line_counts): Likewise.
4746 (accumulate_line_info): Likewise.
4747 (output_line_details): Likewise.
4748 (output_function_details): Likewise.
4750 2017-11-13 Martin Liska <mliska@suse.cz>
4752 * gcov.c (struct function_info): Remove typedef for function_t.
4753 (struct source_info): Likewise.
4754 (source_info::get_functions_at_location): Likewise.
4755 (solve_flow_graph): Likewise.
4756 (find_exception_blocks): Likewise.
4757 (add_line_counts): Likewise.
4758 (output_intermediate_file): Likewise.
4759 (process_file): Likewise.
4760 (generate_results): Likewise.
4761 (release_structures): Likewise.
4762 (read_graph_file): Likewise.
4763 (read_count_file): Likewise.
4764 (accumulate_line_counts): Likewise.
4765 (output_lines): Likewise.
4767 2017-11-13 Martin Liska <mliska@suse.cz>
4769 * gcov.c (function_info::function_info): Remove num_counts
4770 and add vector<gcov_type>.
4771 (function_info::~function_info): Use the vector.
4772 (process_file): Likewise.
4773 (read_graph_file): Likewise.
4774 (read_count_file): Likewise.
4775 (solve_flow_graph): Likewise.
4777 2017-11-13 Martin Liska <mliska@suse.cz>
4779 * gcov.c (function_info::is_artificial): New function.
4780 (process_file): Erase all artificial early.
4781 (generate_results): Skip as all artificial are already
4784 2017-11-13 Martin Liska <mliska@suse.cz>
4786 * gcov.c (read_graph_file): Store to global vector of functions.
4787 (read_count_file): Iterate the vector.
4788 (process_file): Likewise.
4789 (generate_results): Likewise.
4790 (release_structures): Likewise.
4792 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4794 PR tree-optimization/82954
4795 * gimple-ssa-store-merging.c
4796 (imm_store_chain_info::coalesce_immediate_stores): If
4797 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4799 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4801 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4802 Upddate call to ENDIAN_LANE_N.
4803 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4804 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4805 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4806 and use aarch64_endian_lane_rtx.
4807 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4809 2017-11-12 Tom de Vries <tom@codesourcery.com>
4811 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4813 2017-11-12 Tom de Vries <tom@codesourcery.com>
4815 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4817 2017-11-12 Tom de Vries <tom@codesourcery.com>
4819 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4821 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4822 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4823 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4825 2017-11-11 Martin Sebor <msebor@redhat.com>
4828 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4831 * prefic.c (translate_name): Replace strncpy with memcpy to
4832 avoid -Wstringop-truncation.
4834 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4836 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4838 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4840 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4842 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4844 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4847 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4849 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4850 enable generating XXBRH if the value is in a vector register.
4851 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4852 value is in a vector register.
4853 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4854 register to register bswap64's instead of doing the GPR sequence
4855 used on previous machines.
4856 (bswapdi2_xxbrd): New insn.
4857 (bswapdi2_reg): Disallow on ISA 3.0.
4858 (register to register bswap64 splitter): Do not split the insn on
4859 ISA 3.0 systems that use XXBRD.
4861 2017-11-10 Martin Sebor <msebor@redhat.com>
4864 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4866 (find_subframework_file): Same.
4868 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4870 * auto-profile.c (afdo_indirect_call): Drop frequency.
4871 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4872 (cgraph_node::create_edge): Drop frequency argument.
4873 (cgraph_node::create_indirect_edge): Drop frequency argument.
4874 (cgraph_edge::make_speculative): Drop frequency arguments.
4875 (cgraph_edge::resolve_speculation): Do not update frequencies
4876 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4877 (cgraph_node::dump): Check consistency in IPA mode.
4878 (cgraph_edge::maybe_hot_p): Use IPA counter.
4879 (cgraph_edge::verify_count_and_frequency): Rename to ...
4880 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4881 (cgraph_node::verify_node): Update.
4882 * cgraph.h (struct cgraph_edge): Drop frequency.
4883 (cgraph_edge::frequency): New function.
4884 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4886 (cgraph_edge::rebuild_edges): Likewise.
4887 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4888 (duplicate_thunk_for_node): Do not pass frequency.
4889 (cgraph_node::create_clone): Scale only counts.
4890 (cgraph_node::create_virtual_clone): Do not pass frequency.
4891 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4892 (cgraph_node::create_version_clone): Do not pass frequency.
4893 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4894 (cgraph_node::expand_thunk): Do not pass frequency.
4895 (cgraph_node::create_wrapper): Do not pass frequency.
4896 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4898 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4899 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4900 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4901 (gather_caller_stats): Use frequency function.
4902 (ipcp_cloning_candidate_p): Use frequency function.
4903 (ipcp_propagate_stage): Use frequency function.
4904 (get_info_about_necessary_edges): Use frequency function.
4905 (update_profiling_info): Update only IPA profile.
4906 (update_specialized_profile): Use frequency functoin.
4907 (perhaps_add_new_callers): Update only IPA profile.
4908 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4909 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4910 (dump_ipa_call_summary): Use frequency function.
4911 (estimate_edge_size_and_time): Use frequency function.
4912 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4913 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4914 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4915 (update_noncloned_counts): ... ths one; scale counts only.
4916 (clone_inlined_nodes): Do not scale frequency.
4917 (inline_call): Do not pass frequency.
4918 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4919 (compute_inlined_call_time): Use IPA profile.
4920 (want_inline_small_function_p): Use IPA profile.
4921 (want_inline_self_recursive_call_p): Use IPA profile.
4922 (edge_badness): Use IPA profile.
4923 (lookup_recursive_calls): Use IPA profile.
4924 (recursive_inlining): Do not pass frequency.
4925 (resolve_noninline_speculation): Do not update frequency.
4926 (inline_small_functions): Collect max of IPA profile.
4927 (dump_overall_stats): Dump IPA porfile.
4928 (dump_inline_stats): Dump IPA porfile.
4929 (ipa_inline): Collect IPA stats.
4930 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4931 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4932 (ipa_propagate_frequency): Use frequency function.
4933 (ipa_profile): Cleanup.
4934 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4935 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4936 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4937 (input_node): Do not stream frequency.
4938 (input_edge): Do not stream frequency.
4939 (merge_profile_summaries): Scale only IPA profiles.
4940 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4941 * predict.c (drop_profile): Do not recompute frequency.
4942 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4943 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4944 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4945 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4946 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4947 * tree-inline.c (copy_bb): Do not scale frequency.
4948 (expand_call_inline): Do not scale frequency.
4949 (tree_function_versioning): Do not scale frequency.
4950 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4952 2017-11-10 Julia Koval <julia.koval@intel.com>
4954 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4955 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4956 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4957 (_mm256_maskz_gf2p8affine_epi64_epi8)
4958 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4959 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4960 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4961 (__builtin_ia32_vgf2p8affineqb_v32qi)
4962 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4963 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4965 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4969 (arm_option_override): Refactor.
4970 (arm_option_reconfigure_globals): New.
4971 (arm_options_perform_arch_sanity_checks): New.
4972 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4974 (arm_options_perform_arch_sanity_checks): Likewise
4976 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4978 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4979 (power9-qpmul): New.
4980 * rs6000/rs6000.md ("type" attr): Add qmul.
4981 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4982 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4983 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4985 2017-11-10 Martin Sebor <msebor@redhat.com>
4988 * builtins.c (compute_objsize): Handle arrays that
4989 compute_builtin_object_size likes to fail for. Make extern.
4990 * builtins.h (compute_objsize): Declare.
4991 (check_strncpy_sizes): New function.
4992 (expand_builtin_strncpy): Call check_strncpy_sizes.
4993 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4994 -Wstringop-truncation.
4995 (gimple_fold_builtin_strncat): Same.
4996 * gimple.c (gimple_build_call_from_tree): Set call location.
4997 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4998 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4999 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5000 (handle_builtin_strlen): Use strlen_to_stridx.
5001 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5003 Use strlen_to_stridx.
5004 (pass_strlen::execute): Release strlen_to_stridx.
5005 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5006 (-Wstringop-truncation): Document new option.
5008 2017-11-10 Martin Liska <mliska@suse.cz>
5010 PR gcov-profile/82702
5011 * gcov.c (main): Handle intermediate files in a different
5013 (get_gcov_intermediate_filename): New function.
5014 (output_gcov_file): Remove support of intermediate files.
5015 (generate_results): Allocate intermediate file.
5016 (release_structures): Clean-up properly fn_end.
5017 (output_intermediate_file): Start iterating with line 1.
5019 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5021 PR tree-optimization/82929
5022 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5023 ops_swapped_p non-static data member.
5024 (store_immediate_info::store_immediate_info): Clear it.
5025 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5026 ops set ops_swapped_p.
5027 (count_multiple_uses): Handle ops_swapped_p.
5029 2017-11-10 Martin Liska <mliska@suse.cz>
5031 * coverage.c (coverage_init): Stream information about
5032 support of has_unexecuted_blocks.
5033 * doc/gcov.texi: Document that.
5034 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5035 * gcov.c (read_graph_file): Likewise.
5036 (output_line_beginning): Fix a small issue with
5039 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5041 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5042 reference of trivial component.
5044 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5047 * gimple-ssa-store-merging.c
5048 (pass_store_merging::terminate_all_aliasing_chains): For
5049 gimple_store_p stmts also call refs_output_dependent_p.
5051 PR rtl-optimization/82913
5052 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5055 2017-11-09 Jeff Law <law@redhat.com>
5057 * vr-values.h: New file with vr_values class.
5058 * tree-vrp.c: Include vr-values.h
5059 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5060 data objects into the vr_values class.
5061 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5062 (get_value_range): Make it a member function within vr_values class.
5063 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5064 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5065 (op_with_constant_singleton_value_range): Likewise.
5066 (extract_range_for_var_from_comparison_expr): Likewise.
5067 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5068 (extract_range_from_binary_expr): Likewise.
5069 (extract_range_from_unary_expr): Likewise.
5070 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5071 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5072 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5073 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5074 (compare_name_with_value, compare_names): Likewise.
5075 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5076 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5077 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5078 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5079 (extract_range_from_phi_node): Likewise.
5080 (simplify_truth_ops_using_ranges): Likewise.
5081 (simplify_div_or_mod_using_ranges): Likewise.
5082 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5083 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5084 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5085 (simplify_float_conversion_using_ranges): Likewise.
5086 (simplify_internal_call_using_ranges): Likewise.
5087 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5088 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5089 poitner via x_vr_values for calls into gimple folder.
5090 (vrp_initialize_lattice): Make this the vr_values ctor.
5091 (vrp_free_lattice): Make this the vr_values dtor.
5092 (set_vr_value): New function.
5093 (class vrp_prop): Add vr_values data member. Add various member
5094 functions as well as member functions that delegate to vr_values.
5095 (check_array_ref): Make a member function within vrp_prop class.
5096 (search_for_addr_array, vrp_initialize): Likewise.
5097 (vrp_finalize): Likewise. Revamp to avoid direct access to
5098 vr_value, values_propagated, etc.
5099 (check_array_bounds): Extract vrp_prop class instance pointer from
5100 walk info structure. Use it to call member functions.
5101 (check_all_array_refs): Make a member function within vrp_prop class.
5102 Smuggle class instance pointer via walk info structure.
5103 (x_vr_values): New local static.
5104 (vrp_valueize): Use x_vr_values to get class instance.
5105 (vr_valueize_1): Likewise.
5106 (class vrp_folder): Add vr_values data member. Add various member
5107 functions as well as member functions that delegate to vr_values.
5108 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5109 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5110 class instance from vr_values. Use it to call member functions.
5111 (vrp_dom_walker): Add vr_values data member.
5112 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5113 instance via x_vr_values.
5114 (identify_jump_threads): Accept vr_values as argument. Store
5115 it into the walker structure.
5116 (evrp_dom_walker): Add vr_values class data member. Add various
5118 (evrp_dom_walker::try_find_new_range): Use vr_values data
5119 member to access the memory allocator.
5120 (evrp_dom_walker::before_dom_children): Store vr_values class
5121 instance into the vrp_folder class.
5122 (evrp_dom_walker::push_value_range): Rework to avoid direct
5123 access to num_vr_values and vr_value.
5124 (evrp_dom_walker::pop_value_range): Likewise.
5125 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5126 Use vr_values to get to dump_all_value_ranges member function.
5127 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5128 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5129 Pass vr_values class instance down to identify_jump_threads.
5130 Remove call to vrp_free_lattice.
5131 (debug_all_value_ranges): Remove.
5133 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5134 (vrp_folder): Likewise.
5136 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5137 Get it from the existing bitmap instead.
5138 (vrp_intersect_ranges_1): Likewise.
5140 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5142 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5144 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5145 set bit_not_p to it.
5146 (imm_store_chain_info::coalesce_immediate_stores): Break group
5147 if bit_not_p is different.
5148 (count_multiple_uses, split_group,
5149 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5150 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5151 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5152 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5153 to store_immediate_info ctor.
5155 2017-11-09 Jim Wilson <jimw@sifive.com>
5157 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5158 (scan_prog_file): Likewise.
5160 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5162 * bb-reorder.c (max_entry_frequency): Remove.
5163 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5164 connect_traces, push_to_next_round_p): Remove prototypes.
5165 (find_traces_1_round): Use counts only.
5166 (push_to_next_round_p): Likewise.
5167 (find_traces): Likewise.
5168 (rotate_loop): Likewise.
5169 (find_traces_1_round): Likewise.
5170 (connect_traces): Likewise.
5171 (edge_order): Likewise.
5173 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5175 * config/arm/arm.c (output_return_instruction): Add comments to
5176 indicate requirement for cmse_nonsecure_entry return to account
5177 for the size of clearing instruction output here.
5178 (thumb_exit): Likewise.
5179 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5180 return in hardfloat mode.
5182 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5184 * config/rs6000/rs6000.c (machine_function): Add a bool,
5185 "toc_is_wrapped_separately".
5186 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5187 if it wasn't explicitly set or unset, we are optimizing for speed, and
5188 doing separate shrink-wrapping.
5189 (rs6000_get_separate_components): Enable the TOC component if
5190 saving the TOC register in the prologue.
5191 (rs6000_components_for_bb): Handle the TOC component.
5192 (rs6000_emit_prologue_components): Store the TOC register where needed.
5193 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5194 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5196 2017-11-09 Martin Jambor <mjambor@suse.cz>
5198 * ipa-param-manipulation.c: New file.
5199 * ipa-param-manipulation.h: Likewise.
5200 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5201 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5202 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5203 (ipa_parm_adjustment): Likewise.
5204 (ipa_parm_adjustment_vec): Likewise.
5205 (ipa_get_vector_of_formal_parms): Moved declaration to
5206 ipa-param-manipulation.h.
5207 (ipa_get_vector_of_formal_parm_types): Likewise.
5208 (ipa_modify_formal_parameters): Likewise.
5209 (ipa_modify_call_arguments): Likewise.
5210 (ipa_combine_adjustments): Likewise.
5211 (ipa_dump_param_adjustments): Likewise.
5212 (ipa_modify_expr): Likewise.
5213 (ipa_get_adjustment_candidate): Likewise.
5214 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5215 ipa-param-manipulation.c.
5216 (ipa_get_vector_of_formal_parm_types): Likewise.
5217 (ipa_modify_formal_parameters): Likewise.
5218 (ipa_modify_call_arguments): Likewise.
5219 (ipa_modify_expr): Likewise.
5220 (get_ssa_base_param): Likewise.
5221 (ipa_get_adjustment_candidate): Likewise.
5222 (index_in_adjustments_multiple_times_p): Likewise.
5223 (ipa_combine_adjustments): Likewise.
5224 (ipa_dump_param_adjustments): Likewise.
5225 * tree-sra.c: Also include ipa-param-manipulation.h
5226 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5229 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5230 Alan Hayward <alan.hayward@arm.com>
5231 David Sherwood <david.sherwood@arm.com>
5233 * doc/sourcebuild.texi (vect_masked_store): Document.
5235 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5236 Alan Hayward <alan.hayward@arm.com>
5237 David Sherwood <david.sherwood@arm.com>
5239 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5241 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5242 Alan Hayward <alan.hayward@arm.com>
5243 David Sherwood <david.sherwood@arm.com>
5245 * doc/sourcebuild.texi (vect_variable_length): Document.
5247 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5248 Alan Hayward <alan.hayward@arm.com>
5249 David Sherwood <david.sherwood@arm.com>
5251 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5253 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5254 Alan Hayward <alan.hayward@arm.com>
5255 David Sherwood <david.sherwood@arm.com>
5257 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5259 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5260 Alan Hayward <alan.hayward@arm.com>
5261 David Sherwood <david.sherwood@arm.com>
5263 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5264 previously undocumented selectors.
5265 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5267 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5269 * doc/rtl.texi (const_vector): Say that elements can be
5270 const_wide_ints too.
5271 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5272 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5273 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5274 * optabs.c (expand_vector_broadcast): Likewise.
5276 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5277 Alan Hayward <alan.hayward@arm.com>
5278 David Sherwood <david.sherwood@arm.com>
5280 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5281 scaled index even if the unscaled address was invalid.
5282 Don't increase the complexity of using a scale in that case.
5284 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5285 Alan Hayward <alan.hayward@arm.com>
5286 David Sherwood <david.sherwood@arm.com>
5288 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5289 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5290 UNITS_PER_WORD bytes.
5291 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5292 based on REGMODE_NATURAL_SIZE of the inner mode.
5293 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5294 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5295 * expmed.c (lowpart_bit_field_p): Divide the value up into
5296 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5297 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5298 whether something is likely to occupy more than one register.
5300 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5303 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5304 profile_count::adjust_for_ipa_scaling.
5305 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5306 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5308 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5310 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5312 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5313 (split_group): Add total_orig and total_new arguments, estimate the
5314 number of statements related to the store group without store merging
5315 and with store merging.
5316 (imm_store_chain_info::output_merged_store): Adjust split_group
5317 callers, punt if estimated number of statements with store merging
5318 is not smaller than estimated number of statements without it.
5320 (handled_load): Remove has_single_use checks.
5321 (pass_store_merging::process_store): Likewise.
5323 2017-11-09 Richard Biener <rguenther@suse.de>
5325 PR tree-optimization/82902
5326 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5328 2017-11-09 Martin Liska <mliska@suse.cz>
5331 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5334 2017-11-09 Martin Liska <mliska@suse.cz>
5336 PR tree-optimization/82669
5337 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5339 2017-11-09 Martin Liska <mliska@suse.cz>
5341 PR gcov-profile/48463
5342 * coverage.c (coverage_begin_function): Output also end locus
5343 of a function and information whether the function is
5345 * gcov-dump.c (tag_function): Parse and print the information.
5346 * gcov.c (INCLUDE_MAP): Add include.
5347 (INCLUDE_SET): Likewise.
5348 (struct line_info): Move earlier in the source file because
5349 of vector<line_info> in function_info structure.
5350 (line_info::line_info): Likewise.
5351 (line_info::has_block): Likewise.
5352 (struct source_info): Add new member index.
5353 (source_info::get_functions_at_location): New function.
5354 (function_info::group_line_p): New function.
5355 (output_intermediate_line): New function.
5356 (output_intermediate_file): Use the mentioned function.
5357 (struct function_start): New.
5358 (struct function_start_pair_hash): Likewise.
5359 (process_file): Add code that identifies group functions.
5360 Assign lines either to global or function scope.
5361 (generate_results): Skip artificial functions.
5362 (find_source): Assign index for each source file.
5363 (read_graph_file): Read new flag artificial and end_line.
5364 (add_line_counts): Assign it either to global of function scope.
5365 (accumulate_line_counts): Isolate core of the function to
5366 accumulate_line_info and call it for both function and global
5368 (accumulate_line_info): New function.
5369 (output_line_beginning): Fix GNU coding style.
5370 (print_source_line): New function.
5371 (output_line_details): Likewise.
5372 (output_function_details): Likewise.
5373 (output_lines): Iterate both source (global) scope and function
5375 (struct function_line_start_cmp): New class.
5376 * doc/gcov.texi: Reflect changes in documentation.
5378 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5381 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5382 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5383 and similarly for not instead of neg.
5385 2017-11-08 Andi Kleen <ak@linux.intel.com>
5387 * config/i386/i386.opt: Add -mforce-indirect-call.
5388 * config/i386/predicates.md: Check for flag_force_indirect_call.
5389 * doc/invoke.texi: Document -mforce-indirect-call
5391 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5393 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5395 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5396 config/riscv/riscv.c (predict.h): New include.
5397 (riscv_slow_unaligned_access_p): No longer static.
5398 (riscv_block_move_straight): Add require.
5399 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5401 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5404 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5405 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5406 nonimmediate_operand predicate for operand 1 instead of
5409 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5411 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5413 * config/aarch64/constraints.md (Uml): New constraint.
5414 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5417 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5419 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5420 of two vec_duplicates into a vec_concat.
5422 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5424 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5425 Simplify vec_merge of vec_duplicate and vec_concat.
5426 * config/aarch64/constraints.md (Utq): New constraint.
5427 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5430 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5432 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5433 Simplify vec_merge of vec_duplicate and const_vector.
5434 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5436 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5437 mode iterator. Update predicate on operand 1 to
5438 handle non-const_vec constants. Delete constraints.
5439 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5441 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5443 PR tree-optimization/78821
5444 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5446 (store_operand_info::store_operand_info): Initialize it to false.
5447 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5448 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5449 store in the group, and if chain_info is non-NULL, to ignore altogether
5451 (compatible_load_p): Fail if bit_not_p does not match.
5452 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5453 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5454 (pass_store_merging::process_store): Adjust
5455 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5456 call terminate_all_aliasing_chains newly when adding a store into
5457 a chain with non-NULL chain_info.
5459 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5461 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5463 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5465 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5467 (aarch64_layout_frame): Initialise emit_frame_chain.
5468 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5469 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5471 2017-11-08 Martin Liska <mliska@suse.cz>
5473 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5474 of gimple_call_internal_p.
5476 2017-11-07 Tom de Vries <tom@codesourcery.com>
5478 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5480 2017-11-07 Tom de Vries <tom@codesourcery.com>
5482 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5485 2017-11-08 Martin Liska <mliska@suse.cz>
5488 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5490 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5492 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5494 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5497 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5498 (*cmp<mode>_ccz_1): New insn with $k alternative.
5501 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5502 RTX_COMM_COMPARE as commutative as well.
5503 (ix86_binary_operator_ok): Formatting fix.
5504 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5505 *<code><mode>3<mask_name><round_saeonly_name>,
5506 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5507 *<s>mul<mode>3_highpart<mask_name>,
5508 *vec_widen_umult_even_v16si<mask_name>,
5509 *vec_widen_umult_even_v8si<mask_name>,
5510 *vec_widen_umult_even_v4si<mask_name>,
5511 *vec_widen_smult_even_v16si<mask_name>,
5512 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5513 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5514 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5515 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5516 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5517 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5518 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5519 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5520 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5521 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5522 ix86_binary_operator_ok. Formatting fixes.
5523 (*<plusminus_insn><mode>3<mask_name><round_name>,
5524 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5527 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5529 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5530 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5531 eq and ne if TARGET_ISEL.
5532 (cmp): New code_iterator.
5533 (UNS, UNSU_, UNSIK): New code_attrs.
5534 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5535 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5537 ("eq<mode>3"): ... this.
5538 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5540 ("ne<mode>3"): ... this.
5542 2017-11-07 Julia Koval <julia.koval@intel.com>
5545 * common/config/i386/i386-common.c
5546 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5547 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5548 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5549 * config/i386/i386.opt: Ditto.
5550 * config/i386/i386.c (ix86_target_string): Ditto.
5551 (ix86_option_override_internal): Ditto.
5552 (ix86_init_mpx_builtins): Move MPX to args2.
5553 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5554 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5555 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5556 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5557 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5558 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5559 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5561 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5564 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5565 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5566 (zero-extendsidi peephole2): Remove peephole.
5568 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5571 * stor-layout.c: Include attribs.h.
5572 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5573 explicit lookup of "aligned" attribute.
5575 2017-11-07 Andrew Waterman <andrew@sifive.com>
5577 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5578 (riscv_expand_block_move): Likewise.
5579 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5581 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5582 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5583 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5584 (riscv_adjust_block_mem): Likewise.
5585 (riscv_block_move_loop): Likewise.
5586 (riscv_expand_block_move): Likewise.
5587 * config/riscv/riscv.md (movmemsi): New pattern.
5589 2017-11-07 Michael Clark <michaeljclark@mac.com>
5591 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5592 (MUSL_DYNAMIC_LINKER): Likewise.
5594 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5596 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5597 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5600 2017-11-07 Richard Biener <rguenther@suse.de>
5602 * match.pd: Fix build.
5604 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5605 Jackson Woodruff <jackson.woodruff@arm.com>
5607 PR tree-optimization/71026
5608 * match.pd: Canonicalize negate in division.
5610 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5613 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5615 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5617 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5618 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5620 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5622 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5623 non-scalar integral types.
5624 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5625 (-(A-B), -(~A)): New transformations.
5627 2017-11-07 Tom de Vries <tom@codesourcery.com>
5629 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5630 semicolon after "do {} while (0)".
5631 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5632 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5633 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5634 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5635 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5637 2017-11-07 Tom de Vries <tom@codesourcery.com>
5639 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5640 after "do {} while (0)".
5641 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5642 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5643 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5644 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5645 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5647 2017-11-07 Tom de Vries <tom@codesourcery.com>
5650 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5652 (arm_rtx_costs_internal): Add missing semicolon after
5653 HANDLE_NARROW_SHIFT_ARITH call.
5655 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5658 disable isel if it was not set explicitly.
5660 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5662 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5663 (add_type_attribute) likewise.
5665 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5667 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5668 of struct ix86_frame.
5669 (ix86_initial_elimination_offset): Likewise.
5670 (ix86_expand_split_stack_prologue): Likewise.
5672 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5674 * tree-vrp.h (enum value_range_type): Update stale comment.
5676 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5677 Alan Hayward <alan.hayward@arm.com>
5678 David Sherwood <david.sherwood@arm.com>
5680 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5681 (aarch64_expand_vec_perm_const): Take the number of units too.
5682 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5683 (aarch64_expand_vec_perm_const): Likewise.
5684 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5685 (vec_perm<mode>): Update accordingly.
5687 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5688 Alan Hayward <alan.hayward@arm.com>
5689 David Sherwood <david.sherwood@arm.com>
5691 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5692 Take the number of units too.
5693 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5694 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5695 but check for a vector mode before rather than after the call.
5696 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5697 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5698 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5699 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5700 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5701 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5702 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5703 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5704 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5705 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5706 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5707 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5708 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5709 (aarch64_sqdmull2_n<mode>): Update accordingly.
5711 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5712 Alan Hayward <alan.hayward@arm.com>
5713 David Sherwood <david.sherwood@arm.com>
5715 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5716 the number of units too.
5717 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5718 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5719 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5720 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5721 (vec_store_lanesxi<mode>): Update accordingly.
5723 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5724 Alan Hayward <alan.hayward@arm.com>
5725 David Sherwood <david.sherwood@arm.com>
5727 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5728 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5729 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5730 of units rather than the mode.
5731 * config/aarch64/iterators.md (nunits): New mode attribute.
5732 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5733 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5734 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5735 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5736 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5737 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5738 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5739 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5740 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5741 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5742 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5743 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5744 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5745 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5746 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5747 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5748 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5749 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5750 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5751 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5752 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5753 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5754 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5755 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5756 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5757 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5758 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5759 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5760 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5761 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5762 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5763 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5764 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5765 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5766 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5767 (aarch64_simd_vec_setv2di): Likewise.
5769 2017-11-06 Carl Love <cel@us.ibm.com>
5771 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5773 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5774 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5776 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5778 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5779 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5780 Add power 8 macro expansions.
5781 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5782 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5783 power 8 instructions. (VSX_XXBR): Add iterator.
5785 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5787 * config/arm/arm.md (predicable_short_it): Change default to "no",
5788 improve documentation, remove uses that are identical to the default.
5789 (enabled_for_depr_it): Rename to enabled_for_short_it.
5790 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5791 Remove default uses.
5792 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5793 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5794 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5795 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5797 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5800 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5801 float128 helper macros, which are no longer used after deleting
5802 the old 'q' built-in functions, and moving the round to odd
5803 built-in functions to being special built-in functions.
5804 (BU_FLOAT128_2): Likewise.
5805 (BU_FLOAT128_1_HW): Likewise.
5806 (BU_FLOAT128_2_HW): Likewise.
5807 (BU_FLOAT128_3_HW): Likewise.
5808 (FABSQ): Delete old 'q' built-in functions.
5809 (COPYSIGNQ): Likewise.
5810 (SQRTF128_ODD): Move round to odd built-in functions to be
5811 special built-in functions, so that we can handle
5812 -mabi=ieeelongdouble.
5813 (TRUNCF128_ODD): Likewise.
5814 (ADDF128_ODD): Likewise.
5815 (SUBF128_ODD): Likewise.
5816 (MULF128_ODD): Likewise.
5817 (DIVF128_ODD): Likewise.
5818 (FMAF128_ODD): Likewise.
5819 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5820 built-in names to 'f128'.
5821 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5822 old 'q' built-in functions, as the machine independent code for
5823 'f128' built-in functions handles this.
5824 (rs6000_expand_builtin): Add expansion for float128 round to odd
5825 functions, keying off on -mabi=ieeelongdouble of whether to use
5826 the KFmode or TFmode variant.
5827 (rs6000_init_builtins): Initialize the _Float128 round to odd
5829 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5830 _Float128 'q' built-in functions are now mapped into the new
5831 'f128' built-in functions.
5833 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5835 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5836 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5838 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5841 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5842 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5843 * toplev.c: Include "ipa-fnsummary.h".
5844 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5846 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5848 PR tree-optimization/82838
5849 * gimple-ssa-store-merging.c
5850 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5851 on a separate gimple_seq which is then appended to seq.
5853 2017-11-06 Jeff Law <law@redhat.com>
5856 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5857 (get_probe_interval): New functions.
5858 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5859 (ix86_adjust_stack_and_probe): Likewise.
5860 (output_adjust_stack_and_probe): Likewise.
5861 (ix86_emit_probe_stack_range): Likewise.
5862 (ix86_expand_prologue): Likewise.
5864 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5866 PR tree-optimization/82816
5867 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5868 if the modes of the two types are the same.
5869 (convert_plusminus_to_widen): Likewise.
5871 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5873 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5875 (usadv16qi): New define_expand.
5876 (usadv8hi): New define_expand.
5878 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5881 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5883 (inline_transform): Likewise.
5884 * predict.c (counts_to_freqs): Remove useless conditional.
5885 * profile-count.h (profile_count::apply_scale): Move sanity check.
5886 * tree-inline.c (copy_bb): Always scale.
5887 (copy_cfg_body): Likewise.
5889 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5892 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5894 (*cmp_ite0): Add enabled_for_depr_it attribute.
5895 (*cmp_ite1): Likewise.
5897 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5899 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5902 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5904 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5905 return the value by pointer.
5906 (extract_range_from_multiplicative_op_1): Update accordingly.
5907 Return as soon as an operation fails.
5909 2017-11-05 Tom de Vries <tom@codesourcery.com>
5912 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5913 (DEF_SANITIZER_BUILTIN): ... here.
5914 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5915 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5917 2017-11-05 Tom de Vries <tom@codesourcery.com>
5920 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5922 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5923 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5924 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5926 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5927 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5928 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5930 2017-11-05 Tom de Vries <tom@codesourcery.com>
5933 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5936 2017-11-04 Michael Clark <michaeljclark@mac.com>
5938 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5939 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5941 (*addsi3_extended): Likewise.
5942 (*addsi3_extended2): Likewise.
5943 (<optab>si3): Likewise.
5944 (<optab>di3): Likewise.
5945 (<optab><mode>3): Likewise.
5946 (<*optabe>si3_internal): Likewise.
5947 (zero_extendqi<SUPERQI:mode>2): Likewise.
5948 (*add<mode>hi3): Likewise.
5949 (*xor<mode>hi3): Likewise.
5950 (<optab>di3): Likewise.
5951 (*<optab>si3_extend): Likewise.
5952 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5953 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5954 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5956 2017-11-04 Andrew Waterman <andrew@sifive.com>
5958 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5959 TARGET_STRICT_ALIGN based upon -mtune argument.
5961 2017-11-04 Andrew Waterman <andrew@sifive.com>
5963 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5965 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5967 * config/i386/i386.c (choose_basereg): Use optional scratch
5968 register and add assertion.
5969 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5970 needed, and don't allocate stack.
5971 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5972 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5973 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5975 2017-11-03 Jeff Law <law@redhat.com>
5977 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5978 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5979 to probe at the start of a noreturn function.
5981 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5983 PR tree-optimization/78821
5984 * gimple-ssa-store-merging.c: Update the file comment.
5985 (MAX_STORE_ALIAS_CHECKS): Define.
5986 (struct store_operand_info): New type.
5987 (store_operand_info::store_operand_info): New constructor.
5988 (struct store_immediate_info): Add rhs_code and ops data members.
5989 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5990 and op1r arguments to the ctor, initialize corresponding data members.
5991 (struct merged_store_group): Add load_align_base and load_align
5993 (merged_store_group::merged_store_group): Initialize them.
5994 (merged_store_group::do_merge): Update them.
5995 (merged_store_group::apply_stores): Pick the constant for
5996 encode_tree_to_bitpos from one of the two operands, or skip
5997 encode_tree_to_bitpos if neither operand is a constant.
5998 (class pass_store_merging): Add process_store method decl. Remove
5999 bool argument from terminate_all_aliasing_chains method decl.
6000 (pass_store_merging::terminate_all_aliasing_chains): Remove
6001 var_offset_p argument and corresponding handling.
6002 (stmts_may_clobber_ref_p): New function.
6003 (compatible_load_p): New function.
6004 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6005 if there is overlap and rhs_code is not INTEGER_CST. For
6006 non-overlapping stores terminate group if rhs is not mergeable.
6007 (get_alias_type_for_stmts): Change first argument from
6008 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6009 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6010 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6012 (get_location_for_stmts): Change first argument from
6013 auto_vec<gimple *> & to vec<gimple *> &.
6014 (struct split_store): Remove orig_stmts data member, add orig_stores.
6015 (split_store::split_store): Create orig_stores rather than orig_stmts.
6016 (find_constituent_stmts): Renamed to ...
6017 (find_constituent_stores): ... this. Change second argument from
6018 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6019 to info structures rather than the statements.
6020 (split_group): Rename ALLOW_UNALIGNED argument to
6021 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6022 it. Adjust find_constituent_stores caller.
6023 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6024 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6025 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6026 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6027 (mem_valid_for_store_merging): New function.
6028 (handled_load): New function.
6029 (pass_store_merging::process_store): New method.
6030 (pass_store_merging::execute): Use process_store method. Adjust
6031 terminate_all_aliasing_chains caller.
6033 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6035 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6036 Return true for more constants, symbols and label references.
6037 (aarch64_valid_floating_const): Remove unused function.
6039 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6042 * config/aarch64/aarch64.c (aarch64_layout_frame):
6043 Undo forcing of LR at bottom of frame.
6045 2017-11-03 Jeff Law <law@redhat.com>
6048 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6049 for int_registers_saved.
6051 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6052 extracted from tree-ssa-dom.c.
6053 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6054 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6055 (record_equivalences_from_incoming_edge): Add additional argument
6056 to single_pred_edge_ignoring_loop_edges call.
6057 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6058 (uncprop_dom_walker::before_dom_children): Add additional argument
6059 to single_pred_edge_ignoring_loop_edges call.
6060 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6061 single_pred_edge_ignoring_loop_edges rather than open coding.
6062 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6064 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6066 * match.pd (-(-A)): Rewrite.
6068 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6070 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6071 (rs6000_emit_int_cmove): New declaration.
6072 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6073 (rs6000_emit_sISEL): Delete.
6074 (rs6000_emit_int_cmove): Make non-static.
6075 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6076 instead of rs6000_emit_sISEL.
6078 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6080 * asan.c (create_cond_insert_point): Maintain profile.
6081 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6083 * basic-block.h (struct basic_block_def): Remove frequency.
6084 (EDGE_FREQUENCY): Use to_frequency
6085 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6087 (find_traces): Update to use to_frequency.
6088 (find_traces_1_round): Likewise; use only IPA counts.
6089 (bb_to_key): Likewise.
6090 (connect_traces): Use IPA counts only.
6091 (copy_bb_p): Update to use to_frequency.
6092 (fix_up_crossing_landing_pad): Likewise.
6093 (sanitize_hot_paths): Likewise.
6094 * bt-load.c (basic_block_freq): Likewise.
6095 * cfg.c (init_flow): Set count_max to uninitialized.
6096 (check_bb_profile): Remove frequencies; check counts.
6097 (dump_bb_info): Do not dump frequencies.
6098 (update_bb_profile_for_threading): Update counts only.
6099 (scale_bbs_frequencies_int): Likewise.
6100 (MAX_SAFE_MULTIPLIER): Remove.
6101 (scale_bbs_frequencies_gcov_type): Update counts only.
6102 (scale_bbs_frequencies_profile_count): Update counts only.
6103 (scale_bbs_frequencies): Update counts only.
6104 * cfg.h (struct control_flow_graph): Add count-max.
6105 (update_bb_profile_for_threading): Update prototype.
6106 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6107 (find_many_sub_basic_blocks): Likewise.
6108 * cfgcleanup.c (try_forward_edges): Likewise.
6109 (try_crossjump_to_edge): Likewise.
6110 * cfgexpand.c (expand_gimple_cond): Likewise.
6111 (expand_gimple_tailcall): Likewise.
6112 (construct_init_block): Likewise.
6113 (construct_exit_block): Likewise.
6114 * cfghooks.c (verify_flow_info): Check consistency of counts.
6115 (dump_bb_for_graph): Do not dump frequencies.
6116 (split_block_1): Do not update frequencies.
6117 (split_edge): Do not update frequencies.
6118 (make_forwarder_block): Do not update frequencies.
6119 (duplicate_block): Do not update frequencies.
6120 (account_profile_record): Do not update frequencies.
6121 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6122 for global heuristics.
6123 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6124 (expected_loop_iterations_unbounded): Use counts only.
6125 * cfgloopmanip.c (scale_loop_profile): Simplify.
6126 (create_empty_loop_on_edge): Simplify
6128 (duplicate_loop_to_header_edge): Simplify
6129 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6130 (update_br_prob_note): Take care of removing note when profile
6132 (relink_block_chain): Do not dump frequency.
6133 (rtl_account_profile_record): Use to_frequency.
6134 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6135 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6136 (cgraph_update_edges_for_call_stmt_node): Likewise.
6137 (cgraph_edge::verify_count_and_frequency): Update.
6138 (cgraph_node::verify_node): Temporarily disable frequency verification.
6139 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6140 to_cgraph_frequency.
6141 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6142 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6144 (cgraph_node::expand_thunk): Update profile.
6145 * except.c (dw2_build_landing_pads): Do not update frequency.
6146 * final.c (compute_alignments): Use to_frequency.
6147 (dump_basic_block_info): Do not dump frequency.
6148 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6149 (dump_gimple_bb_header): Do not dump frequency.
6150 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6152 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6153 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6154 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6155 (init_before_recovery): Do not update frequency.
6156 (sched_create_recovery_edges): Do not update frequency.
6157 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6158 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6159 (ipa_cp_c_finalize): Set max_count to uninitialized.
6160 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6161 (param_change_prob): Use counts.
6162 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6164 * ipa-split.c (consider_split): Use to_frequency.
6165 (split_function): Use to_frequency.
6166 * ira-build.c (loop_compare_func): Likewise.
6167 (mark_loops_for_removal): Likewise.
6168 (mark_all_loops_for_removal): Likewise.
6169 * loop-doloop.c (doloop_modify): Do not update frequency.
6170 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6172 * lto-streamer-in.c (input_function): Update count_max.
6173 * omp-expand.c (expand_omp_taskreg): Update count_max.
6174 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6175 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6176 (maybe_hot_count_p): Use ipa counts only.
6177 (maybe_hot_bb_p): Simplify.
6178 (maybe_hot_edge_p): Simplify.
6179 (probably_never_executed): Do not take frequency argument.
6180 (probably_never_executed_bb_p): Do not pass frequency.
6181 (probably_never_executed_edge_p): Likewise.
6182 (combine_predictions_for_bb): Check that profile is nonzero.
6183 (propagate_freq): Do not set frequency.
6184 (drop_profile): Simplify.
6185 (counts_to_freqs): Simplify.
6186 (expensive_function_p): Use to_frequency.
6187 (propagate_unlikely_bbs_forward): Simplify.
6188 (determine_unlikely_bbs): Simplify.
6189 (estimate_bb_frequencies): Add hack to silence graphite issues.
6190 (compute_function_frequency): Use ipa counts.
6191 (pass_profile::execute): Update.
6192 (rebuild_frequencies): Use counts only.
6193 (force_edge_cold): Use counts only.
6194 * profile-count.c (profile_count::dump): Dump new count types.
6195 (profile_count::differs_from_p): Check compatiblity.
6196 (profile_count::to_frequency): New function.
6197 (profile_count::to_cgraph_frequency): New function.
6198 * profile-count.h (struct function): Declare.
6199 (enum profile_quality): Add profile_guessed_local and
6200 profile_guessed_global0.
6201 (class profile_proability): Decrease number of bits to 29;
6202 update from_reg_br_prob_note and to_reg_br_prob_note.
6203 (class profile_count: Update comment; decrease number of bits
6204 to 61. Check compatibility.
6205 (profile_count::compatible_p): New private member function.
6206 (profile_count::ipa_p): New member function.
6207 (profile_count::operator<): Handle global zero correctly.
6208 (profile_count::operator>): Handle global zero correctly.
6209 (profile_count::operator<=): Handle global zero correctly.
6210 (profile_count::operator>=): Handle global zero correctly.
6211 (profile_count::nonzero_p): New member function.
6212 (profile_count::force_nonzero): New member function.
6213 (profile_count::max): New member function.
6214 (profile_count::apply_scale): Handle IPA scalling.
6215 (profile_count::guessed_local): New member function.
6216 (profile_count::global0): New member function.
6217 (profile_count::ipa): New member function.
6218 (profile_count::to_frequency): Declare.
6219 (profile_count::to_cgraph_frequency): Declare.
6220 * profile.c (OVERLAP_BASE): Delete.
6221 (compute_frequency_overlap): Delete.
6222 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6223 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6224 * sched-ebb.c (rank): Use counts only.
6225 * shrink-wrap.c (handle_simple_exit): Use counts only.
6226 (try_shrink_wrapping): Use counts only.
6227 (place_prologue_for_one_component): Use counts only.
6228 * tracer.c (find_best_predecessor): Use to_frequency.
6229 (find_trace): Use to_frequency.
6230 (tail_duplicate): Use to_frequency.
6231 * trans-mem.c (expand_transaction): Do not update frequency.
6232 * tree-call-cdce.c: Do not update frequency.
6233 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6234 (gimple_merge_blocks): Likewise.
6235 (gimple_split_edge): Likewise.
6236 (gimple_duplicate_sese_region): Likewise.
6237 (gimple_duplicate_sese_tail): Likewise.
6238 (move_sese_region_to_fn): Likewise.
6239 (gimple_account_profile_record): Likewise.
6240 (insert_cond_bb): Likewise.
6241 * tree-complex.c (expand_complex_div_wide): Likewise.
6242 * tree-eh.c (lower_resx): Update profile.
6243 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6245 (initialize_cfun): Do not initialize frequencies
6246 (freqs_to_counts): Delete.
6247 (copy_cfg_body): Ignore count parameter.
6248 (copy_body): Update.
6249 (expand_call_inline): Update count_max.
6250 (optimize_inline_calls): Update count_max.
6251 (tree_function_versioning): Update count_max.
6252 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6253 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6255 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6256 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6257 (try_peel_loop): Likewise.
6258 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6260 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6261 (tree_transform_and_unroll_loop): Do not use frequencies
6262 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6263 Use reliable prediction only.
6264 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6265 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6266 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6267 probability scaling.
6268 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6269 not update frequency
6270 (any_remaining_duplicated_blocks): Likewise.
6271 (update_profile): Likewise.
6272 (estimated_freqs_path): Delete.
6273 (freqs_to_counts_path): Delete.
6274 (clear_counts_path): Delete.
6275 (ssa_fix_duplicate_block_edges): Likewise.
6276 (duplicate_thread_path): Likewise.
6277 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6278 * tree-tailcall.c (decrease_profile): Do not update frequency.
6279 (eliminate_tail_call): Likewise.
6280 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6281 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6282 (optimize_mask_stores): Likewise.
6283 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6284 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6285 (ubsan_expand_ptr_ifn): Update profile.
6286 * value-prof.c (gimple_ic): Simplify.
6287 * value-prof.h (gimple_ic): Update prototype.
6288 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6289 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6291 (want_inline_self_recursive_call_p): Likewise.
6292 (resolve_noninline_speculation): Only cummulate defined counts.
6293 (inline_small_functions): Use nonzero_p.
6294 (ipa_inline): Do not access freed node.
6296 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6298 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6299 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6301 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6303 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6304 non-legitimate address.
6306 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6308 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6309 (*lt0_<mode>di, *lt0_<mode>si): New.
6311 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6313 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6314 TARGET_PAIRED_FLOAT.
6316 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6317 Jim Wilson <jim.wilson@linaro.org>
6319 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6320 * config/aarch64/aarch64-tune.md: Regenerated.
6321 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6322 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6324 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6326 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6327 (arc_expand_prologue): Restore blink for millicode.
6328 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6330 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6333 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6334 gen_vec_duplicate after forcing the scalar into a register.
6336 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6338 * combine (try_combine): Print the insns input to try_combine to the
6341 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6344 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6345 Remove second argument from aarch64_process_target_attr call.
6346 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6348 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6350 (aarch64_handle_attr_arch): Remove second argument.
6351 (aarch64_handle_attr_cpu): Ditto.
6352 (aarch64_handle_attr_tune): Ditto.
6353 (aarch64_handle_attr_isa_flags): Ditto.
6354 (aarch64_process_one_target_attr): Ditto.
6355 (aarch64_process_target_attr): Ditto.
6356 (aarch64_option_valid_attribute_p): Remove second argument.
6357 on aarch64_process_target_attr call.
6359 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6361 * diagnostic.c: Include "selftest-diagnostic.h".
6362 (selftest::assert_location_text): New function.
6363 (selftest::test_diagnostic_get_location_text): New function.
6364 (selftest::diagnostic_c_tests): Call it.
6366 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6368 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6369 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6370 (class selftest::test_diagnostic_context): Move to...
6371 * selftest-diagnostic.c: New file.
6372 * selftest-diagnostic.h: New file.
6374 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6376 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6377 offset range for FT32B.
6378 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6379 * config/ft32/ft32.md: Add TARGET_NOPM.
6380 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6381 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6383 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6385 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6387 2017-11-02 Jeff Law <law@redhat.com>
6389 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6390 virtual keyword on FINAL OVERRIDE members.
6392 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6393 virtuals together. Add virtual destructor.
6394 (substitute_and_fold_engine): Similarly.
6396 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6398 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6400 2017-11-02 Richard Biener <rguenther@suse.de>
6402 PR tree-optimization/82795
6403 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6405 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6407 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6409 (gcc_GAS_CHECK_FEATURE): Remove.
6410 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6411 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6412 * configure: Regenerate.
6414 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6416 * config/arc/arc.c (hwloop_optimize): Account for empty
6419 2017-11-02 Richard Biener <rguenther@suse.de>
6422 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6423 Truncate ARRAY_REF index and element size.
6425 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6427 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6429 2017-11-01 Jeff Law <law@redhat.com>
6431 * tree-ssa-ccp.c (ccp_folder): New class derived from
6432 substitute_and_fold_engine.
6433 (ccp_folder::get_value): New member function.
6434 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6435 (ccp_fold_stmt): Remove prototype.
6436 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6437 * tree-ssa-copy.c (copy_folder): New class derived from
6438 substitute_and_fold_engine.
6439 (copy_folder::get_value): Renamed from get_value.
6440 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6441 * tree-vrp.c (vrp_folder): New class derived from
6442 substitute_and_fold_engine.
6443 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6444 (vrp_folder::get_value): New member function.
6445 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6446 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6447 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6448 provide a class interface to folder/substitute routines.
6449 (ssa_prop_fold_stmt_fn): Remove typedef.
6450 (ssa_prop_get_value_fn): Likewise.
6451 (subsitute_and_fold): Remove prototype.
6452 (replace_uses_in): Likewise.
6453 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6454 Renamed from replace_uses_in. Call the virtual member function
6455 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6456 (substitute_and_fold_dom_walker): Remove initialization of
6457 data member entries for calbacks. Add substitute_and_fold_engine
6458 member and initialize it.
6459 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6460 member functions for get_value, replace_phi_args_in c
6461 replace_uses_in, and fold_stmt calls.
6462 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6463 substitute_and_fold. Remove assert. Update ctor call.
6465 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6466 (ssa_prop_visit_phi_fn): Likewise.
6467 (class ssa_propagation_engine): New class to provide an interface
6469 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6471 (ssa_prop_visit_phi): Likewise.
6472 (ssa_propagation_engine::simulate_stmt): Moved into class.
6473 Call visit_phi/visit_stmt from the class rather than via
6474 file scoped static variables.
6475 (ssa_propagation_engine::simulate_block): Moved into class.
6476 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6477 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6478 set file scoped statics for the visit_stmt/visit_phi callbacks.
6479 * tree-complex.c (complex_propagate): New class derived from
6480 ssa_propagation_engine.
6481 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6482 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6483 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6485 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6486 ssa_propagation_engine.
6487 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6488 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6489 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6490 * tree-ssa-copy.c (copy_prop): New class derived from
6491 ssa_propagation_engine.
6492 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6493 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6494 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6495 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6496 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6497 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6498 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6500 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6502 PR rtl-optimization/82778
6503 PR rtl-optimization/82597
6504 * compare-elim.c (struct comparison): Add in_a_setter field.
6505 (find_comparison_dom_walker::before_dom_children): Remove killed
6506 bitmap and df_simulate_find_defs call, instead walk the defs.
6507 Compute last_setter and initialize in_a_setter. Merge definitions
6508 with first initialization for a few variables.
6509 (try_validate_parallel): Use insn_invalid_p instead of
6510 recog_memoized. Return insn rather than just the pattern.
6511 (try_merge_compare): Fix up comment. Don't uselessly test if
6512 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6514 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6517 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6522 aarch64_hard_regno_nregs to get the number of registers
6525 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6526 Alan Hayward <alan.hayward@arm.com>
6527 David Sherwood <david.sherwood@arm.com>
6529 * config/aarch64/constraints.md (Upl): Rename to...
6531 * config/aarch64/aarch64.md
6532 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6535 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6540 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6543 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6544 Alan Hayward <alan.hayward@arm.com>
6545 David Sherwood <david.sherwood@arm.com>
6547 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6548 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6549 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6551 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6552 of a single element.
6553 * config/aarch64/iterators.md: Add a comment above the permute
6554 unspecs to say that they are generated directly by
6555 aarch64_expand_vec_perm_const.
6556 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6558 2017-11-01 Nathan Sidwell <nathan@acm.org>
6560 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6562 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6564 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6565 and medany code models, and describe what they do.
6567 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6569 Revert accidental duplicate:
6571 * combine.c (can_change_dest_mode): Reject changes in
6572 REGMODE_NATURAL_SIZE.
6574 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6576 PR rtl-optimization/64682
6577 PR rtl-optimization/69567
6578 PR rtl-optimization/69737
6579 PR rtl-optimization/82683
6580 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6581 register mentioned in the note, drop the note, unless it came from I3,
6582 in which case it should go to I3 again.
6584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6586 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6587 and return false if not.
6588 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6590 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6592 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6593 range is known to be empty.
6595 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6596 Alan Hayward <alan.hayward@arm.com>
6597 David Sherwood <david.sherwood@arm.com>
6599 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6600 and CONST_VECTOR_NUNITS instead of computing the number of units from
6601 the byte sizes of the vector and element.
6602 (simplify_binary_operation_1): Likewise.
6603 (simplify_const_binary_operation): Likewise.
6604 (simplify_ternary_operation): Likewise.
6606 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6607 Alan Hayward <alan.hayward@arm.com>
6608 David Sherwood <david.sherwood@arm.com>
6610 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6611 (int_mem_offset): ...this new function.
6612 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6613 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6614 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6617 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6618 Alan Hayward <alan.hayward@arm.com>
6619 David Sherwood <david.sherwood@arm.com>
6621 * lower-subreg.c (interesting_mode_p): New function.
6622 (compute_costs, find_decomposable_subregs, decompose_register)
6623 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6624 (resolve_clobber, dump_choices): Use it.
6626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6627 Alan Hayward <alan.hayward@arm.com>
6628 David Sherwood <david.sherwood@arm.com>
6630 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6632 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6633 Alan Hayward <alan.hayward@arm.com>
6634 David Sherwood <david.sherwood@arm.com>
6636 * alias.c (find_base_value, find_base_term): Only process integer
6637 truncations. Check the precision rather than the size.
6639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6640 Alan Hayward <alan.hayward@arm.com>
6641 David Sherwood <david.sherwood@arm.com>
6643 * machmode.h (is_narrower_int_mode): New function
6644 * optabs.c (expand_float, expand_fix): Use it.
6645 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6647 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6648 Alan Hayward <alan.hayward@arm.com>
6649 David Sherwood <david.sherwood@arm.com>
6651 * rtl.h (narrower_subreg_mode): New function.
6652 * ira-color.c (update_costs_from_allocno): Use it.
6654 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6655 Alan Hayward <alan.hayward@arm.com>
6656 David Sherwood <david.sherwood@arm.com>
6658 * optabs-query.h (convert_optab_p): New function, split out from...
6659 (convert_optab_handler): ...here.
6660 (widening_optab_handler): Delete.
6661 (find_widening_optab_handler): Remove permit_non_widening parameter.
6662 (find_widening_optab_handler_and_mode): Likewise. Provide an
6663 override that operates on mode class wrappers.
6664 * optabs-query.c (widening_optab_handler): Delete.
6665 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6666 parameter. Assert that the two modes are the same class and that
6667 the "from" mode is narrower than the "to" mode. Use
6668 convert_optab_handler instead of widening_optab_handler.
6669 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6670 instead of widening_optab_handler.
6671 * expr.c (expand_expr_real_2): Update calls to
6672 find_widening_optab_handler.
6673 * optabs.c (expand_widen_pattern_expr): Likewise.
6674 (expand_binop_directly): Take the insn_code as a parameter.
6675 (expand_binop): Only call find_widening_optab_handler for
6676 conversion optabs; use optab_handler otherwise. Update calls
6677 to find_widening_optab_handler and expand_binop_directly.
6678 Use convert_optab_handler instead of widening_optab_handler.
6679 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6680 find_widening_optab_handler and use scalar_mode rather than
6682 (convert_plusminus_to_widen): Likewise.
6684 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6685 Alan Hayward <alan.hayward@arm.com>
6686 David Sherwood <david.sherwood@arm.com>
6688 * machmode.h (fixed_size_mode): New class.
6689 * rtl.h (get_pool_mode): Return fixed_size_mode.
6690 * gengtype.c (main): Add fixed_size_mode.
6691 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6692 (get_raw_arg_mode): Likewise.
6693 * doc/tm.texi: Regenerate.
6694 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6695 * targhooks.c (default_get_reg_raw_mode): Likewise.
6696 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6697 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6698 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6699 (msp430_get_raw_result_mode): Likewise.
6700 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6701 * dbxout.c (dbxout_parms): Require fixed-size modes.
6702 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6703 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6704 * omp-low.c (lower_oacc_reductions): Likewise.
6705 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6706 (simplify_subreg): Update accordingly.
6707 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6708 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6709 that aren't fixed-size.
6710 (get_pool_mode): Return a fixed_size_mode.
6711 (output_constant_pool_2): Take a fixed_size_mode.
6713 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6714 Alan Hayward <alan.hayward@arm.com>
6715 David Sherwood <david.sherwood@arm.com>
6717 * doc/rtl.texi (vec_series): Document.
6718 (const): Say that the operand can be a vec_series.
6719 * rtl.def (VEC_SERIES): New rtx code.
6720 * rtl.h (const_vec_series_p_1): Declare.
6721 (const_vec_series_p): New function.
6722 * emit-rtl.h (gen_const_vec_series): Declare.
6723 (gen_vec_series): Likewise.
6724 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6725 (gen_vec_series): Likewise.
6726 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6727 * simplify-rtx.c (simplify_unary_operation): Handle negations
6729 (simplify_binary_operation_series): New function.
6730 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6731 (test_vector_ops_series): New function.
6732 (test_vector_ops): Call it.
6733 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6734 gen_const_vec_series.
6735 (altivec_lvsr): Likewise.
6736 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6738 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6739 Alan Hayward <alan.hayward@arm.com>
6740 David Sherwood <david.sherwood@arm.com>
6742 * doc/rtl.texi (const): Update description of address constants.
6743 Say that vector constants are allowed too.
6744 * common.md (E, F): Use CONSTANT_P instead of checking for
6746 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6747 checking for CONST_VECTOR.
6748 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6750 * expr.c (expand_expr_real_2): Check for vector modes instead
6751 of checking for CONST_VECTOR.
6752 * rtl.h (const_vec_p): New function.
6753 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6754 (unwrap_const_vec_duplicate): Handle them here too.
6756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6757 David Malcolm <dmalcolm@redhat.com>
6758 Alan Hayward <alan.hayward@arm.com>
6759 David Sherwood <david.sherwood@arm.com>
6761 * rtl.h (vec_duplicate_p): New function.
6762 * selftest-rtl.c (assert_rtx_eq_at): New function.
6763 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6764 (assert_rtx_eq_at): Declare.
6765 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6766 * selftest-run-tests.c (selftest::run_tests): Call it.
6767 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6768 (simplify_unary_operation_1): Recursively handle vector duplicates.
6769 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6771 (simplify_subreg): Handle subregs of vector duplicates.
6772 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6773 (selftest::simplify_rtx_c_tests): New functions.
6775 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6776 Alan Hayward <alan.hayward@arm.com>
6777 David Sherwood <david.sherwood@arm.com>
6779 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6780 (gen_vec_duplicate): Likewise.
6781 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6783 (gen_const_vector): ...here.
6784 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6785 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6786 whose elements are all equal.
6787 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6788 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6789 (simplify_relational_operation): Likewise.
6790 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6792 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6793 (aarch64_expand_vector_init): Likewise.
6794 * config/arm/arm.c (neon_vdup_constant): Likewise.
6795 (neon_expand_vector_init): Likewise.
6796 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6797 (arm_block_set_unaligned_vect): Likewise.
6798 (arm_block_set_aligned_vect): Likewise.
6799 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6800 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6801 (expand_vec_perm_even_odd_pack): Likewise.
6802 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6803 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6804 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6805 gen_const_vec_duplicate.
6806 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6807 * config/mips/mips.c (mips_gen_const_int_vector): Use
6808 gen_const_vec_duplicate.
6809 (mips_expand_vector_init): Use CONST0_RTX.
6810 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6811 (define_split): Use gen_const_vec_duplicate.
6812 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6813 (define_split): Use gen_const_vec_duplicate.
6814 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6815 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6816 * config/spu/spu.c (spu_const): Likewise.
6818 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6819 Alan Hayward <alan.hayward@arm.com>
6820 David Sherwood <david.sherwood@arm.com>
6822 * combine.c (can_change_dest_mode): Reject changes in
6823 REGMODE_NATURAL_SIZE.
6825 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6827 * configure.ac (--enable-libssp): New.
6828 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6829 trying to determine target-specific default. Adjust indentation.
6830 * configure: Regenerated.
6831 * doc/install.texi (Configuration): Expand --disable-libssp
6834 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6836 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6839 2017-10-31 Martin Jambor <mjambor@suse.cz>
6842 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6844 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6846 * auto-profile.c (autofdo_source_profile::read): Use
6847 UNKNOWN_LOCATION rather than 0.
6848 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6849 (warning_at): ...this overload.
6850 (warning_at_rich_loc_n): Rename to...
6851 (warning_n): ...this overload.
6852 (error_at_rich_loc): Rename to...
6853 (error_at): ...this overload.
6854 (pedwarn_at_rich_loc): Rename to...
6855 (pedwarn): ...this overload.
6856 (permerror_at_rich_loc): Rename to...
6857 (permerror): ...this overload.
6858 (inform_at_rich_loc): Rename to...
6859 (inform): ...this overload.
6860 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6861 (diagnostic_n_impl_richloc): Rename to...
6862 (diagnostic_n_impl): ...this rich_location *-based decl.
6863 (inform_at_rich_loc): Rename to...
6864 (inform): ...this, and add an assertion.
6865 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6866 (warning_at_rich_loc): Rename to...
6867 (warning_at): ...this, and add an assertion.
6868 (warning_at_rich_loc_n): Rename to...
6869 (warning_n): ...this, and add an assertion.
6870 (warning_n): Update location_t-based implementation for removal of
6871 location_t-based diagnostic_n_impl.
6872 (pedwarn_at_rich_loc): Rename to...
6873 (pedwarn): ...this, and add an assertion.
6874 (permerror_at_rich_loc): Rename to...
6875 (permerror): ...this, and add an assertion.
6876 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6877 (error_at_rich_loc): Rename to...
6878 (error_at): ...this, and add an assertion.
6879 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6880 (driver::do_spec_on_infiles): Likewise.
6881 * substring-locations.c (format_warning_va): Update for renaming
6882 of inform_at_rich_loc.
6884 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6886 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6887 _Float<N>X built-in functions so that the variant without the
6888 "__builtin_" prefix is only enabled for the GNU C and Objective C
6889 languages when they are in non-strict ANSI/ISO mode.
6890 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6891 * target.def (floatn_builtin_p): Add a target hook to control
6892 whether _Float<N> and _Float<N>X built-in functions without the
6893 "__builtin_" prefix are enabled, and return true for C and
6894 Objective C in the default hook. Include langhooks.h in
6896 * targhooks.h (default_floatn_builtin_p): Likewise.
6897 * targhooks.c (default_floatn_builtin_p): Likewise.
6898 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6899 floatn_builtin_p target hook.
6900 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6902 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6903 Eric Botcazou <ebotcazou@adacore.com>
6905 PR rtl-optimization/81803
6906 * lra-constraints.c (curr_insn_transform): Also reload the whole
6907 register for a strict subreg no wider than a word if this is for
6908 a WORD_REGISTER_OPERATIONS target.
6910 2017-10-31 Jason Merrill <jason@redhat.com>
6912 * gdbinit.in: Skip over inlines from timevar.h.
6914 2017-10-31 Martin Liska <mliska@suse.cz>
6916 * doc/gcov.texi: Document new option.
6917 * gcov.c (print_usage): Likewise print it.
6918 (process_args): Support the argument.
6919 (format_count): New function.
6920 (format_gcov): Use the function.
6922 2017-10-31 Martin Liska <mliska@suse.cz>
6924 * gcov.c (struct name_map): do not use typedef.
6925 Define operator== and operator<.
6926 (name_search): Remove.
6927 (name_sort): Remove.
6928 (main): Do not allocate names.
6929 (process_file): Add vertical space.
6930 (generate_results): Use std::find.
6931 (release_structures): Do not release memory.
6932 (find_source): Use std::find.
6934 2017-10-31 Martin Liska <mliska@suse.cz>
6936 * gcov.c (struct line_info): Remove it's typedef.
6937 (line_info::line_info): Add proper ctor.
6938 (line_info::has_block): Do not use a typedef.
6939 (struct source_info): Do not use typedef.
6940 (circuit): Likewise.
6941 (get_cycles_count): Likewise.
6942 (output_intermediate_file): Iterate via vector iterator.
6943 (add_line_counts): Use std::vector methods.
6944 (accumulate_line_counts): Likewise.
6945 (output_lines): Likewise.
6947 2017-10-31 Martin Liska <mliska@suse.cz>
6949 * gcov.c (struct source_info): Remove typedef.
6950 (source_info::source_info): Add proper ctor.
6951 (accumulate_line_counts): Use struct, not it's typedef.
6952 (output_gcov_file): Likewise.
6953 (output_lines): Likewise.
6954 (main): Do not allocate an array.
6955 (output_intermediate_file): Use size of vector container.
6956 (process_file): Resize the vector.
6957 (generate_results): Do not preallocate, use newly added vector
6959 (release_structures): Do not release sources.
6960 (find_source): Use vector methods.
6961 (add_line_counts): Do not use typedef.
6963 2017-10-31 Martin Liska <mliska@suse.cz>
6965 * doc/gcov.texi: Document that.
6966 * gcov.c (add_line_counts): Mark lines with a non-executed
6968 (output_line_beginning): Handle such lines.
6969 (output_lines): Pass new argument.
6970 (output_intermediate_file): Print it in intermediate format.
6972 2017-10-31 Martin Liska <mliska@suse.cz>
6974 * color-macros.h: New file.
6975 * diagnostic-color.c: Factor out color related to macros to
6977 * doc/gcov.texi: Document -k option.
6978 * gcov.c (INCLUDE_STRING): Include string.h.
6979 (print_usage): Add -k option.
6980 (process_args): Parse it.
6981 (pad_count_string): New function.
6982 (output_line_beginning): Likewise.
6983 (DEFAULT_LINE_START): New macro.
6984 (output_lines): Support color output.
6986 2017-10-31 Martin Liska <mliska@suse.cz>
6988 PR gcov-profile/82633
6989 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6990 their interaction with GCOV infrastructure.
6991 * configure.ac: Add -fkeep-{inline,static}-functions to
6993 * configure: Regenerate.
6995 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6998 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7000 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7003 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7004 into a register if it does not fit into an immediate offset field.
7006 2017-10-31 Olivier Hainque <hainque@adacore.com>
7008 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7010 2017-10-31 Julia Koval <julia.koval@intel.com>
7012 * config.gcc: Add gfniintrin.h.
7013 * config/i386/gfniintrin.h: New.
7014 * config/i386/i386-builtin-types.def
7015 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7016 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7017 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7018 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7019 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7020 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7021 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7022 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7023 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7024 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7025 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7026 * config/i386/immintrin.h: Include gfniintrin.h.
7027 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7029 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7031 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7033 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7035 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7036 (arm_ashldi3_1bit): Remove pattern.
7037 (ashrdi3): Remove shift by 1 expansion.
7038 (arm_ashrdi3_1bit): Remove pattern.
7039 (lshrdi3): Remove shift by 1 expansion.
7040 (arm_lshrdi3_1bit): Remove pattern.
7041 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7042 cost of ashldi3 by 1.
7043 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7044 (<shift>di3_neon): Likewise.
7046 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7048 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7049 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7050 and (*aarch64_simd_mov<VQ:mode>).
7051 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7052 pattern alternative.
7053 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7054 attributes to match pattern alternative.
7056 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7058 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7059 * config/rs6000/emmintrin.h: New file.
7060 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7062 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7064 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7065 * (movdi_vfp_cortexa8): Remove pattern.
7067 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7069 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7071 (Specific, alpha*-dec-osf5.1): Remove.
7072 (Specific, mips-sgi-irix5): Remove.
7073 (Specific, mips-sgi-irix6): Remove.
7075 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7078 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7079 arguments to clear_bit_region_be.
7081 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7083 * gimplify.c: Include memmodel.h.
7085 2017-10-30 Martin Jambor <mjambor@suse.cz>
7087 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7088 condition whether loop should be executed at all.
7090 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7092 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7093 gimple folding of vec_madd() intrinsics.
7094 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7095 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7096 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7098 2017-10-30 Richard Biener <rguenther@suse.de>
7100 PR tree-optimization/82762
7102 2017-10-23 Richard Biener <rguenther@suse.de>
7104 PR tree-optimization/82129
7106 2017-08-01 Richard Biener <rguenther@suse.de>
7108 PR tree-optimization/81181
7109 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7110 (compute_antic): ... end of iteration here.
7112 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7114 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7115 -std=iso9899:2017 and -std=gnu17.
7116 * doc/standards.texi (C Language): Document C17 support.
7117 * doc/cpp.texi (Overview): Mention -std=c17.
7118 (Standard Predefined Macros): Document C11 and C17 values of
7119 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7120 * doc/extend.texi (Inline): Do not list individual options for
7121 standards newer than C99.
7122 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7124 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7127 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7129 * asan.c (asan_finish_file): Align asan globals array by shadow
7132 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7135 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7136 (struct store_immediate_info): Add bitregion_start and bitregion_end
7138 (store_immediate_info::store_immediate_info): Add brs and bre
7139 arguments and initialize bitregion_{start,end} from those.
7140 (struct merged_store_group): Add bitregion_start, bitregion_end,
7141 align_base and mask fields. Drop unnecessary struct keyword from
7142 struct store_immediate_info. Add do_merge method.
7143 (clear_bit_region_be): Use memset instead of loop storing zeros.
7144 (merged_store_group::do_merge): New method.
7145 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7146 stores as long as the surrounding bitregions have no gaps.
7147 (merged_store_group::merge_overlapping): Use do_merge.
7148 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7149 is byte aligned, rather than requiring that start and width are
7150 byte aligned. Drop unnecessary struct keyword from
7151 struct store_immediate_info. Allocate and populate also mask array.
7152 Make start of the arrays relative to bitregion_start rather than
7153 start and size them according to bitregion_{end,start} difference.
7154 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7155 struct store_immediate_info.
7156 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7157 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7158 struct keyword from struct store_immediate_info.
7159 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7160 between stores as long as the surrounding bitregions have no gaps.
7162 (struct split_store): Add orig non-static data member.
7163 (split_store::split_store): Initialize orig to false.
7164 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7165 if there is exactly a single original stmt. Change stmts argument
7166 to pointer from reference, if NULL, don't push anything to it. Add
7167 first argument, use it to optimize skipping over orig stmts that
7168 are known to be before bitpos already. Simplify.
7169 (split_group): Return unsigned int count how many stores are or
7170 would be needed rather than a bool. Add allow_unaligned argument.
7171 Change split_stores argument from reference to pointer, if NULL,
7172 only do a dry run computing how many stores would be produced.
7173 Rewritten algorithm to use both alignment and misalign if
7174 !allow_unaligned and handle bitfield stores with gaps.
7175 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7176 from bitregion_start instead of start. Compute allow_unaligned
7177 here, if true, do 2 split_group dry runs to compute which one
7178 produces fewer stores and prefer aligned if equal. Punt if
7179 new count is bigger or equal than original before emitting any
7180 statements, rather than during that. Remove no longer needed
7181 new_ssa_names tracking. Replace num_stmts with
7182 split_stores.length (). Use 32-bit stack allocated entries
7183 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7184 if possible. Handle bitfields with gaps.
7185 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7186 Compute bitregion_{start,end} for the stores and construct
7187 store_immediate_info with that. Formatting fixes.
7189 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7192 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7193 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7195 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7197 * gimplify.c: Include tm_p.h.
7199 * common.opt (gcoff): Re-add as ignored option.
7200 (gcoff1, gcoff2, gcoff3): Likewise.
7202 * Makefile.in (OBJS): Delete sdbout.o.
7203 (GTFILES): Delete $(srcdir)/sdbout.c.
7204 * debug.h: Delete sdb_debug_hooks.
7205 * final.c: Delete sdbout.h include.
7206 (final_scan_insn): Delete SDB_DEBUG check.
7207 (rest_of_clean_state): Likewise.
7208 * output.h: Delete sdb_begin_function_line.
7211 * toplev.c: Delete sdbout.h include.
7212 (process_options): Delete SDB_DEBUG check.
7213 * tree-core.h (tree_type_common): Delete pointer field of
7215 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7216 TYPE_SYMTAB_POINTER.
7217 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7218 (TYPE_SYMTAB_IS_POINTER): Delete.
7219 (TYPE_SYMTAB_IS_DIE): Renumber.
7220 * xcoffout.c: Refer to former sdbout.c file.
7221 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7223 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7224 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7225 (Debugging Options): Delete -gcoff.
7226 (-gstabs): Delete SDB reference.
7228 (-gcoff@var{level}): Delete.
7229 * doc/passes.texi (Debugging information output): Delete SDB and
7230 sdbout.c references.
7231 * doc/tm.texi: Regenerate.
7232 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7233 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7235 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7236 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7237 (SDB_DEBUGGING_INFO): Delete.
7238 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7239 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7240 * target.def (output_source_filename): Delete COFF reference.
7242 * common.opt (gcoff): Delete.
7243 (gxcoff+): Update Negative chain.
7244 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7246 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7247 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7248 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7250 (expand_function_start): Change sdb reference to past tense.
7251 (expand_function_end): Change sdb reference to past tense.
7252 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7253 * opts.c (debug_type_names): Delete coff entry.
7254 (common_handle_option): Delete OPT_gcoff case.
7255 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7257 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7258 * config/cris/cris.h: Delete SDB reference in comment.
7259 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7260 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7261 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7262 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7264 (ix86_expand_prologue): Likewise.
7265 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7266 * config/ia64/ia64.h: Likewise.
7267 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7268 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7270 * config/mmix/mmix.h: Likewise.
7271 * config/nds32/nds32.c: Likewise.
7272 * config/stormy/storym16.h: Likewise.
7273 * config/visium/visium.h: Likewise.
7274 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7276 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7278 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7279 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7280 FRAME_POINTER_REGNUM point at high end of local var area.
7282 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7284 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7285 Move comment around. Do not reset best_edge for a copiable
7286 destination if the copy would cause a partition change.
7287 (better_edge_p): Remove redundant check.
7289 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7291 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7293 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7295 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7296 for math functions that have _Float<N> and _Float<N>X variants.
7297 (mathfn_built_in_2): Add support for math functions that have
7298 _Float<N> and _Float<N>X variants.
7299 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7300 (expand_builtin_mathfn_ternary): Add support for fma with
7301 _Float<N> and _Float<N>X variants.
7302 (expand_builtin): Likewise.
7303 (fold_builtin_3): Likewise.
7304 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7305 create math function _Float<N> and _Float<N>X variants as external
7307 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7308 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7309 the __builtin_ prefix and if not strict ansi, without the prefix.
7310 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7311 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7312 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7313 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7314 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7315 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7316 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7317 function signatures for fma _Float<N> and _Float<N>X variants.
7318 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7319 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7320 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7321 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7322 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7323 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7324 * gencfn-macros.c (print_case_cfn): Add support for math functions
7325 that have _Float<N> and _Float<N>X variants.
7326 (print_define_operator_list): Likewise.
7327 (fltfn_suffixes): Likewise.
7329 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7330 for math functions that have _Float<N> and _Float<N>X variants.
7331 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7332 and _Float<N>X variants.
7333 (COPYSIGN): Likewise.
7336 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7337 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7339 (integer_valued_read_call_p): Likewise.
7340 * fold-const-call.c (fold_const_call_ss): Likewise.
7341 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7342 _Float<N> and _Float<N>X variants.
7343 (fold_const_call_ssss): Add support for fma _Float<N> and
7344 _Float<N>X variants.
7345 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7346 support for copysign and fma _Float<N> and _Float<N>X variants.
7347 (backprop::process_builtin_call_use): Likewise.
7348 * tree-call-cdce.c (can_test_argument_range); Add support for
7349 sqrt _Float<N> and _Float<N>X variants.
7350 (edom_only_function): Likewise.
7351 (get_no_error_domain): Likewise.
7352 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7353 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7354 copysign _Float<N> and _Float<N>X variants.
7355 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7356 handled by machine independent code.
7357 (FMAF128): Likewise.
7358 * doc/cpp.texi (Common Predefined Macros): Document defining
7359 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7360 fma _Float<N> and _Float<N>X variants.
7362 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7365 * config/i386/i386-modes.def (CCFPU): Remove definition.
7366 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7367 (ix86_cc_modes_compatible): Ditto.
7368 (ix86_expand_carry_flag_compare): Ditto.
7369 (ix86_expand_int_movcc): Ditto.
7370 (ix86_expand_int_addcc): Ditto.
7371 (ix86_reverse_condition): Ditto.
7372 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7373 Return true/false for unordered/ordered fp comparisons.
7374 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7375 (ix86_prepare_fp_compare_args): Update for rename.
7376 (ix86_expand_fp_compare): Update for rename. Generate unordered
7377 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7378 (ix86_expand_sse_compare_and_jump): Ditto.
7379 * config/i386/predicates.md (fcmov_comparison_operator):
7380 Remove CCFPU mode handling.
7381 (ix86_comparison_operator): Ditto.
7382 (ix86_carry_flag_operator): Ditto.
7383 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7384 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7385 (*cmpu<mode>_cc_i387): Ditto.
7386 (FPCMP): Remove mode iterator.
7387 (unord): Remove mode attribute.
7388 (unord_subst): New define_subst transformation
7389 (unord): New define_subst attribute.
7391 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7392 (*cmpi<unord>xf_i387): Ditto.
7393 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7394 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7395 using unord_subst transformation.
7396 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7397 (round_saeonly): Also handle CCFP mode.
7398 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7399 Remove UNSPEC_SAHF unspec handling.
7401 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7403 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7405 2017-10-27 Jeff Law <law@redhat.com>
7407 * gimple-ssa-sprintf.c: Include domwalk.h.
7408 (class sprintf_dom_walker): New class, derived from dom_walker.
7409 (sprintf_dom_walker::before_dom_children): New function.
7410 (struct call_info): Moved into sprintf_dom_walker class
7411 (compute_formath_length, handle_gimple_call): Likewise.
7412 (sprintf_length::execute): Call the dominator walker rather
7413 than walking the statements.
7415 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7416 gimple statement locations.
7417 (check_array_bounds): Corresponding changes. Get the statement's
7418 location directly from wi->stmt.
7420 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7423 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7425 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7427 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7428 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7430 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7433 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7434 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7435 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7436 maybe_get_pool_constant.
7437 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7440 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7442 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7443 2.26 caveat. Update gas and gld versions.
7444 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7447 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7449 * cgraph.h (set_malloc_flag): Declare.
7450 * cgraph.c (set_malloc_flag_1): New function.
7451 (set_malloc_flag): Likewise.
7452 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7453 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7455 (read_ipa_call_summary): Add support for reading is_return_callee.
7456 (write_ipa_call_summary): Stream is_return_callee.
7457 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7458 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7459 ipa-prop.h, ipa-fnsummary.h.
7460 (pure_const_names): Change to static.
7461 (malloc_state_e): Define.
7462 (malloc_state_names): Define.
7463 (funct_state_d): Add field malloc_state.
7464 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7465 (check_retval_uses): New function.
7466 (malloc_candidate_p): Likewise.
7467 (analyze_function): Add support for malloc attribute.
7468 (pure_const_write_summary): Stream malloc_state.
7469 (pure_const_read_summary): Add support for reading malloc_state.
7470 (dump_malloc_lattice): New function.
7471 (propagate_malloc): New function.
7472 (warn_function_malloc): New function.
7473 (ipa_pure_const::execute): Call propagate_malloc and
7474 ipa_free_fn_summary.
7475 (pass_local_pure_const::execute): Add support for malloc attribute.
7476 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7477 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7479 2017-10-27 Martin Liska <mliska@suse.cz>
7481 PR gcov-profile/82457
7482 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7483 for fork-like functions to be properly instrumented.
7485 2017-10-27 Richard Biener <rguenther@suse.de>
7488 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7489 info when we redirected EH.
7491 2017-10-26 Michael Collison <michael.collison@arm.com>
7493 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7495 (<optab>_trunchf<GPI:mode>2: New pattern.
7496 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7497 * config/aarch64/iterators.md (wv): New mode attribute.
7498 (vf, VF): New mode attributes.
7499 (vgp, VGP): New mode attributes.
7500 (s): Update attribute with SImode and DImode prefixes.
7502 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7504 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7505 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7506 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7507 (nios2_option_overide): Initialize it. Don't allow R0-relative
7508 addressing with PIC.
7509 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7510 (nios2_symbolic_constant_p): Likewise.
7511 (nios2_legitimate_address_p): Likewise.
7512 (nios2_r0rel_section_name_p): New.
7513 (nios2_symbol_ref_in_r0rel_data_p): New.
7514 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7515 (r0rel_constant_p): New.
7516 (nios2_print_operand_address): Handle r0rel_constant_p.
7517 (nios2_cdx_narrow_form_p): Likewise.
7518 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7519 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7520 (Nios II Options): Document -mr0rel-sec.
7522 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7524 * config/nios2/nios2.c: Include xregex.h.
7525 (nios2_gprel_sec_regex): New.
7526 (nios2_option_overide): Initialize it. Don't allow GP-relative
7527 addressing with PIC.
7528 (nios2_small_section_name_p): Check for regex match.
7529 * config/nios2/nios2.opt (mgprel-sec=): New option.
7530 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7531 (Nios II Options): Document -mgprel-sec.
7533 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7535 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7537 2017-10-26 Tom de Vries <tom@codesourcery.com>
7539 PR tree-optimization/82707
7540 * gimple.c (gimple_copy): Fix unsharing of
7541 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7543 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7545 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7546 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7547 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7548 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7549 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7550 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7551 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7552 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7553 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7554 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7555 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7556 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7557 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7558 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7559 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7560 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7561 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7563 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7565 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7567 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7568 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7569 warning to rs6000.c. Remove the Undocumented flag, since it has
7571 (-mabi=ibmlongdouble): Likewise.
7572 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7573 already set, set the default format for long double.
7574 (rs6000_debug_reg_global): Print whether long double is IBM or
7576 (rs6000_option_override_internal): Rework setting long double
7577 format. Only warn if the user is changing the long double default
7578 and they did not use -Wno-psabi.
7579 * doc/invoke.texi (PowerPC options): Update the documentation for
7580 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7582 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7583 Alan Hayward <alan.hayward@arm.com>
7584 David Sherwood <david.sherwood@arm.com>
7586 * rtl.h (wider_subreg_mode): New function.
7587 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7588 rather than an unsigned int *.
7589 * ira-color.c (regno_max_ref_width): Replace with...
7590 (regno_max_ref_mode): ...this new variable.
7591 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7592 Use wider_subreg_mode.
7593 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7594 rather than an unsigned int *.
7595 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7596 (process_alt_operands): Likewise.
7597 (invariant_p): Likewise.
7598 * lra-spills.c (assign_mem_slot): Likewise.
7599 (add_pseudo_to_slot): Likewise.
7600 * lra.c (collect_non_operand_hard_regs): Likewise.
7601 (add_regs_to_insn_regno_info): Likewise.
7602 * reload1.c (regno_max_ref_width): Replace with...
7603 (regno_max_ref_mode): ...this new variable.
7604 (reload): Update accordingly. Update call to
7605 ira_sort_regnos_for_alter_reg.
7606 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7607 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7608 (scan_paradoxical_subregs): Likewise.
7610 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7612 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7613 (aarch64_frame): Add emit_frame_chain boolean.
7614 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7615 Move eh_return case to aarch64_layout_frame.
7616 (aarch64_layout_frame): Initialize emit_frame_chain.
7617 (aarch64_expand_prologue): Use emit_frame_chain.
7619 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7621 * config/aarch64/aarch64.c (aarch64_layout_frame):
7622 Ensure LR is always stored at the bottom of the callee-saves.
7623 Remove rarely used frame layout which saves callee-saves at top of
7624 frame, so the store of LR can be used as a valid probe in all cases.
7626 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7628 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7629 Improve unaligned TImode/TFmode base/offset split.
7631 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7632 Alan Hayward <alan.hayward@arm.com>
7633 David Sherwood <david.sherwood@arm.com>
7635 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7636 * combine.c (find_single_use_1): Likewise.
7637 (expand_field_assignment): Likewise.
7638 (move_deaths): Likewise.
7639 * lra-constraints.c (simplify_operand_subreg): Likewise.
7640 (curr_insn_transform): Likewise.
7641 * lra.c (collect_non_operand_hard_regs): Likewise.
7642 (add_regs_to_insn_regno_info): Likewise.
7643 * rtlanal.c (reg_referenced_p): Likewise.
7644 (covers_regno_no_parallel_p): Likewise.
7646 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7648 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7649 Don't print any bits outside the precision of the value.
7650 * wide-int.cc (test_printing): Add some new tests.
7652 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7654 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7655 supports -xbrace_comment option.
7656 * configure: Regenerate.
7657 * config.in: Regenerate.
7658 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7659 (ASM_CPU_SPEC): Use it.
7661 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7663 * target.def (static_rtx_alignment): New hook.
7664 * targhooks.h (default_static_rtx_alignment): Declare.
7665 * targhooks.c (default_static_rtx_alignment): New function.
7666 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7667 * doc/tm.texi: Regenerate.
7668 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7669 instead of targetm.constant_alignment. Remove call to
7671 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7672 (cris_preferred_mininum_alignment): New function, split out from...
7673 (cris_constant_alignment): ...here.
7674 (cris_static_rtx_alignment): New function.
7675 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7677 (ix86_constant_alignment): ...here.
7678 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7679 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7680 (mmix_static_rtx_alignment): New function.
7681 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7682 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7684 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7687 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7688 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7690 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7692 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7693 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7694 mark as a sign-extending load.
7695 (local_pic_load_u): Define.
7697 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7700 * fold-const.c (operand_equal_for_comparison_p): Also return true
7701 if ARG0 is a simple variant of ARG1 with narrower precision.
7702 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7704 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7706 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7708 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7709 scatter_static, scatter_per_elt.
7710 * x86-tune-costs.h: Add new cost entries.
7712 2017-10-25 Richard Biener <rguenther@suse.de>
7714 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7715 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7716 class pass_fre): Move to ...
7717 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7718 class pass_fre): ... here and adjust for statistics.
7720 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7723 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7724 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7725 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7726 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7729 2017-10-25 Richard Biener <rguenther@suse.de>
7731 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7732 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7733 eliminate_push_avail, eliminate_insert): Move inside...
7734 (class eliminate_dom_walker): ... this class in preparation
7736 (fini_eliminate): Remove by merging with ...
7737 (eliminate): ... this function. Adjust for class changes.
7738 (pass_pre::execute): Remove fini_eliminate call.
7739 (pass_fre::execute): Likewise.
7741 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7744 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7745 (VPERMI2, VPERMI2I): New mode iterators.
7746 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7747 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7749 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7750 mode iterator. Remove 3 old define_insn patterns.
7751 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7752 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7753 VPERMI2 mode iterator, remove the other two expanders.
7754 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7755 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7756 instructions, remove the other two patterns.
7757 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7758 mode iterator, remove the other two patterns.
7759 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7760 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7761 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7762 and adjust argument order accordingly.
7763 (ix86_expand_vec_perm): Adjust caller.
7764 (expand_vec_perm_1): Likewise.
7765 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7766 (expand_vec_perm_vpermt2_vpshub2): ... this.
7767 (ix86_expand_vec_perm_const_1): Adjust caller.
7768 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7771 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7772 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7773 (vec_shl_<mode>): Remove unused expander.
7774 (avx512bw_<shift_insn><mode>3): New define_insn.
7775 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7776 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7778 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7781 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7784 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7786 PR rtl-optimization/82396
7787 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7788 (autopref_multipass_init): Simplify initialization.
7789 (autopref_rank_data): Simplify sort order.
7790 * gcc/sched-int.h (autopref_multipass_data_): Remove
7791 multi_mem_insn_p, min_offset and max_offset.
7793 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7796 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7797 Check special value of flag_omit_frame_pointer.
7798 (aarch64_can_eliminate): Likewise.
7799 (aarch64_override_options_after_change_1): Simplify handling of
7800 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7802 2017-10-24 Richard Biener <rguenther@suse.de>
7804 PR tree-optimization/82697
7805 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7806 zero for conditional load and unconditional store.
7808 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7810 * doc/install.texi: Document bootstrap-cet.
7812 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7815 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7816 ENDBR instruction at function entrance if function is only
7819 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7822 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7823 patterns to better describe from which operation the CF is computed.
7824 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7825 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7826 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7827 is 0, use _0 suffixed expanders instead of emitting a comparison
7830 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7832 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7833 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7835 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7838 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7839 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7840 * loop-iv.c (iv_get_reaching_def): Likewise.
7841 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7842 variable is promoted and the partition contains undefined values.
7844 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7846 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7848 (nios2_address_cost): Define.
7849 (nios2_legitimize_address): Recognize (exp + constant) directly.
7850 (TARGET_ADDRESS_COST): Define.
7852 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7854 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7855 (nios2_symbolic_memory_operand_p): Declare.
7856 (nios2_split_large_constant): Declare.
7857 (nios2_split_symbolic_memory_operand): Declare.
7858 * config/nios2/nios2.c: Adjust includes.
7859 (nios2_symbolic_constant_allowed): New.
7860 (nios2_symbolic_constant_p): New.
7861 (nios2_plus_symbolic_constant_p): New.
7862 (nios2_valid_addr_expr_p): Recognize addresses involving
7864 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7865 (nios2_symbolic_memory_operand_p): New.
7866 (nios2_large_constant_p): New.
7867 (nios2_split_large_constant): New.
7868 (nios2_split_plus_large_constant): New.
7869 (nios2_split_symbolic_memory_operand): New.
7870 (nios2_legitimize_address): Code refactoring. Handle addresses
7871 involving symbolic constants.
7872 (nios2_emit_move_sequence): Likewise.
7873 (nios2_print_operand): Improve error output.
7874 (nios2_print_operand_address): Handle LO_SUM.
7875 (nios2_cdx_narrow_form_p): Likewise.
7876 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7877 operands involving symbolic constants.
7878 (movhi_internal, movsi_internal): Likewise.
7879 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7880 (extendhisi2, extendqi<mode>2): Likewise.
7882 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7884 * tree-pass.h (PROP_rtl_split_insns): Define.
7885 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7887 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7889 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7891 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7894 * target.def (const_not_ok_for_debug_p): Default to
7895 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7896 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7897 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7898 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7899 which targetm.const_not_ok_for_debug_p returned true.
7900 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7902 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7904 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7905 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7906 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7908 (ix86_const_not_ok_for_debug_p): New function.
7909 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7910 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7912 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7915 * system.h: Conditionally include "unique-ptr.h" if
7916 INCLUDE_UNIQUE_PTR is defined.
7917 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7918 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7920 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7922 * config/rl78/rl78.md: New define_expand "subdi3".
7924 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7927 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7928 DF_REF_INSN if DF_REF_INSN_INFO is false.
7930 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7932 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7933 xmm_move instead of sse_move.
7934 (sse_store_index): New function.
7935 (ix86_register_move_cost): Be more sensible about mismatch stall;
7936 model AVX moves correctly; make difference between sse->integer and
7938 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7939 moves; make difference between SSE and AVX.
7940 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7941 and zmm_move. Increase size of sse load and store tables;
7942 add unaligned load and store tables; add ssemmx_to_integer.
7943 * x86-tune-costs.h: Update all entries according to real
7944 move latencies from Agner Fog's manual and chip documentation.
7946 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7949 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7950 * config/i386/constraints.md (Wf): New constraint.
7951 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7952 (cmp<dwi>_doubleword): Removed.
7953 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7954 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7955 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7956 expand with cmp<dwi>_doubleword. For LTU and GEU use
7957 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7959 * common.opt (gcolumn-info): Enable by default.
7960 * doc/invoke.texi (gcolumn-info): Document new default.
7962 2017-10-23 Richard Biener <rguenther@suse.de>
7964 PR tree-optimization/82672
7965 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7966 Fold the stmt if we propagated into it.
7968 2017-10-23 Richard Biener <rguenther@suse.de>
7970 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7971 (bitmap_remove_expr_from_set): ... this. All callers call this
7972 for non-constant values.
7973 (bitmap_set_subtract): Rename to...
7974 (bitmap_set_subtract_expressions): ... this. Adjust and
7976 (bitmap_set_contains_value): Remove superfluous check.
7977 (bitmap_set_replace_value): Inline into single caller ...
7978 (bitmap_value_replace_in_set): ... here and simplify.
7979 (dependent_clean): Merge into ...
7980 (clean): ... this using an overload. Adjust.
7981 (prune_clobbered_mems): Adjust.
7982 (compute_antic_aux): Likewise.
7983 (compute_partial_antic_aux): Likewise.
7985 2017-10-23 Richard Biener <rguenther@suse.de>
7987 PR tree-optimization/82129
7989 2017-08-01 Richard Biener <rguenther@suse.de>
7991 PR tree-optimization/81181
7992 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7993 (compute_antic): ... end of iteration here.
7995 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7997 * target.def (starting_frame_offset): New hook.
7998 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7999 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8000 * doc/tm.texi.in: Regenerate.
8001 * hooks.h (hook_hwi_void_0): Declare.
8002 * hooks.c (hook_hwi_void_0): New function.
8003 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8004 STARTING_FRAME_OFFSET.
8005 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8006 * reload1.c (reload): Likewise.
8007 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8008 instead of STARTING_FRAME_OFFSET.
8009 * function.c (try_fit_stack_local): Likewise.
8010 (assign_stack_local_1): Likewise
8011 (instantiate_virtual_regs): Likewise.
8012 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8013 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8014 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8015 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8016 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8017 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8018 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8019 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8020 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8021 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8022 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8023 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8024 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8025 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8026 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8027 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8028 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8029 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8030 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8031 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8032 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8033 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8034 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8035 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8036 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8037 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8038 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8039 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8040 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8041 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8042 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8043 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8044 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8045 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8046 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8047 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8048 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8049 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8050 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8051 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8052 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8053 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8054 return a HOST_WIDE_INT.
8055 (avr_builtin_setjmp_frame_value): Use it instead of
8056 STARTING_FRAME_OFFSET.
8057 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8058 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8059 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8061 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8062 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8063 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8064 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8065 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8066 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8067 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8068 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8069 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8070 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8071 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8072 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8074 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8075 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8076 * config/mips/mips.c (mips_compute_frame_info): Refer to
8077 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8078 (mips_starting_frame_offset): New function.
8079 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8080 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8081 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8082 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8083 and return a HOST_WIDE_INT.
8084 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8085 (mmix_initial_elimination_offset): Refer to
8086 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8087 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8088 * config/pa/pa.c (pa_starting_frame_offset): New function.
8089 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8090 (pa_expand_prologue): Likewise.
8091 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8092 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8093 !FRAME_GROWS_DOWNWARD handling to...
8094 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8095 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8096 !FRAME_GROWS_DOWNWARD handling to...
8097 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8098 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8099 !FRAME_GROWS_DOWNWARD handling to...
8100 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8101 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8103 (rs6000_starting_frame_offset): New function.
8104 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8105 !FRAME_GROWS_DOWNWARD handling to...
8106 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8107 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8108 !FRAME_GROWS_DOWNWARD handling to...
8109 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8110 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8111 !FRAME_GROWS_DOWNWARD handling to...
8112 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8113 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8114 (rs6000_starting_frame_offset): New function.
8115 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8116 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8117 * config/vax/vax.c (vax_starting_frame_offset): New function.
8118 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8119 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8120 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8121 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8122 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8123 * system.h (STARTING_FRAME_OFFSET): Poison.
8125 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8127 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8128 SCALAR_TYPE_MODE instead of TYPE_MODE.
8130 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8131 Alan Hayward <alan.hayward@arm.com>
8132 David Sherwood <david.sherwood@arm.com>
8134 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8136 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8137 Alan Hayward <alan.hayward@arm.com>
8138 David Sherwood <david.sherwood@arm.com>
8140 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8142 2017-10-23 Richard Biener <rguenther@suse.de>
8144 PR tree-optimization/82129
8145 * tree-ssa-pre.c (bitmap_set_and): Remove.
8146 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8147 canonicalizing expressions in the set to those with lowest
8148 ID rather than taking that from the first edge.
8150 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8152 * combine.c (rtx_equal_for_field_assignment_p): Use
8153 byte_lowpart_offset.
8155 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8156 Alan Hayward <alan.hayward@arm.com>
8157 David Sherwood <david.sherwood@arm.com>
8159 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8160 to a SUBREG_PROMOTED_VAR.
8162 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8163 Alan Hayward <alan.hayward@arm.com>
8164 David Sherwood <david.sherwood@arm.com>
8166 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8167 (expand_debug_source_expr): Likewise.
8168 * combine.c (combine_simplify_rtx): Likewise.
8169 * cse.c (fold_rtx): Likewise.
8170 * optabs.c (expand_float): Likewise.
8171 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8172 (simplify_binary_operation_1): Likewise.
8174 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8179 (record_promoted_value): Likewise.
8180 * expr.c (expand_expr_real_2): Likewise.
8181 * ree.c (update_reg_equal_equiv_notes): Likewise.
8182 (combine_set_extension): Likewise.
8183 * rtlanal.c (low_bitmask_len): Likewise.
8184 * simplify-rtx.c (neg_const_int): Likewise.
8185 (simplify_binary_operation_1): Likewise.
8187 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8188 Alan Hayward <alan.hayward@arm.com>
8189 David Sherwood <david.sherwood@arm.com>
8191 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8192 * regcprop.c (maybe_mode_change): Likewise.
8193 * reload1.c (alter_reg): Likewise.
8195 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8197 * inchash.h (inchash::hash::add_wide_int): New function.
8198 * lto-streamer-out.c (hash_tree): Use it.
8200 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8202 * inchash.h (inchash::hash::add_wide_int): Rename to...
8203 (inchash::hash::add_hwi): ...this.
8204 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8205 (polymorphic_call_target_hasher::hash): Likewise.
8206 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8207 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8208 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8209 * lto-streamer-out.c (hash_tree): Likewise.
8210 * optc-save-gen.awk: Likewise.
8211 * tree.c (add_expr): Likewise.
8213 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8216 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8217 for ordered inequality comparisons even with TARGET_IEEE_FP.
8219 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8222 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8223 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8224 Expand with cmp<dwi>_doubleword.
8226 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8228 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8229 List CET intrinsics.
8230 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8231 specific to -fcf-protection option.
8233 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8235 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8236 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8237 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8238 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8239 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8240 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8241 (extra_objs): Add cet.o for Linux/x86 targets.
8242 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8243 * config/i386/cet.c: New file.
8244 * config/i386/cetintrin.h: Likewise.
8245 * config/i386/t-cet: Likewise.
8246 * config/i386/cpuid.h (bit_SHSTK): New.
8247 (bit_IBT): Likewise.
8248 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8249 pass IBT and SHSTK bits.
8250 * config/i386/i386-builtin-types.def
8251 (VOID_FTYPE_UNSIGNED_PVOID): New.
8252 (VOID_FTYPE_UINT64_PVOID): Likewise.
8253 * config/i386/i386-builtin.def: Add CET intrinsics.
8254 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8255 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8256 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8257 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8259 * config/i386/i386.c (rest_of_insert_endbranch): New.
8260 (pass_data_insert_endbranch): Likewise.
8261 (pass_insert_endbranch): Likewise.
8262 (make_pass_insert_endbranch): Likewise.
8263 (ix86_notrack_prefixed_insn_p): Likewise.
8264 (ix86_target_string): Add -mibt, -mshstk flags.
8265 (ix86_option_override_internal): Add flag_cf_protection
8267 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8268 (ix86_print_operand): Add 'notrack' prefix output.
8269 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8270 (ix86_expand_builtin): Expand CET intrinsics.
8271 (x86_output_mi_thunk): Add 'endbranch' instruction.
8272 * config/i386/i386.h (TARGET_IBT): New.
8273 (TARGET_IBT_P): Likewise.
8274 (TARGET_SHSTK): Likewise.
8275 (TARGET_SHSTK_P): Likewise.
8276 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8277 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8278 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8279 (builtin_setjmp_setup): New pattern.
8280 (builtin_longjmp): Likewise.
8281 (rdssp<mode>): Likewise.
8282 (incssp<mode>): Likewise.
8283 (saveprevssp): Likewise.
8284 (rstorssp): Likewise.
8285 (wrss<mode>): Likewise.
8286 (wruss<mode>): Likewise.
8287 (setssbsy): Likewise.
8288 (clrssbsy): Likewise.
8289 (nop_endbr): Likewise.
8290 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8292 * config/i386/immintrin.h: Include <cetintrin.h>.
8293 * config/i386/linux-common.h
8294 (file_end_indicate_exec_stack_and_cet): New prototype.
8295 (TARGET_ASM_FILE_END): New.
8297 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8299 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8300 latencies instead of having separate table; make difference between
8301 integer and float costs.
8302 * i386.h (processor_costs): Remove scalar_stmt_cost,
8303 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8304 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8306 * x86-tune-costs.h: Remove entries which has been removed in
8307 procesor_costs from all tables; make cond_taken_branch_cost
8308 and cond_not_taken_branch_cost COST_N_INSNS based.
8310 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8312 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8314 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8316 * config/i386/i386.md (isa): Remove fma_avx512f.
8317 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8318 <avx512>_fmadd_<mode>_mask3<round_name>,
8319 <avx512>_fmsub_<mode>_mask<round_name>,
8320 <avx512>_fmsub_<mode>_mask3<round_name>,
8321 <avx512>_fnmadd_<mode>_mask<round_name>,
8322 <avx512>_fnmadd_<mode>_mask3<round_name>,
8323 <avx512>_fnmsub_<mode>_mask<round_name>,
8324 <avx512>_fnmsub_<mode>_mask3<round_name>,
8325 <avx512>_fmaddsub_<mode>_mask<round_name>,
8326 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8327 <avx512>_fmsubadd_<mode>_mask<round_name>,
8328 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8329 (*vec_widen_umult_even_v16si<mask_name>,
8330 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8331 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8333 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8335 * extend.texi: Add 'nocf_check' documentation.
8336 * gimple.texi: Add second parameter to
8337 gimple_build_call_from_tree.
8338 * invoke.texi: Add -fcf-protection documentation.
8339 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8341 2017-10-20 Richard Biener <rguenther@suse.de>
8343 PR tree-optimization/82473
8344 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8345 the largest input type.
8347 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8349 * c-attribs.c (handle_nocf_check_attribute): New function.
8350 (c_common_attribute_table): Add 'nocf_check' handling.
8351 * gimple-parser.c: Add second argument NULL to
8352 gimple_build_call_from_tree.
8353 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8354 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8356 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8357 * common.opt: Add fcf-protection flag.
8358 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8359 * flag-types.h: Add enum cf_protection_level.
8360 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8361 Add 'nocf_check' attribute propagation to gimple call.
8362 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8363 (gimple_build_call_from_tree): Update prototype.
8364 (gimple_call_nocf_check_p): New function.
8365 (gimple_call_set_nocf_check): Likewise.
8366 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8367 * ipa-icf.c: Add nocf_check attribute in statement hash.
8368 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8369 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8370 * toplev.c (process_options): Add flag_cf_protection handling.
8372 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8374 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8376 2017-10-20 Richard Biener <rguenther@suse.de>
8378 PR tree-optimization/82603
8379 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8380 remove false predicated stores.
8382 2017-10-20 Richard Biener <rguenther@suse.de>
8384 * graphite-isl-ast-to-gimple.c
8385 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8386 Remove return value and simplify, dump copied stmt after lhs
8388 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8389 Reduce dump verbosity.
8390 (gsi_insert_earliest): Likewise.
8391 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8392 * graphite.c (print_global_statistics): Adjust dumping.
8393 (print_graphite_scop_statistics): Likewise.
8394 (print_graphite_statistics): Do not dump loops here.
8395 (graphite_transform_loops): But here.
8397 2017-10-20 Nicolas Roche <roche@adacore.com>
8399 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8400 * configure: Regenerate.
8402 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8405 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8406 functions when optimizing replace GIMPLE_RETURN stmts with
8407 calls to __builtin_unreachable ().
8410 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8411 for -fsanitize=thread link of executables.
8412 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8413 link of executables.
8416 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8418 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8419 define_insns for logical vector shifts to use VI248_AVX512BW
8420 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8421 condition, useless isa and prefix attributes. Change the first
8422 2 of these define_insns to ...
8423 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8424 define_insn for avx512vl.
8425 (<shift_insn><mode>3): ... and this, new define_insn without
8426 masking for non-avx512vl.
8429 * config/i386/sse.md (*andnot<mode>3,
8430 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8431 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8432 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8433 not applied use empty suffix even for TARGET_AVX512VL.
8434 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8435 is applied, supply evex,evex or evex,evex,evex instead of just
8438 2017-10-20 Julia Koval <julia.koval@intel.com>
8440 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8441 (OPTION_MASK_ISA_GFNI_UNSET): New.
8442 (ix86_handle_option): Handle OPT_mgfni.
8443 * config/i386/cpuid.h (bit_GFNI): New.
8444 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8445 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8446 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8447 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8448 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8449 * config/i386/i386.opt: Add mgfni.
8451 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8453 * config/msp430/msp430.c (msp430_option_override): Disable
8454 -fdelete-null-pointer-checks.
8455 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8457 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8459 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8460 of x87 and SSE instructions.
8462 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8464 * asan.c (create_cond_insert_point): Do not update edge count.
8465 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8466 (afdo_propagate_circuit): Likewise.
8467 (afdo_calculate_branch_prob): Likewise.
8468 (afdo_annotate_cfg): Likewise.
8469 * basic-block.h (struct edge_def): Remove count.
8470 (edge_def::count): New accessor.
8471 * bb-reorder.c (rotate_loop): Update.
8472 (find_traces_1_round): Update.
8473 (connect_traces): Update.
8474 (sanitize_hot_paths): Update.
8475 * cfg.c (unchecked_make_edge): Update.
8476 (make_single_succ_edge): Update.
8477 (check_bb_profile): Update.
8478 (dump_edge_info): Update.
8479 (update_bb_profile_for_threading): Update.
8480 (scale_bbs_frequencies_int): Update.
8481 (scale_bbs_frequencies_gcov_type): Update.
8482 (scale_bbs_frequencies_profile_count): Update.
8483 (scale_bbs_frequencies): Update.
8484 * cfganal.c (connect_infinite_loops_to_exit): Update.
8485 * cfgbuild.c (compute_outgoing_frequencies): Update.
8486 (find_many_sub_basic_blocks): Update.
8487 * cfgcleanup.c (try_forward_edges): Update.
8488 (try_crossjump_to_edge): Update
8489 * cfgexpand.c (expand_gimple_cond): Update
8490 (expand_gimple_tailcall): Update
8491 (construct_exit_block): Update
8492 * cfghooks.c (verify_flow_info): Update
8493 (redirect_edge_succ_nodup): Update
8494 (split_edge): Update
8495 (make_forwarder_block): Update
8496 (duplicate_block): Update
8497 (account_profile_record): Update
8498 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8499 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8500 * cfgloopmanip.c (scale_loop_profile): Update.
8502 (lv_adjust_loop_entry_edge): Update.
8503 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8504 (force_nonfallthru_and_redirect): Update.
8505 (purge_dead_edges): Update.
8506 (rtl_flow_call_edges_add): Update.
8507 * cgraphunit.c (init_lowered_empty_function): Update.
8508 (cgraph_node::expand_thunk): Update.
8509 * gimple-pretty-print.c (dump_probability): Update.
8510 (dump_edge_probability): Update.
8511 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8512 * haifa-sched.c (sched_create_recovery_edges): Update.
8513 * hsa-gen.c (convert_switch_statements): Update.
8514 * ifcvt.c (dead_or_predicable): Update.
8515 * ipa-inline-transform.c (inline_transform): Update.
8516 * ipa-split.c (split_function): Update.
8517 * ipa-utils.c (ipa_merge_profiles): Update.
8518 * loop-doloop.c (add_test): Update.
8519 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8520 * lto-streamer-in.c (input_cfg): Update.
8521 (input_function): Update.
8522 * lto-streamer-out.c (output_cfg): Update.
8523 * modulo-sched.c (sms_schedule): Update.
8524 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8525 * predict.c (maybe_hot_edge_p): Update.
8526 (unlikely_executed_edge_p): Update.
8527 (probably_never_executed_edge_p): Update.
8528 (dump_prediction): Update.
8529 (drop_profile): Update.
8530 (propagate_unlikely_bbs_forward): Update.
8531 (determine_unlikely_bbs): Update.
8532 (force_edge_cold): Update.
8533 * profile.c (compute_branch_probabilities): Update.
8534 * reg-stack.c (better_edge): Update.
8535 * shrink-wrap.c (handle_simple_exit): Update.
8536 * tracer.c (better_p): Update.
8537 * trans-mem.c (expand_transaction): Update.
8538 (split_bb_make_tm_edge): Update.
8539 * tree-call-cdce.c: Update.
8540 * tree-cfg.c (gimple_find_sub_bbs): Update.
8541 (gimple_split_edge): Update.
8542 (gimple_duplicate_sese_region): Update.
8543 (gimple_duplicate_sese_tail): Update.
8544 (gimple_flow_call_edges_add): Update.
8545 (insert_cond_bb): Update.
8546 (execute_fixup_cfg): Update.
8547 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8548 * tree-complex.c (expand_complex_div_wide): Update.
8549 * tree-eh.c (lower_resx): Update.
8550 (unsplit_eh): Update.
8551 (cleanup_empty_eh_move_lp): Update.
8552 * tree-inline.c (copy_edges_for_bb): Update.
8553 (freqs_to_counts): Update.
8554 (copy_cfg_body): Update.
8555 * tree-ssa-dce.c (remove_dead_stmt): Update.
8556 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8557 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8558 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8559 (unloop_loops): Update.
8560 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8561 * tree-ssa-loop-split.c (connect_loops): Update.
8562 (split_loop): Update.
8563 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8564 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8565 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8566 * tree-ssa-reassoc.c (branch_fixup): Update.
8567 * tree-ssa-tail-merge.c (replace_block_by): Update.
8568 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8569 (compute_path_counts): Update.
8570 (update_profile): Update.
8571 (recompute_probabilities): Update.
8572 (update_joiner_offpath_counts): Update.
8573 (estimated_freqs_path): Update.
8574 (freqs_to_counts_path): Update.
8575 (clear_counts_path): Update.
8576 (ssa_fix_duplicate_block_edges): Update.
8577 (duplicate_thread_path): Update.
8578 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8579 (case_bit_test_cmp): Update.
8580 (collect_switch_conv_info): Update.
8581 (gen_inbound_check): Update.
8582 (do_jump_if_equal): Update.
8583 (emit_cmp_and_jump_insns): Update.
8584 * tree-tailcall.c (decrease_profile): Update.
8585 (eliminate_tail_call): Update.
8586 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8587 (vect_do_peeling): Update.
8588 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8589 * ubsan.c (ubsan_expand_null_ifn): Update.
8590 (ubsan_expand_ptr_ifn): Update.
8591 * value-prof.c (gimple_divmod_fixed_value): Update.
8592 (gimple_mod_pow2): Update.
8593 (gimple_mod_subtract): Update.
8594 (gimple_ic): Update.
8595 (gimple_stringop_fixed_value): Update.
8597 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8600 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8602 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8604 PR rtl-optimization/82395
8605 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8606 based on non_spilled_static_chain_regno_p.
8608 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8610 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8611 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8612 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8614 2017-10-19 Martin Sebor <msebor@redhat.com>
8616 PR tree-optimization/82596
8617 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8619 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8621 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8622 (get_mem_refs_of_builtin_call): Likewise.
8623 * builtins.c (expand_builtin_apply): Adjust call to
8624 allocate_dynamic_stack_space.
8625 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8626 the third argument to allocate_dynamic_stack_space, otherwise -1.
8627 (expand_builtin): Deal with all alloca variants.
8628 (is_inexpensive_builtin): Likewise.
8629 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8630 * calls.c (special_function_p): Deal with all alloca variants.
8631 (initialize_argument_information): Adjust call to
8632 allocate_dynamic_stack_space.
8633 (expand_call): Likewise.
8634 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8635 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8636 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8637 use it for the stack usage computation.
8638 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8639 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8640 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8641 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8642 (in_loop_p): Remove first argument and useless check.
8643 (pass_walloca::execute): Remove useless test and adjust call to above.
8644 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8645 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8646 (gimplify_call_expr): Deal with all alloca variants.
8647 * hsa-gen.c (gen_hsa_alloca): Likewise.
8648 (gen_hsa_insns_for_call): Likewise.
8649 * ipa-pure-const.c (special_builtin_state): Likewise.
8650 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8651 * tree-object-size.c (alloc_object_size): Likewise.
8652 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8653 (call_may_clobber_ref_p_1): Likewise.
8654 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8655 (ccp_fold_stmt): Likewise.
8656 (optimize_stack_restore): Likewise.
8657 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8658 (mark_all_reaching_defs_necessary_1): Likewise.
8659 (propagate_necessity): Likewise.
8660 (eliminate_unnecessary_stmts): Likewise.
8661 * tree.c (build_common_builtin_nodes): Build
8662 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8663 (build_alloca_call_expr): New function.
8664 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8665 (CASE_BUILT_IN_ALLOCA): Likewise.
8666 (build_alloca_call_expr): Declare.
8667 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8669 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8672 * dwarf2out.c (new_die_raw): New static inline function.
8673 (new_die): Use it to create the DIE.
8674 (add_AT_external_die_ref): Likewise.
8675 (clone_die): Likewise.
8676 (clone_as_declaration): Likewise.
8677 (dwarf2out_vms_debug_main_pointer): Likewise.
8678 (base_type_die): Likewise. Remove early return for corner cases.
8679 Do not call add_pubtype on the DIE here.
8680 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8681 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8682 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8683 native order exists for base types, attach the DIE manually and call
8684 add_pubtype on it. Do not equate a reverse order DIE to the type.
8686 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8688 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8689 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8691 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8693 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8695 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8696 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8697 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8698 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8699 * opts.c (sanitizer_opts): Add builtin.
8700 * ubsan.c (instrument_builtin): New function.
8701 (pass_ubsan::execute): Call it.
8702 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8703 * doc/invoke.texi: Document -fsanitize=builtin.
8705 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8706 builtins, store max (log2 (align), 0) into uchar field instead of
8707 align into uptr field.
8708 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8709 store uchar 0 field instead of uptr 0 field.
8710 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8711 instead of passing one address of struct with 2 locations pass
8712 two addresses of structs with 1 location each.
8713 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8714 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8715 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8716 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8717 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8718 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8719 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8720 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8722 2017-10-19 Martin Liska <mliska@suse.cz>
8725 * file-find.c (remove_prefix): Remove.
8726 * file-find.h (remove_prefix): Likewise.
8727 * gcc-ar.c: Remove smartness of lookup.
8729 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8731 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8732 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8733 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8735 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8738 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8739 (setcc + and to xor + setcc): New peephole2.
8741 2017-10-19 Tom de Vries <tom@codesourcery.com>
8743 * doc/sourcebuild.texi (Test Directives, Variants of
8744 dg-require-support): Add dg-require-stack-size.
8746 2017-10-19 Martin Liska <mliska@suse.cz>
8749 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8750 that have a large alignment.
8751 (gimplify_target_expr): Likewise.
8753 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8755 PR rtl-optimization/82602
8756 * ira.c (rtx_moveable_p): Return false for volatile asm.
8758 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8761 * config/i386/i386-modes.def (CCGZ): New CC mode.
8762 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8763 * config/i386/predicates.md (ix86_comparison_operator):
8765 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8766 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8767 with double-word subtraction.
8768 (put_condition_code): Handle CCGZmode.
8770 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8772 * wide-int.cc (debug (const wide_int &)): New.
8773 (debug (const wide_int *)): New.
8774 (debug (const widest_int &)): New.
8775 (debug (const widest_int *)): New.
8777 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8780 * lra-constraints.c (curr_insn_transform): Use non-input operand
8781 instead of output one for matched reload.
8783 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8785 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8786 (tree-ssa-loop-ivopts.h): New header file.
8787 (struct builtin_info): New fields.
8788 (classify_builtin_1): Compute and record base and offset parts for
8789 memset builtin partition by calling strip_offset.
8790 (offset_cmp, fuse_memset_builtins): New functions.
8791 (finalize_partitions): Fuse adjacent memset partitions by calling
8793 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8794 Expose the interface.
8795 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8797 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8799 PR tree-optimization/82574
8800 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8801 that data reference must be executed exactly once per iteration
8802 against the outermost loop in nest.
8803 (classify_partition): Update call to above function.
8805 2017-10-18 Richard Biener <rguenther@suse.de>
8807 PR tree-optimization/82591
8808 * graphite.c (graphite_transform_loops): Move code gen message
8810 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8811 Here. Handle scop_to_isl_ast failing.
8812 (scop_to_isl_ast): Limit the number of ISL operations.
8814 2017-10-18 Richard Biener <rguenther@suse.de>
8816 * graphite-isl-ast-to-gimple.c
8817 (translate_isl_ast_to_gimple::set_rename): Simplify.
8818 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8819 (graphite_copy_stmts_from_block): ... here.
8820 (copy_bb_and_scalar_dependences): Simplify.
8821 (add_parameters_to_ivs_params): Canonicalize.
8822 (generate_entry_out_of_ssa_copies): Simplify.
8823 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8824 by passing in ISL dimension.
8825 (parameter_index_in_region_1): Rename to ...
8826 (parameter_index_in_region): ... this.
8827 (extract_affine): Adjust assert, pass down parameter index.
8828 (add_param_constraints): Use range-info when available.
8829 (build_scop_context): Adjust.
8830 * sese.c (new_sese_info): Adjust.
8831 (free_sese_info): Likewise.
8832 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8833 Remove unused typedefs.
8834 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8836 2017-10-18 Martin Liska <mliska@suse.cz>
8838 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8840 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8842 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8843 (s390_sched_init): Do not reset s390_sched_state if we entered the
8844 current basic block via a fallthru edge and all others are unlikely.
8846 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8848 * config/s390/s390.c (NUM_SIDES): New variable.
8849 (LONGRUNNING_THRESHOLD): New variable.
8850 (LATENCY_FACTOR): New variable.
8851 (s390_sched_score): Decrease score for long-running instructions on
8853 (s390_sched_variable_issue): Perform bookkeeping for long-running
8856 2017-10-18 Richard Biener <rguenther@suse.de>
8858 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8859 Simplify with removal of the parameter rename map.
8860 (set_rename): Likewise.
8861 (should_copy_to_new_region): Likewise.
8862 (graphite_copy_stmts_from_block): Likewise.
8863 (copy_bb_and_scalar_dependences): Remove initialization of
8864 unused copied_bb_map.
8866 (copy_internal_parameters): Likewise.
8867 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8868 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8869 Use INTEGRAL_TYPE_P.
8870 (parameter_index_in_region_1): Rename to ...
8871 (assign_parameter_index_in_region): ... this. Assert we have
8872 a parameter we handle.
8873 (scan_tree_for_params): Adjust.
8874 * sese.h (parameter_rename_map_t): Remove.
8875 (struct sese_info_t): Remove unused parameter_rename_map and
8876 copied_bb_map members.
8877 * sese.c (new_sese_info): Adjust.
8878 (free_sese_info): Likewise.
8880 2017-10-18 Martin Liska <mliska@suse.cz>
8883 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8884 on an abnormal edge.
8886 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8888 * doc/invoke.texi (ffunction-sections and fdata-sections):
8891 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8893 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8894 the use statement can throw internally.
8896 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8898 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8899 any RTX present on the RHS of a SET.
8900 * compare-elim.c (try_eliminate_compare): Restore comment.
8902 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8904 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8905 may be also called on tcc_type nodes.
8906 * langhooks.c (lhd_tree_size): Likewise.
8908 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8910 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8911 format_warning_at_substring.
8912 (maybe_warn): Convert source_range * param to a location_t. Pass
8913 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8914 (format_directive): Remove code to extract source_ranges and
8915 source_range * in favor of just a location_t.
8916 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8918 * substring-locations.c (format_warning_va): Convert
8919 source_range * param to a location_t.
8920 (format_warning_at_substring): Likewise.
8921 * substring-locations.h (format_warning_va): Likewise.
8922 (format_warning_at_substring): Likewise.
8924 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8926 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8928 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8929 and scatter/gather ops.
8931 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8932 vec_gather_load and vec_scatter_store.
8933 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8934 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8935 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8936 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8937 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8938 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8940 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8942 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8943 Detect FTST instruction and handle its register pops. Only pop
8944 second operand if can_pop_second_op is true.
8945 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8946 set can_pop_second_op to false in the compare_for_stack_reg call.
8948 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8949 output_fp_compare for stack register operands.
8950 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8951 instructions here. Do not emit stack register pops here. Assert
8952 that FCOMPP pops next to top stack register. Rewrite function.
8954 2017-10-17 Nathan Sidwell <nathan@acm.org>
8957 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8958 use DECL_ASSEMBLER_NAME_RAW.
8961 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8964 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8965 Wilco Dijkstra <wilco.dijkstra@arm.com>
8967 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8968 converstion to Pmode from the buf_addr.
8970 2017-10-17 Richard Biener <rguenther@suse.de>
8972 * graphite-dependences.c (scop_get_reads_and_writes): Change
8973 output parameters to references.
8975 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8977 PR 71026/tree-optimization
8978 * fold-const.c (distribute_real_division): Removed.
8979 (fold_binary_loc): Remove calls to distribute_real_divison.
8981 2017-10-17 Richard Biener <rguenther@suse.de>
8983 * graphite-scop-detection.c
8984 (scop_detection::stmt_has_simple_data_refs_p): Always use
8985 the full nest as region.
8986 (try_generate_gimple_bb): Likewise.
8987 * sese.c (scalar_evolution_in_region): Simplify now that
8988 SCEV can handle instantiation in regions.
8989 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8990 in the non-loop part of a function if requested.
8992 2017-10-17 Richard Biener <rguenther@suse.de>
8994 PR tree-optimization/82563
8995 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8997 (graphite_regenerate_ast_isl): Call it.
8998 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9000 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9002 PR tree-optimization/82549
9003 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9004 Formatting fixes. Instead of calling make_bit_field_ref with negative
9007 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9009 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9010 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9011 _mm_maskz_reduce_ss): New.
9012 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9013 __builtin_ia32_reducess_mask): Ditto..
9014 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9015 * config/i386/sse.md (reduces<mode>): Renamed to ...
9016 (reduces<mode><mask_scalar_name>): ... this.
9017 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9019 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9020 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9022 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9024 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9025 * selftest-run-tests.c (selftest::run_tests): Call
9026 selftest::unique_ptr_tests_cc_tests.
9027 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9028 * unique-ptr-tests.cc: New file.
9030 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9033 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9035 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9036 (make_hard_regno_born, make_hard_regno_dead): Update
9037 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9039 2017-10-16 Jeff Law <law@redhat.com>
9041 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9043 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9045 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9047 2017-10-16 Olivier Hainque <hainque@adacore.com>
9049 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9050 with_cpu if we were configured for an e500v2 target cpu name.
9052 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9054 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9055 * doc/invoke.texi: Document +nodsp as a valid extension for
9058 2017-10-16 Martin Liska <mliska@suse.cz>
9060 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9061 (test_set_range): Likewise.
9062 (test_range_functions): Rename to ...
9063 (test_bit_in_range): ... this.
9064 (sbitmap_c_tests): Add new test.
9066 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9068 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9070 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9071 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9073 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9075 * config/aarch64/aarch64-builtins.c
9076 (aarch64_types_quadopu_lane_qualifiers): New.
9077 (TYPES_QUADOPU_LANE): New.
9078 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9079 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9080 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9081 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9082 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9083 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9084 (Vdottype, DOTPROD): New.
9085 (sur): Add SDOT and UDOT.
9087 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9089 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9090 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9091 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9093 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9094 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9095 Enable TARGET_DOTPROD.
9096 (cortex-a75.cortex-a55): Likewise.
9097 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9099 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9101 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9102 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9104 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9106 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9107 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9108 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9109 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9110 * config/arm/types.md (neon_dot, neon_dot_q): New.
9111 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9113 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9115 * config/arm/arm.h (TARGET_DOTPROD): New.
9116 * config/arm/arm.c (arm_arch_dotprod): New.
9117 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9118 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9119 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9120 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9121 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9122 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9123 * doc/invoke.texi (armv8.2-a): Document dotprod
9125 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9127 * i386.c (ix86_vec_cost): New function.
9128 (ix86_rtx_costs): Handle vector operations better.
9129 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9130 * x86-tune-costs.h: Add new costs to all tables.
9132 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9134 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9136 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9137 divsd, sqrtss and sqrtsd
9138 * x86-tune-costs.h: Add new entries to all costs.
9139 (znver1_cost): Fix to match real instruction latencies.
9141 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9142 Michael Collison <michael.collison@arm.com>
9144 * compare-elim.c: Include emit-rtl.h.
9145 (can_merge_compare_into_arith): New function.
9146 (try_validate_parallel): Likewise.
9147 (try_merge_compare): Likewise.
9148 (try_eliminate_compare): Call the above when no previous clobber
9150 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9153 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9157 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9158 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9162 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9163 to be any operand_equal_p operands. For & (B - 1) require
9164 B to be power of 2. Recognize
9165 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9167 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9170 * optabs.c (expand_memory_blockage): Fix call of
9171 targetm.have_memory_blockage.
9173 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9176 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9177 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9178 objects to extra_objs instead of overwriting it.
9180 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9182 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9183 Use any_fp_register_operand as operand[3] predicate. Simplify
9184 equality test for operands[2] and operands[4] memory location.
9185 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9186 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9187 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9188 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9189 any_fp_register_operand as operand[1] predicate. Simplify
9190 equality test for operands[0] and operands[3] memory location.
9191 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9192 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9193 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9195 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9197 * target-insns.def: Add memory_blockage.
9198 * optabs.c (expand_memory_blockage): New function.
9199 (expand_asm_memory_barrier): Rename ...
9200 (expand_asm_memory_blockage): ... to this.
9201 (expand_mem_thread_fence): Call expand_memory_blockage
9202 instead of expand_asm_memory_barrier.
9203 (expand_mem_singnal_fence): Ditto.
9204 (expand_atomic_load): Ditto.
9205 (expand_atomic_store): Ditto.
9206 * doc/md.texi (Standard Pattern Names For Generation):
9207 Document memory_blockage instruction pattern.
9209 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9211 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9212 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9213 * config/rl78/rl78.md: New define_expand "adddi3".
9215 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9217 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9220 2017-10-13 Jeff Law <law@redhat.com>
9222 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9224 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9227 * internal-fn.c (expand_mul_overflow): If both operands have
9228 the same highpart of -1 or 0 and the topmost bit of lowpart
9229 is different, overflow is if res <= 0 rather than res < 0.
9231 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9233 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9234 TARGET_P9_VECTOR code for unaligned_load case.
9236 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9238 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9240 2017-10-13 Nathan Sidwell <nathan@acm.org>
9242 * tree-core.h (tree_contains_struct): Make bool.
9243 * tree.c (tree_contains_struct): Likewise.
9244 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9245 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9246 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9247 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9248 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9250 2017-10-13 Richard Biener <rguenther@suse.de>
9252 * graphite-isl-ast-to-gimple.c
9253 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9254 parameters and dominance check.
9255 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9256 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9257 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9258 Do not update SSA form here or do intermediate IL verification.
9259 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9260 (graphite_initialize): Remove check on the number of loops in
9261 the function and inline into graphite_transform_loops.
9262 (graphite_finalize): Inline into graphite_transform_loops.
9263 (graphite_transform_loops): Perform SSA update and IL verification
9265 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9267 2017-10-13 Richard Biener <rguenther@suse.de>
9269 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9270 graphite_expression_type_precision): Avoid global constructor
9272 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9273 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9274 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9275 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9276 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9277 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9279 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9282 * config/i386/i386.h (ix86_red_zone_size): New.
9283 * config/i386/i386.md (push peephole2s): Replace
9284 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9286 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9287 Alan Hayward <alan.hayward@arm.com>
9288 David Sherwood <david.sherwood@arm.com>
9290 * combine.c (can_change_dest_mode): Reject changes in
9291 REGMODE_NATURAL_SIZE.
9293 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9294 Alan Hayward <alan.hayward@arm.com>
9295 David Sherwood <david.sherwood@arm.com>
9297 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9298 (expand_debug_source_expr): Likewise.
9299 * combine.c (combine_simplify_rtx): Likewise.
9300 * cse.c (fold_rtx): Likewise.
9301 * fwprop.c (canonicalize_address): Likewise.
9302 * targhooks.c (default_shift_truncation_mask): Likewise.
9304 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9305 Alan Hayward <alan.hayward@arm.com>
9306 David Sherwood <david.sherwood@arm.com>
9308 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9309 (widened_mode): Likewise.
9310 (expand_unop): Likewise.
9311 * ree.c (transform_ifelse): Likewise.
9312 (merge_def_and_ext): Likewise.
9313 (combine_reaching_defs): Likewise.
9314 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9316 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9317 Alan Hayward <alan.hayward@arm.com>
9318 David Sherwood <david.sherwood@arm.com>
9320 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9321 * combine.c (gen_lowpart_for_combine): Likewise.
9322 * dwarf2out.c (rtl_for_decl_location): Likewise.
9323 * final.c (alter_subreg): Likewise.
9324 * rtlhooks.c (gen_lowpart_general): Likewise.
9325 (gen_lowpart_if_possible): Likewise.
9327 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9328 Alan Hayward <alan.hayward@arm.com>
9329 David Sherwood <david.sherwood@arm.com>
9331 * calls.c (expand_call): Use subreg_lowpart_offset.
9332 * cse.c (cse_insn): Likewise.
9333 * regcprop.c (copy_value): Likewise.
9334 (copyprop_hardreg_forward_1): Likewise.
9336 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9339 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9340 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9341 =Q constraints to +Q and into insn condition add check
9342 that operands[0] and operands[1] are equal.
9343 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9344 =Q constraints to +Q and into insn condition add check
9345 that operands[0] is equal to either operands[1] or operands[2].
9348 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9349 instead of handling MINUS_EXPR twice (once for each argument),
9350 canonicalize operand order and handle just once, use rtype where
9351 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9354 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9355 any values of __C while still being pattern recognizable as a simple
9358 2017-10-13 Richard Biener <rguenther@suse.de>
9360 PR tree-optimization/82451
9362 2017-10-02 Richard Biener <rguenther@suse.de>
9364 PR tree-optimization/82355
9365 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9366 a mapping for the enclosing loop but avoid generating one for
9368 (copy_bb_and_scalar_dependences): Remove premature codegen
9369 error on PHIs in blocks duplicated into multiple places.
9370 * graphite-scop-detection.c
9371 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9372 in the region use it as loop and nest to analyze the DR in.
9373 (try_generate_gimple_bb): Likewise.
9374 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9375 (add_loop_constraints): For blocks in a loop not in the region
9376 create a dimension with a single iteration.
9377 * sese.h (gbb_loop_at_index): Remove assert.
9379 * cfgloop.c (loop_preheader_edge): For the loop tree root
9380 return the single successor of the entry block.
9381 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9382 Reset the SCEV hashtable and niters.
9383 * graphite-scop-detection.c
9384 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9385 assert that we only have POLYNOMIAL_CHREC that vary in loops
9386 contained in the region.
9387 (scop_detection::graphite_can_represent_expr): Adjust.
9388 (scop_detection::stmt_has_simple_data_refs_p): For loops
9389 not in the region set loop to NULL. The nest is now the
9390 entry edge to the region.
9391 (try_generate_gimple_bb): Likewise.
9392 * sese.c (scalar_evolution_in_region): Adjust for
9393 instantiate_scev change.
9394 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9395 Make nest parameter the edge into the region.
9396 (create_data_ref): Likewise.
9397 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9398 entry edge into a region and adjust instantiate_scev calls.
9399 (create_data_ref): Likewise.
9400 (graphite_find_data_references_in_stmt): Likewise.
9401 (find_data_references_in_stmt): Pass the loop preheader edge
9402 from the nest argument.
9403 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9404 parameter the edge into the region.
9405 (instantiate_parameters): Use the loop preheader edge as entry.
9406 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9408 (get_instantiated_value_entry): Make instantiate_below parameter
9409 the edge into the region.
9410 (instantiate_scev_name): Likewise. Adjust dominance checks,
9411 when we cannot use loop-based instantiation instantiate by
9412 walking use-def chains.
9413 (instantiate_scev_poly): Adjust.
9414 (instantiate_scev_binary): Likewise.
9415 (instantiate_scev_convert): Likewise.
9416 (instantiate_scev_not): Likewise.
9417 (instantiate_array_ref): Remove.
9418 (instantiate_scev_3): Likewise.
9419 (instantiate_scev_2): Likewise.
9420 (instantiate_scev_1): Likewise.
9421 (instantiate_scev_r): Do not blindly handle N-operand trees.
9422 Do not instantiate array-refs. Handle all constants and invariants.
9423 (instantiate_scev): Make instantiate_below parameter
9424 the edge into the region.
9425 (resolve_mixers): Use the loop preheader edge for the region
9426 parameter to instantiate_scev_r.
9427 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9429 2017-10-13 Richard Biener <rguenther@suse.de>
9431 PR tree-optimization/82525
9432 * graphite-isl-ast-to-gimple.c
9433 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9435 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9436 Fail code generation when we cannot represent the isl integer.
9437 (binary_op_to_tree): Elide modulo operations that are no-ops
9438 in the type we code generate. Remove now superfluous code
9441 2017-10-13 Richard Biener <rguenther@suse.de>
9443 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9444 (scop_detection::harmful_loop_in_region): Remove premature
9445 IV type restriction.
9446 (scop_detection::graphite_can_represent_scev): We can handle
9447 pointer IVs just fine.
9449 2017-10-13 Alan Modra <amodra@gmail.com>
9451 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9452 "Clobbers and Scratch Registers". Add paragraph on
9453 alternative to clobbers for scratch registers and OpenBLAS
9456 2017-10-13 Alan Modra <amodra@gmail.com>
9458 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9459 example of a memory input for a string of known length. Move
9460 commentary out of table. Add a number of new examples
9461 covering array memory inputs.
9463 2017-10-12 Martin Liska <mliska@suse.cz>
9465 PR tree-optimization/82493
9466 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9467 (test_range_functions): New function.
9468 (sbitmap_c_tests): Likewise.
9469 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9470 * selftest.h (sbitmap_c_tests): New function.
9472 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9474 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9476 * config/rs6000/amo.h: Fix spacing issue.
9478 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9481 * config/i386/i386.md (*ashl<mode>3_mask_1,
9482 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9483 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9486 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9488 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9489 (profile_probability): Set max_probability
9490 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9492 * profile-count.c (profile_probability::differs_from_p): Do not
9493 rely on max_probaiblity == 10000
9495 2017-10-12 Jeff Law <law@redhat.com>
9497 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9500 2017-10-12 Martin Sebor <msebor@redhat.com>
9504 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9505 (handle_alias_pairs): Call it.
9506 * common.opt (-Wattribute-alias): New option.
9507 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9508 * doc/invoke.texi (-Wattribute-alias): Document.
9510 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9513 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9515 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9517 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9518 (make_hard_regno_born, make_hard_regno_dead): Update
9519 bb_killed_pseudos and bb_gen_pseudos.
9521 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9523 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9525 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9527 * config/alpha/alpha.c (alpha_split_conditional_move):
9528 Use std::swap instead of manually swapping.
9529 (alpha_stdarg_optimize_hook): Ditto.
9530 (alpha_canonicalize_comparison): Ditto.
9532 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9534 * tree-loop-distribution.c (struct builtin_info): New struct.
9535 (struct partition): Refactor fields into struct builtin_info.
9536 (partition_free): Free struct builtin_info.
9537 (build_size_arg_loc, build_addr_arg_loc): Delete.
9538 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9539 information from struct builtin_info.
9540 (find_single_drs): New function refactored from classify_partition.
9541 Also moved builtin validity checks to this function.
9542 (compute_access_range, alloc_builtin): New functions.
9543 (classify_builtin_st, classify_builtin_ldst): New functions.
9544 (classify_partition): Refactor code into functions find_single_drs,
9545 classify_builtin_st and classify_builtin_ldst.
9546 (distribute_loop): Don't do runtime alias check when distributing
9548 (find_seed_stmts_for_distribution): New function.
9549 (pass_loop_distribution::execute): Refactor code finding seed
9550 stmts into above function. Support distribution for the innermost
9551 two-level loop nest. Adjust dump information.
9553 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9555 * tree-loop-distribution.c: Adjust the general comment.
9556 (NUM_PARTITION_THRESHOLD): New macro.
9557 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9558 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9559 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9560 in call to build_partition_graph.
9561 (finalize_partitions): New parameter. Make loop distribution more
9562 conservative by fusing more partitions.
9563 (distribute_loop): Don't do runtime alias check in case of loop nest
9565 (find_seed_stmts_for_distribution): New function.
9566 (prepare_perfect_loop_nest): New function.
9567 (pass_loop_distribution::execute): Refactor code finding seed stmts
9568 and loop nest into above functions. Support loop nest distribution.
9569 Adjust dump information accordingly.
9571 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9573 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9574 and set PTYPE_SEQUENTIAL for merged partition.
9576 2017-10-12 Richard Biener <rguenther@suse.de>
9578 PR tree-optimization/69728
9580 2017-09-19 Richard Biener <rguenther@suse.de>
9582 PR tree-optimization/69728
9583 * graphite-sese-to-poly.c (schedule_error): New global.
9584 (add_loop_schedule): Handle empty domain by failing the
9586 (build_original_schedule): Handle schedule_error.
9588 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9589 domain by returning an unchanged schedule.
9591 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9593 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9594 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9596 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9598 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9601 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9604 * expr.c (store_field): Don't optimize away bitsize == 0 store
9605 from CALL_EXPR with addressable return type.
9607 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9609 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9610 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9611 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9612 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9613 TARGET_ISEL instead of TARGET_ISEL<sel>.
9615 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9617 * config/rs6000/rs6000.c
9618 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9620 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9622 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9624 (reg_or_cint_operand): Fix comment.
9625 (reg_or_zero_operand): New predicate.
9626 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9627 * config/rs6000/rs6000.c (output_isel): Delete.
9628 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9629 instead of reg_or_cint_operand. Output instruction directly (not via
9631 (isel_unsigned_<mode>): Ditto.
9632 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9633 gpc_reg_operand. Add an instruction alternative for this. Output
9634 instruction directly.
9635 (*isel_reversed_unsigned_<mode>): Ditto.
9637 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9639 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9640 (TARGET_CANONICALIZE_COMPARISON): Define.
9642 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9645 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9646 Check whether the dest is REG before adding REG_EQUIV note.
9648 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9651 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9653 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9654 (make_hard_regno_born, make_hard_regno_dead): Update
9655 bb_killed_pseudos and bb_gen_pseudos.
9657 2017-10-11 Nathan Sidwell <nathan@acm.org>
9659 * incpath.h (enum incpath_kind): Name enum, prefix values.
9660 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9661 * incpath.c (heads, tails): Use INC_MAX.
9662 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9663 (merge_include_chains, split_quote_chain,
9664 register_include_chains): Update incpath_kind names.
9665 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9666 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9668 (add_framework_path, darwin_register_objc_includes): Likewise.
9669 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9671 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9673 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9674 Do not use float_operator operator predicate.
9675 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9676 * config/i386/predicates.md (float_operator): Remove predicate.
9678 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9680 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9681 (*jccxf_i387): Ditto.
9682 (*jcc<mode>_i387): Ditto.
9683 (*jccu<mode>_i387): Ditto.
9684 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9685 (*jcc_*_i387 splitters): Remove.
9686 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9687 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9688 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9691 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9693 * profile-count.h (slow_safe_scale_64bit): New function.
9694 (safe_scale_64bit): New inline.
9695 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9696 * profile-count.c: Include wide-int.h
9697 (slow_safe_scale_64bit): New.
9699 2017-10-11 Nathan Sidwell <nathan@acm.org>
9701 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9702 HAS_DECL_ASSEMBLER_NAME_P.
9703 * gimple-expr.c (gimple_decl_printable_name: Check
9704 HAS_DECL_ASSEMBLER_NAME_P too.
9705 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9706 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9707 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9708 * passes.c (rest_of_decl_compilation): Check
9709 HAS_DECL_ASSEMBLER_NAME_P too.
9710 * recog.c (verify_changes): Likewise.
9711 * tree-pretty-print.c (dump_decl_name): Likewise.
9712 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9714 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9715 (SET_DECL_ASSEMBLER_NAME): Use it.
9716 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9717 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9718 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9720 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9722 * config.gcc (i386, x86_64): Add extra objects.
9723 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9724 (ix86_min_insn_size): Declare.
9725 (ix86_issue_rate): Declare.
9726 (ix86_adjust_cost): Declare.
9727 (ia32_multipass_dfa_lookahead): Declare.
9728 (ix86_macro_fusion_p): Declare.
9729 (ix86_macro_fusion_pair_p): Declare.
9730 (ix86_bd_has_dispatch): Declare.
9731 (ix86_bd_do_dispatch): Declare.
9732 (ix86_core2i7_init_hooks): Declare.
9733 (ix86_atom_sched_reorder): Declare.
9734 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9735 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9736 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9737 (rip_relative_addr_p): Rename to ...
9738 (ix86_rip_relative_addr_p): ... this one; export.
9739 (memory_address_length): Update.
9740 (ix86_issue_rate): Move to x86-tune-sched.c.
9741 (ix86_flags_dependent): Move to x86-tune-sched.c.
9742 (ix86_agi_dependent): Move to x86-tune-sched.c.
9743 (exact_dependency_1): Move to x86-tune-sched.c.
9744 (exact_store_load_dependency): Move to x86-tune-sched.c.
9745 (ix86_adjust_cost): Move to x86-tune-sched.c.
9746 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9747 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9748 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9749 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9750 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9751 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9752 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9753 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9754 (min_insn_size): Rename to ...
9755 (ix86_min_insn_size): ... this one; export.
9756 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9757 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9758 (core2i7_first_cycle_multipass_backtrack): Move to
9759 x86-tune-sched-core.c.
9760 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9761 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9762 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9763 (ix86_avoid_jump_mispredicts): Update.
9764 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9765 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9766 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9767 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9768 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9769 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9770 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9771 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9772 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9773 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9774 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9775 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9776 (BIG): Move to ix86-tune-sched-bd.c.
9777 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9778 (enum insn_path): Move to ix86-tune-sched-bd.c.
9779 (get_mem_group): Move to ix86-tune-sched-bd.c.
9780 (is_cmp): Move to ix86-tune-sched-bd.c.
9781 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9782 (is_branch): Move to ix86-tune-sched-bd.c.
9783 (is_prefetch): Move to ix86-tune-sched-bd.c.
9784 (init_window): Move to ix86-tune-sched-bd.c.
9785 (allocate_window): Move to ix86-tune-sched-bd.c.
9786 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9787 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9788 (process_end_window): Move to ix86-tune-sched-bd.c.
9789 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9790 (find_constant): Move to ix86-tune-sched-bd.c.
9791 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9792 (has_immediate): Move to ix86-tune-sched-bd.c.
9793 (get_insn_path): Move to ix86-tune-sched-bd.c.
9794 (get_insn_group): Move to ix86-tune-sched-bd.c.
9795 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9796 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9797 (add_insn_window): Move to ix86-tune-sched-bd.c.
9798 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9799 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9800 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9801 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9802 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9803 (do_dispatch): Move to ix86-tune-sched-bd.c.
9804 (has_dispatch): Move to ix86-tune-sched-bd.c.
9805 * config/i386/t-i386: Add new object files.
9806 * config/i386/x86-tune-costs.h: New file.
9807 * config/i386/x86-tune-sched-atom.c: New file.
9808 * config/i386/x86-tune-sched-bd.c: New file.
9809 * config/i386/x86-tune-sched-core.c: New file.
9810 * config/i386/x86-tune-sched.c: New file.
9812 2017-10-11 Liu Hao <lh_mouse@126.com>
9814 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9815 the generic version below instead.
9816 (should_colorize): Recognize Windows consoles as terminals
9818 * pretty-print.c [__MINGW32__] (write_all): New function.
9819 [__MINGW32__] (find_esc_head): Likewise.
9820 [__MINGW32__] (find_esc_terminator): Likewise.
9821 [__MINGW32__] (eat_esc_sequence): Likewise.
9822 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9824 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9827 2017-10-11 Richard Biener <rguenther@suse.de>
9829 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9830 Properly call analyze_scalar_evolution with the loop of the stmt.
9832 2017-10-11 Richard Biener <rguenther@suse.de>
9834 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9835 * tree-core.h (tree_base): Add chrec_var union member.
9836 * tree.h (CHREC_VAR): Remove.
9837 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9838 * tree-chrec.h (build_polynomial_chrec): Adjust.
9839 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9840 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9842 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9844 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9845 * match.pd: ... here.
9846 ((T) X == (T) Y): Relax condition.
9848 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9850 PR tree-optimization/82472
9851 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9853 (break_alias_scc_partitions): Update postorder number.
9855 2017-10-11 Martin Liska <mliska@suse.cz>
9858 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9860 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9862 2017-10-11 Martin Liska <mliska@suse.cz>
9864 * print-rtl.c (print_insn): Move declaration of idbuf
9865 to same scope as name.
9867 2017-10-11 Martin Liska <mliska@suse.cz>
9872 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9875 2017-10-11 Martin Liska <mliska@suse.cz>
9878 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9881 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9883 * config.gcc (default_gnu_indirect_function): Default to yes for
9884 arm*-*-linux* with glibc.
9886 2017-10-11 Richard Biener <rguenther@suse.de>
9888 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9889 default-defs and types we do not want to analyze.
9890 (interpret_loop_phi): Replace unreachable code with an assert.
9891 (compute_scalar_evolution_in_loop): Remove and inline ...
9892 (analyze_scalar_evolution_1): ... here, replacing condition with
9893 what makes the intent clearer. Remove handling of cases
9894 get_scalar_evolution now handles.
9896 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9898 PR rtl-optimization/81434
9899 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9900 comment for main loop. In sched_group_found if, also add checks for
9901 pass and min_cost_group.
9903 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9905 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9906 (rs6000_insn_cost): New function.
9907 * config/rs6000/rs6000.md (cost): New attribute.
9909 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9910 H.J. Lu <hongjiu.lu@intel.com>
9914 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9915 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9916 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9917 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9918 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9919 OPTION_MASK_ISA_MMX and some other bit set require both
9920 mmx and the ISAs without the mmx bit.
9921 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9922 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9923 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9924 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9925 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9926 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9927 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9928 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9929 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9930 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9931 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9932 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9933 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9934 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9935 Add OPTION_MASK_ISA_MMX.
9937 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9939 * config.gcc (armv7*-*-freebsd*): New target.
9940 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9942 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9944 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9945 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9948 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9950 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9951 a template parameter.
9952 (WIDE_INT_REF_FOR): Update accordingly.
9953 * tree.h (wi::int_traits <const_tree>): Delete.
9954 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9955 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9956 (wi::tree_to_wide_ref): New typedef.
9957 (wi::to_wide): New function.
9958 * calls.c (get_size_range): Use wi::to_wide when operating on
9960 * cgraph.c (cgraph_node::create_thunk): Likewise.
9961 * config/i386/i386.c (ix86_data_alignment): Likewise.
9962 (ix86_local_alignment): Likewise.
9963 * dbxout.c (stabstr_O): Likewise.
9964 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9965 * expr.c (const_vector_from_tree): Likewise.
9966 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9967 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9968 (fold_negate_expr_1, int_const_binop_1, const_binop)
9969 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9970 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9971 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9972 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9973 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9974 (fold_not_const, round_up_loc): Likewise.
9975 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9976 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9977 (alloca_call_type): Likewise.
9978 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9979 * godump.c (go_output_typedef): Likewise.
9980 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9981 * internal-fn.c (get_min_precision): Likewise.
9982 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9983 * ipa-polymorphic-call.c
9984 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9985 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9986 (ipa_modify_call_arguments): Likewise.
9987 * match.pd: Likewise.
9988 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9989 * print-tree.c (print_node_brief, print_node): Likewise.
9990 * stmt.c (expand_case): Likewise.
9991 * stor-layout.c (layout_type): Likewise.
9992 * tree-affine.c (tree_to_aff_combination): Likewise.
9993 * tree-cfg.c (group_case_labels_stmt): Likewise.
9994 * tree-data-ref.c (dr_analyze_indices): Likewise.
9995 (prune_runtime_alias_test_list): Likewise.
9996 * tree-dump.c (dequeue_and_dump): Likewise.
9997 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9998 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9999 * tree-pretty-print.c (dump_generic_node): Likewise.
10000 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10001 (simple_iv_with_niters): Likewise.
10002 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10003 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10004 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10005 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10006 (refine_value_range_using_guard, number_of_iterations_ne_max)
10007 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10008 (get_cst_init_from_scev, record_nonwrapping_iv)
10009 (scev_var_range_cant_overflow): Likewise.
10010 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10011 * tree-ssa-pre.c (compute_avail): Likewise.
10012 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10013 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10014 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10015 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10016 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10017 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10018 (dump_case_nodes, try_switch_expansion): Likewise.
10019 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10020 (vect_do_peeling): Likewise.
10021 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10022 * tree-vect-stmts.c (vectorizable_load): Likewise.
10023 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10024 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10025 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10026 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10027 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10028 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10029 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10030 (evrp_dom_walker::before_dom_children): Likewise.
10031 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10032 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10033 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10034 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10035 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10036 (get_range_pos_neg): Likewise.
10037 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10038 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10039 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10040 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10041 * config/avr/avr.c (avr_fold_builtin): Likewise.
10042 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10043 * config/msp430/msp430.c (msp430_attr): Likewise.
10044 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10045 * config/powerpcspe/powerpcspe-c.c
10046 (altivec_resolve_overloaded_builtin): Likewise.
10047 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10048 (rs6000_expand_ternop_builtin): Likewise.
10049 * config/rs6000/rs6000-c.c
10050 (altivec_resolve_overloaded_builtin): Likewise.
10051 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10052 (rs6000_expand_ternop_builtin): Likewise.
10053 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10055 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10057 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10058 when copying loop nest with only one inner loop.
10060 2017-10-10 Richard Biener <rguenther@suse.de>
10062 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10063 blocks if SCEV is active.
10064 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10066 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10067 (scev_initialize): Assert we are not yet initialized.
10069 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10071 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10072 inner loop's exit stmt by making it always exit the loop, otherwise
10073 we would generate an infinite empty loop.
10075 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10077 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10078 renaming variables in new preheader if it's deleted.
10080 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10082 * tree-loop-distribution.c (struct partition): Remove unused field
10083 loops of the structure.
10084 (partition_alloc, partition_free): Ditto.
10085 (build_rdg_partition_for_vertex): Ditto.
10087 2017-10-09 Jeff Law <law@redhat.com>
10089 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10090 return type to match prototype and documentation.
10092 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10094 * config/rs6000/rs6000.c (processor_costs): Move to ...
10095 * config/rs6000/rs6000.h: ... here.
10096 (rs6000_cost): Declare.
10098 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10100 * except.c (setjmp_fn): New global variable.
10101 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10102 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10103 if DONT_USE_BUILTIN_SETJMP is defined.
10105 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10107 * target.def (insn_cost): New hook.
10108 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10109 * doc/tm.texi: Regenerate.
10110 * rtlanal.c (insn_cost): Use the new hook.
10112 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10114 * combine.c (combine_validate_cost): Compute the new insn_cost,
10115 not just pattern_cost.
10116 (try_combine): Adjust comment.
10118 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10120 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10122 * combine.c (uid_insn_cost): Adjust comment.
10123 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10125 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10126 * dse.c (find_shift_sequence): Ditto.
10127 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10128 (bb_valid_for_noce_process_p): Use pattern_cost.
10129 * rtl.h (insn_rtx_cost): Delete.
10130 (pattern_cost): New prototype.
10131 (insn_cost): New prototype.
10132 * rtlanal.c (insn_rtx_cost): Rename to...
10133 (pattern_cost): ... this.
10136 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10138 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10139 (*jcc<mode>_0_r_i387): Ditto.
10140 (*jccxf_r_i387): Ditto.
10141 (*jcc<mode>_r_i387): Ditto.
10142 (*jccu<mode>_r_i387): Ditto.
10143 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10144 (*jcc): Rename from *jcc_1.
10146 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10148 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10149 deferred rescans after the lvx/stvx recombination pre-pass.
10151 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10153 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10154 memory operation instruction support.
10155 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10156 (rs6000-ibm-aix[789]*): Likewise.
10157 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10158 Document new functions.
10160 2017-10-09 Richard Biener <rguenther@suse.de>
10162 PR tree-optimization/82397
10163 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10164 equality only for semantically equal trees.
10166 2017-10-09 Richard Biener <rguenther@suse.de>
10168 PR tree-optimization/82449
10169 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10170 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10171 allow constant addresses.
10172 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10175 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10177 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10180 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10183 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10186 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10189 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10191 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10194 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10195 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10197 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10199 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10200 (WI_BINARY_PREDICATE_RESULT): Likewise.
10201 (wi::binary_traits::operator_result): New type.
10202 (wi::binary_traits::predicate_result): Likewise.
10203 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10204 (generic_wide_int::operator==, generic_wide_int::operator!=)
10205 (generic_wide_int::operator&, generic_wide_int::and_not)
10206 (generic_wide_int::operator|, generic_wide_int::or_not)
10207 (generic_wide_int::operator^, generic_wide_int::operator+
10208 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10210 (operator~, unary operator-, operator==, operator!=, operator&)
10211 (operator|, operator^, operator+, binary operator-, operator*): New
10213 * expr.c (get_inner_reference): Use wi::bit_and_not.
10214 * fold-const.c (fold_binary_loc): Likewise.
10215 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10216 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10217 (bit_value_binop): Likewise.
10218 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10219 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10220 (extract_range_from_binary_expr_1): Likewise.
10221 (masked_increment): Likewise.
10222 (simplify_bit_ops_using_ranges): Likewise.
10224 2017-10-09 Martin Jambor <mjambor@suse.cz>
10227 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10228 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10229 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10231 (hsa_fixup_mov_insn_type): New function.
10232 (hsa_op_with_type::get_in_type): Use it.
10233 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10234 immediates in an assert.
10235 (hsa_op_with_type::extend_int_to_32bit): New method.
10236 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10237 types. Convert to dest type if necessary.
10238 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10239 (reg_for_gimple_ssa): Pass false as min32int to
10240 hsa_type_for_scalar_tree_type.
10241 (gen_hsa_addr): Fixup type when creating addresable temporary.
10242 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10243 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10244 necessary. Call hsa_fixup_mov_insn_type.
10245 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10246 extend operands and convert to dest type if necessary.
10247 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10248 to dest type if necessary.
10249 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10250 if conversion nt necessary and size matches.
10251 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10252 to dest type if necessary.
10253 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10254 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10256 (gen_hsa_clrsb): Likewise.
10257 (gen_hsa_ffs): Likewise.
10258 (gen_hsa_divmod): Extend operands and convert to dest type if
10260 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10262 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10264 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10265 Remove empty default arguments. Use a brace block as output
10267 (conditional return): Ditto.
10269 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10270 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10272 (group_ending_nop): Ditto.
10273 (doloop_end): Ditto.
10274 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10275 (splitters for those): Ditto.
10277 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10279 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10280 a conditional jump (and the compare for it) so that pc_rtx is the
10282 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10283 for the deleted and renamed ctr<mode>_internal[234] patterns.
10284 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10285 Delete second conditional return pattern.
10286 (ctr<mode>_internal2): Delete this second bdnz pattern.
10287 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10288 (ctr<mode>_internal4): Delete this second bdz pattern.
10290 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10292 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10293 (always_initialized_rtx_for_ssa_name_p): New predicate.
10294 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10295 (finish_out_of_ssa): Free new field of SA.
10296 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10297 * tree-ssa-coalesce.c: Include tree-ssa.h.
10298 (get_parm_default_def_partitions): Remove extern keyword.
10299 (get_undefined_value_partitions): New function.
10300 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10301 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10302 uninitialized bits.
10303 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10305 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10307 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10309 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10311 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10312 for targets that preffer 128bit.
10314 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10316 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10318 2017-10-08 Olivier Hainque <hainque@adacore.com>
10320 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10321 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10323 (thumb_set_return_address): Likewise.
10325 2017-10-08 Olivier Hainque <hainque@adacore.com>
10327 * common/config/arm/arm-common.c (arm_except_unwind_info):
10328 Handle DWARF2_UNWIND_INFO.
10330 2017-10-07 Michael Collison <michael.collison@arm.com>
10332 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10335 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10337 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10338 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10339 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10340 defined, force the creation of a new block for a dispatch label.
10342 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10344 * invoke.texi (Wsuggest-attribute=cold): Document.
10345 * common.opt (Wsuggest-attribute=cold): New
10346 * ipa-pure-const.c (warn_function_cold): New function.
10347 * predict.c (compute_function_frequency): Use it.
10348 * predict.h (warn_function_cold): Declare.
10350 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10352 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10355 2017-10-06 Martin Liska <mliska@suse.cz>
10357 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10358 keyword for member functions.
10359 (struct sanopt_tree_couple): New struct.
10360 (struct sanopt_tree_couple_hash): New function.
10361 (struct sanopt_ctx): Add new hash_map.
10362 (has_dominating_ubsan_ptr_check): New function.
10363 (record_ubsan_ptr_check_stmt): Likewise.
10364 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10365 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10366 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10368 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10371 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10372 aarch64_simd_valid_immediate on CONST_VECTORs.
10373 (aarch64_reg_or_bic_imm): Likewise.
10375 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10377 PR rtl-optimization/82396
10378 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10380 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10382 * graphite-dependences.c (scop_get_reads): Move code to...
10383 (scop_get_must_writes): Move code to...
10384 (scop_get_may_writes): Move code to...
10385 (scop_get_reads_and_writes): ... here.
10386 (scop_get_dependences): Call scop_get_reads_and_writes.
10388 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10390 PR tree-optimization/82434
10391 * fold-const.h (can_native_encode_type_p,
10392 can_native_encode_string_p): Remove.
10393 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10394 don't encode anything, just return what would be otherwise returned.
10395 (native_encode_fixed, native_encode_complex, native_encode_vector):
10397 (native_encode_string): Likewise. Inline by hand
10398 can_native_encode_string_p.
10399 (can_native_encode_type_p): Remove.
10400 (can_native_encode_string_p): Remove.
10401 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10402 STRING_CSTs using can_native_encode_string_p, test all
10403 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10404 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10405 argument from native_encode_expr.
10406 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10407 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10410 2017-10-06 Richard Biener <rguenther@suse.de>
10412 PR tree-optimization/82397
10413 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10414 operand_equal_p but rely on data_ref_compare_tree for detecting
10416 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10417 to match up with dr_group_sort_cmp.
10419 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10422 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10424 * config/s390/s390-builtin-types.def: Regenerate.
10426 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10429 * config/s390/s390-builtin-types.def: Regenerate.
10430 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10431 Change flag from B_VXE to B_VX.
10432 (s390_vec_min_dbl): Remove B_VXE flag.
10434 2017-10-06 Richard Biener <rguenther@suse.de>
10436 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10437 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10438 translate_isl_ast_to_gimple::is_valid_rename,
10439 translate_isl_ast_to_gimple::get_rename,
10440 translate_isl_ast_to_gimple::get_def_bb_for_const,
10441 translate_isl_ast_to_gimple::get_new_name,
10442 translate_isl_ast_to_gimple::collect_all_ssa_names,
10443 translate_isl_ast_to_gimple::copy_loop_phi_args,
10444 translate_isl_ast_to_gimple::collect_all_ssa_names,
10445 translate_isl_ast_to_gimple::copy_loop_phi_args,
10446 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10447 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10448 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10449 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10450 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10451 translate_isl_ast_to_gimple::copy_cond_phi_args,
10452 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10453 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10454 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10455 translate_isl_ast_to_gimple::rename_uses,
10456 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10457 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10458 (set_rename_for_each_def): Likewise.
10459 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10460 here. Handle rewriting SCEV analyzable uses here.
10461 (copy_bb_and_scalar_dependences): Generate code for PHI
10463 (graphite_regenerate_ast_isl): Adjust.
10464 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10465 (add_write, add_read): New functions.
10466 (build_cross_bb_scalars_def): Use it and simplify.
10467 (build_cross_bb_scalars_use): Likewise.
10468 (graphite_find_cross_bb_scalar_vars): Inline into...
10469 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10470 simulating out-of-SSA. Compute liveout and add dependencies.
10471 (build_scops): Force an empty entry block.
10472 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10474 (sese_build_liveouts): Declare.
10475 (sese_trivially_empty_bb_p): Likewise.
10476 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10477 compute liveout and debug_liveout.
10478 (sese_bad_liveouts_use): Remove.
10479 (sese_reset_debug_liveouts_bb): Likewise.
10480 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10481 (sese_build_liveouts): Build liveout and debug_liveout and store
10483 (new_sese_info): Adjust.
10484 (free_sese_info): Likewise.
10485 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10486 do not build liveout here.
10487 (move_sese_in_condition): Adjust region entry.
10488 (scev_analyzable_p): Match up with chrec_apply requirements.
10489 (sese_trivially_empty_bb_p): New.
10490 * tree-into-ssa.c (get_reaching_def): Properly support generating
10491 default-defs for incremental rewrite of anonymous names.
10493 2017-10-06 Richard Biener <rguenther@suse.de>
10495 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10496 precision do not perform modulo reduction.
10498 2017-10-06 Richard Biener <rguenther@suse.de>
10500 PR tree-optimization/82436
10501 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10502 conservatively choose the vectorization factor when checking
10503 whether we can perform the required load permutation.
10504 (vect_transform_slp_perm_load): Assert when we may not fail.
10506 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10508 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10509 message for incompatible -msdata=* and -mcall-* options.
10511 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10513 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10514 rate for post-reload scheduling.
10516 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10518 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10520 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10522 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10523 to improve monte carlo in scimark.
10525 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10527 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10528 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10529 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10530 pentium4_cost, nocona_cost): Set reassociation width to 1.
10531 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10532 width to 2 for fp operations and 1 otherwise.
10533 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10535 (atom_cost): Set reassociation width to 2.
10536 (slm_cost, generic_cost): Set fp reassociation width
10537 to 2 and 1 otherwise.
10538 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10539 (core_cost): Set fp reassociation width to 4 and vector to 2.
10540 (ix86_reassociation_width): Rewrite using cost table; special case
10541 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10542 and TARGET_AVX128_OPTIMAL.
10543 * config/i386/i386.h (processor_costs): Add
10544 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10545 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10546 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10547 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10548 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10549 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10551 2017-10-05 Nathan Sidwell <nathan@acm.org>
10553 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10555 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10557 * config/arm/arm.c (arm_test_fpu_data): New.
10558 (arm_run_selftests): Call arm_test_fpu_data.
10560 2017-10-04 Nathan Sidwell <nathan@acm.org>
10562 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10564 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10566 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10568 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10569 check type for aarch64_simd_valid_immediate.
10570 (aarch64_output_simd_mov_immediate): Update prototype.
10571 (aarch64_simd_valid_immediate): Update prototype.
10572 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10573 support for ORR-immediate.
10574 (and<mode>3): modified pattern to add support for BIC-immediate.
10575 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10576 now checks for valid immediate for BIC and ORR based on new enum
10578 (aarch64_output_simd_mov_immediate): Function now used to output
10579 BIC/ORR imm as well based on new enum argument.
10580 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10582 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10583 (aarch64_reg_or_bic_imm): Likewise.
10585 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10587 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10588 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10590 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10594 PR rtl-optimization/82396
10595 * haifa-sched.c (autopref_multipass_init): Simplify
10597 (autopref_rank_data): Simplify sort order.
10598 * sched-int.h (autopref_multipass_data_): Remove
10599 multi_mem_insn_p, min_offset and max_offset.
10601 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10603 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10605 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10607 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10608 vect_doubleint_cvt.
10610 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10612 * doc/sourcebuild.texi: Document vect_long_mult.
10614 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10616 PR tree-optimization/82413
10617 * fold-const.c (build_range_check): Use widest_int when comparing
10618 the maximum ETYPE value with HIGH.
10620 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10622 PR rtl-optimization/82396
10623 * haifa-sched.c (autopref_multipass_init): Simplify
10625 (autopref_rank_data): Simplify sort order.
10626 * sched-int.h (autopref_multipass_data_): Remove
10627 multi_mem_insn_p, min_offset and max_offset.
10629 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10631 PR tree-optimization/82381
10632 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10633 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10636 PR tree-optimization/82374
10637 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10638 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10639 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10640 current_function_decl to the new decl.
10642 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10644 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10645 helper macro for IEEE float128 hardware built-in functions.
10646 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10648 (TRUNCF128_ODD): Likewise.
10649 (ADDF128_ODD): Likewise.
10650 (SUBF128_ODD): Likewise.
10651 (MULF128_ODD): Likewise.
10652 (DIVF128_ODD): Likewise.
10653 (FMAF128_ODD): Likewise.
10654 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10655 UNSPEC_TRUNC_ROUND_TO_ODD.
10656 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10657 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10658 floating point round to odd instructions.
10659 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10660 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10661 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10662 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10663 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10664 (trunc<mode>sf2_hw): Change the truncate with round to odd
10665 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10666 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10667 to odd hardware instructions.
10668 (sub<mode>3_odd): Likewise.
10669 (mul<mode>3_odd): Likewise.
10670 (div<mode>3_odd): Likewise.
10671 (sqrt<mode>2_odd): Likewise.
10672 (fma<mode>4_odd): Likewise.
10673 (fms<mode>4_odd): Likewise.
10674 (nfma<mode>4_odd): Likewise.
10675 (nfms<mode>4_odd): Likewise.
10676 (trunc<mode>df2_odd): Change the truncate with round to odd
10677 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10679 * doc/extend.texi (PowerPC built-in functions): Update documentation
10680 for existing IEEE float128-bit built-in functions. Add built-in
10681 functions that generate the IEEE 128-bit floating point round to
10684 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10686 PR rtl-optimization/77729
10687 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10688 to (X&(C1&~C2))|C2 transformations.
10690 2017-10-03 Martin Jambor <mjambor@suse.cz>
10692 PR tree-optimization/82363
10693 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10694 mismatch, mark lacc written regardless of racc.
10696 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10698 PR tree-optimization/82381
10699 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10700 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10701 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10702 fallthrough into reassoc_stmt_dominates_stmt_p.
10705 * combine.c (combine_instructions): Don't combine in unreachable
10708 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10711 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10712 function to not use the have_cpu variable. Do not set cpu_index,
10713 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10714 or the default cpu.
10715 (rs6000_valid_attribute_p): Remove duplicate initializations of
10716 old_optimize and func_optimize.
10717 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10718 (rs6000_activate_target_options): Make global.
10719 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10722 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10724 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10725 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10726 Set *poffset to 0 and *psize and *pmax_size to -1 if
10727 *poffset + *psize overflows in HOST_WIDE_INT.
10729 PR tree-optimization/82387
10730 PR tree-optimization/82388
10731 PR tree-optimization/82389
10732 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10733 instead of live_bytes non-NULL.
10735 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10738 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10741 2017-10-02 Richard Biener <rguenther@suse.de>
10743 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10744 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10745 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10747 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10749 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10750 requested precision matches the type's.
10751 * calls.c (alloc_max_size): Calculate the new candidate size as
10752 a widest_int and use wi::to_widest when comparing it with the
10753 current candidate size.
10754 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10755 zero rather than integer_zero_node.
10756 * match.pd: Check for a no-op conversion before using wi::add
10757 rather than after. Use tree_to_uhwi when summing small shift
10758 counts into an unsigned int.
10760 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10761 Alan Hayward <alan.hayward@arm.com>
10762 David Sherwood <david.sherwood@arm.com>
10765 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10766 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10767 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10768 POINTER_AND_FP_REGS.
10770 2017-10-02 Richard Biener <rguenther@suse.de>
10772 PR tree-optimization/82355
10773 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10774 a mapping for the enclosing loop but avoid generating one for
10775 the loop tree root.
10776 (copy_bb_and_scalar_dependences): Remove premature codegen
10777 error on PHIs in blocks duplicated into multiple places.
10778 * graphite-scop-detection.c
10779 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10780 in the region use it as loop and nest to analyze the DR in.
10781 (try_generate_gimple_bb): Likewise.
10782 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10783 (add_loop_constraints): For blocks in a loop not in the region
10784 create a dimension with a single iteration.
10785 * sese.h (gbb_loop_at_index): Remove assert.
10787 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10789 * omp-expand.c (adjust_context_scope): New function.
10790 (expand_parallel_call): Call adjust_context_scope.
10792 2017-10-01 Jeff Law <law@redhat.com>
10794 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10795 dom_opt_dom_walker class with direct access to private members.
10796 Add comments. Call test_for_singularity.
10797 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10798 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10799 m_dummy_cond anymore.
10800 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10802 (pass_dominator:execute): Build the dummy_cond here and pass it
10803 to the dom_opt_dom_walker ctor.
10804 (test_for_singularity): New function.
10806 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10807 Maya Rashish <coypu@sdf.org>
10809 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10810 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10811 (alpha*-*-netbsd*) Use nbsd_tm_file.
10812 (arm*-*-netbsdelf*) Likewise.
10813 (i[34567]86-*-netbsdelf*) Likewise.
10814 (x86_64-*-netbsd*) Likewise.
10815 (mips*-*-netbsd*) Likewise.
10816 (powerpc-*-netbsd*) Likewise.
10817 (sh*-*-netbsd*) Likewise.
10818 (sparc-*-netbsdelf*) Likewise.
10819 (sparc64-*-netbsd*) Likewise.
10820 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10822 (vax-*-netbsdelf*) Likewise.
10823 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10824 (UINT_FAST8_TYPE) Likewise.
10825 (INT_FAST16_TYPE) Check CHAR_FAST16.
10826 (UINT_FAST16_TYPE) Likewise.
10828 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10831 * config/i386/i386.md
10832 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10833 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10834 *divmodsi4_zext_2): New define_insn_and_split.
10835 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10836 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10837 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10838 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10839 New define_insn_and_split.
10840 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10841 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10842 operands[1] having DImode when mode is SImode.
10844 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10845 always SImode for DIV and MOD in REG_EQUAL notes.
10847 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10849 PR middle-end/82319
10850 * match.pd: Fix handling of NaNs in pattern.
10852 2017-09-29 Jeff Law <law@redhat.com>
10854 * sbitmap.c (bitmap_bit_in_range_p): New function.
10855 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10856 * tree-ssa-dse.c (live_bytes_read): New function.
10857 (dse_classify_store): Ignore reads of dead bytes.
10859 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10860 typos and whitespace errors.
10861 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10862 * config/s390/s390.c (s390_emit_prologue): Likewise.
10864 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10867 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10868 with a symbol for LRA.
10870 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10872 PR rtl-optimization/82338
10873 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10875 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10877 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10878 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10879 (qsort_chk): Declare.
10880 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10881 (qsort_chk): New function.
10883 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10885 PR tree-optimization/82337
10886 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10887 phi definition if the PHI result appears in an abnormal PHI.
10888 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10889 basis appears in an abnormal PHI.
10891 2017-09-29 Richard Biener <rguenther@suse.de>
10893 * graphite-isl-ast-to-gimple.c
10894 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10895 (binary_op_to_tree): Use it.
10896 (get_rename_from_scev): Likewise.
10897 (copy_loop_phi_nodes): Likewise.
10898 (copy_bb_and_scalar_dependences): Likewise.
10899 (translate_pending_phi_nodes): Likewise.
10901 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10904 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10905 for movabsq $(i32 << shift), r64.
10907 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10909 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10910 index when encoding %esp as %rsp to avoid 0x67 prefix.
10912 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10914 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10915 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10917 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10919 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10920 Extensions with more than 16 double VFP registers.
10921 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10922 to_clear_mask and all code related to it. Replace the remaining
10923 entry by a sbitmap and adapt code accordingly.
10925 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10927 * brig-builtins.def: Change pure attributes to const.
10929 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10931 * config.gcc (default_gnu_indirect_function): Default to yes for
10932 sparc*-*-linux* with glibc.
10934 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10936 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10937 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10938 when creating .init_array and .fini_array sections with priority
10941 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10944 * config/aarch64/aarch64.c
10945 (aarch64_builtin_support_vector_misalignment): Always return false
10946 when misalignment is unknown.
10948 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10950 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10951 this function to return false if the definition used by the swap
10952 instruction is artificial, or if the memory address from which the
10953 constant value is loaded is not represented by a base address held
10954 in a register or if the base address register is a frame or stack
10955 pointer. Additionally, return false if the base address of the
10956 loaded constant is a SYMBOL_REF but is not considered to be a
10958 (replace_swapped_load_constant): New function.
10959 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10960 loaded constant vector with a load of a swapped constant vector.
10962 2017-09-27 Carl Love <cel@us.ibm.com>
10964 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10965 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10966 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10969 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10971 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10972 first, always call autopref_rank_data otherwise.
10974 2017-09-27 Richard Biener <rguenther@suse.de>
10976 * graphite-scop-detection.c (find_scop_parameters): Move
10977 loop bound handling ...
10978 (gather_bbs::before_dom_children): ... here, avoiding the need
10979 to build scop_info->loop_nest.
10980 (record_loop_in_sese): Remove.
10981 * sese.h (sese_info_t::loop_nest): Remove.
10982 * sese.c (new_sese_info): Do not allocate loop_nest.
10983 (free_sese_info): Do not free loop_nest.
10985 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10988 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10989 lhs from calls if the lhs has addressable type.
10991 2017-09-27 Richard Biener <rguenther@suse.de>
10993 * graphite.h (scop::max_alias_set): New member.
10994 * graphite-scop-detection.c: Remove references to non-existing
10995 --param in comments.
10996 (build_alias_sets): Record the maximum alias set used for drs.
10997 (build_scops): Support zero as unlimited for
10998 --param graphite-max-arrays-per-scop.
10999 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11000 and inline into ...
11001 (build_poly_sr_1): ... here. Compute alias set based on the
11002 maximum alias set used for drs rather than
11003 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11005 2017-09-27 Richard Biener <rguenther@suse.de>
11007 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11008 --param loop-block-tile-size=0 to disable tiling.
11010 2017-09-27 Richard Biener <rguenther@suse.de>
11012 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11013 (graphite-max-nb-scop-params): Document special value zero.
11014 * domwalk.h (dom_walker::STOP): New symbolical constant.
11015 (dom_walker::dom_walker): Add optional parameter for bb to
11017 (dom_walker::~dom_walker): Declare.
11018 (dom_walker::before_dom_children): Document STOP return value.
11019 (dom_walker::m_user_bb_to_rpo): New member.
11020 (dom_walker::m_bb_to_rpo): Likewise.
11021 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11022 mapping here if not provided by the user.
11023 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11024 provided by the user.
11025 (dom_walker::STOP): Define.
11026 (dom_walker::walk): Do not compute bb to RPO mapping here.
11027 Support STOP return value from before_dom_children to stop
11029 * graphite-optimize-isl.c (optimize_isl): If the schedule
11030 is the same still generate code if -fgraphite-identity
11031 or -floop-parallelize-all are given.
11032 * graphite-scop-detection.c: Include cfganal.h.
11033 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11035 (gather_bbs::before_dom_children): Return STOP for BBs
11037 (build_scops): Compute bb to RPO mapping and pass it to
11038 the domwalk. Treat --param graphite-max-nb-scop-params=0
11039 as not limiting the number of params.
11040 * graphite.c (graphite_initialize): Remove limit on the
11041 number of basic-blocks in a function.
11042 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11043 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11044 default value of 10.
11046 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11048 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11049 Adjust code to eliminate needing to do the shift right 32-bits
11050 operation after XSCVDPSPN.
11052 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11054 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11055 ((X / Y) != 0 -> X >= Y): Likewise.
11057 2017-09-26 Carl Love <cel@us.ibm.com>
11059 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11060 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11061 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11062 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11063 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11064 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11065 definitions and overloading.
11066 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11067 statement for P9V_BUILTIN_XST_LEN_R.
11068 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11069 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11070 define_expand and define_insn for the instructions and builtins.
11071 * doc/extend.texi: Update the built-in documentation file for the new
11072 built-in functions.
11073 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11074 define_insn for the instructions
11076 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11079 * gcc/config/netbsd-protos.h: New file.
11080 * gcc/config/netbsd.c: New file.
11081 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11082 * gcc/config/t-netbsd: New file.
11083 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11084 (tmake_file) Add t-netbsd.
11085 (extra_objs) Add netbsd.o.
11087 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11091 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11093 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11095 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11096 sign extension from a vector register to a GPR by doing a 32-bit
11097 direct move and then an EXTSW.
11098 (extendsi<mode>2 splitter): Likewise.
11099 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11100 right or vector extract after doing XSCVDPSPN. Use
11101 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11103 (movdi_from_sf_zero_ext): Likewise.
11104 (reload_gpr_from_vsxsf): Likewise.
11105 (p8_mfvsrd_4_disf): Delete, no longer used.
11106 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11107 then needing to move the SFmode to a GPR to use the XSCVDPSP
11108 instruction instead of FRSP and XSCVDPSPN.
11109 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11110 it is adjacent to the other XSCVSPDP insns.
11111 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11112 SFmode to be in traditional Altivec registers.
11113 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11114 (vsx_xscvspdpn): Likewise.
11115 (vsx_xscvspdpn_scalar): Likewise.
11117 2017-09-26 Martin Jambor <mjambor@suse.cz>
11119 * tree-sra.c (compare_access_positions): Put integral types first,
11120 stabilize sorting of integral types, remove conditions putting
11121 non-full-precision integers last.
11122 (sort_and_splice_var_accesses): Disable scalarization if a
11123 non-integert would be represented by a non-full-precision integer.
11125 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11127 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11128 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11129 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11130 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11131 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11132 conditionals inside the function instead of around it. Call
11133 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11134 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11136 2017-09-26 Richard Biener <rguenther@suse.de>
11138 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11140 (scop_detection::build_scop_breadth): ... this. Removed.
11141 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11142 (scop_detection::harmful_stmt_in_bb): Likewise.
11143 (scop_detection::graphite_can_represent_stmt): Likewise.
11144 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11145 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11146 (scop_detection::can_represent_loop_1): ... this. Removed.
11147 (scop_detection::harmful_loop_in_region): Simplify after inlining
11148 the above and remove more quadraticness.
11149 (build_scops): Adjust.
11150 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11153 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11156 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11157 REGNO (base) == SP_REG if base is a REG.
11159 PR middle-end/35691
11160 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11161 if it is different SSA_NAME.
11162 (optimize_range_tests_cmp_bitwise): New function.
11163 (optimize_range_tests): Call it.
11165 2017-09-26 Richard Biener <rguenther@suse.de>
11167 PR tree-optimization/82321
11168 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11169 for the def being inside the loop.
11171 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11173 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11175 * config/s390/s390-builtins.def: Fix constraint on op4.
11177 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11179 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11180 independent expanders.
11181 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11182 ("vec_ordered", "vec_unordered"): New expanders.
11184 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11186 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11189 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11191 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11192 vec_unpacks_lo_v16qi.
11193 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11195 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11197 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11198 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11199 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11201 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11203 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11205 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11207 ("*vec_slb<mode>"): New insn pattern.
11208 ("vec_shr_<mode>"): New expander.
11209 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11210 and force the shift count operand to V16QImode.
11211 ("vec_srb<mode>"): Set shift count mode to V16QI.
11213 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11215 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11216 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11217 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11219 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11222 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11224 2017-09-26 Richard Biener <rguenther@suse.de>
11226 PR tree-optimization/82320
11227 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11230 2017-09-25 Jeff Law <law@redhat.com>
11232 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11233 prototype for new argument.
11234 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11235 mostly extracted from rs6000_emit_allocate_stack.
11236 (rs6000_emit_probe_stack_range_stack_clash): New function.
11237 (rs6000_emit_allocate_stack): Call
11238 rs6000_emit_probe_stack_range_stack_clash as needed.
11239 (rs6000_emit_probe_stack_range): Add additional argument
11240 to call to gen_probe_stack_range{si,di}.
11241 (output_probe_stack_range): New.
11242 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11243 (output_probe_stack_range_stack_clash): New.
11244 (rs6000_emit_prologue): Emit notes into dump file as requested.
11245 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11246 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11247 Add additional operand and pass it to output_probe_stack_range.
11249 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11251 PR tree-optimization/82163
11252 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11253 (checking_verify_loop_closed_ssa): New parameter.
11254 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11255 (check_loop_closed_ssa_stmt): Delete.
11256 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11257 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11258 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11261 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11263 * brig-builtins.def: Treat HSAIL barrier builtins as
11264 setjmp/longjump style functions.
11266 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11268 * target.def (constant_alignment): New hook.
11269 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11270 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11271 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11272 * doc/tm.texi: Regenerate.
11273 * targhooks.h (default_constant_alignment): Declare.
11274 (constant_alignment_word_strings): Likewise.
11275 * targhooks.c (default_constant_alignment): New function.
11276 (constant_alignment_word_strings): Likewise.
11277 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11278 instead of CONSTANT_ALIGNMENT.
11279 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11280 (force_const_mem): Likewise.
11281 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11282 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11283 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11284 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11285 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11287 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11288 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11289 constant_alignment_word_strings.
11290 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11291 (CONSTANT_ALIGNMENT): Likewise.
11292 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11293 (arm_constant_alignment): New function.
11294 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11295 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11296 constant_alignment_word_strings.
11297 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11298 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11299 constant_alignment_word_strings.
11300 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11301 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11302 (cris_constant_alignment): New function.
11303 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11304 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11305 (epiphany_constant_alignment): New function.
11306 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11307 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11308 constant_alignment_word_strings.
11309 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11310 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11311 constant_alignment_word_strings.
11312 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11313 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11314 constant_alignment_word_strings.
11315 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11316 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11317 * config/i386/i386.c (ix86_constant_alignment): Make static.
11318 Use the same interface as the target hook.
11319 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11320 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11321 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11322 constant_alignment_word_strings.
11323 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11324 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11325 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11326 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11327 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11328 constant_alignment_word_strings.
11329 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11330 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11331 constant_alignment_word_strings.
11332 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11333 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11334 constant_alignment_word_strings.
11335 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11336 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11338 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11339 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11340 * config/mips/mips.c (mips_constant_alignment): New function.
11341 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11342 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11343 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11344 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11345 (mmix_constant_alignment): Make static. Use the same interface
11346 as the target hook.
11347 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11348 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11349 constant_alignment_word_strings.
11350 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11351 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11352 constant_alignment_word_strings.
11353 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11354 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11355 constant_alignment_word_strings.
11356 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11357 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11358 (rs6000_constant_alignment): New function.
11359 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11360 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11361 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11362 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11363 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11364 (rs6000_constant_alignment): New function.
11365 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11366 * config/s390/s390.c (s390_constant_alignment): New function.
11367 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11368 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11369 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11370 constant_alignment_word_strings.
11371 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11372 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11373 (sparc_constant_alignment): New function.
11374 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11375 * config/spu/spu.c (spu_constant_alignment): New function.
11376 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11377 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11378 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11379 constant_alignment_word_strings.
11380 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11381 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11382 constant_alignment_word_strings.
11383 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11384 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11385 constant_alignment_word_strings.
11386 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11387 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11388 (visium_constant_alignment): New function.
11389 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11390 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11391 (xtensa_constant_alignment): New function.
11392 * system.h (CONSTANT_ALIGNMENT): Poison.
11394 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11396 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11397 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11398 (rs6000_builtin_valid_without_lhs): New helper function.
11399 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11400 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11402 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11404 * target.h (vec_perm_indices): Use unsigned short rather than
11406 (auto_vec_perm_indices): Likewise.
11407 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11408 Use unsigned int rather than unsigned char.
11409 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11411 2017-09-25 Richard Biener <rguenther@suse.de>
11413 * cfgloop.h (sort_sibling_loops): Declare.
11414 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11415 (sort_sibling_loops): New function sorting the sibling loop list
11417 * graphite.c (graphite_transform_loops): Sort sibling loops.
11419 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11421 * target.def (vec_perm_const_ok): Change sel parameter to
11423 * optabs-query.c (can_vec_perm_p): Update accordingly.
11424 * doc/tm.texi: Regenerate.
11425 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11426 auto_vec_perm_indices and remove separate nelt field.
11427 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11428 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11429 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11430 (aarch64_expand_vec_perm_const): Update accordingly.
11431 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11432 to vec_perm_indices.
11433 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11434 auto_vec_perm_indices and remove separate nelt field.
11435 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11436 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11437 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11439 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11440 to vec_perm_indices.
11441 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11442 sel to vec_perm_indices.
11443 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11444 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11445 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11447 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11450 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11453 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11454 on the FUNCTION_DECL function context if it has a DIE that is a
11457 2017-09-25 Richard Biener <rguenther@suse.de>
11459 PR tree-optimization/82285
11460 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11463 2017-09-25 Tom de Vries <tom@codesourcery.com>
11467 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11470 2017-09-25 Richard Biener <rguenther@suse.de>
11472 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11473 ISL errors other than isl_error_quota happen. Dump if the
11474 schedule is the same.
11475 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11476 errors instead of aborting inside ISL.
11478 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11481 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11482 of libgcc_eh for m64.
11483 * config/i386/darwin64.h: Likewise.
11485 2017-09-25 Richard Biener <rguenther@suse.de>
11487 PR middle-end/82144
11488 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11489 attribute for incomplete types nor twice for complete ones.
11491 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11494 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11495 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11498 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11501 * config/i386/i386.opt (mprefer-avx256): Use
11502 ix86_target_flags variable.
11503 * config/i386/i386.c (ix86_target_string): Move
11504 -mprefer-avx256 to flag2_opts.
11506 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11508 PR middle-end/35691
11509 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11510 and x != -1 | y != -1 into (x & y) != -1.
11512 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11514 * config.gcc: Add new case statement to set
11515 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11516 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11517 s390x-*-linux* case statements. Added aarch64 to the list of
11518 supported architectures.
11520 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11522 PR tree-optimization/82289
11523 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11524 STMT_VINFO_RELEVANT_P.
11526 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11527 Alan Hayward <alan.hayward@arm.com>
11528 David Sherwood <david.sherwood@arm.com>
11530 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11531 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11532 (extract_range_from_binary_expr_1): Don't call
11533 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11535 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11536 Alan Hayward <alan.hayward@arm.com>
11537 David Sherwood <david.sherwood@arm.com>
11539 * target.def (preferred_vector_alignment): New hook.
11540 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11542 * doc/tm.texi: Regenerate.
11543 * targhooks.h (default_preferred_vector_alignment): Declare.
11544 * targhooks.c (default_preferred_vector_alignment): New function.
11545 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11547 (DR_TARGET_ALIGNMENT): New macro.
11548 (aligned_access_p): Update commentary.
11549 (vect_known_alignment_in_bytes): New function.
11550 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11552 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11553 Calculate the misalignment based on the target alignment rather than
11555 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11556 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11557 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11558 the target alignment, rather than masking the element misalignment
11559 with the number of elements in a vector. Also use the target
11560 alignment when calculating the maximum number of peels.
11561 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11562 instead of TYPE_ALIGN_UNIT.
11563 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11564 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11565 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11566 (vect_create_data_ref_ptr): Likewise.
11567 (vect_setup_realignment): Realign by ANDing with
11568 -DR_TARGET_MISALIGNMENT.
11569 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11570 the number of peels based on DR_TARGET_ALIGNMENT.
11571 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11572 with the guaranteed alignment boundary when deciding whether
11574 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11575 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11576 (ensure_base_align): Remove stmt_info parameter. Get the
11577 target base alignment from DR_TARGET_ALIGNMENT.
11578 (vectorizable_store): Update call accordingly. Interpret
11579 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11581 (vectorizable_load): Likewise.
11583 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11584 Alan Hayward <alan.hayward@arm.com>
11585 David Sherwood <david.sherwood@arm.com>
11587 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11588 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11589 (vect_enhance_data_refs_alignment): Likewise.
11591 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11593 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11594 error. Only quit immediately if parsing is complete.
11595 (BEGIN): Initialize fatal_err and parse_done.
11596 (begin fpu, end fpu): Check number of arguments.
11597 (begin arch, end arch): Likewise.
11598 (begin cpu, end cpu): Likewise.
11599 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11600 (optalias): Likewise.
11602 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11604 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11605 * config/arm/arm-isa.h: Delete. Move definitions to ...
11606 * arm-cpus.in: ... here. Use new feature and fgroup values.
11607 * config/arm/arm.c (arm_option_override): Use lower case for feature
11609 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11610 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11611 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11613 (print_isa_bits_for): New function.
11614 (gen_isa): New function.
11615 (gen_comm_data): Use print_isa_bits_for.
11616 (define feature): New keyword.
11617 (define fgroup): New keyword.
11618 * config/arm/t-arm (TM_H): Remove.
11619 (GTM_H): Add arm-isa.h.
11620 (arm-isa.h): Add rule to generate file.
11621 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11622 case for feature bit names.
11624 2017-09-22 Richard Biener <rguenther@suse.de>
11626 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11628 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11629 print of no dependency loops ...
11630 * graphite.c (graphite_transform_loops): ... here.
11631 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11633 (same_close_phi_node, remove_duplicate_close_phi,
11634 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11635 (canonicalize_loop_closed_ssa): ... here and simplify.
11636 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11637 (optimize_isl): Use dump_printf_loc to tell when we stopped
11638 optimizing because of an ISL timeout.
11640 2017-09-22 Richard Biener <rguenther@suse.de>
11642 PR tree-optimization/82291
11643 * tree-if-conv.c (predicate_mem_writes): Make sure to
11644 remove writes in blocks predicated with false.
11646 2017-09-22 Richard Biener <rguenther@suse.de>
11648 * sese.c: Include cfganal.h.
11649 (if_region_set_false_region): Remove.
11650 (create_if_region_on_edge): Likewise.
11651 (move_sese_in_condition): Re-implement without destroying
11654 2017-09-22 Richard Biener <rguenther@suse.de>
11656 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11657 Verify both BBs contain loop PHI nodes before dispatching to
11658 copy_loop_phi_args.
11659 (graphite_regenerate_ast_isl): Do not recompute dominators,
11660 do not verify three times. Restructure for clarity.
11661 * graphite-scop-detection.c (same_close_phi_node,
11662 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11663 defined_in_loop_p, canonicalize_loop_closed_ssa,
11664 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11665 checking and SSA rewrite, move to ...
11666 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11667 (graphite_initialize): Do not pass in ctx, do not reset the
11668 SCEV cache, compute only dominators.
11669 (graphite_transform_loops): Allocate ISL ctx after
11670 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11671 Maintain post-dominators only around build_scops.
11672 * sese.c (if_region_set_false_region): Make static. Free
11673 and recompute dominators.
11674 (move_sese_in_condition): Assert we don't get called with
11675 post-dominators computed.
11676 * sese.h (if_region_set_false_region): Remove.
11678 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11680 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11681 mode attribute for TARGET_AVX512VL.
11683 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11685 * config/i386/i386.opt (mprefer-avx256): New option.
11686 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11688 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11689 for TARGET_PREFER_AVX256.
11690 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11692 2017-09-21 Jeff Law <law@redhat.com>
11694 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11695 Fix dump output if the only stack space is for pushed registers.
11697 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11699 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11702 2017-09-21 Martin Sebor <msebor@redhat.com>
11705 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11706 code (in C++) or code that triggers warnings.
11708 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11710 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11712 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11714 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11715 * sched-rgn.c: Ditto.
11716 * sel-sched-ir.c: Ditto.
11718 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11720 * toplev.h (set_random_seed): Adjust return type.
11721 * toplev.c (init_local_tick): Move eager initialization of random_seed
11722 to get_random_seed. Adjust comment.
11723 (init_random_seed): Inline to get_random_seed, delete.
11724 (get_random_seed): Initialize random_seed lazily.
11725 (set_random_seed): Do not return previous value.
11726 (print_switch_value): Do not call get_random_seed.
11728 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11730 * cgraph.c (delete_function_version): New, broken out from...
11731 (cgraph_node::delete_function_version): ...here. Rename to
11732 cgraph_node::delete_function_version_by_decl. Update all uses.
11733 (cgraph_node::remove): Call delete_function_version.
11735 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11738 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11739 VAR_DECLs to which addressable non-volatile parameters are mapped
11740 and for id->retvar after the return value assignment. Clear
11741 id->retval and id->retbnd after inlining.
11743 2017-09-21 Richard Biener <rguenther@suse.de>
11745 PR tree-optimization/82276
11746 PR tree-optimization/82244
11747 * tree-vrp.c (build_assert_expr_for): Set
11748 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11750 (remove_range_assertions): Revert earlier change.
11752 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11755 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11757 2017-09-21 Richard Biener <rguenther@suse.de>
11759 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11760 Restore valid IL after code generation errors.
11761 * graphite.c (graphite_transform_loops): Diagnose code
11762 generation issues as MSG_MISSED_OPTIMIZATION and continue
11763 with processing SCOPs.
11765 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11766 Alan Hayward <alan.hayward@arm.com>
11767 David Sherwood <david.sherwood@arm.com>
11769 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11770 rather than choosing between plus_constant and gen_rtx_<CODE>.
11771 * expr.c (emit_push_insn): Likewise.
11772 (expand_expr_real_2): Likewise.
11774 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11775 Alan Hayward <alan.hayward@arm.com>
11776 David Sherwood <david.sherwood@arm.com>
11778 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11780 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11781 Alan Hayward <alan.hayward@arm.com>
11782 David Sherwood <david.sherwood@arm.com>
11784 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11785 calling tree_to_uhwi.
11787 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11788 Alan Hayward <alan.hayward@arm.com>
11789 David Sherwood <david.sherwood@arm.com>
11791 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11792 INTEGER_CST rather than a negative test for ADDR_EXPR.
11794 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11795 Alan Hayward <alan.hayward@arm.com>
11796 David Sherwood <david.sherwood@arm.com>
11798 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11799 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11801 2017-09-21 Richard Biener <rguenther@suse.de>
11803 PR tree-optimization/71351
11804 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11805 graphite_create_new_loop_guard): Remove, fold remaining parts
11807 (translate_isl_ast_node_for): ... here and simplify.
11809 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11812 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11813 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11814 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11815 alternative always use QI mode, for -Os imov (=R,R) alternative
11816 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11819 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11820 Jeff Law <law@redhat.com>
11822 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11823 (allocate_stack_space): New function, partially extracted from
11824 s390_emit_prologue.
11825 (s390_emit_prologue): Track offset to most recent stack probe.
11826 Code to allocate space moved into allocate_stack_space.
11827 Dump actions when no stack is allocated.
11828 (s390_prologue_plus_offset): New function.
11829 (s390_emit_stack_probe): Likewise.
11831 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11833 * common.opt (Wa, Wl, Wp, g, gz=): Add
11835 (gno-column-info): Remove.
11836 (gcolumn-info): Drop RejectNegative.
11837 (gno-): New prefix.
11838 (gno-record-gcc-switches): Remove.
11839 (grecord-gcc-switches): Drop RejectNegative.
11840 (gno-split-dwarf): Remove.
11841 (gsplit-dwarf): Drop RejectNegative.
11842 (gno-strict-dwarf): Remove.
11843 (gstrict-dwarf): Drop RejectNegative.
11844 * config/darwin.opt (gfull, gused): Add RejectNegative.
11845 * dwarf2out.c (gen_producer_string): Drop
11846 gno-record-gcc-switches handler.
11847 * optc-gen.awk: Add g to prefixes with negative forms.
11848 * opts-common.c (remapping_prefix_p): New.
11849 (find_opt): Check it.
11850 (generate_canonical_option): Test g prefix.
11851 (option_map): Add -gno- mapping.
11852 (add_misspelling_candidates): Check remapping_prefix_p.
11854 2017-09-20 Jeff Law <law@redhat.com>
11856 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11857 thinko in stack clash protection support.
11859 * explow.c (compute_stack_clash_protection_loop_data): Use
11860 CONST_INT_P instead of explicit test. Verify object is a
11861 CONST_INT_P before looking at INTVAL.
11862 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11863 instead of explicit test.
11865 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11868 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11869 address instead of to r1 and r11.
11871 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11873 * config.gcc: Support "knm".
11874 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11875 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11877 * config/i386/i386.c (m_KNM): Define.
11878 (processor_target_table): Add "knm".
11880 (ix86_option_override_internal): Add "knm".
11881 (ix86_issue_rate): Add PROCESSOR_KNM.
11882 (ix86_adjust_cost): Ditto.
11883 (ia32_multipass_dfa_lookahead): Ditto.
11884 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11885 (fold_builtin_cpu): Add M_INTEL_KNM.
11886 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11887 (processor_type): Add PROCESSOR_KNM.
11888 * config/i386/x86-tune.def: Add m_KNM.
11889 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11891 2017-09-20 Richard Biener <rguenther@suse.de>
11893 PR tree-optimization/80213
11894 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11895 are allowed in empty BBs as well.
11896 (canonicalize_loop_closed_ssa): Also look for other complex
11898 (scop_detection::get_sese): Include the loop-closed PHI block
11900 (scop_detection::merge_sese): Remove code adding extra blocks.
11901 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11902 (build_scops): Assert the final returned scop is invalid.
11904 2017-09-20 Richard Biener <rguenther@suse.de>
11906 PR tree-optimization/82264
11907 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11909 (vn_phi_lookup): Likewise.
11910 (vn_phi_insert): Likewise.
11912 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11914 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11915 that fits into uhwi or shwi, add DW_AT_const_value regardless
11916 of early_dwarf without going through RTL, using add_AT_unsigned
11919 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11920 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11921 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11922 (DEBUG_MACRO_SECTION): Likewise.
11923 (DEBUG_LTO_MACRO_SECTION): Likewise.
11924 (DEBUG_STR_DWO_SECTION): Likewise.
11925 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11926 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11927 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11928 (DEBUG_LTO_LINE_STR_SECTION): Define.
11929 (init_sections_and_labels): Initialize debug_line_str_section
11930 variable. Initialize debug_loc_section for -gdwarf-5 to
11931 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11933 2017-09-20 Richard Biener <rguenther@suse.de>
11935 * graphite-sese-to-poly.c (extract_affine): Properly handle
11936 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11938 2017-09-20 Richard Biener <rguenther@suse.de>
11940 PR tree-optimization/81373
11941 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11942 Force SESE live-out defs to be handled even if they are
11945 2017-09-19 Jeff Law <law@redhat.com>
11947 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11948 nothing for stack adjustments with REG_STACK_CHECK.
11949 * sched-deps.c (parse_add_or_inc): Reject insns with
11950 REG_STACK_CHECK from dependency breaking.
11951 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11952 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11953 * reg-notes.def (STACK_CHECK): New note.
11955 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11956 (ix86_expand_prologue): Dump stack clash info as needed.
11957 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11959 * function.c (dump_stack_clash_frame_info): New function.
11960 * function.h (dump_stack_clash_frame_info): Prototype.
11961 (enum stack_clash_probes): New enum.
11963 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11964 flag_stack_clash_protection.
11965 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11966 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11967 (arm_frame_pointer_required): Likewise.
11968 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11969 (ia64_expand_prologue): Likewise.
11970 * config/mips/mips.c (mips_expand_prologue): Likewise.
11971 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11972 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11973 (sparc_flat_expand_prologue): Likewise.
11974 * config/spu/spu.c (spu_expand_prologue): Likewise.
11976 * explow.c: Include "params.h".
11977 (anti_adjust_stack_and_probe_stack_clash): New function.
11978 (get_stack_check_protect): Likewise.
11979 (compute_stack_clash_protection_loop_data): Likewise.
11980 (emit_stack_clash_protection_loop_start): Likewise.
11981 (emit_stack_clash_protection_loop_end): Likewise.
11982 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11983 Use anti_adjust_stack_and_probe_stack_clash.
11984 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11985 (emit_stack_clash_protection_loop_start): Likewise.
11986 (emit_stack_clash_protection_loop_end): Likewise.
11987 * rtl.h (get_stack_check_protect): Prototype.
11988 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11989 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11990 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11992 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11994 * doc/tm.texi: Rebuilt.
11995 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11996 get_stack_check_protect.
11997 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11998 * config/arm/arm.c (arm_expand_prologue): Likewise.
11999 (arm_frame_pointer_required): Likewise.
12000 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12001 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12002 * config/mips/mips.c (mips_expand_prologue): Likewise.
12003 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12004 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12005 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12006 (sparc_flat_expand_prologue): Likewise.
12008 * common.opt (-fstack-clash-protection): New option.
12009 * flag-types.h (enum stack_check_type): Note difference between
12010 -fstack-check= and -fstack-clash-protection.
12011 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12012 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12013 * toplev.c (process_options): Issue warnings/errors for cases
12014 not handled with -fstack-clash-protection.
12015 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12016 (-fstack-check): Note additional problem with -fstack-check=generic.
12017 Note that -fstack-check is primarily for Ada and refer users
12018 to -fstack-clash-protection for stack-clash-protection.
12019 Document new params for stack clash protection.
12021 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12023 * config/i386/i386.c (ix86_split_long_move): Do not handle
12024 address used for LEA in a special way.
12026 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12028 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12030 2017-09-19 Martin Sebor <msebor@redhat.com>
12033 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12034 of incompatible types.
12036 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12038 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12039 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12040 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12041 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12043 2017-09-19 Richard Biener <rguenther@suse.de>
12045 PR tree-optimization/82244
12046 * tree-vrp.c (remove_range_assertions): Do not propagate
12047 a constant to abnormals but replace the assert with a copy.
12049 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12051 PR rtl-optimization/57878
12052 PR rtl-optimization/68988
12053 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12054 avoidance test involving non_reload_pseudos. Move frequency test
12055 below the general fragmentation avoidance test.
12057 2017-09-19 Richard Biener <rguenther@suse.de>
12059 PR tree-optimization/69728
12060 * graphite-sese-to-poly.c (schedule_error): New global.
12061 (add_loop_schedule): Handle empty domain by failing the
12063 (build_original_schedule): Handle schedule_error.
12065 2017-09-19 Richard Biener <rguenther@suse.de>
12067 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12068 Do not iterate to sibling loops but only to siblings of inner
12071 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12074 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12075 that the registers are different.
12077 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12079 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12080 to processor_model and "amdfam17h" to arch_names_table.
12081 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12083 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12086 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12088 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12089 Alan Hayward <alan.hayward@arm.com>
12090 David Sherwood <david.sherwood@arm.com>
12092 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12094 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12096 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12097 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12098 vect_schedule_slp_instance.
12099 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12100 Update call to vect_slp_analyze_node_operations. Simplify return
12102 (vect_slp_analyze_bb_1): Update call accordingly.
12103 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12104 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12105 (vect_schedule_slp): Update call accordingly.
12107 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12108 Alan Hayward <alan.hayward@arm.com>
12109 David Sherwood <david.sherwood@arm.com>
12111 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12112 with types that aren't in fact scalar.
12114 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12116 * tree-vect-slp.c (vect_record_max_nunits): New function,
12118 (vect_build_slp_tree_1): ...here.
12119 (vect_build_slp_tree_2): Call it for phis too.
12121 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12123 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12124 to vect_get_vec_def_for_operand when getting the mask operand.
12126 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12127 Alan Hayward <alan.hayward@arm.com>
12128 David Sherwood <david.sherwood@arm.com>
12130 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12133 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12134 Alan Hayward <alan.hayward@arm.com>
12135 David Sherwood <david.sherwood@arm.com>
12137 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12138 calculation for vector booleans.
12140 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12141 Alan Hayward <alan.hayward@arm.com>
12142 David Sherwood <david.sherwood@arm.com>
12144 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12146 (vect_transform_stmt): ...here.
12147 (vect_analyze_stmt): Use it instead of calling
12148 vectorizable_live_operation directly.
12150 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12152 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12153 non-SIMT targets in acc vector loops.
12155 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12157 * configure.ac: Add arc and check if assembler supports gdwarf2.
12158 * configure: Regenerate.
12160 2017-09-18 Richard Biener <rguenther@suse.de>
12162 PR tree-optimization/82220
12163 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12164 epilogue niters from the min_profitable_iters compute.
12166 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12169 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12170 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12171 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12173 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12176 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12177 switching to a new text section.
12179 2017-09-18 Richard Biener <rguenther@suse.de>
12181 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12183 (build_alias_set): Reject aliases with no access function.
12185 2017-09-18 Richard Biener <rguenther@suse.de>
12187 PR tree-optimization/79622
12188 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12190 (build_cross_bb_scalars_use): Likewise.
12192 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12194 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12196 2017-09-18 Alan Modra <amodra@gmail.com>
12199 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12200 stack_pointer_rtx for count 0. Update comments. Break up
12201 large rtl expression.
12203 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12205 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12206 Increase to 20 bytes.
12207 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12208 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12209 or avx version of the stub.
12211 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12214 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12215 compute the minimum stack alignment. Also update preferred stack
12216 boundary for leaf functions.
12218 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12220 PR tree-optimization/82228
12221 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12224 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12226 * common/config/nds32/nds32-common.c
12227 (nds32_option_optimization_table): Refine formatting.
12228 (nds32_option_optimization_table): Use -fsched-pressure and
12229 -fomit-frame-pointer for specific optimization level.
12231 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12233 * config/nds32/nds32.c: Refine formatting and comments.
12234 * config/nds32/nds32.h: Likewise.
12235 * config/nds32/nds32.md: Likewise.
12236 * config/nds32/nds32-cost.c: Likewise.
12237 * config/nds32/nds32-isr.c: Likewise.
12238 * config/nds32/nds32-md-auxiliary.c: Likewise.
12239 * config/nds32/nds32-multiple.md: Likewise.
12240 * config/nds32/nds32-predicates.c: Likewise.
12242 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12243 Jakub Jelinek <jakub@redhat.com>
12245 Add support for -std=c++2a.
12246 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12248 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12250 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12253 * doc/extend.texi (Common Function Attributes): Add
12254 references to ARM, AArch64, and S/390 specific attributes.
12255 (Function Specific Option Pragmas): Add AArch64 and S/390
12256 to list of back ends that support the target pragma.
12258 2017-09-15 Nathan Sidwell <nathan@acm.org>
12260 * doc/standards.texi: Fix C++17 description. Update URLs for
12263 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12265 * common.opt (Wcast-align=strict): New warning option.
12266 * doc/invoke.texi: Document -Wcast-align=strict.
12268 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12270 * cgraph.h (cgraph_thunk_info): Add comments.
12271 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12272 assert for VIRTUAL_* arguments stricter.
12274 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12276 PR tree-optimization/71026
12277 * match.pd: Move RDIV patterns from fold-const.c
12278 * fold-const.c (distribute_real_division): Removed.
12279 (fold_binary_loc): Remove calls to distribute_real_divison.
12281 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12283 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12284 c++1z and gnu++1z as deprecated. Change other references to
12285 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12286 Change -Wc++1z-compat to -Wc++17-compat.
12287 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12288 * dwarf2out.c (highest_c_language): Handle C++17.
12289 (gen_compile_unit_die): Likewise.
12291 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12293 PR rtl-optimization/82192
12294 * combine.c (make_extraction): Don't look through non-paradoxical
12295 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12298 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12299 Alan Hayward <alan.hayward@arm.com>
12300 David Sherwood <david.sherwood@arm.com>
12302 * target.def (function_arg_offset): New hook.
12303 * targhooks.h (default_function_arg_offset): Declare.
12304 * targhooks.c (default_function_arg_offset): New function.
12305 * function.c (locate_and_pad_parm): Use
12306 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12307 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12308 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12309 * doc/tm.texi: Regenerate.
12310 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12311 * config/spu/spu.c (spu_function_arg_offset): New function.
12312 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12313 * system.h (FUNCTION_ARG_OFFSET): Poison.
12315 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12316 Alan Hayard <alan.hayward@arm.com>
12317 David Sherwood <david.sherwood@arm.com>
12319 * target.def (truly_noop_truncation): New hook.
12320 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12321 than TRULY_NOOP_TRUNCATION.
12322 * hooks.h (hook_bool_uint_uint_true): Declare.
12323 * hooks.c (hook_bool_uint_uint_true): New function.
12324 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12325 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12326 * doc/tm.texi: Regenerate.
12327 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12328 rather than TRULY_NOOP_TRUNCATION in comments.
12329 (simplify_comparison): Likewise.
12330 (record_truncated_value): Likewise.
12331 * expmed.c (extract_bit_field_1): Likewise.
12332 (extract_split_bit_field): Likewise.
12333 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12334 instead of TRULY_NOOP_TRUNCATION.
12335 * function.c (assign_parm_setup_block): Likewise.
12336 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12337 * rtlhooks.c: Include target.h.
12338 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12339 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12340 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12341 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12342 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12343 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12344 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12345 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12346 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12347 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12348 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12349 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12350 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12351 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12352 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12353 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12354 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12355 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12356 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12357 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12358 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12359 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12360 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12361 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12362 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12363 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12364 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12365 rather than TRULY_NOOP_TRUNCATION in comments.
12366 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12367 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12368 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12369 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12370 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12371 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12372 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12373 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12374 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12375 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12376 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12377 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12378 rather than TRULY_NOOP_TRUNCATION in comments.
12379 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12380 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12381 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12382 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12383 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12384 TRULY_NOOP_TRUNCATION condition.
12385 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12386 (TRULY_NOOP_TRUNCATION): Delete.
12387 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12388 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12389 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12390 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12391 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12392 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12393 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12394 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12395 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12396 rather than TRULY_NOOP_TRUNCATION in comments.
12397 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12398 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12399 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12400 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12401 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12402 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12404 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12407 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12408 (*cmp_ior): Likewise.
12409 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12410 (*ior_scc_scc_cmp): Likewise.
12411 (*and_scc_scc): Likewise.
12412 (*and_scc_scc_cmp): Likewise.
12414 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12415 Alan Hayard <alan.hayward@arm.com>
12416 David Sherwood <david.sherwood@arm.com>
12418 * target.def (can_change_mode_class): New hook.
12419 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12420 (hard_regno_nregs): Likewise.
12421 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12422 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12423 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12424 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12425 (LOAD_EXTEND_OP): Update accordingly.
12426 * doc/tm.texi: Regenerate.
12427 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12428 CANNOT_CHANGE_MODE_CLASS.
12429 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12430 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12431 * combine.c (simplify_set): Update accordingly.
12432 * emit-rtl.c (validate_subreg): Likewise.
12433 * recog.c (general_operand): Likewise.
12434 * regcprop.c (mode_change_ok): Likewise.
12435 * reload1.c (choose_reload_regs): Likewise.
12436 (inherit_piecemeal_p): Likewise.
12437 * rtlanal.c (simplify_subreg_regno): Likewise.
12438 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12439 instead of CANNOT_CHANGE_MODE_CLASS.
12440 (reload_cse_simplify_operands): Likewise.
12441 * reload.c (push_reload): Use targetm.can_change_mode_class
12442 instead of CANNOT_CHANGE_MODE_CLASS.
12443 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12444 REG_CANNOT_CHANGE_MODE_P.
12445 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12446 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12447 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12448 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12449 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12450 (arm_can_change_mode_class): New function.
12451 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12452 than CANNOT_CHANGE_MODE_CLASS in comments.
12453 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12454 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12455 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12456 (ix86_can_change_mode_class): ...this new function, inverting the
12457 sense of the return value.
12458 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12459 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12460 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12461 (ia64_can_change_mode_class): New function.
12462 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12463 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12464 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12465 (m32c_can_change_mode_class): ...this new function, inverting the
12466 sense of the return value.
12467 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12468 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12469 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12470 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12471 (mips_can_change_mode_class): ...this new function, inverting the
12472 sense of the return value.
12473 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12474 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12475 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12476 (msp430_can_change_mode_class): New function.
12477 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12478 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12479 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12480 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12481 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12482 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12483 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12484 (pa_cannot_change_mode_class): Replace with...
12485 (pa_can_change_mode_class): ...this new function, inverting the
12486 sense of the return value.
12487 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12488 than CANNOT_CHANGE_MODE_CLASS in comments.
12489 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12490 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12491 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12492 (pdp11_cannot_change_mode_class): Replace with...
12493 (pdp11_can_change_mode_class): ...this new function, inverting the
12494 sense of the return value.
12495 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12496 * config/powerpcspe/powerpcspe-protos.h
12497 (rs6000_cannot_change_mode_class_ptr): Delete.
12498 * config/powerpcspe/powerpcspe.c
12499 (rs6000_cannot_change_mode_class_ptr): Delete.
12500 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12501 (rs6000_option_override_internal): Assign to
12502 targetm.can_change_mode_class instead of
12503 rs6000_cannot_change_mode_class_ptr.
12504 (rs6000_cannot_change_mode_class): Replace with...
12505 (rs6000_can_change_mode_class): ...this new function, inverting the
12506 sense of the return value.
12507 (rs6000_debug_cannot_change_mode_class): Replace with...
12508 (rs6000_debug_can_change_mode_class): ...this new function.
12509 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12510 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12511 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12512 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12513 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12515 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12516 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12517 (rs6000_option_override_internal): Assign to
12518 targetm.can_change_mode_class instead of
12519 rs6000_cannot_change_mode_class_ptr.
12520 (rs6000_cannot_change_mode_class): Replace with...
12521 (rs6000_can_change_mode_class): ...this new function, inverting the
12522 sense of the return value.
12523 (rs6000_debug_cannot_change_mode_class): Replace with...
12524 (rs6000_debug_can_change_mode_class): ...this new function.
12525 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12526 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12527 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12528 (s390_can_change_mode_class): ...this new function, inverting the
12529 sense of the return value.
12530 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12531 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12532 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12533 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12534 (sh_cannot_change_mode_class): Replace with...
12535 (sh_can_change_mode_class): ...this new function, inverting the
12536 sense of the return value.
12537 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12538 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12539 (sparc_can_change_mode_class): New function.
12540 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12541 * config/spu/spu.c (spu_can_change_mode_class): New function.
12542 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12543 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12544 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12545 (visium_can_change_mode_class): New function.
12546 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12548 2017-09-15 Richard Biener <rguenther@suse.de>
12550 PR tree-optimization/82217
12551 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12552 but not undefined case.
12554 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12557 * postreload.c (reload_cse_simplify_operands): Skip
12558 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12560 2017-09-15 Richard Biener <rguenther@suse.de>
12562 PR tree-optimization/68823
12563 * graphite-scop-detection.c (build_alias_set): If we have a
12564 possible dependence check whether we can handle them by just
12565 looking at the DRs DR_ACCESS_FNs.
12566 (build_scops): If build_alias_set fails, fail the SCOP.
12568 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12570 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12571 to support float128 built-in functions that require the ISA 3.0
12573 (BU_FLOAT128_3_HW): Likewise.
12574 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12575 built-in functions.
12576 (FMAF128): Likewise.
12578 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12579 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12580 floating point instructions.
12581 (rs6000_invalid_builtin): Likewise.
12582 (rs6000_builtin_mask_names): Likewise.
12583 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12584 (RS6000_BTM_FLOAT128_HW): Likewise.
12585 (RS6000_BTM_COMMON): Likewise.
12586 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12588 * doc/extend.texi (RS/6000 built-in functions): Document the
12589 IEEE 128-bit floating point square root and fused multiply-add
12590 built-in functions.
12592 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12594 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12595 reg (r2) isn't in the set of registers defined in the prologue.
12597 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12598 Alan Hayward <alan.hayward@arm.com>
12599 David Sherwood <david.sherwood@arm.com>
12601 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12602 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12603 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12604 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12605 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12607 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12608 max_vectorization_factor.
12609 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12611 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12612 Alan Hayward <alan.hayward@arm.com>
12613 David Sherwood <david.sherwood@arm.com>
12615 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12616 (vect_worthwhile_without_simd_p): Declare.
12617 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12618 (vectorizable_reduction): Use it.
12619 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12620 (vectorizable_operation): Likewise.
12622 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12623 Alan Hayward <alan.hayward@arm.com>
12624 David Sherwood <david.sherwood@arm.com>
12626 * tree-vectorizer.h (vect_get_num_copies): New function.
12627 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12628 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12629 (vectorizable_induction): Likewise.
12630 (vectorizable_live_operation): Likewise.
12631 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12632 (vectorizable_bswap): Likewise.
12633 (vectorizable_call): Likewise.
12634 (vectorizable_conversion): Likewise.
12635 (vectorizable_assignment): Likewise.
12636 (vectorizable_shift): Likewise.
12637 (vectorizable_operation): Likewise.
12638 (vectorizable_store): Likewise.
12639 (vectorizable_load): Likewise.
12640 (vectorizable_condition): Likewise.
12641 (vectorizable_comparison): Likewise.
12642 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12644 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12645 Alan Hayward <alan.hayward@arm.com>
12646 David Sherwood <david.sherwood@arm.com>
12648 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12649 of vect_init_vector.
12651 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12652 Alan Hayward <alan.hayward@arm.com>
12653 David Sherwood <david.sherwood@arm.com>
12655 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12656 an inline wrapper that provides a location.
12657 (gimple_build_vector): Likewise.
12658 * gimple-fold.c (gimple_build_vector_from_val): New function.
12659 (gimple_build_vector): Likewise.
12660 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12661 functions to build the initial value. Always return a gimple value.
12662 (get_initial_defs_for_reduction): Likewise. Only compute
12664 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12665 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12666 (vectorizable_induction): Use gimple_build_vector rather than
12669 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12670 Alan Hayward <alan.hayward@arm.com>
12671 David Sherwood <david.sherwood@arm.com>
12673 * target.h (vec_perm_indices): New typedef.
12674 (auto_vec_perm_indices): Likewise.
12675 * optabs-query.h: Include target.h
12676 (can_vec_perm_p): Take a vec_perm_indices *.
12677 * optabs-query.c (can_vec_perm_p): Likewise.
12678 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12679 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12680 * tree-vect-generic.c (lower_vec_perm): Likewise.
12681 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12682 (vect_grouped_load_supported): Likewise.
12683 (vect_shift_permute_load_chain): Likewise.
12684 (vect_permute_store_chain): Use auto_vec_perm_indices.
12685 (vect_permute_load_chain): Likewise.
12686 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12687 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12688 Update uses of can_vec_perm_p.
12689 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12690 mode with a number of elements. Take a vec_perm_indices *.
12691 (vect_create_epilog_for_reduction): Update accordingly.
12692 Use auto_vec_perm_indices.
12693 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12694 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12695 (vect_transform_slp_perm_load): Likewise.
12696 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12697 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12698 (vect_gen_perm_mask_checked): Likewise.
12699 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12700 (vect_gen_perm_mask_checked): Likewise.
12701 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12702 (vectorizable_store): Likewise.
12703 (vectorizable_load): Likewise.
12704 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12705 (vectorizable_bswap): Likewise.
12707 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12708 Alan Hayward <alan.hayward@arm.com>
12709 David Sherwood <david.sherwood@arm.com>
12711 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12712 * tree.c (build_vector): Likewise.
12713 (build_vector_from_ctor): Update accordingly.
12714 (build_vector_from_val): Likewise.
12715 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12716 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12717 * tree-vect-generic.c (add_rshift): Likewise.
12718 (expand_vector_divmod): Likewise.
12719 (optimize_vector_constructor): Likewise.
12720 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12721 (vect_transform_slp_perm_load): Likewise.
12722 (vect_schedule_slp_instance): Likewise.
12723 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12724 (vectorizable_call): Likewise.
12725 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12726 * expmed.c (make_tree): Likewise.
12727 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12728 a vector passed to build_vector.
12729 (fold_convert_const): Likewise.
12730 (exact_inverse): Likewise.
12731 (fold_ternary_loc): Likewise.
12732 (fold_relational_const): Likewise.
12733 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12734 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12735 (const_unop): Likewise. Store the reduction accumulator in a
12736 variable rather than an array.
12737 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12738 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12739 the new vector, rather than constructing it after the input arrays.
12740 (native_interpret_vector): Use auto_vec<tree> when building
12741 a vector passed to build_vector. Add elements in order.
12742 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12743 auto_vec<tree> when building a vector passed to build_vector.
12744 (vect_create_epilog_for_reduction): Likewise.
12745 (vectorizable_induction): Likewise.
12746 (get_initial_def_for_reduction): Likewise. Fix indentation of
12748 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12750 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12751 passed to build_vector.
12753 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12754 Alan Hayward <alan.hayward@arm.com>
12755 David Sherwood <david.sherwood@arm.com>
12757 * tree-core.h (tree_base::u): Add an "nelts" field.
12758 (tree_vector): Use VECTOR_CST_NELTS as the length.
12759 * tree.c (tree_size): Likewise.
12760 (make_vector): Initialize VECTOR_CST_NELTS.
12761 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12762 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12763 TYPE_VECTOR_SUBPARTS.
12764 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12765 as the number of units, setting it from VECTOR_CST_NELTS.
12766 (const_vector_from_tree): Likewise.
12767 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12768 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12769 (fold_negate_expr_1): Likewise.
12770 (fold_convert_const): Likewise.
12771 (const_binop): Likewise. Differentiate the number of output and
12773 (const_unop): Likewise.
12774 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12775 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12776 in cases that did the opposite.
12778 2017-09-14 Richard Biener <rguenther@suse.de>
12780 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12783 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12785 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12787 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12790 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12791 if and where to split a bb, except for splitting before debug insn
12792 sequences followed by non-label real insn. Delete debug insns
12793 in between basic blocks.
12795 * combine.c (make_compound_operation_int): Formatting fixes.
12797 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12798 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12799 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12800 * config/sol2.h (LINK_EH_SPEC): Likewise.
12801 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12802 * config/s390/linux.h (LINK_SPEC): Likewise.
12803 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12804 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12805 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12806 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12807 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12808 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12809 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12810 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12812 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12814 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12816 (ENDFILE_LINUX_SPEC): Likewise.
12817 (LINK_EH_SPEC): Likewise.
12818 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12819 (LINK_OS_LINUX_SPEC32): Likewise.
12820 (LINK_OS_LINUX_SPEC64): Likewise.
12821 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12822 (LINK_OS_LINUX_SPEC): Likewise.
12824 2017-09-13 Martin Liska <mliska@suse.cz>
12826 PR middle-end/82154
12827 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12828 CASE_HIGH is NULL_TREE.
12830 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12831 Alan Hayward <alan.hayward@arm.com>
12832 David Sherwood <david.sherwood@arm.com>
12834 * target.def (secondary_memory_needed): New hook.
12835 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12836 instead of SECONDARY_MEMORY_NEEDED.
12837 (secondary_memory_needed_mode): Likewise.
12838 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12839 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12840 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12841 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12842 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12843 * doc/tm.texi: Regenerate.
12844 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12845 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12846 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12847 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12848 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12849 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12850 mode argument first and change the reg_class arguments to reg_class_t.
12851 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12852 Make static. Update the call to inline_secondary_memory_needed.
12853 (ix86_register_move_cost): Update the call to
12854 inline_secondary_memory_needed.
12855 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12856 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12858 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12859 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12861 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12862 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12863 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12864 and match hook interface. Add comment from mips.h.
12865 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12866 * config/mmix/mmix.md (truncdfsf2): Refer to
12867 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12869 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12870 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12871 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12872 (pa_secondary_memory_needed): New function.
12873 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12874 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12875 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12876 (pdp11_secondary_memory_needed): Make static and match hook interface.
12877 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12878 * config/powerpcspe/powerpcspe-protos.h
12879 (rs6000_secondary_memory_needed_ptr): Delete.
12880 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12882 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12883 (rs6000_option_override_internal): Assign to
12884 targetm.secondary_memory_needed rather than
12885 rs6000_secondary_memory_needed_ptr.
12886 (rs6000_secondary_memory_needed): Match hook interface.
12887 (rs6000_debug_secondary_memory_needed): Likewise.
12888 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12889 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12890 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12891 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12892 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12893 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12895 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12896 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12897 (rs6000_option_override_internal): Assign to
12898 targetm.secondary_memory_needed rather than
12899 rs6000_secondary_memory_needed_ptr.
12900 (rs6000_secondary_memory_needed): Match hook interface.
12901 (rs6000_debug_secondary_memory_needed): Likewise.
12902 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12903 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12904 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12905 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12906 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12907 (sparc_secondary_memory_needed): New function.
12908 * lra-constraints.c (check_and_process_move): Refer to
12909 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12911 (curr_insn_transform): Likewise.
12912 (process_alt_operands): Use targetm.secondary_memory_needed
12913 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12914 (check_secondary_memory_needed_p): Likewise.
12915 (choose_split_class): Likewise.
12916 * reload.c: Unconditionally include code that was previously
12917 conditional on SECONDARY_MEMORY_NEEDED.
12918 (push_secondary_reload): Use targetm.secondary_memory_needed
12919 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12920 (push_reload): Likewise.
12921 * reload1.c: Unconditionally include code that was previously
12922 conditional on SECONDARY_MEMORY_NEEDED.
12923 (choose_reload_regs): Use targetm.secondary_memory_needed
12924 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12925 (gen_reload): Likewise.
12926 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12928 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12929 Alan Hayward <alan.hayward@arm.com>
12930 David Sherwood <david.sherwood@arm.com>
12932 * target.def (secondary_memory_needed_mode): New hook:
12933 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12934 * targhooks.h (default_secondary_memory_needed_mode): New function.
12935 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12936 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12937 * doc/tm.texi: Regenerate.
12938 * lra-constraints.c (check_and_process_move): Use
12939 targetm.secondary_memory_needed_mode instead of
12940 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12941 (curr_insn_transform): Likewise.
12942 * reload.c (get_secondary_mem): Likewise.
12943 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12944 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12946 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12947 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12948 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12949 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12950 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12952 * config/powerpcspe/powerpcspe-protos.h
12953 (rs6000_secondary_memory_needed_mode): Delete.
12954 * config/powerpcspe/powerpcspe.c
12955 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12956 (rs6000_secondary_memory_needed_mode): Make static.
12957 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12958 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12960 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12962 (rs6000_secondary_memory_needed_mode): Make static.
12963 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12964 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12965 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12966 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12967 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12969 (sparc_secondary_memory_needed_mode): New function.
12970 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12972 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12974 * config/aarch64/constraints.md (Umq): New constraint.
12975 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12977 (mov<mode>): Update condition.
12979 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12981 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12982 when bitposition is the same.
12984 2017-09-13 Richard Biener <rguenther@suse.de>
12986 * dwarf2out.c (output_die_symbol): Remove.
12987 (output_die): Do not output a DIEs symbol.
12989 2017-09-13 Richard Biener <rguenther@suse.de>
12991 PR middle-end/82128
12992 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12993 default-def to avoid breaking iterator update with the weird
12994 interaction with cgraph_update_edges_for_call_stmt_node.
12996 2017-09-13 Richard Biener <rguenther@suse.de>
12998 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12999 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13000 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13001 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13003 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13005 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13006 Disable pc relative literal load irrespective of
13007 TARGET_FIX_ERR_A53_84341 for default.
13009 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13011 * config/sparc/sparc.c (output_return): Output the source location of
13012 the insn in the delay slot, if any.
13013 (output_sibcall): Likewise.
13015 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13018 * common.opt (-static-pie): New alias.
13019 (shared): Negate static-pie.
13020 (-no-pie): Update help text.
13022 (static-pie): New option.
13023 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13024 -static-pie support.
13025 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13026 (LINK_EH_SPEC): Likewise.
13027 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13028 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13029 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13030 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13031 (init_gcc_specs): Likewise.
13032 (init_spec): Likewise.
13033 (display_help): Update help message for -pie.
13034 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13037 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13039 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13040 (movdi_aarch64): Likewise.
13041 (movti_aarch64): Likewise.
13043 2017-09-12 Simon Wright <simon@pushface.org>
13046 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13047 calculation of the minor version, always output as 0.
13049 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13052 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13053 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13054 on it early, rather than manual conversion late. For
13055 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13056 instead of performing manual conversion.
13058 2017-09-12 Carl Love <cel@us.ibm.com>
13060 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13061 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13063 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13064 VMULOSW): Add definitions.
13065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13066 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13067 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13068 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13069 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13071 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13073 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13075 (movti_aarch64): Likewise.
13076 (movdf_aarch64): Likewise.
13077 (movtf_aarch64): Likewise.
13078 (load_pairdi): Likewise.
13079 (store_pairdi): Likewise.
13080 (load_pairdf): Likewise.
13081 (store_pairdf): Likewise.
13082 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13083 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13084 (ldr_got_small_<mode>): Likewise.
13085 (ldr_got_small_28k_<mode>): Likewise.
13086 (ldr_got_tiny): Likewise.
13087 * config/aarch64/iterators.md (ldst_sz): New.
13088 (ldpstp_sz): Likewise.
13089 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13091 (thunderx_load): Split load_8 to load_16.
13092 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13094 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13095 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13096 (xgene1_store_pair): Split store_8 to store_16.
13097 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13098 (falkor_st_0_st_sd): Split store_8 to store_16.
13100 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13102 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13103 and store1/2/3/4 to store_4/8/12/16.
13104 * config/aarch64/aarch64.md: Update for rename.
13105 * config/arm/arm.md: Likewise.: Likewise.
13106 * config/arm/arm.c: Likewise.
13107 * config/arm/thumb1.md: Likewise.
13108 * config/arm/thumb2.md: Likewise.
13109 * config/arm/vfp.md: Likewise.
13110 * config/arm/arm-generic.md: Likewise.
13111 * config/arm/arm1020e.md: Likewise.
13112 * config/arm/arm1026ejs.md: Likewise.
13113 * config/arm/arm1136jfs.md: Likewise.
13114 * config/arm/arm926ejs.md: Likewise.
13115 * config/arm/cortex-a15.md: Likewise.
13116 * config/arm/cortex-a17.md: Likewise.
13117 * config/arm/cortex-a5.md: Likewise.
13118 * config/arm/cortex-a53.md: Likewise.
13119 * config/arm/cortex-a57.md: Likewise.
13120 * config/arm/cortex-a7.md: Likewise.
13121 * config/arm/cortex-a8.md: Likewise.
13122 * config/arm/cortex-a9.md: Likewise.
13123 * config/arm/cortex-m4.md: Likewise.
13124 * config/arm/cortex-m7.md: Likewise.
13125 * config/arm/cortex-r4.md: Likewise.
13126 * config/arm/exynos-m1.md: Likewise.
13127 * config/arm/fa526.md: Likewise.
13128 * config/arm/fa606te.md: Likewise.
13129 * config/arm/fa626te.md: Likewise.
13130 * config/arm/fa726te.md: Likewise.
13131 * config/arm/fmp626.md: Likewise.
13132 * config/arm/iwmmxt.md: Likewise.
13133 * config/arm/ldmstm.md: Likewise.
13134 * config/arm/marvell-pj4.md: Likewise.
13135 * config/arm/xgene1.md: Likewise.
13136 * config/aarch64/thunderx.md: Likewise.
13137 * config/aarch64/thunderx2t99.md: Likewise.
13138 * config/aarch64/falkor.md: Likewise.
13140 2017-09-12 Martin Liska <mliska@suse.cz>
13142 * attribs.c (private_lookup_attribute): New function.
13143 * attribs.h (private_lookup_attribute): Declared here.
13144 (lookup_attribute): Called from this place.
13146 2017-09-12 Richard Biener <rguenther@suse.de>
13148 PR tree-optimization/82157
13149 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13150 stmts with side-effects.
13152 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13153 Alan Hayward <alan.hayward@arm.com>
13154 David Sherwood <david.sherwood@arm.com>
13156 * target.def (hard_regno_nregs): New hook.
13157 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13158 * targhooks.h (default_hard_regno_nregs): Declare.
13159 * targhooks.c (default_hard_regno_nregs): New function.
13160 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13161 (TARGET_HARD_REGNO_NREGS): ...this hook.
13162 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13163 (CLASS_MAX_NREGS): Likewise.
13164 * doc/tm.texi: Regenerate.
13165 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13166 instead of HARD_REGNO_NREGS.
13167 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13168 HARD_REGNO_NREGS in the comment.
13169 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13170 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13171 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13172 Return an unsigned int.
13173 (TARGET_HARD_REGNO_NREGS): Redefine.
13174 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13175 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13176 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13177 (arc_hard_regno_nregs): New function.
13178 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13179 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13180 (arm_hard_regno_nregs): New function.
13181 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13182 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13183 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13184 (TARGET_HARD_REGNO_NREGS): Redefine.
13185 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13186 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13187 (HARD_REGNO_NREGS): Delete.
13188 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13189 (cr16_hard_regno_nregs): New function.
13190 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13191 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13192 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13193 (cris_hard_regno_nregs): New function.
13194 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13195 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13196 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13197 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13198 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13199 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13200 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13201 (frv_hard_regno_nregs): Make static. Take and return an
13203 (frv_class_max_nregs): Remove outdated copy of documentation.
13204 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13205 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13206 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13207 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13208 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13209 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13210 (TARGET_HARD_REGNO_NREGS): Redefine.
13211 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13212 (CLASS_MAX_NREGS): Update comment.
13213 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13214 (ia64_hard_regno_nregs): New function.
13215 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13216 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13217 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13218 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13219 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13221 (m32c_hard_regno_nregs): Likewise. Make static.
13222 (TARGET_HARD_REGNO_NREGS): Redefine.
13223 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13224 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13225 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13226 (m68k_hard_regno_nregs): New function.
13227 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13228 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13229 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13230 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13231 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13232 Take and return an unsigned int.
13233 (TARGET_HARD_REGNO_NREGS): Redefine.
13234 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13235 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13236 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13237 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13238 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13239 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13240 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13241 (msp430_hard_regno_nregs): Make static. Take and return an
13243 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13244 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13245 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13246 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13247 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13248 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13249 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13250 (TARGET_HARD_REGNO_NREGS): Redefine.
13251 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13252 (PA_HARD_REGNO_NREGS): ...this.
13253 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13254 (PA_HARD_REGNO_NREGS): ...this.
13255 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13256 (pa_hard_regno_nregs): New function.
13257 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13258 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13259 (pdp11_hard_regno_nregs): New function.
13260 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13261 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13262 (rs6000_hard_regno_nregs_hook): New function.
13263 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13264 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13265 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13266 Take and return an unsigned int. Move earlier in file.
13267 (TARGET_HARD_REGNO_NREGS): Redefine.
13268 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13269 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13270 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13271 (rl78_hard_regno_nregs): Make static. Take and return an
13273 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13274 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13275 (rs6000_hard_regno_nregs_hook): New function.
13276 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13277 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13278 (TARGET_HARD_REGNO_NREGS): Redefine.
13279 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13280 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13281 instead of HARD_REGNO_NREGS.
13282 (s390_hard_regno_nregs): New function.
13283 (s390_hard_regno_mode_ok): Add comment from s390.h.
13284 (TARGET_HARD_REGNO_NREGS): Redefine.
13285 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13286 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13287 (sh_hard_regno_nregs): New function.
13288 (sh_pass_in_reg_p): Use it.
13289 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13290 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13291 (sparc_hard_regno_nregs): New function.
13292 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13293 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13294 (spu_function_arg_advance): Use it, supplying a valid register number.
13295 (TARGET_HARD_REGNO_NREGS): Redefine.
13296 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13297 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13298 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13299 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13300 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13301 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13302 (CLASS_MAX_NREGS): Remove copy of old documentation.
13303 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13304 (visium_hard_regno_nregs): New function.
13305 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13306 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13307 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13308 xtensa_hard_regno_nregs): New function.
13309 * system.h (HARD_REGNO_NREGS): Poison.
13311 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13313 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13314 hard_regno_nregs instead of HARD_REGNO_NREGS.
13315 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13316 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13317 (c6x_expand_epilogue): Likewise.
13318 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13319 (frv_read_iacc_argument): Likewise.
13320 * config/sh/sh.c: Include regs.h.
13321 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13322 (regs_used): Likewise.
13323 (output_stack_adjust): Likewise.
13324 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13325 * expmed.c: Include regs.h.
13326 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13327 * ree.c: Include regs.h.
13328 (combine_reaching_defs): Use hard_regno_nregs instead of
13330 (add_removable_extension): Likewise.
13332 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13334 * regs.h (hard_regno_nregs): Turn into a function.
13335 (end_hard_regno): Update accordingly.
13336 * caller-save.c (setup_save_areas): Likewise.
13337 (save_call_clobbered_regs): Likewise.
13338 (replace_reg_with_saved_mem): Likewise.
13339 (insert_restore): Likewise.
13340 (insert_save): Likewise.
13341 * combine.c (can_change_dest_mode): Likewise.
13342 (move_deaths): Likewise.
13343 (distribute_notes): Likewise.
13344 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13345 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13346 (rs6000_split_multireg_move): Likewise.
13347 (rs6000_register_move_cost): Likewise.
13348 (rs6000_memory_move_cost): Likewise.
13349 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13350 (rs6000_split_multireg_move): Likewise.
13351 (rs6000_register_move_cost): Likewise.
13352 (rs6000_memory_move_cost): Likewise.
13353 * cselib.c (cselib_reset_table): Likewise.
13354 (cselib_lookup_1): Likewise.
13355 * emit-rtl.c (set_mode_and_regno): Likewise.
13356 * function.c (aggregate_value_p): Likewise.
13357 * ira-color.c (setup_profitable_hard_regs): Likewise.
13358 (check_hard_reg_p): Likewise.
13359 (calculate_saved_nregs): Likewise.
13360 (assign_hard_reg): Likewise.
13361 (improve_allocation): Likewise.
13362 (calculate_spill_cost): Likewise.
13363 * ira-emit.c (modify_move_list): Likewise.
13364 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13365 (ira_hard_reg_in_set_p): Likewise.
13366 * ira.c (setup_reg_mode_hard_regset): Likewise.
13367 (clarify_prohibited_class_mode_regs): Likewise.
13368 (check_allocation): Likewise.
13369 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13370 (lra_setup_reg_renumber): Likewise.
13371 (setup_try_hard_regno_pseudos): Likewise.
13372 (spill_for): Likewise.
13373 (assign_hard_regno): Likewise.
13374 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13375 * lra-constraints.c (in_class_p): Likewise.
13376 (lra_constraint_offset): Likewise.
13377 (simplify_operand_subreg): Likewise.
13378 (lra_constraints): Likewise.
13379 (split_reg): Likewise.
13380 (split_if_necessary): Likewise.
13381 (invariant_p): Likewise.
13382 (inherit_in_ebb): Likewise.
13383 * lra-lives.c (process_bb_lives): Likewise.
13384 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13385 (get_hard_regs): Likewise.
13386 (do_remat): Likewise.
13387 * lra-spills.c (assign_spill_hard_regs): Likewise.
13388 * mode-switching.c (create_pre_exit): Likewise.
13389 * postreload.c (reload_combine_recognize_pattern): Likewise.
13390 * recog.c (peep2_find_free_register): Likewise.
13391 * regcprop.c (kill_value_regno): Likewise.
13392 (set_value_regno): Likewise.
13393 (copy_value): Likewise.
13394 (maybe_mode_change): Likewise.
13395 (find_oldest_value_reg): Likewise.
13396 (copyprop_hardreg_forward_1): Likewise.
13397 * regrename.c (check_new_reg_p): Likewise.
13398 (regrename_do_replace): Likewise.
13399 * reload.c (push_reload): Likewise.
13400 (combine_reloads): Likewise.
13401 (find_dummy_reload): Likewise.
13402 (operands_match_p): Likewise.
13403 (find_reloads): Likewise.
13404 (find_equiv_reg): Likewise.
13405 (reload_adjust_reg_for_mode): Likewise.
13406 * reload1.c (count_pseudo): Likewise.
13407 (count_spilled_pseudo): Likewise.
13408 (find_reg): Likewise.
13409 (clear_reload_reg_in_use): Likewise.
13410 (free_for_value_p): Likewise.
13411 (allocate_reload_reg): Likewise.
13412 (choose_reload_regs): Likewise.
13413 (reload_adjust_reg_for_temp): Likewise.
13414 (emit_reload_insns): Likewise.
13415 (delete_output_reload): Likewise.
13416 * rtlanal.c (subreg_get_info): Likewise.
13417 * sched-deps.c (sched_analyze_reg): Likewise.
13418 * sel-sched.c (init_regs_for_mode): Likewise.
13419 (mark_unavailable_hard_regs): Likewise.
13420 (choose_best_reg_1): Likewise.
13421 (verify_target_availability): Likewise.
13422 * valtrack.c (dead_debug_insert_temp): Likewise.
13423 * var-tracking.c (track_loc_p): Likewise.
13424 (emit_note_insn_var_location): Likewise.
13425 * varasm.c (make_decl_rtl): Likewise.
13426 * reginfo.c (choose_hard_reg_mode): Likewise.
13427 (init_reg_modes_target): Refer directly to
13428 this_target_regs->x_hard_regno_nregs.
13430 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13432 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13433 instead of hard_regno_nregs.
13435 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13437 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13438 end_hard_regno instead of hard_regno_nregs.
13439 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13440 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13441 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13442 * ira-color.c (improve_allocation): Likewise.
13443 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13444 * lra-lives.c (mark_regno_live): Likewise.
13445 (mark_regno_dead): Likewise.
13446 * lra-remat.c (operand_to_remat): Likewise.
13447 * lra.c (collect_non_operand_hard_regs): Likewise.
13448 * postreload.c (reload_combine_note_store): Likewise.
13449 (move2add_valid_value_p): Likewise.
13450 * reload.c (regno_clobbered_p): Likewise.
13452 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13454 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13456 * config/v850/v850.c (v850_reorg): Likewise.
13457 * reload.c (refers_to_regno_for_reload_p): Likewise.
13458 (find_equiv_reg): Likewise.
13459 * reload1.c (reload_reg_reaches_end_p): Likewise.
13461 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13463 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13465 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13466 * config/arm/arm.c (output_move_neon): Likewise.
13467 (arm_attr_length_move_neon): Likewise.
13468 (neon_split_vcombine): Likewise.
13469 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13470 (c6x_mark_reg_written): Likewise.
13471 (c6x_dwarf_register_span): Likewise.
13472 * config/i386/i386.c (ix86_save_reg): Likewise.
13473 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13474 (rws_access_reg): Likewise.
13475 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13476 * mode-switching.c (create_pre_exit): Likewise.
13477 * ree.c (combine_reaching_defs): Likewise.
13478 (add_removable_extension): Likewise.
13479 * regcprop.c (find_oldest_value_reg): Likewise.
13480 (copyprop_hardreg_forward_1): Likewise.
13481 * reload.c (reload_inner_reg_of_subreg): Likewise.
13482 (push_reload): Likewise.
13483 (combine_reloads): Likewise.
13484 (find_dummy_reload): Likewise.
13485 (reload_adjust_reg_for_mode): Likewise.
13486 * reload1.c (find_reload_regs): Likewise.
13487 (forget_old_reloads_1): Likewise.
13488 (reload_reg_free_for_value_p): Likewise.
13489 (reload_adjust_reg_for_temp): Likewise.
13490 (emit_reload_insns): Likewise.
13491 (delete_output_reload): Likewise.
13492 * sel-sched.c (choose_best_reg_1): Likewise.
13493 (choose_best_pseudo_reg): Likewise.
13495 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13496 Alan Hayward <alan.hayward@arm.com>
13497 David Sherwood <david.sherwood@arm.com>
13499 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13500 * target.def (slow_unaligned_access): New hook.
13501 * targhooks.h (default_slow_unaligned_access): Declare.
13502 * targhooks.c (default_slow_unaligned_access): New function.
13503 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13504 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13505 * doc/tm.texi: Regenerate.
13506 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13507 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13508 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13510 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13511 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13513 (rs6000_slow_unaligned_access): New function.
13514 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13515 (expand_block_compare): Likewise.
13516 (expand_strn_compare): Likewise.
13517 (rs6000_rtx_costs): Likewise.
13518 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13519 (riscv_slow_unaligned_access): Likewise.
13520 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13521 (riscv_slow_unaligned_access_p): ...this and make static.
13522 (riscv_option_override): Update accordingly.
13523 (riscv_slow_unaligned_access): New function.
13524 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13525 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13526 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13527 (rs6000_slow_unaligned_access): New function.
13528 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13529 (rs6000_rtx_costs): Likewise.
13530 * config/rs6000/rs6000-string.c (expand_block_compare)
13531 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13532 of SLOW_UNALIGNED_ACCESS.
13533 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13534 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13535 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13536 of SLOW_UNALIGNED_ACCESS.
13537 * expmed.c (simple_mem_bitfield_p): Likewise.
13538 * expr.c (alignment_for_piecewise_move): Likewise.
13539 (emit_group_load_1): Likewise.
13540 (emit_group_store): Likewise.
13541 (copy_blkmode_from_reg): Likewise.
13542 (emit_push_insn): Likewise.
13543 (expand_assignment): Likewise.
13544 (store_field): Likewise.
13545 (expand_expr_real_1): Likewise.
13546 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13547 * lra-constraints.c (simplify_operand_subreg): Likewise.
13548 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13549 * gimple-ssa-store-merging.c: Likewise in block comment at start
13551 * tree-ssa-strlen.c: Include target.h.
13552 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13553 of SLOW_UNALIGNED_ACCESS.
13554 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13556 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13558 PR rtl-optimization/82185
13559 * expmed.c (emit_store_flag_int): Only test tem if it has been
13562 2017-09-12 Richard Biener <rguenther@suse.de>
13564 PR middle-end/82149
13565 * match.pd ((FTYPE) N CMP CST): Fix typo.
13567 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13569 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13571 (mips_near_type_p): Add 'short_call' attribute as a synonym
13573 * doc/extend.texi (short_call): Document new function attribute.
13575 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13578 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13579 assertion check that in the condition.
13580 (get_atomic_generic_size): Likewise. Before testing if parameter
13581 has pointer type, if it has array type, call for C++
13582 default_conversion to perform array-to-pointer conversion.
13584 2017-09-12 Richard Biener <rguenther@suse.de>
13586 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13587 for operations we cannot scalarize.
13589 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13591 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13592 vectors heap vectors. Clean up comments.
13593 Make visited_bbs a reference.
13594 (profitable_jump_thread_path): Make GC
13595 vectors heap vectors. Clean up comments.
13597 (convert_and_register_jump_thread_path): Make GC vectors heap
13599 (check_subpath_and_update_thread_path): Same. Clean up comments.
13600 Make visited_bbs a reference.
13601 (handle_phi): Abstract common code to to
13602 register_jump_thread_path_if_profitable.
13603 Rename VAR_BB to DEF_BB.
13605 Make GC vectors heap vectors.
13606 Make visited_bbs a reference.
13607 (handle_assignment): Same.
13608 (register_jump_thread_path_if_profitable): New.
13609 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13611 Make GC vectors heap vectors. Clean up comments.
13612 Make visited_bbs a reference.
13613 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13614 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13617 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13620 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13621 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13623 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13625 Revert r251800 and r251799.
13627 2017-09-11 Martin Jambor <mjambor@suse.cz>
13630 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13631 arguments in advance.
13632 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13633 use it to find predecessor edges.
13634 (naive_outof_ssa): Collect vector of predecessors.
13636 2017-09-08 Jason Merrill <jason@redhat.com>
13638 PR c++/70029 - ICE with ref-qualifier and -flto
13639 * langhooks.h (struct lang_hooks_for_types): Add
13640 copy_lang_qualifiers.
13641 * attribs.c (build_type_attribute_qual_variant): Use it.
13642 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13644 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13645 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13647 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13650 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13651 (*mulsi3_sp64): New instruction.
13652 (mulsi3): New expander.
13654 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13656 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13658 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13660 * sancov.c: Include memmodel.h.
13662 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13665 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13668 2017-09-07 Carl Love <cel@us.ibm.com>
13670 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13671 the sldi instruction.
13673 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13675 * sancov.c: Include tm_p.h.
13677 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13680 * output.h (switch_to_other_text_partition): New declaration.
13681 * varasm.c (switch_to_other_text_partition): New function.
13682 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13683 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13684 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13685 to the other text partition before emitting LCL label and switch back
13686 after emitting the word after it.
13688 2017-09-07 Richard Biener <rguenther@suse.de>
13690 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13691 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13692 critical edges here, after loop init.
13693 (pass_data_pre): Remove PROP_no_crit_edges flags.
13694 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13695 for valueization of call args to avoid leaking VN_TOP.
13696 (visit_use): Assert we do not visit default defs.
13697 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13698 Use error_mark_node to more easily detect leaking VN_TOP.
13699 All default-defs are varying, not VN_TOP. Mark them visited.
13700 (run_scc_vn): Make code match comment.
13702 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13704 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13705 OPTION_MASK_FLOAT128_KEYWORD.
13706 (POWERPC_MASKS): Likewise.
13707 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13708 support for the -mfloat128-type option, and make -mfloat128
13709 default on PowerPC Linux systems. Define or undefine
13710 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13711 Define __float128 to be __ieee128 if IEEE 128-bit support is
13712 enabled, or undefine it.
13713 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13714 Delete defining __FLOAT128_TYPE__.
13715 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13716 -mfloat128-type option and make -mfloat128 default on PowerPC
13718 (TARGET_FLOAT128_TYPE): Likewise.
13719 (-mfloat128-type): Likewise.
13720 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13721 Delete the -mfloat128-type option and make -mfloat128 default on
13722 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13723 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13724 128-bit floating point is enabled. Change tests from using
13725 -mfloat128-type to -mfloat128.
13726 (rs6000_mangle_type): Use the correct mangling for the __float128
13727 type even if normal long double is restricted to 64-bits.
13728 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13730 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13731 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13732 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13734 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13735 documentation for -mfloat128.
13737 2017-09-06 Olivier Hainque <hainque@adacore.com>
13739 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13741 2017-09-06 Wish Wu <wishwu007@gmail.com>
13742 Jakub Jelinek <jakub@redhat.com>
13744 * asan.c (initialize_sanitizer_builtins): Add
13745 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13746 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13747 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13748 BT_FN_VOID_UINT64_PTR variables.
13749 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13750 (BT_FN_VOID_UINT16_UINT16): Likewise.
13751 (BT_FN_VOID_UINT32_UINT32): Likewise.
13752 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13753 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13754 (BT_FN_VOID_UINT64_PTR): Likewise.
13755 * common.opt (flag_sanitize_coverage): New variable.
13756 (fsanitize-coverage=trace-pc): Remove.
13757 (fsanitize-coverage=): Add.
13758 * flag-types.h (enum sanitize_coverage_code): New enum.
13759 * fold-const.c (fold_range_test): Disable non-short-circuit
13760 optimization if flag_sanitize_coverage.
13761 (fold_truth_andor): Likewise.
13762 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13763 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13764 (coverage_sanitizer_opts): New array.
13765 (get_closest_sanitizer_option): Add OPTS argument, handle also
13766 OPT_fsanitize_coverage_.
13767 (parse_sanitizer_options): Adjusted to also handle
13768 OPT_fsanitize_coverage_.
13769 (common_handle_option): Add OPT_fsanitize_coverage_.
13770 * sancov.c (instrument_comparison, instrument_switch): New function.
13771 (sancov_pass): Add trace-cmp support.
13772 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13773 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13774 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13775 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13776 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13777 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13778 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13779 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13780 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13781 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13783 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13785 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13786 error. Only quit immediately if parsing is complete.
13787 (BEGIN): Initialize fatal_err and parse_done.
13788 (begin fpu, end fpu): Check number of arguments.
13789 (begin arch, end arch): Likewise.
13790 (begin cpu, end cpu): Likewise.
13791 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13792 (optalias): Likewise.
13794 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13796 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13797 * config/arm/arm-isa.h: Delete. Move definitions to ...
13798 * arm-cpus.in: ... here. Use new feature and fgroup values.
13799 * config/arm/arm.c (arm_option_override): Use lower case for feature
13801 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13802 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13803 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13805 (print_isa_bits_for): New function.
13806 (gen_isa): New function.
13807 (gen_comm_data): Use print_isa_bits_for.
13808 (define feature): New keyword.
13809 (define fgroup): New keyword.
13810 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13811 (arm-isa.h): Add rule to generate file.
13812 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13813 case for feature bit names.
13815 2017-09-06 Richard Biener <rguenther@suse.de>
13817 * tree-ssa-pre.c (NECESSARY): Remove.
13818 (create_expression_by_pieces): Do not touch pass-local flags.
13819 (insert_into_preds_of_block): Likewise.
13820 (do_pre_regular_insertion): Likewise.
13821 (eliminate_insert): Likewise.
13822 (eliminate_dom_walker::before_dom_children): Likewise.
13823 (fini_eliminate): Do not look at inserted_exprs.
13824 (mark_operand_necessary): Remove.
13825 (remove_dead_inserted_code): Replace with simple work-list
13826 algorithm based on inserted_exprs and SSA uses.
13827 (pass_pre::execute): Re-order fini_eliminate and
13828 remove_dead_inserted_code.
13830 2017-09-06 Olivier Hainque <hainque@adacore.com>
13832 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13833 for VxWorks 7. Adjust surrounding comments.
13835 2017-09-06 Richard Biener <rguenther@suse.de>
13837 * gimple-ssa-strength-reduction.c
13838 (find_candidates_dom_walker::before_dom_children): Also allow
13841 2017-09-06 Richard Biener <rguenther@suse.de>
13843 PR tree-optimization/82108
13844 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13845 for gap in the non-permutation SLP case.
13847 2017-09-06 Martin Jambor <mjambor@suse.cz>
13849 PR tree-optimization/82078
13850 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13851 add_access_to_work_queue...
13852 (build_accesses_from_assign): ...here.
13853 (propagate_all_subaccesses): Make sure racc is the group
13854 representative, if there is one.
13856 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13858 PR middle-end/82095
13859 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13862 2017-09-06 Richard Biener <rguenther@suse.de>
13864 * gimple-ssa-strength-reduction.c
13865 (find_candidates_dom_walker::before_doom_children): Use a
13866 type and not a mode check.
13868 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13871 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13873 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13875 2017-09-05 Jeff Law <law@redhat.com>
13877 PR tree-optimization/64910
13878 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13879 cases where we have 3 or more operands.
13881 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13883 PR middle-end/81768
13884 * omp-low.c (lower_omp_for): Recompute tree invariant if
13885 gimple_omp_for_initial/final is ADDR_EXPR.
13887 PR middle-end/81768
13888 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13889 into gimple val before gimplification fo the COND_EXPR.
13891 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13893 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13894 REGION_COPY argument.
13895 (thread_through_all_blocks): Remove unused argument to
13896 duplicate_thread_path.
13898 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13899 Alan Hayward <alan.hayward@arm.com>
13900 David Sherwood <david.sherwood@arm.com>
13902 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13903 Take a scalar_mode rather than a machine_mode.
13904 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13905 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13906 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13907 (aarch64_gen_adjusted_ldpstp): Likewise.
13908 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13910 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13911 Alan Hayward <alan.hayward@arm.com>
13912 David Sherwood <david.sherwood@arm.com>
13914 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13915 Take a scalar_int_mode instead of a machine_mode.
13916 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13917 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13918 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13919 (aarch64_simd_attr_length_rglist): Delete.
13920 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13921 a scalar_int_mode instead of a machine_mode.
13922 (aarch64_add_offset): Likewise.
13923 (aarch64_internal_mov_immediate): Likewise
13924 (aarch64_add_constant_internal): Likewise.
13925 (aarch64_add_constant): Likewise.
13926 (aarch64_movw_imm): Likewise.
13927 (aarch64_rtx_arith_op_extract_p): Likewise.
13928 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13929 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13930 Remove assert that the mode isn't a vector.
13931 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13932 (aarch64_expand_mov_immediate): Update calls after above changes.
13933 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13934 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13935 (aarch64_move_imm): Likewise.
13936 (aarch64_can_const_movi_rtx_p): Likewise.
13937 (aarch64_strip_extend): Likewise.
13938 (aarch64_extr_rtx_p): Likewise.
13939 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13940 a CONST_INT when the mode parameter is VOIDmode.
13941 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13945 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13946 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13947 (bitwise_type_for_mode): Update accordingly.
13949 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13951 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13952 * stor-layout.c (mode_for_size_tree): Likewise.
13953 (mode_for_array): Update accordingly.
13954 (layout_decl): Likewise.
13955 (compute_record_mode): Likewise. Only set the mode once.
13957 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13959 * target.def (get_mask_mode): Change return type to opt_mode.
13961 * doc/tm.texi: Regenerate.
13962 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13963 * targhooks.c (default_get_mask_mode): Likewise.
13964 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13965 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13966 targetm.get_mask_mode.
13967 * tree.c (build_truth_vector_type): Likewise.
13969 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13971 * machmode.h (mode_for_vector): Return an opt_mode.
13972 * stor-layout.c (mode_for_vector): Likewise.
13973 (mode_for_int_vector): Update accordingly.
13974 (layout_type): Likewise.
13975 * config/i386/i386.c (emit_memmov): Likewise.
13976 (ix86_expand_set_or_movmem): Likewise.
13977 (ix86_expand_vector_init): Likewise.
13978 (ix86_get_mask_mode): Likewise.
13979 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13981 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13982 * expmed.c (extract_bit_field_1): Likewise.
13983 * expr.c (expand_expr_real_2): Likewise.
13984 * optabs-query.c (can_vec_perm_p): Likewise.
13985 (can_vec_mask_load_store_p): Likewise.
13986 * optabs.c (expand_vec_perm): Likewise.
13987 * targhooks.c (default_get_mask_mode): Likewise.
13988 * tree-vect-stmts.c (vectorizable_store): Likewise.
13989 (vectorizable_load): Likewise.
13990 (get_vectype_for_scalar_type_and_size): Likewise.
13992 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13994 * machmode.h (mode_for_int_vector): New function.
13995 * stor-layout.c (mode_for_int_vector): Likewise.
13996 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13997 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13998 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13999 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14000 (s390_expand_vcond): Likewise.
14002 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14004 * machmode.h (opt_machine_mode): New type.
14005 (opt_mode<T>): Allow construction from anything that can be
14007 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14008 (mode_for_size): Return an opt_machine_mode.
14009 * stor-layout.c (mode_for_size): Likewise.
14010 (mode_for_size_tree): Update call accordingly.
14011 (bitwise_mode_for_mode): Likewise.
14012 (make_fract_type): Likewise.
14013 (make_accum_type): Likewise.
14014 * caller-save.c (replace_reg_with_saved_mem): Update call
14016 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14017 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14018 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14019 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14020 * expmed.c (extract_bit_field_1): Likewise.
14021 * reload.c (get_secondary_mem): Likewise.
14022 * varasm.c (assemble_integer): Likewise.
14023 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14026 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14028 * machmode.h (decimal_float_mode_for_size): New function.
14029 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14030 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14031 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14032 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14033 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14034 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14036 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14038 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14039 (get_builtin_sync_mode): Likewise.
14040 (expand_ifn_atomic_compare_exchange): Likewise.
14041 (expand_builtin_atomic_clear): Likewise.
14042 (expand_builtin_atomic_test_and_set): Likewise.
14043 (fold_builtin_atomic_always_lock_free): Likewise.
14044 * calls.c (compute_argument_addresses): Likewise.
14045 (emit_library_call_value_1): Likewise.
14046 (store_one_arg): Likewise.
14047 * combine.c (combine_instructions): Likewise.
14048 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14049 * config/arm/arm.c (arm_function_value): Likewise.
14050 (aapcs_allocate_return_reg): Likewise.
14051 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14052 * config/i386/i386.c (construct_container): Likewise.
14053 (ix86_gimplify_va_arg): Likewise.
14054 (ix86_expand_sse_cmp): Likewise.
14055 (emit_memmov): Likewise.
14056 (emit_memset): Likewise.
14057 (expand_small_movmem_or_setmem): Likewise.
14058 (ix86_expand_pextr): Likewise.
14059 (ix86_expand_pinsr): Likewise.
14060 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14061 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14063 * config/mips/mips.c (mips_function_value_1) Likewise.
14064 (mips_block_move_straight): Likewise.
14065 (mips_expand_ins_as_unaligned_store): Likewise.
14066 * config/powerpcspe/powerpcspe.c
14067 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14068 (rs6000_darwin64_record_arg_flush): Likewise.
14069 * config/rs6000/rs6000.c
14070 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14071 (rs6000_darwin64_record_arg_flush): Likewise.
14072 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14073 (sparc_function_value_1): Likewise.
14074 * config/spu/spu.c (adjust_operand): Likewise.
14075 (spu_emit_branch_or_set): Likewise.
14076 (arith_immediate_p): Likewise.
14077 * emit-rtl.c (gen_lowpart_common): Likewise.
14078 * expr.c (expand_expr_real_1): Likewise.
14079 * function.c (assign_parm_setup_block): Likewise.
14080 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14081 * reload1.c (alter_reg): Likewise.
14082 * stor-layout.c (mode_for_vector): Likewise.
14083 (layout_type): Likewise.
14085 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14087 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14088 (spu_convert_move): Likewise.
14089 * lower-subreg.c (resolve_simple_move): Likewise.
14091 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14094 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14095 define_insn to a define_expand.
14096 (altivec_vsum2sws_direct): New define_insn.
14097 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14099 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14101 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14104 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14109 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14110 realign stack if stack alignment needed is less than incoming
14113 2017-09-05 Marek Polacek <polacek@redhat.com>
14116 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14119 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14121 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14123 2017-09-05 Richard Biener <rguenther@suse.de>
14125 PR tree-optimization/82084
14126 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14128 2017-09-05 Richard Biener <rguenther@suse.de>
14130 PR tree-optimization/82102
14131 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14133 2017-09-05 Martin Liska <mliska@suse.cz>
14135 PR tree-optimization/82032
14136 * tree-cfg.c (generate_range_test): New function.
14137 * tree-cfg.h (generate_range_test): Declared here.
14138 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14139 (cleanup_control_expr_graph): Use it.
14140 * tree-switch-conversion.c (try_switch_expansion): Remove
14142 (emit_case_nodes): Use generate_range_test.
14144 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14147 * config/i386/i386.md (*<btsc><mode>_mask): Add
14148 TARGET_USE_BT to insn constraint.
14149 (*btr<mode>_mask): Ditto.
14151 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14153 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14154 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14156 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14159 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14160 TARGET_NEON and TARGET_IWMMXT.
14161 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14162 TARGET_NEON and TARGET_IWMMXT.
14163 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14165 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14167 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14168 (ix86_rewrite_tls_address): Ditto.
14169 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14170 (ix86_rewrite_tls_address_1): Ditto.
14171 (ix86_rewrite_tls_address): Ditto.
14172 * config/i386/predicates.md (tls_address_pattern): New predicate.
14173 * config/i386/i386.md (TLS address splitter): New splitter.
14175 2017-09-04 Richard Biener <rguenther@suse.de>
14177 PR tree-optimization/82084
14178 * fold-const.h (can_native_encode_string_p): Declare.
14179 * fold-const.c (can_native_encode_string_p): Factor out from ...
14180 (native_encode_string): ... here.
14181 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14182 vectorizing stores from constants we later cannot handle.
14184 2017-09-04 Marek Polacek <polacek@redhat.com>
14187 * doc/invoke.texi: Update -Wtautological-compare documentation.
14189 2017-09-04 Jeff Law <law@redhat.com>
14191 PR tree-optimization/64910
14192 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14193 swap the first and last operand if the last is a constant.
14195 2017-09-04 Marek Polacek <polacek@redhat.com>
14198 * convert.c (do_narrow): When sanitizing signed integer overflows,
14199 bail out for signed types.
14200 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14202 2017-09-04 Richard Biener <rguenther@suse.de>
14204 PR tree-optimization/82060
14205 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14206 Move devirtualization after stmt folding and before EH/AB/noreturn
14207 cleanup to get the stmt refs canonicalized. Use a bool instead
14208 of gimple_modified_p since that doesn't work for NOPs. Schedule
14209 NOPs generated by folding for removal.
14211 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14212 Alan Hayward <alan.hayward@arm.com>
14213 David Sherwood <david.sherwood@arm.com>
14215 * coretypes.h (pad_direction): New enum.
14216 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14217 (FUNCTION_ARG_PADDING): Likewise.
14218 * target.def (function_arg_padding): New hook.
14219 * targhooks.h (default_function_arg_padding): Declare.
14220 * targhooks.c (default_function_arg_padding): New function.
14221 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14222 (TARGET_FUNCTION_ARG_PADDING): ...this.
14223 * doc/tm.texi: Regenerate.
14224 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14225 instead of direction.
14226 (compute_argument_addresses): Likewise.
14227 (load_register_parameters): Likewise.
14228 (emit_library_call_value_1): Likewise.
14229 (store_one_arg): Use targetm.calls.function_arg_padding instead
14230 of FUNCTION_ARG_PADDING.
14231 (must_pass_in_stack_var_size_or_pad): Likewise.
14232 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14233 (emit_group_store): Likewise.
14234 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14235 instead of FUNCTION_ARG_PADDING.
14236 (emit_push_insn): Likewise, and propagate enum change throughout
14238 * function.h (direction): Delete.
14239 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14241 * function.c (assign_parm_find_stack_rtl): Likewise.
14242 (assign_parm_setup_block_p): Likewise.
14243 (assign_parm_setup_block): Likewise.
14244 (gimplify_parameters): Likewise.
14245 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14246 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14248 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14249 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14250 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14251 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14252 (aarch64_function_arg_padding): ...this new function.
14253 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14254 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14255 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14256 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14257 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14258 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14259 (arm_pad_arg_upward): Replace with...
14260 (arm_function_arg_padding): ...this new function.
14261 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14263 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14264 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14265 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14266 (ia64_hpux_function_arg_padding): Replace with...
14267 (ia64_function_arg_padding): ...this new function. Use pad_direction
14268 instead of direction. Check for TARGET_HPUX.
14269 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14270 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14271 (iq2000_function_arg_padding): New function.
14272 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14273 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14274 (mips_function_arg_padding): ...this new function.
14275 (mips_pad_reg_upward): Update accordingly.
14276 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14277 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14278 targetm.calls.function_arg_padding.
14279 (FUNCTION_ARG_PADDING): Delete.
14280 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14281 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14282 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14283 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14284 (nios2_block_reg_padding): Return pad_direction instead of direction.
14285 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14286 instead of direction.
14287 (nios2_function_arg_padding): Likewise. Make static.
14288 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14289 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14290 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14291 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14292 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14293 (pa_function_arg_padding): Make static. Return pad_direction instead
14295 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14296 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14297 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14298 instead of direction. Use targetm.calls.function_arg_padding.
14299 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14300 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14301 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14302 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14303 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14305 (function_arg_padding): Rename to...
14306 (rs6000_function_arg_padding): ...this. Make static. Return
14307 pad_direction instead of direction.
14308 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14309 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14310 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14311 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14312 instead of direction. Use targetm.calls.function_arg_padding.
14313 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14314 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14315 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14316 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14317 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14318 (function_arg_padding): Rename to...
14319 (rs6000_function_arg_padding): ...this. Make static. Return
14320 pad_direction instead of direction.
14321 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14322 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14323 * config/s390/s390.c (s390_function_arg_padding): New function.
14324 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14325 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14326 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14327 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14328 (function_arg_padding): Rename to...
14329 (sparc_function_arg_padding): ...this. Make static. Return
14330 pad_direction instead of direction.
14331 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14332 * config/spu/spu.c (spu_function_arg_padding): New function.
14333 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14334 * system.h (FUNCTION_ARG_PADDING): Poison.
14336 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14337 Alan Hayward <alan.hayward@arm.com>
14338 David Sherwood <david.sherwood@arm.com>
14340 * target.def (modes_tieable_p): New hook.
14341 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14342 (TARGET_MODES_TIEABLE_P): ...this.
14343 * doc/tm.texi.in: Regenerate.
14344 * hooks.h (hook_bool_mode_mode_true): Declare.
14345 * hooks.c (hook_bool_mode_mode_true): New function.
14346 * combine.c (subst): Use targetm.modes_tieable_p instead of
14348 * dse.c (find_shift_sequence): Likewise.
14349 * expmed.c (extract_low_bits): Likewise.
14350 * lower-subreg.c: Include target.h.
14351 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14353 * rtlanal.c (rtx_cost): Likewise.
14354 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14355 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14356 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14357 (TARGET_MODES_TIEABLE_P): Redefine.
14358 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14359 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14360 (TARGET_MODES_TIEABLE_P): Redefine.
14361 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14362 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14363 (arc_modes_tieable_p): New function.
14364 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14365 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14366 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14367 (arm_modes_tieable_p): Make static.
14368 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14369 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14370 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14371 (TARGET_MODES_TIEABLE_P): Redefine.
14372 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14373 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14374 (TARGET_MODES_TIEABLE_P): Redefine.
14375 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14376 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14377 (cr16_modes_tieable_p): New function.
14378 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14379 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14380 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14381 (TRULY_NOOP_TRUNCATION): Update comment.
14382 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14383 (TRULY_NOOP_TRUNCATION): Update comment.
14384 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14385 (frv_modes_tieable_p): New function.
14386 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14387 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14388 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14389 (TARGET_MODES_TIEABLE_P): Redefine.
14390 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14391 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14392 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14393 (TARGET_MODES_TIEABLE_P): Redefine.
14394 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14395 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14396 (ia64_modes_tieable_p): New function.
14397 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14398 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14399 (iq2000_modes_tieable_p): New function.
14400 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14401 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14402 (lm32_modes_tieable_p): New function.
14403 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14404 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14405 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14406 (TARGET_MODES_TIEABLE_P): Redefine.
14407 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14408 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14409 (m32r_modes_tieable_p): New function.
14410 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14411 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14412 (m68k_modes_tieable_p): New function.
14413 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14414 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14415 (mcore_modes_tieable_p): New function.
14416 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14417 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14419 (TARGET_MODES_TIEABLE_P): Redefine.
14420 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14421 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14422 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14423 (TARGET_MODES_TIEABLE_P): Redefine.
14424 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14425 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14426 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14427 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14428 (mn10300_modes_tieable_p): ...this and make static.
14429 (TARGET_MODES_TIEABLE_P): Redefine.
14430 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14431 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14432 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14433 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14434 (msp430_modes_tieable_p): Make static.
14435 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14436 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14437 (TARGET_MODES_TIEABLE_P): Redefine.
14438 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14439 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14440 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14441 (TARGET_MODES_TIEABLE_P): Redefine.
14442 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14443 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14444 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14445 (TARGET_MODES_TIEABLE_P): Redefine.
14446 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14447 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14448 (pdp11_modes_tieable_p): New function.
14449 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14450 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14451 (rs6000_modes_tieable_p): New function.
14452 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14453 * config/powerpcspe/powerpcspe.md: Update comment.
14454 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14455 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14456 (TARGET_MODES_TIEABLE_P): Redefine.
14457 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14458 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14459 (rl78_modes_tieable_p): New function.
14460 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14461 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14462 (rs6000_modes_tieable_p): New function.
14463 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14464 * config/rs6000/rs6000.md: Update comment.
14465 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14466 * config/rx/rx.c (rx_modes_tieable_p): New function.
14467 (TARGET_MODES_TIEABLE_P): Redefine.
14468 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14469 * config/s390/s390.c (s390_modes_tieable_p): New function.
14470 (TARGET_MODES_TIEABLE_P): Redefine.
14471 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14472 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14473 (sh_modes_tieable_p): New function.
14474 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14475 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14476 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14477 (sparc_modes_tieable_p): Make static.
14478 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14479 * config/spu/spu.c (spu_modes_tieable_p): New function.
14480 (TARGET_MODES_TIEABLE_P): Redefine.
14481 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14482 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14483 (TARGET_MODES_TIEABLE_P): Redefine.
14484 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14485 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14486 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14487 * config/v850/v850.c (v850_modes_tieable_p): New function.
14488 (TARGET_MODES_TIEABLE_P): Redefine.
14489 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14490 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14491 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14492 (visium_modes_tieable_p): New function.
14493 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14494 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14495 (xtensa_modes_tieable_p): New function.
14496 * system.h (MODES_TIEABLE_P): Poison.
14498 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14499 Alan Hayward <alan.hayward@arm.com>
14500 David Sherwood <david.sherwood@arm.com>
14502 * target.def (hard_regno_mode_ok): New hook.
14503 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14504 (TARGET_HARD_REGNO_MODE_OK): ...this.
14505 * doc/tm.texi.in: Regenerate.
14506 * hooks.h (hook_bool_uint_mode_true): Declare.
14507 * hooks.c (hook_bool_uint_mode_true): New function.
14508 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14509 HARD_REGNO_MODE_OK.
14510 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14511 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14512 instead of HARD_REGNO_MODE_OK.
14513 * caller-save.c: Include target.h.
14514 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14515 HARD_REGNO_MODE_OK.
14516 * combine.c (can_combine_p): Likewise.
14517 (combinable_i3pat): Likewise.
14518 (can_change_dest_mode): Likewise.
14519 * expr.c (init_expr_target): Likewise.
14520 (convert_move): Likewise.
14521 (convert_modes): Likewise.
14522 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14523 (setup_prohibited_mode_move_regs): Likewise.
14524 * ira.h (target_ira): Likewise.
14525 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14526 * lra-constraints.c (process_alt_operands): Likewise.
14527 (split_reg): Likewise.
14528 * recog.c (peep2_find_free_register): Likewise.
14529 * ree.c (combine_reaching_defs): Likewise.
14530 * regcprop.c (maybe_mode_change): Likewise.
14531 * reginfo.c (init_reg_sets_1): Likewise.
14532 (choose_hard_reg_mode): Likewise.
14533 (simplifiable_subregs): Likewise.
14534 * regrename.c (check_new_reg_p): Likewise.
14535 * reload.c (find_valid_class): Likewise.
14536 (find_valid_class_1): Likewise.
14537 (reload_inner_reg_of_subreg): Likewise.
14538 (push_reload): Likewise.
14539 (combine_reloads): Likewise.
14540 (find_dummy_reload): Likewise.
14541 (find_reloads): Likewise.
14542 * reload1.c (find_reg): Likewise.
14543 (set_reload_reg): Likewise.
14544 (allocate_reload_reg): Likewise.
14545 (choose_reload_regs): Likewise.
14546 (reload_adjust_reg_for_temp): Likewise.
14547 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14548 (simplify_subreg_regno): Likewise.
14549 * sel-sched.c (init_regs_for_mode): Likewise.
14550 * varasm.c (make_decl_rtl): Likewise.
14551 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14552 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14553 HARD_REGNO_MODE_OK.
14554 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14555 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14556 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14557 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14558 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14560 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14561 (arc_mode_class): Delete.
14562 (HARD_REGNO_MODE_OK): Delete.
14563 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14564 (arc_hard_regno_mode_ok): Rename old array to...
14565 (arc_hard_regno_mode_ok_modes): ...this.
14566 (arc_conditional_register_usage): Update accordingly.
14567 (arc_mode_class): Make static.
14568 (arc_hard_regno_mode_ok): New function.
14569 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14570 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14571 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14572 (arm_hard_regno_mode_ok): Make static.
14573 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14574 HARD_REGNO_MODE_OK.
14575 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14576 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14577 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14579 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14580 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14581 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14582 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14583 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14585 * config/bfin/predicates.md (valid_reg_operand): Use
14586 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14587 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14588 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14589 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14590 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14591 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14592 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14593 (cr16_hard_regno_mode_ok): Make static and return a bool.
14594 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14595 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14596 (cris_hard_regno_mode_ok): New function.
14597 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14598 (epiphany_mode_class): Delete.
14599 (HARD_REGNO_MODE_OK): Delete.
14600 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14601 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14602 (hard_regno_mode_ok): Rename to...
14603 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14604 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14605 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14606 HARD_REGNO_MODE_OK.
14607 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14608 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14609 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14610 (frv_hard_regno_mode_ok): Make static and return a bool.
14611 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14612 HARD_REGNO_MODE_OK.
14613 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14614 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14615 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14616 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14619 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14620 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14621 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14624 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14625 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14626 (ia64_hard_regno_mode_ok): New function.
14627 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14628 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14629 (iq2000_hard_regno_mode_ok): New function.
14630 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14631 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14632 (lm32_hard_regno_mode_ok): New function.
14633 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14634 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14635 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14636 instead of HARD_REGNO_MODE_OK.
14637 (m32c_hard_regno_ok): Rename to...
14638 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14639 (m32c_cannot_change_mode_class): Update accordingly.
14640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14641 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14642 (m32r_mode_class): Delete.
14643 (HARD_REGNO_MODE_OK): Delete.
14644 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14645 (m32r_hard_regno_mode_ok): Rename to...
14646 (m32r_hard_regno_modes): ...this.
14647 (m32r_mode_class): Make static.
14648 (m32r_hard_regno_mode_ok): New function.
14649 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14650 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14651 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14652 (m68k_hard_regno_mode_ok): Make static.
14653 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14654 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14655 (mcore_hard_regno_mode_ok): New function.
14656 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14657 (HARD_REGNO_MODE_OK): Delete.
14658 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14660 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14661 (microblaze_hard_regno_mode_ok): New function.
14662 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14663 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14664 (mips_hard_regno_mode_ok): Delete.
14665 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14666 (mips_hard_regno_mode_ok_p): ...this and make static.
14667 (mips_hard_regno_mode_ok_p): Rename to...
14668 (mips_hard_regno_mode_ok_uncached): ...this.
14669 (mips_hard_regno_mode_ok): New function.
14670 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14671 of HARD_REGNO_MODE_OK.
14672 (mips_option_override): Update after above name changes.
14673 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14674 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14675 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14676 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14677 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14678 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14679 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14680 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14681 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14682 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14683 (msp430_hard_regno_mode_ok): Make static and return a bool.
14684 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14685 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14686 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14689 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14690 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14691 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14692 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14693 (PA_HARD_REGNO_MODE_OK): ...this
14694 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14695 (PA_HARD_REGNO_MODE_OK): ...this.
14696 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14697 (pa_hard_regno_mode_ok): New function.
14698 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14699 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14700 (pdp11_hard_regno_mode_ok): New function.
14701 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14702 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14704 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14706 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14707 (rs6000_hard_regno_mode_ok): Rename to...
14708 (rs6000_hard_regno_mode_ok_uncached): ...this.
14709 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14710 (rs6000_hard_regno_mode_ok): New function.
14711 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14712 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14713 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14714 (riscv_hard_regno_mode_ok): ...this and make static.
14715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14716 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14717 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14718 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14719 (rl78_hard_regno_mode_ok): Make static and return bool.
14720 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14721 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14723 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14725 (rs6000_hard_regno_mode_ok): Rename to...
14726 (rs6000_hard_regno_mode_ok_uncached): ...this.
14727 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14728 (rs6000_hard_regno_mode_ok): New function.
14729 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14730 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14731 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14732 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14733 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14734 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14735 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14736 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14737 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14738 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14739 (sh_hard_regno_mode_ok): Make static.
14740 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14741 instead of HARD_REGNO_MODE_OK.
14742 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14743 (sparc_mode_class): Delete.
14744 (HARD_REGNO_MODE_OK): Delete.
14745 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14746 (hard_regno_mode_classes): Make static.
14747 (sparc_mode_class): Likewise.
14748 (sparc_hard_regno_mode_ok): New function.
14749 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14750 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14751 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14754 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14755 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14756 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14757 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14758 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14759 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14760 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14761 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14762 (visium_hard_regno_mode_ok): New function.
14763 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14764 instead of HARD_REGNO_MODE_OK.
14765 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14766 (HARD_REGNO_MODE_OK): Delete.
14767 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14768 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14769 (xtensa_option_override): Update accordingly.
14770 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14771 (xtensa_hard_regno_mode_ok): New function.
14772 * system.h (HARD_REGNO_MODE_OK): Poison.
14774 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14775 Alan Hayward <alan.hayward@arm.com>
14776 David Sherwood <david.sherwood@arm.com>
14778 * target.def (hard_regno_call_part_clobbered): New hook.
14779 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14780 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14781 * doc/tm.texi: Regenerate.
14782 * hooks.h (hook_bool_uint_mode_false): Declare.
14783 * hooks.c (hook_bool_uint_mode_false): New function.
14784 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14785 * cselib.c (cselib_process_insn): Use
14786 targetm.hard_regno_call_part_clobbered instead of
14787 HARD_REGNO_CALL_PART_CLOBBERED.
14788 * ira-conflicts.c (ira_build_conflicts): Likewise.
14789 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14790 * lra-constraints.c (need_for_call_save_p): Likewise.
14791 * lra-lives.c: Include target.h.
14792 (check_pseudos_live_through_calls): Use
14793 targetm.hard_regno_call_part_clobbered instead of
14794 HARD_REGNO_CALL_PART_CLOBBERED.
14795 * regcprop.c: Include target.h.
14796 (copyprop_hardreg_forward_1): Use
14797 targetm.hard_regno_call_part_clobbered instead of
14798 HARD_REGNO_CALL_PART_CLOBBERED.
14799 * reginfo.c (choose_hard_reg_mode): Likewise.
14800 * regrename.c (check_new_reg_p): Likewise.
14801 * reload.c (find_equiv_reg): Likewise.
14802 * reload1.c (emit_reload_insns): Likewise.
14803 * sched-deps.c (deps_analyze_insn): Likewise.
14804 * sel-sched.c (init_regs_for_mode): Likewise.
14805 (mark_unavailable_hard_regs): Likewise.
14806 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14807 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14808 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14810 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14811 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14812 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14814 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14816 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14817 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14818 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14820 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14821 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14822 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14824 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14825 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14827 * config/powerpcspe/powerpcspe.c
14828 (rs6000_hard_regno_call_part_clobbered): New function.
14829 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14830 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14831 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14833 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14834 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14835 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14837 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14838 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14839 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14841 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14842 Alan Hayward <alan.hayward@arm.com>
14843 David Sherwood <david.sherwood@arm.com>
14845 * rtl.h (subreg_memory_offset): Declare.
14846 * emit-rtl.c (subreg_memory_offset): New function.
14847 * expmed.c (store_bit_field_1): Use it.
14848 * expr.c (undefined_operand_subword_p): Likewise.
14849 * simplify-rtx.c (simplify_subreg): Likewise.
14851 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14853 PR rtl-optimization/57448
14856 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14857 using atomic_load pattern.
14858 (expand_atomic_store): Likewise.
14860 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14863 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14864 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14865 handling. Use replace_call_with_value with NULL instead of
14866 gsi_replace, unlink_stmt_vdef and release_defs.
14868 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14871 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14873 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14876 * rtl.h (emit_library_call_value_1): Declare.
14877 (emit_library_call): Replace declaration with a series of overloads.
14878 Remove the parameter count argument.
14879 (emit_library_call_value): Likewise.
14880 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14881 with an "rtx_mode_t *".
14882 (emit_library_call_value): Delete.
14883 (emit_library_call): Likewise.
14884 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14885 (asan_emit_allocas_unpoison): Likewise.
14886 * builtins.c (expand_builtin_powi): Likewise.
14887 (expand_asan_emit_allocas_unpoison): Likewise.
14888 * cfgexpand.c (expand_main_function): Likewise.
14889 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14890 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14891 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14892 * config/arm/arm.c (arm_trampoline_init): Likewise.
14893 (arm_call_tls_get_addr): Likewise.
14894 (arm_expand_divmod_libfunc): Likewise.
14895 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14896 (smulsi3_highpart): Likewise.
14897 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14898 (c6x_expand_compare): Likewise.
14899 (c6x_expand_movmem): Likewise.
14900 * config/frv/frv.c (frv_trampoline_init): Likewise.
14901 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14902 (ix86_expand_divmod_libfunc): Likewise.
14903 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14904 (ia64_expand_compare): Likewise.
14905 (ia64_profile_hook): Likewise.
14906 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14907 (nonlocal_goto): Likewise.
14908 (restore_stack_nonlocal): Likewise.
14909 * config/m32r/m32r.c (block_move_call): Likewise.
14910 (m32r_trampoline_init): Likewise.
14911 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14912 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14913 (m68k_call_m68k_read_tp): Likewise.
14914 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14915 (microblaze_expand_divide): Likewise.
14916 * config/mips/mips.h (mips_args): Likewise.
14917 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14918 (MIPS_ICACHE_SYNC): Likewise.
14919 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14920 (nios2_trampoline_init): Likewise.
14921 * config/pa/pa.c (hppa_tls_call): Likewise.
14922 (pa_trampoline_init): Likewise.
14923 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14924 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14925 (expand_strn_compare): Likewise.
14926 (rs6000_generate_compare): Likewise.
14927 (rs6000_expand_float128_convert): Likewise.
14928 (output_profile_hook): Likewise.
14929 (rs6000_trampoline_init): Likewise.
14930 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14931 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14932 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14933 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14934 (rs6000_generate_compare): Likewise.
14935 (rs6000_expand_float128_convert): Likewise.
14936 (output_profile_hook): Likewise.
14937 (rs6000_trampoline_init): Likewise.
14938 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14939 * config/sh/sh.c (sh_trampoline_init): Likewise.
14940 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14941 (sparc_emit_float_lib_cmp): Likewise.
14942 (sparc32_initialize_trampoline): Likewise.
14943 (sparc64_initialize_trampoline): Likewise.
14944 (sparc_profile_hook): Likewise.
14945 * config/spu/spu.c (ea_load_store): Likewise.
14946 * config/spu/spu.md (floatunssidf2): Likewise.
14947 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14948 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14949 * config/visium/visium.c (expand_block_move_4): Likewise.
14950 (expand_block_move_2): Likewise.
14951 (expand_block_move_1): Likewise.
14952 (expand_block_set_4): Likewise.
14953 (expand_block_set_2): Likewise.
14954 (expand_block_set_1): Likewise.
14955 (visium_trampoline_init): Likewise.
14956 (visium_profile_hook): Likewise.
14957 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14958 (xtensa_setup_frame_addresses): Likewise.
14959 (xtensa_trampoline_init): Likewise.
14960 * except.c (sjlj_emit_function_enter): Likewise.
14961 (sjlj_emit_function_exit): Likewise.
14962 * explow.c (allocate_dynamic_stack_space): Likewise.
14963 (probe_stack_range): Likewise.
14964 * expr.c (convert_mode_scalar): Likewise.
14965 * optabs.c (expand_binop): Likewise.
14966 (expand_twoval_binop_libfunc): Likewise.
14967 (expand_unop): Likewise.
14968 (prepare_cmp_insn): Likewise.
14969 (prepare_float_lib_cmp): Likewise.
14970 (expand_float): Likewise.
14971 (expand_fix): Likewise.
14972 (expand_fixed_convert): Likewise.
14973 (maybe_emit_sync_lock_test_and_set): Likewise.
14974 (expand_atomic_compare_and_swap): Likewise.
14975 (expand_mem_thread_fence): Likewise.
14976 (expand_atomic_fetch_op): Likewise.
14978 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14980 * doc/generic.texi (OpenACC): Adjust URL.
14981 * doc/invoke.texi (C Dialect Options): Ditto.
14983 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14985 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14986 predicate for operand 1. Add (m,<S>) constraint.
14987 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14988 Prevent memory operand 1 with register operand 2.
14990 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14992 PR rtl-optimization/82024
14993 * combine.c (try_combine): If the combination result is a PARALLEL,
14994 and we only need to retain the SET in there that would be placed
14995 at I2, check that we can place that at I3 instead, before doing so.
14997 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15000 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15002 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15003 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15006 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15007 Jeff Law <law@redhat.com>
15009 * varasm.c (bss_initializer_p): Do not put constants into .bss
15010 (categorize_decl_for_section): Handle bss_initializer_p returning
15011 false when DECL_INITIAL is NULL.
15013 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15016 * config/s390/s390.c (s390_can_inline_p): New function.
15018 2017-09-01 Jeff Law <law@redhat.com>
15020 PR tree-optimization/82052
15021 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15022 Always initialize the returned slot after a hash table miss
15023 when INSERT is true.
15025 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15027 * config/s390/s390.md (mem_signal_fence): Remove.
15028 * doc/md.texi (mem_signal_fence): Remove.
15029 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15031 * target-insns.def (mem_signal_fence): Remove.
15033 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15036 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15039 * asan.c (create_odr_indicator): Strip name encoding from assembler
15040 name before appending it after __odr_asan_.
15042 2017-09-01 Martin Liska <mliska@suse.cz>
15044 PR tree-optimization/82059
15045 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15046 frequency only when an edge is redirected.
15048 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15050 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15051 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15052 (arc_conditional_register_usage): Remove ARC600 lp_count
15054 (arc_file_start): Emit Tag_ARC_CPU_variation.
15055 (arc_can_use_doloop_p): New conditions to use ZOLs.
15056 (hwloop_fail): New function.
15057 (hwloop_optimize): Likewise.
15058 (hwloop_pattern_reg): Likewise.
15059 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15060 (arc_reorg_loops): New function, calls reorg_loops.
15061 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15062 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15064 (arc_loop_hazard): Remove function, functionality moved into
15066 (arc_hazard): Remove arc_loop_hazard call.
15067 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15068 into hwloop_optimize.
15069 (arc_label_align): Remove ZOL handling.
15070 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15071 * config/arc/arc.md (doloop_begin): Remove pattern.
15072 (doloop_begin_i): Likewise.
15073 (doloop_end_i): Likewise.
15074 (doloop_fallback): Likewise.
15075 (doloop_fallback_m): Likewise.
15076 (doloop_end): Reimplement expand.
15077 (arc_lp): New pattern for LP instruction.
15078 (loop_end): New pattern.
15079 (loop_fail): Likewise.
15080 (decrement_and_branch_until_zero): Likewise.
15081 * config/arc/arc.opt (mlpc-width): New option.
15082 * doc/invoke.texi (mlpc-width): Document option.
15084 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15086 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15087 (arc_ccfsm_advance): Fix checking for delay slots.
15088 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15090 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15092 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15094 (movhi_insn): Update store instruction constraint which are saving
15095 6-bit short immediates.
15096 (movsi_insn): Consider also short scaled load operations.
15097 (zero_extendhisi2_i): Use Usd constraint instead of T.
15098 (extendhisi2_i): Add q constraint.
15099 (arc_clzsi2): Add type and length attributes.
15100 (arc_ctzsi2): Likewise.
15101 * config/arc/constraints.md (Usc): Update constraint, the
15102 assembler can parse two relocations for a single instruction.
15104 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15106 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15107 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15109 2017-08-31 Olivier Hainque <hainque@adacore.com>
15111 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15112 match as powerpc-wrs-vxworks*.
15114 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15116 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15117 register constraint for by-element operand.
15118 (aarch64_mls_elt_merge<mode>): Likewise.
15120 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15122 * config/arc/arc.c (arc_can_follow_jump): Check for short
15125 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15127 * config.gcc: Use g.opt for arc.
15128 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15129 functionality moved to ...
15130 (legitimate_scaled_address_p): New function, ...here.
15131 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15132 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15133 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15135 (arc_override_options): Handle G option.
15136 (arc_output_pic_addr_const): Correct function definition.
15137 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15138 (arc_decl_anon_ns_mem_p): Delete.
15139 (arc_in_small_data_p): Overhaul this function to take into
15140 consideration the value given via G option.
15141 (arc_rewrite_small_data_1): Renamed and corrected old
15142 arc_rewrite_small_data function.
15143 (arc_rewrite_small_data): New function.
15144 (small_data_pattern): Don't use pic_offset_table_rtx.
15145 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15146 * config/arc/simdext.md (movmisalignv2hi): Use
15147 prepare_move_operands function.
15149 (movmisalign*): Likewise.
15150 * doc/invoke.texi (ARC options): Document -G option.
15152 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15154 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15156 * config/arc/arc.c (arc_print_operand): Output scalled address for
15157 sdata whenever is possible.
15158 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15159 load/stores are available.
15160 (compact_sda_memory_operand): Check for the alignment required by
15161 code density instructions.
15162 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15164 * config/arc/constraints.md (Usd): Update constraint.
15165 (Us0): New constraint.
15166 (Usc): Update constraint.
15168 2017-08-31 Richard Biener <rguenther@suse.de>
15170 PR middle-end/82054
15171 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15172 function only once.
15174 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15176 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15177 Resize type_signature.
15179 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15180 Alan Hayward <alan.hayward@arm.com>
15181 David Sherwood <david.sherwood@arm.com>
15183 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15184 subregs whose inner modes can be stored in GPRs.
15185 (aarch64_classify_index): Likewise.
15187 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15188 Alan Hayward <alan.hayward@arm.com>
15189 David Sherwood <david.sherwood@arm.com>
15191 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15192 (V_INT_EQUIV): ...this.
15193 (v_cmp_result): Rename to...
15194 (v_int_equiv): ...this.
15195 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15196 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15197 (copysign<mode>3): Likewise.
15198 (aarch64_simd_bsl<mode>_internal): Likewise.
15199 (aarch64_simd_bsl<mode>): Likewise.
15200 (vec_cmp<mode><mode>): Likewise.
15201 (vcond<mode><mode>): Likewise.
15202 (vcond<v_cmp_mixed><mode>): Likewise.
15203 (vcondu<mode><v_cmp_mixed>): Likewise.
15204 (aarch64_cm<optab><mode>): Likewise.
15205 (aarch64_cmtst<mode>): Likewise.
15206 (aarch64_fac<optab><mode>): Likewise.
15207 (vec_perm_const<mode>): Likewise.
15208 (vcond_mask_<mode><v_cmp_result>): Rename to...
15209 (vcond_mask_<mode><v_int_equiv>): ...this.
15210 (vec_cmp<mode><v_cmp_result>): Rename to...
15211 (vec_cmp<mode><v_int_equiv>): ...this.
15213 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15214 Alan Hayward <alan.hayward@arm.com>
15215 David Sherwood <david.sherwood@arm.com>
15217 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15219 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15220 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15221 (UNSPEC_LD4_DREG): New unspecs.
15222 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15223 (aarch64_ld2<mode>_dreg_be): Replace with...
15224 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15226 (aarch64_ld3<mode>_dreg_le)
15227 (aarch64_ld3<mode>_dreg_be): Replace with...
15228 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15230 (aarch64_ld4<mode>_dreg_le)
15231 (aarch64_ld4<mode>_dreg_be): Replace with...
15232 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15235 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15237 PR tree-optimization/81987
15238 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15239 insert an initializer in a location not dominated by the stride
15242 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15244 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15245 on the entire header of the finally block in the fallthru case.
15247 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15249 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15251 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15253 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15254 rs6000_emit_move_from_cr and call renamed function.
15255 (rs6000_emit_prologue): Call renamed functions.
15256 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15257 movesi_from_cr, remove volatile CRs.
15259 2017-08-30 Jon Beniston <jon@beniston.com>
15260 Richard Biener <rguenther@suse.de>
15262 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15263 of VECTOR_MODE_P check.
15264 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15265 element vector types.
15267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15269 * df.h (df_read_modify_subreg_p): Remove in favor of...
15270 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15271 const_rtx instead of an rtx.
15272 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15273 * df-problems.c (df_word_lr_mark_ref): Likewise.
15274 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15275 (mark_pseudo_reg_dead): Likewise.
15276 (mark_ref_dead): Likewise.
15277 * reginfo.c (init_subregs_of_mode): Likewise.
15278 * sched-deps.c (sched_analyze_1): Likewise.
15279 * df-scan.c (df_def_record_1): Likewise.
15280 (df_uses_record): Likewise.
15281 (df_read_modify_subreg_p): Remove in favor of...
15282 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15283 const_rtx instead of an rtx.
15285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15286 Alan Hayward <alan.hayward@arm.com>
15287 David Sherwood <david.sherwood@arm.com>
15289 * rtl.h (partial_subreg_p): New function.
15290 * caller-save.c (save_call_clobbered_regs): Use it.
15291 * calls.c (expand_call): Likewise.
15292 * combine.c (combinable_i3pat): Likewise.
15293 (simplify_set): Likewise.
15294 (make_extraction): Likewise.
15295 (make_compound_operation_int): Likewise.
15296 (gen_lowpart_or_truncate): Likewise.
15297 (force_to_mode): Likewise.
15298 (make_field_assignment): Likewise.
15299 (reg_truncated_to_mode): Likewise.
15300 (record_truncated_value): Likewise.
15301 (move_deaths): Likewise.
15302 * cse.c (record_jump_cond): Likewise.
15303 (cse_insn): Likewise.
15304 * cselib.c (cselib_lookup_1): Likewise.
15305 * expmed.c (extract_bit_field_using_extv): Likewise.
15306 * function.c (assign_parm_setup_reg): Likewise.
15307 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15308 * ira-build.c (create_insn_allocnos): Likewise.
15309 * lra-coalesce.c (merge_pseudos): Likewise.
15310 * lra-constraints.c (match_reload): Likewise.
15311 (simplify_operand_subreg): Likewise.
15312 (curr_insn_transform): Likewise.
15313 * lra-lives.c (process_bb_lives): Likewise.
15314 * lra.c (new_insn_reg): Likewise.
15315 (lra_substitute_pseudo): Likewise.
15316 * regcprop.c (mode_change_ok): Likewise.
15317 (maybe_mode_change): Likewise.
15318 (copyprop_hardreg_forward_1): Likewise.
15319 * reload.c (push_reload): Likewise.
15320 (find_reloads): Likewise.
15321 (find_reloads_subreg_address): Likewise.
15322 * reload1.c (alter_reg): Likewise.
15323 (eliminate_regs_1): Likewise.
15324 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15326 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15328 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15329 back to if statements, including unpack.
15331 2017-08-30 Martin Liska <mliska@suse.cz>
15333 PR inline-asm/82001
15334 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15336 (func_checker::compare_asm_inputs_outputs): ... this function.
15337 (func_checker::compare_gimple_asm): Use the function to compare
15338 also ASM constrains.
15339 * ipa-icf-gimple.h: Rename the function.
15341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15342 Alan Hayward <alan.hayward@arm.com>
15343 David Sherwood <david.sherwood@arm.com>
15345 * coretypes.h (complex_mode): New type.
15346 * gdbhooks.py (build_pretty_printer): Handle it.
15347 * machmode.h (complex_mode): New class.
15348 (complex_mode::includes_p): New function.
15349 (is_complex_int_mode): Likewise.
15350 (is_complex_float_mode): Likewise.
15351 * genmodes.c (get_mode_class): Handle complex mode classes.
15352 * function.c (expand_function_end): Use is_complex_int_mode.
15354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15355 Alan Hayward <alan.hayward@arm.com>
15356 David Sherwood <david.sherwood@arm.com>
15358 * coretypes.h (scalar_mode_pod): New typedef.
15359 * gdbhooks.py (build_pretty_printer): Handle it.
15360 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15361 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15362 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15363 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15364 as_a <scalar_mode>.
15366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15367 Alan Hayward <alan.hayward@arm.com>
15368 David Sherwood <david.sherwood@arm.com>
15370 * machmode.h (mode_for_vector): Take a scalar_mode instead
15372 * stor-layout.c (mode_for_vector): Likewise.
15373 * explow.c (promote_mode): Use as_a <scalar_mode>.
15374 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15377 Alan Hayward <alan.hayward@arm.com>
15378 David Sherwood <david.sherwood@arm.com>
15380 * target.def (preferred_simd_mode): Take a scalar_mode
15381 instead of a machine_mode.
15382 * targhooks.h (default_preferred_simd_mode): Likewise.
15383 * targhooks.c (default_preferred_simd_mode): Likewise.
15384 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15385 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15386 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15387 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15388 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15389 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15390 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15391 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15393 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15394 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15395 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15396 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15397 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15398 * doc/tm.texi: Regenerate.
15399 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15403 Alan Hayward <alan.hayward@arm.com>
15404 David Sherwood <david.sherwood@arm.com>
15406 * target.def (scalar_mode_supported_p): Take a scalar_mode
15407 instead of a machine_mode.
15408 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15409 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15410 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15411 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15412 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15413 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15414 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15415 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15416 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15417 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15418 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15419 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15420 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15421 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15423 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15424 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15425 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15426 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15427 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15429 * doc/tm.texi: Regenerate.
15431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15432 Alan Hayward <alan.hayward@arm.com>
15433 David Sherwood <david.sherwood@arm.com>
15435 * coretypes.h (opt_scalar_mode): New typedef.
15436 * gdbhooks.py (build_pretty_printers): Handle it.
15437 * machmode.h (mode_iterator::get_2xwider): Add overload for
15439 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15441 * expr.c (convert_mode_scalar): Likewise.
15442 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15443 * optabs.c (expand_float): Likewise.
15444 (expand_fix): Likewise.
15445 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15448 Alan Hayward <alan.hayward@arm.com>
15449 David Sherwood <david.sherwood@arm.com>
15451 * optabs.c (expand_float): Explicitly check for scalars before
15452 using a branching expansion.
15453 (expand_fix): Likewise.
15455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15456 Alan Hayward <alan.hayward@arm.com>
15457 David Sherwood <david.sherwood@arm.com>
15459 * expr.c (convert_mode): Split scalar handling out into...
15460 (convert_mode_scalar): ...this new function. Treat the modes
15463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15464 Alan Hayward <alan.hayward@arm.com>
15465 David Sherwood <david.sherwood@arm.com>
15467 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15469 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15472 Alan Hayward <alan.hayward@arm.com>
15473 David Sherwood <david.sherwood@arm.com>
15475 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15476 rather than a machine_mode.
15477 (fixed_from_string): Likewise.
15478 (fixed_convert): Likewise.
15479 (fixed_convert_from_int): Likewise.
15480 (fixed_convert_from_real): Likewise.
15481 (real_convert_from_fixed): Likewise.
15482 * fixed-value.c (fixed_from_double_int): Likewise.
15483 (fixed_from_string): Likewise.
15484 (fixed_convert): Likewise.
15485 (fixed_convert_from_int): Likewise.
15486 (fixed_convert_from_real): Likewise.
15487 (real_convert_from_fixed): Likewise.
15488 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15491 Alan Hayward <alan.hayward@arm.com>
15492 David Sherwood <david.sherwood@arm.com>
15494 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15495 of separate mode class checks. Do not allow vector modes here.
15496 (immed_wide_int_const): Use as_a <scalar_mode>.
15497 * explow.c (trunc_int_for_mode): Likewise.
15498 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15499 (wi::shwi): Likewise.
15500 (wi::min_value): Likewise.
15501 (wi::max_value): Likewise.
15502 * dwarf2out.c (loc_descriptor): Likewise.
15503 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15504 for CONST_WIDE_INT.
15506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15507 Alan Hayward <alan.hayward@arm.com>
15508 David Sherwood <david.sherwood@arm.com>
15510 * tree.h (SCALAR_TYPE_MODE): New macro.
15511 * expr.c (expand_expr_addr_expr_1): Use it.
15512 (expand_expr_real_2): Likewise.
15513 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15514 (fold_convert_const_fixed_from_int): Likewise.
15515 (fold_convert_const_fixed_from_real): Likewise.
15516 (native_encode_fixed): Likewise
15517 (native_encode_complex): Likewise
15518 (native_encode_vector): Likewise.
15519 (native_interpret_fixed): Likewise.
15520 (native_interpret_real): Likewise.
15521 (native_interpret_complex): Likewise.
15522 (native_interpret_vector): Likewise.
15523 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15524 (simd_clone_adjust_argument_types): Likewise.
15525 (simd_clone_init_simd_arrays): Likewise.
15526 (simd_clone_adjust): Likewise.
15527 * stor-layout.c (layout_type): Likewise.
15528 * tree.c (build_minus_one_cst): Likewise.
15529 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15530 * tree-inline.c (estimate_move_cost): Likewise.
15531 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15532 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15533 (vectorizable_reduction): Likewise.
15534 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15535 (vect_recog_mixed_size_cond_pattern): Likewise.
15536 (check_bool_pattern): Likewise.
15537 (adjust_bool_pattern): Likewise.
15538 (search_type_for_mask_1): Likewise.
15539 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15540 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15541 (vectorizable_load): Likewise.
15542 (vectorizable_store): Likewise.
15543 * ubsan.c (ubsan_encode_value): Likewise.
15544 * varasm.c (output_constant): Likewise.
15546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15547 Alan Hayward <alan.hayward@arm.com>
15548 David Sherwood <david.sherwood@arm.com>
15550 * coretypes.h (scalar_mode): New class.
15551 * machmode.h (scalar_mode): Likewise.
15552 (scalar_mode::includes_p): New function.
15553 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15554 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15555 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15556 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15557 * expmed.c (store_bit_field_1): Likewise.
15558 (extract_bit_field_1): Likewise.
15559 * expr.c (write_complex_part): Likewise.
15560 (read_complex_part): Likewise.
15561 (emit_move_complex_push): Likewise.
15562 (expand_expr_real_2): Likewise.
15563 * function.c (assign_parm_setup_reg): Likewise.
15564 (assign_parms_unsplit_complex): Likewise.
15565 * optabs.c (expand_binop): Likewise.
15566 * rtlanal.c (subreg_get_info): Likewise.
15567 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15568 * varasm.c (output_constant_pool_2): Likewise.
15570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15571 Alan Hayward <alan.hayward@arm.com>
15572 David Sherwood <david.sherwood@arm.com>
15574 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15576 (expmed_mult_highpart_optab): Likewise.
15577 (expmed_mult_highpart): Likewise.
15579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15580 Alan Hayward <alan.hayward@arm.com>
15581 David Sherwood <david.sherwood@arm.com>
15583 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15584 instead of a machine_mode.
15585 (builtin_memset_read_str): Likewise.
15586 * builtins.c (c_readstr): Likewise.
15587 (builtin_memcpy_read_str): Likewise.
15588 (builtin_strncpy_read_str): Likewise.
15589 (builtin_memset_read_str): Likewise.
15590 (builtin_memset_gen_str): Likewise.
15591 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15592 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15593 instead of a machine_mode.
15594 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15596 (make_extraction): Likewise.
15597 (try_widen_shift_mode): Take and return scalar_int_modes instead
15599 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15600 a scalar_int_mode instead of a machine_mode.
15601 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15602 (avr_addr_space_pointer_mode): Likewise.
15603 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15604 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15605 (msp430_unwind_word_mode): Likewise.
15606 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15607 (spu_addr_space_pointer_mode): Likewise.
15608 (spu_addr_space_address_mode): Likewise.
15609 (spu_libgcc_cmp_return_mode): Likewise.
15610 (spu_libgcc_shift_count_mode): Likewise.
15611 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15612 (rl78_addr_space_pointer_mode): Likewise.
15613 (fl78_unwind_word_mode): Likewise.
15614 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15616 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15617 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15618 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15619 (mips_valid_pointer_mode): Likewise.
15620 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15621 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15622 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15624 (ft32_addr_space_address_mode): Likewise.
15625 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15626 scalar_int_mode instead of a machine_mode.
15627 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15629 (m32c_addr_space_address_mode): Likewise.
15630 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15631 (rs6000_eh_return_filter_mode): Likewise.
15632 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15633 (rs6000_eh_return_filter_mode): Likewise.
15634 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15635 (s390_libgcc_shift_count_mode): Likewise.
15636 (s390_unwind_word_mode): Likewise.
15637 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15639 * target.def (mode_rep_extended): Likewise.
15640 (valid_pointer_mode): Likewise.
15641 (addr_space.valid_pointer_mode): Likewise.
15642 (eh_return_filter_mode): Return a scalar_int_mode rather than
15644 (libgcc_cmp_return_mode): Likewise.
15645 (libgcc_shift_count_mode): Likewise.
15646 (unwind_word_mode): Likewise.
15647 (addr_space.pointer_mode): Likewise.
15648 (addr_space.address_mode): Likewise.
15649 * doc/tm.texi: Regenerate.
15650 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15652 (do_jump): Use scalar_int_mode for local variables.
15653 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15654 rather than a machine_mode.
15655 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15656 (scompare_loc_descriptor_wide): Likewise.
15657 (scompare_loc_descriptor_narrow): Likewise.
15658 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15660 * except.c (sjlj_emit_dispatch_table): Likewise.
15661 (expand_builtin_eh_copy_values): Likewise.
15662 * explow.c (convert_memory_address_addr_space_1): Likewise.
15663 Take a scalar_int_mode rather than a machine_mode.
15664 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15665 than a machine_mode.
15666 (memory_address_addr_space): Use scalar_int_mode for local variables.
15667 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15668 rather than a machine_mode.
15669 * expmed.c (mask_rtx): Likewise.
15670 (init_expmed_one_conv): Likewise.
15671 (expand_mult_highpart_adjust): Likewise.
15672 (extract_high_half): Likewise.
15673 (expmed_mult_highpart_optab): Likewise.
15674 (expmed_mult_highpart): Likewise.
15675 (expand_smod_pow2): Likewise.
15676 (expand_sdiv_pow2): Likewise.
15677 (emit_store_flag_int): Likewise.
15678 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15680 (extract_low_bits): Likewise.
15681 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15683 * expr.c (pieces_addr::adjust): Likewise.
15684 (can_store_by_pieces): Likewise.
15685 (store_by_pieces): Likewise.
15686 (clear_by_pieces_1): Likewise.
15687 (expand_expr_addr_expr_1): Likewise.
15688 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15689 (expand_expr_real_1): Likewise.
15690 (try_casesi): Likewise.
15691 * final.c (shorten_branches): Likewise.
15692 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15693 type of "mode" to machine_mode.
15694 * internal-fn.c (expand_arith_overflow_result_store): Take a
15695 scalar_int_mode rather than a machine_mode.
15696 (expand_mul_overflow): Use scalar_int_mode for local variables.
15697 * loop-doloop.c (doloop_modify): Likewise.
15698 (doloop_optimize): Likewise.
15699 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15700 than a machine_mode.
15701 (expand_doubleword_shift_condmove): Likewise.
15702 (expand_doubleword_shift): Likewise.
15703 (expand_doubleword_clz): Likewise.
15704 (expand_doubleword_popcount): Likewise.
15705 (expand_doubleword_parity): Likewise.
15706 (expand_absneg_bit): Use scalar_int_mode for local variables.
15707 (prepare_float_lib_cmp): Likewise.
15708 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15709 rather than a machine_mode.
15710 (convert_memory_address_addr_space): Likewise.
15711 (get_mode_bounds): Likewise.
15712 (get_address_mode): Return a scalar_int_mode rather than a
15714 * rtlanal.c (get_address_mode): Likewise.
15715 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15716 than a machine_mode.
15717 * targhooks.c (default_mode_rep_extended): Likewise.
15718 (default_valid_pointer_mode): Likewise.
15719 (default_addr_space_valid_pointer_mode): Likewise.
15720 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15721 than a machine_mode.
15722 (default_libgcc_cmp_return_mode): Likewise.
15723 (default_libgcc_shift_count_mode): Likewise.
15724 (default_unwind_word_mode): Likewise.
15725 (default_addr_space_pointer_mode): Likewise.
15726 (default_addr_space_address_mode): Likewise.
15727 * targhooks.h (default_eh_return_filter_mode): Likewise.
15728 (default_libgcc_cmp_return_mode): Likewise.
15729 (default_libgcc_shift_count_mode): Likewise.
15730 (default_unwind_word_mode): Likewise.
15731 (default_addr_space_pointer_mode): Likewise.
15732 (default_addr_space_address_mode): Likewise.
15733 (default_mode_rep_extended): Take a scalar_int_mode rather than
15735 (default_valid_pointer_mode): Likewise.
15736 (default_addr_space_valid_pointer_mode): Likewise.
15737 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15739 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15740 rather than a machine_mode.
15741 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15742 for local variables.
15743 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15744 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15745 than a machine_mode.
15747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15748 Alan Hayward <alan.hayward@arm.com>
15749 David Sherwood <david.sherwood@arm.com>
15751 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15752 the mode argument to scalar_int_mode.
15753 (do_jump_by_parts_zero_rtx): Likewise.
15754 (do_jump_by_parts_equality_rtx): Likewise.
15755 (do_jump_by_parts_greater): Take a mode argument.
15756 (do_jump_by_parts_equality): Likewise.
15757 (do_jump_1): Update calls accordingly.
15759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15760 Alan Hayward <alan.hayward@arm.com>
15761 David Sherwood <david.sherwood@arm.com>
15763 * is-a.h (safe_dyn_cast): New function.
15764 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15765 (jump_table_for_label): Likewise.
15766 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15767 instead of an rtx_insn *.
15768 (shorten_branches): Use dyn_cast instead of LABEL_P and
15769 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15770 rtx_jump_table_data::get_data_mode.
15771 (final_scan_insn): Likewise.
15773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15774 Alan Hayward <alan.hayward@arm.com>
15775 David Sherwood <david.sherwood@arm.com>
15777 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15778 trying to combine a full-register integer set with a subreg
15781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15782 Alan Hayward <alan.hayward@arm.com>
15783 David Sherwood <david.sherwood@arm.com>
15785 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15786 that is always either address_mode or pointer_mode.
15788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15789 Alan Hayward <alan.hayward@arm.com>
15790 David Sherwood <david.sherwood@arm.com>
15792 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15793 when the two are known to be equal.
15795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15796 Alan Hayward <alan.hayward@arm.com>
15797 David Sherwood <david.sherwood@arm.com>
15799 * simplify-rtx.c (simplify_const_unary_operation): Use
15800 is_a <scalar_int_mode> instead of checking for a nonzero
15801 precision. Forcibly convert op_mode to a scalar_int_mode
15802 in that case. More clearly differentiate the operand and
15803 result modes and use the former when deciding what the value
15804 of a count-bits operation should be. Use is_int_mode instead
15805 of checking for a MODE_INT. Remove redundant check for whether
15806 this mode has a zero precision.
15808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15809 Alan Hayward <alan.hayward@arm.com>
15810 David Sherwood <david.sherwood@arm.com>
15812 * optabs.c (widen_leading): Change the type of the mode argument
15813 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15814 (widen_bswap): Likewise.
15815 (expand_parity): Likewise.
15816 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15817 (expand_ffs): Likewise.
15818 (epand_unop): Check for scalar integer modes before calling the
15821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15822 Alan Hayward <alan.hayward@arm.com>
15823 David Sherwood <david.sherwood@arm.com>
15825 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15827 (expand_expr_real_1): Update call accordingly.
15829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15830 Alan Hayward <alan.hayward@arm.com>
15831 David Sherwood <david.sherwood@arm.com>
15833 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15834 value_mode arguments. Use scalar_int_mode internally.
15835 (store_bit_field_1): Rename the new integer mode from imode
15836 to op0_mode and use it instead of GET_MODE (op0). Update calls
15837 to store_split_bit_field, store_bit_field_using_insv and
15838 store_fixed_bit_field.
15839 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15840 Use scalar_int_mode internally. Use a bit count rather than a mode
15841 when calculating the largest bit size for get_best_mode.
15842 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15843 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15844 Remove assertion that OP0 has a scalar integer mode.
15845 (store_split_bit_field): Add op0_mode and value_mode arguments.
15846 Update calls to extract_fixed_bit_field.
15847 (extract_bit_field_using_extv): Add an op0_mode argument.
15848 Use scalar_int_mode internally.
15849 (extract_bit_field_1): Rename the new integer mode from imode to
15850 op0_mode and use it instead of GET_MODE (op0). Update calls to
15851 extract_split_bit_field, extract_bit_field_using_extv and
15852 extract_fixed_bit_field.
15853 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15854 to extract_split_bit_field and extract_fixed_bit_field_1.
15855 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15856 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15857 on the target mode.
15858 (extract_split_bit_field): Add an op0_mode argument. Update call
15859 to extract_fixed_bit_field.
15861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15862 Alan Hayward <alan.hayward@arm.com>
15863 David Sherwood <david.sherwood@arm.com>
15865 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15866 * explow.c (hard_function_value): Likewise.
15867 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15868 convert_to_mode call outside the loop.
15869 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15870 for the mode iterator. Require the mode specified by max_pieces
15872 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15874 (copy_blkmode_to_reg): Likewise.
15875 (set_storage_via_setmem): Likewise.
15876 * optabs.c (prepare_cmp_insn): Likewise.
15877 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15878 * stor-layout.c (finish_bitfield_representative): Likewise.
15880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15881 Alan Hayward <alan.hayward@arm.com>
15882 David Sherwood <david.sherwood@arm.com>
15884 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15885 * expr.c (convert_move): Use them.
15886 (convert_modes): Likewise.
15887 (store_expr_with_bounds): Likewise.
15889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15890 Alan Hayward <alan.hayward@arm.com>
15891 David Sherwood <david.sherwood@arm.com>
15893 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15894 parameter for the mode of "x". Remove the "known_x", "known_mode"
15895 and "known_ret" arguments. Change the type of the mode argument
15896 to scalar_int_mode.
15897 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15898 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15899 (reg_num_sign_bit_copies_for_combine): Likewise.
15900 * rtlanal.c (nonzero_bits1): Likewise.
15901 (num_sign_bit_copies1): Likewise.
15902 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15903 (reg_num_sign_bit_copies_general): Likewise.
15904 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15905 (reg_nonzero_bits_general): Likewise.
15907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15908 Alan Hayward <alan.hayward@arm.com>
15909 David Sherwood <david.sherwood@arm.com>
15911 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15912 than in subroutines. Return 1 for non-integer modes.
15913 (cached_num_sign_bit_copies): Change the type of the mode parameter
15914 to scalar_int_mode.
15915 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15916 classes. Handle CONST_INT_P first and then check whether X also
15917 has a scalar integer mode. Check the same thing for inner registers
15918 of a SUBREG and for values that are being extended or truncated.
15920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15921 Alan Hayward <alan.hayward@arm.com>
15922 David Sherwood <david.sherwood@arm.com>
15924 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15925 in subroutines. Return the mode mask for non-integer modes.
15926 (cached_nonzero_bits): Change the type of the mode parameter
15927 to scalar_int_mode.
15928 (nonzero_bits1): Likewise. Remove early exit for other mode
15929 classes. Handle CONST_INT_P first and then check whether X
15930 also has a scalar integer mode.
15932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15933 Alan Hayward <alan.hayward@arm.com>
15934 David Sherwood <david.sherwood@arm.com>
15936 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15937 Return a scalar_int_mode and assert that the size is greater than
15939 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15940 (op_by_pieces_d::op_by_pieces_d): Likewise.
15941 (op_by_pieces_d::run): Likewise.
15942 (can_store_by_pieces): Likewise.
15944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15945 Alan Hayward <alan.hayward@arm.com>
15946 David Sherwood <david.sherwood@arm.com>
15948 * combine.c (extract_left_shift): Add a mode argument and update
15950 (make_compound_operation_int): Change the type of the mode parameter
15951 to scalar_int_mode and update the call to extract_left_shift.
15953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15954 Alan Hayward <alan.hayward@arm.com>
15955 David Sherwood <david.sherwood@arm.com>
15957 * combine.c (simplify_and_const_int): Change the type of the mode
15958 parameter to scalar_int_mode.
15959 (simplify_and_const_int_1): Likewise. Update recursive call.
15961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15962 Alan Hayward <alan.hayward@arm.com>
15963 David Sherwood <david.sherwood@arm.com>
15965 * combine.c (simplify_compare_const): Check that the mode is a
15966 scalar_int_mode (rather than VOIDmode) before testing its
15968 (simplify_comparison): Move COMPARISON_P handling out of the
15969 loop and restrict the latter part of the loop to scalar_int_modes.
15970 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15971 and when considering SUBREG_REGs. Use is_int_mode instead of
15972 checking GET_MODE_CLASS against MODE_INT.
15974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15975 Alan Hayward <alan.hayward@arm.com>
15976 David Sherwood <david.sherwood@arm.com>
15978 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15979 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15980 shift_unit_mode and for modes involved in scalar shifts.
15982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15983 Alan Hayward <alan.hayward@arm.com>
15984 David Sherwood <david.sherwood@arm.com>
15986 * combine.c (force_int_to_mode): New function, split out from...
15987 (force_to_mode): ...here. Keep xmode up-to-date and use it
15988 instead of GET_MODE (x).
15990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15991 Alan Hayward <alan.hayward@arm.com>
15992 David Sherwood <david.sherwood@arm.com>
15994 * optabs-query.h (extraction_insn::struct_mode): Change type to
15995 opt_scalar_int_mode and update comment.
15996 (extraction_insn::field_mode): Change type to scalar_int_mode.
15997 (extraction_insn::pos_mode): Likewise.
15998 * combine.c (make_extraction): Update accordingly.
15999 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16000 (get_optab_extraction_insn): Likewise.
16001 * recog.c (simplify_while_replacing): Likewise.
16002 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16003 parameter to opt_scalar_int_mode.
16005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16006 Alan Hayward <alan.hayward@arm.com>
16007 David Sherwood <david.sherwood@arm.com>
16009 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16010 to a scalar_int_mode instead of a machine_mode.
16011 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16012 (get_best_mode): Return a boolean and use a pointer argument to store
16013 the selected mode. Replace the limit mode parameter with a bit limit.
16014 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16015 for the values returned by bit_field_mode_iterator::next_mode.
16016 (store_bit_field): Update call to get_best_mode.
16017 (store_fixed_bit_field): Likewise.
16018 (extract_fixed_bit_field): Likewise.
16019 * expr.c (optimize_bitfield_assignment_op): Likewise.
16020 * fold-const.c (optimize_bit_field_compare): Likewise.
16021 (fold_truth_andor_1): Likewise.
16022 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16023 Update for new type of m_mode.
16024 (get_best_mode): As above.
16026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16027 Alan Hayward <alan.hayward@arm.com>
16028 David Sherwood <david.sherwood@arm.com>
16030 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16031 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16032 (store_bit_field): Check is_a <scalar_int_mode> before calling
16033 strict_volatile_bitfield_p.
16034 (extract_bit_field): Likewise.
16036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16037 Alan Hayward <alan.hayward@arm.com>
16038 David Sherwood <david.sherwood@arm.com>
16040 * target.def (cstore_mode): Return a scalar_int_mode.
16041 * doc/tm.texi: Regenerate.
16042 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16043 * targhooks.h (default_cstore_mode): Likewise.
16044 * targhooks.c (default_cstore_mode): Likewise, using a forced
16046 * expmed.c (emit_cstore): Expect the target of the cstore to be
16049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16050 Alan Hayward <alan.hayward@arm.com>
16051 David Sherwood <david.sherwood@arm.com>
16053 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16055 (niter_desc): Likewise mode.
16056 (iv_analyze): Add a mode parameter.
16058 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16059 and change its type to scalar_int_mode.
16060 * loop-iv.c: Update commentary at head of file.
16061 (iv_constant): Pass the mode paraeter before the rtx it describes
16062 and change its type to scalar_int_mode. Remove VOIDmode handling.
16063 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16064 (iv_extend): Likewise.
16065 (shorten_into_mode): Likewise.
16066 (iv_add): Use scalar_int_mode.
16067 (iv_mult): Likewise.
16068 (iv_shift): Likewise.
16069 (canonicalize_iv_subregs): Likewise.
16070 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16071 it describes and change its mode to scalar_int_mode. Also change
16072 the type of the returned inner_mode to scalar_int_mode.
16073 (get_biv_step): Likewise, turning outer_mode from a pointer
16074 into a direct parameter. Update call to get_biv_step_1.
16075 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16076 iv_constant and get_biv_step.
16077 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16078 and change its type to scalar_int_mode. Don't initialise iv->mode
16079 to VOIDmode and remove later checks for its still being VOIDmode.
16080 Update calls to iv_analyze_op and iv_analyze_expr. Check
16081 is_a <scalar_int_mode> when changing the mode under consideration.
16082 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16083 Update call to iv_analyze_expr.
16084 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16085 inner register is not also a scalar_int_mode. Update call to
16087 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16088 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16089 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16090 separate mode class checks. Update calls to iv_analyze. Remove
16091 fix-up of VOIDmodes after iv_analyze_biv.
16092 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16093 don't have a scalar_int_mode. Update call to biv_p.
16095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16096 Alan Hayward <alan.hayward@arm.com>
16097 David Sherwood <david.sherwood@arm.com>
16099 * cfgexpand.c (convert_debug_memory_address): Use
16100 as_a <scalar_int_mode>.
16101 * combine.c (expand_compound_operation): Likewise.
16102 (make_extraction): Likewise.
16103 (change_zero_ext): Likewise.
16104 (simplify_comparison): Likewise.
16105 * cse.c (cse_insn): Likewise.
16106 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16107 (mem_loc_descriptor): Likewise.
16108 (loc_descriptor): Likewise.
16109 * expmed.c (init_expmed_one_mode): Likewise.
16110 (synth_mult): Likewise.
16111 (emit_store_flag_1): Likewise.
16112 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16113 of a comparison with size.
16114 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16115 (reduce_to_bit_field_precision): Likewise.
16116 * function.c (expand_function_end): Likewise.
16117 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16118 * loop-doloop.c (doloop_modify): Likewise.
16119 * optabs.c (expand_binop): Likewise.
16120 (expand_unop): Likewise.
16121 (expand_copysign_absneg): Likewise.
16122 (prepare_cmp_insn): Likewise.
16123 (maybe_legitimize_operand): Likewise.
16124 * recog.c (const_scalar_int_operand): Likewise.
16125 * rtlanal.c (get_address_mode): Likewise.
16126 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16127 (simplify_cond_clz_ctz): Likewise.
16128 * tree-nested.c (get_nl_goto_field): Likewise.
16129 * tree.c (build_vector_type_for_mode): Likewise.
16130 * var-tracking.c (use_narrower_mode): Likewise.
16132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16133 Alan Hayward <alan.hayward@arm.com>
16134 David Sherwood <david.sherwood@arm.com>
16136 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16137 * builtins.c (expand_builtin_signbit): Use it.
16138 * cfgexpand.c (expand_debug_expr): Likewise.
16139 * dojump.c (do_jump): Likewise.
16140 (do_compare_and_jump): Likewise.
16141 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16142 * expmed.c (make_tree): Likewise.
16143 * expr.c (expand_expr_real_2): Likewise.
16144 (expand_expr_real_1): Likewise.
16145 (try_casesi): Likewise.
16146 * fold-const-call.c (fold_const_call_ss): Likewise.
16147 * fold-const.c (unextend): Likewise.
16148 (extract_muldiv_1): Likewise.
16149 (fold_single_bit_test): Likewise.
16150 (native_encode_int): Likewise.
16151 (native_encode_string): Likewise.
16152 (native_interpret_int): Likewise.
16153 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16154 * internal-fn.c (expand_addsub_overflow): Likewise.
16155 (expand_neg_overflow): Likewise.
16156 (expand_mul_overflow): Likewise.
16157 (expand_arith_overflow): Likewise.
16158 * match.pd: Likewise.
16159 * stor-layout.c (layout_type): Likewise.
16160 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16161 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16162 * tree-ssanames.c (get_range_info): Likewise.
16163 * tree-switch-conversion.c (array_value_type) Likewise.
16164 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16165 (vect_recog_divmod_pattern): Likewise.
16166 (vect_recog_mixed_size_cond_pattern): Likewise.
16167 * tree-vrp.c (extract_range_basic): Likewise.
16168 (simplify_float_conversion_using_ranges): Likewise.
16169 * tree.c (int_fits_type_p): Likewise.
16170 * ubsan.c (instrument_bool_enum_load): Likewise.
16171 * varasm.c (mergeable_string_section): Likewise.
16172 (narrowing_initializer_constant_valid_p): Likewise.
16173 (output_constant): Likewise.
16175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16176 Alan Hayward <alan.hayward@arm.com>
16177 David Sherwood <david.sherwood@arm.com>
16179 * machmode.h (NARROWEST_INT_MODE): New macro.
16180 * expr.c (alignment_for_piecewise_move): Use it instead of
16181 GET_CLASS_NARROWEST_MODE (MODE_INT).
16182 (push_block): Likewise.
16183 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16185 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16188 Alan Hayward <alan.hayward@arm.com>
16189 David Sherwood <david.sherwood@arm.com>
16191 * postreload.c (move2add_valid_value_p): Change the type of the
16192 mode parameter to scalar_int_mode.
16193 (move2add_use_add2_insn): Add a mode parameter and use it instead
16195 (move2add_use_add3_insn): Likewise.
16196 (reload_cse_move2add): Update accordingly.
16198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16199 Alan Hayward <alan.hayward@arm.com>
16200 David Sherwood <david.sherwood@arm.com>
16202 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16204 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16205 * optabs.c (expand_unop): Likewise.
16207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16208 Alan Hayward <alan.hayward@arm.com>
16209 David Sherwood <david.sherwood@arm.com>
16211 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16212 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16213 (popcount_loc_descriptor): Likewise.
16214 (bswap_loc_descriptor): Likewise.
16215 (rotate_loc_descriptor): Likewise.
16216 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16217 calling the functions above.
16219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16220 Alan Hayward <alan.hayward@arm.com>
16221 David Sherwood <david.sherwood@arm.com>
16223 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16225 (try_combine): Likewise.
16226 (simplify_if_then_else): Likewise.
16227 * cse.c (cse_insn): Likewise.
16228 * dwarf2out.c (mem_loc_descriptor): Likewise.
16229 * emit-rtl.c (gen_lowpart_common): Likewise.
16230 * simplify-rtx.c (simplify_truncation): Likewise.
16231 (simplify_binary_operation_1): Likewise.
16232 (simplify_const_relational_operation): Likewise.
16233 (simplify_ternary_operation): Likewise.
16234 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16237 Alan Hayward <alan.hayward@arm.com>
16238 David Sherwood <david.sherwood@arm.com>
16240 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16241 * reload.c (push_reload): Likewise.
16242 (find_reloads): Likewise.
16244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16245 Alan Hayward <alan.hayward@arm.com>
16246 David Sherwood <david.sherwood@arm.com>
16248 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16249 (make_compound_operation_int): Likewise.
16250 (change_zero_ext): Likewise.
16251 * expr.c (convert_move): Likewise.
16252 (convert_modes): Likewise.
16253 * fwprop.c (forward_propagate_subreg): Likewise.
16254 * loop-iv.c (get_biv_step_1): Likewise.
16255 * optabs.c (widen_operand): Likewise.
16256 * postreload.c (move2add_valid_value_p): Likewise.
16257 * recog.c (simplify_while_replacing): Likewise.
16258 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16259 (simplify_binary_operation_1): Likewise. Remove redundant
16260 mode equality check.
16262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16263 Alan Hayward <alan.hayward@arm.com>
16264 David Sherwood <david.sherwood@arm.com>
16266 * combine.c (combine_simplify_rtx): Add checks for
16267 is_a <scalar_int_mode>.
16268 (simplify_if_then_else): Likewise.
16269 (make_field_assignment): Likewise.
16270 (simplify_comparison): Likewise.
16271 * ifcvt.c (noce_try_bitop): Likewise.
16272 * loop-invariant.c (canonicalize_address_mult): Likewise.
16273 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16276 Alan Hayward <alan.hayward@arm.com>
16277 David Sherwood <david.sherwood@arm.com>
16279 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16280 is_a <scalar_int_mode> instead of != BLKmode.
16282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16283 Alan Hayward <alan.hayward@arm.com>
16284 David Sherwood <david.sherwood@arm.com>
16286 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16287 instead of != VOIDmode.
16288 * combine.c (if_then_else_cond): Likewise.
16289 (change_zero_ext): Likewise.
16290 * dwarf2out.c (mem_loc_descriptor): Likewise.
16291 (loc_descriptor): Likewise.
16292 * rtlanal.c (canonicalize_condition): Likewise.
16293 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16296 Alan Hayward <alan.hayward@arm.com>
16297 David Sherwood <david.sherwood@arm.com>
16299 * simplify-rtx.c (simplify_binary_operation_1): Use
16300 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16303 Alan Hayward <alan.hayward@arm.com>
16304 David Sherwood <david.sherwood@arm.com>
16306 * wide-int.h (int_traits<unsigned char>) New class.
16307 (int_traits<unsigned short>) Likewise.
16308 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16309 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16311 * combine.c (set_nonzero_bits_and_sign_copies): Use
16312 is_a <scalar_int_mode>.
16313 (find_split_point): Likewise.
16314 (combine_simplify_rtx): Likewise.
16315 (simplify_logical): Likewise.
16316 (expand_compound_operation): Likewise.
16317 (expand_field_assignment): Likewise.
16318 (make_compound_operation): Likewise.
16319 (extended_count): Likewise.
16320 (change_zero_ext): Likewise.
16321 (simplify_comparison): Likewise.
16322 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16323 (ucompare_loc_descriptor): Likewise.
16324 (minmax_loc_descriptor): Likewise.
16325 (mem_loc_descriptor): Likewise.
16326 (loc_descriptor): Likewise.
16327 * expmed.c (init_expmed_one_mode): Likewise.
16328 * lra-constraints.c (lra_constraint_offset): Likewise.
16329 * optabs.c (prepare_libcall_arg): Likewise.
16330 * postreload.c (move2add_note_store): Likewise.
16331 * reload.c (operands_match_p): Likewise.
16332 * rtl.h (load_extend_op): Likewise.
16333 * rtlhooks.c (gen_lowpart_general): Likewise.
16334 * simplify-rtx.c (simplify_truncation): Likewise.
16335 (simplify_unary_operation_1): Likewise.
16336 (simplify_binary_operation_1): Likewise.
16337 (simplify_const_binary_operation): Likewise.
16338 (simplify_const_relational_operation): Likewise.
16339 (simplify_subreg): Likewise.
16340 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16341 * var-tracking.c (adjust_mems): Likewise.
16342 (prepare_call_arguments): Likewise.
16344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16345 Alan Hayward <alan.hayward@arm.com>
16346 David Sherwood <david.sherwood@arm.com>
16348 * machmode.h (is_int_mode): New fuction.
16349 * combine.c (find_split_point): Use it.
16350 (combine_simplify_rtx): Likewise.
16351 (simplify_if_then_else): Likewise.
16352 (simplify_set): Likewise.
16353 (simplify_shift_const_1): Likewise.
16354 (simplify_comparison): Likewise.
16355 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16356 * cse.c (notreg_cost): Likewise.
16357 (cse_insn): Likewise.
16358 * cselib.c (cselib_lookup_1): Likewise.
16359 * dojump.c (do_jump_1): Likewise.
16360 (do_compare_rtx_and_jump): Likewise.
16361 * dse.c (get_call_args): Likewise.
16362 * dwarf2out.c (rtl_for_decl_init): Likewise.
16363 (native_encode_initializer): Likewise.
16364 * expmed.c (emit_store_flag_1): Likewise.
16365 (emit_store_flag): Likewise.
16366 * expr.c (convert_modes): Likewise.
16367 (store_field): Likewise.
16368 (expand_expr_real_1): Likewise.
16369 * fold-const.c (fold_read_from_constant_string): Likewise.
16370 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16371 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16372 * optabs.c (expand_binop): Likewise.
16373 (expand_unop): Likewise.
16374 (expand_abs_nojump): Likewise.
16375 (expand_one_cmpl_abs_nojump): Likewise.
16376 * simplify-rtx.c (mode_signbit_p): Likewise.
16377 (val_signbit_p): Likewise.
16378 (val_signbit_known_set_p): Likewise.
16379 (val_signbit_known_clear_p): Likewise.
16380 (simplify_relational_operation_1): Likewise.
16381 * tree.c (vector_type_mode): Likewise.
16383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16384 Alan Hayward <alan.hayward@arm.com>
16385 David Sherwood <david.sherwood@arm.com>
16387 * machmode.h (smallest_mode_for_size): Fix formatting.
16388 (smallest_int_mode_for_size): New function.
16389 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16390 instead of smallest_mode_for_size.
16391 * combine.c (make_extraction): Likewise.
16392 * config/arc/arc.c (arc_expand_movmem): Likewise.
16393 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16394 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16395 * config/s390/s390.c (s390_expand_insv): Likewise.
16396 * config/sparc/sparc.c (assign_int_registers): Likewise.
16397 * config/spu/spu.c (spu_function_value): Likewise.
16398 (spu_function_arg): Likewise.
16399 * coverage.c (get_gcov_type): Likewise.
16400 (get_gcov_unsigned_t): Likewise.
16401 * dse.c (find_shift_sequence): Likewise.
16402 * expmed.c (store_bit_field_1): Likewise.
16403 * expr.c (convert_move): Likewise.
16404 (store_field): Likewise.
16405 * internal-fn.c (expand_arith_overflow): Likewise.
16406 * optabs-query.c (get_best_extraction_insn): Likewise.
16407 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16408 * stor-layout.c (layout_type): Likewise.
16409 (initialize_sizetypes): Likewise.
16410 * targhooks.c (default_get_mask_mode): Likewise.
16411 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16414 Alan Hayward <alan.hayward@arm.com>
16415 David Sherwood <david.sherwood@arm.com>
16417 * machmode.h (opt_mode::else_blk): New function.
16418 (int_mode_for_mode): Declare.
16419 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16420 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16422 * cfgexpand.c (expand_debug_expr): Likewise.
16423 * combine.c (gen_lowpart_or_truncate): Likewise.
16424 (gen_lowpart_for_combine): Likewise.
16425 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16426 * config/avr/avr.c (avr_to_int_mode): Likewise.
16427 (avr_out_plus_1): Likewise.
16428 (avr_out_plus): Likewise.
16429 (avr_out_round): Likewise.
16430 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16431 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16432 (s390_expand_vcond): Likewise.
16433 * config/spu/spu.c (spu_split_immediate): Likewise.
16434 (spu_expand_mov): Likewise.
16435 * dse.c (get_stored_val): Likewise.
16436 * expmed.c (store_bit_field_1): Likewise.
16437 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16439 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16440 (extract_low_bits): Likewise.
16441 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16442 handling rather than repeating the check.
16443 (emit_group_store): Likewise.
16444 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16445 * optabs.c (expand_absneg_bit): Likewise.
16446 (expand_copysign_absneg): Likewise.
16447 (expand_copysign_bit): Likewise.
16448 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16449 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16450 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16451 * var-tracking.c (prepare_call_arguments): Likewise.
16452 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16453 int_mode_for_mode instead of mode_for_size.
16454 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16457 Alan Hayward <alan.hayward@arm.com>
16458 David Sherwood <david.sherwood@arm.com>
16460 * machmode.h (int_mode_for_size): New function.
16461 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16462 instead of mode_for_size.
16463 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16465 * combine.c (expand_field_assignment): Use int_mode_for_size
16466 instead of mode_for_size.
16467 (make_extraction): Likewise.
16468 (simplify_shift_const_1): Likewise.
16469 (simplify_comparison): Likewise.
16470 * dojump.c (do_jump): Likewise.
16471 * dwarf2out.c (mem_loc_descriptor): Likewise.
16472 * emit-rtl.c (init_derived_machine_modes): Likewise.
16473 * expmed.c (flip_storage_order): Likewise.
16474 (convert_extracted_bit_field): Likewise.
16475 * expr.c (copy_blkmode_from_reg): Likewise.
16476 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16477 * internal-fn.c (expand_mul_overflow): Likewise.
16478 * lower-subreg.c (simple_move): Likewise.
16479 * optabs-libfuncs.c (init_optabs): Likewise.
16480 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16481 * tree.c (vector_type_mode): Likewise.
16482 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16483 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16484 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16485 * tree-vect-stmts.c (vectorizable_load): Likewise.
16486 (vectorizable_store): Likewise.
16488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16489 Alan Hayward <alan.hayward@arm.com>
16490 David Sherwood <david.sherwood@arm.com>
16492 * coretypes.h (pod_mode): New type.
16493 (scalar_int_mode_pod): New typedef.
16494 * machmode.h (pod_mode): New class.
16495 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16496 * genmodes.c (emit_mode_int_n): Update accordingly.
16497 * lower-subreg.h (target_lower_subreg): Change type to
16498 scalar_int_mode_pod.
16499 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16500 scalar_int_mode_pod.
16502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16503 Alan Hayward <alan.hayward@arm.com>
16504 David Sherwood <david.sherwood@arm.com>
16506 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16507 machine_mode to scalar_int_mode.
16508 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16509 (rs6000_option_override_internal): Remove cast to int.
16510 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16511 machine_mode to scalar_int_mode.
16512 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16513 (rs6000_option_override_internal): Remove cast to int.
16514 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16515 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16517 * config/s390/s390.c (s390_expand_builtin): Likewise.
16518 * coretypes.h (scalar_int_mode): New type.
16519 (opt_scalar_int_mode): New typedef.
16520 * machmode.h (scalar_int_mode): New class.
16521 (scalar_int_mode::includes_p): New function.
16522 (byte_mode): Change type to scalar_int_mode.
16523 (word_mode): Likewise.
16524 (ptr_mode): Likewise.
16525 * emit-rtl.c (byte_mode): Likewise.
16526 (word_mode): Likewise.
16527 (ptr_mode): Likewise.
16528 (init_derived_machine_modes): Update accordingly.
16529 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16530 and MODE_PARTIAL_INT.
16531 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16532 opt_scalar_int_mode.
16534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16535 Alan Hayward <alan.hayward@arm.com>
16536 David Sherwood <david.sherwood@arm.com>
16538 * target.def (libgcc_floating_mode_supported_p): Take a
16540 * doc/tm.texi: Regenerate.
16541 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16543 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16544 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16548 Alan Hayward <alan.hayward@arm.com>
16549 David Sherwood <david.sherwood@arm.com>
16551 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16552 * doc/tm.texi: Regenerate.
16553 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16554 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16555 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16556 * targhooks.h (default_floatn_mode): Likewise.
16557 * targhooks.c (default_floatn_mode): Likewise.
16558 * tree.c (build_common_tree_nodes): Update accordingly.
16560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16561 Alan Hayward <alan.hayward@arm.com>
16562 David Sherwood <david.sherwood@arm.com>
16564 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16565 (mode_iterator::iterate_p): Likewise.
16566 (mode_iterator::get_wider): Likewise.
16567 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16570 Alan Hayward <alan.hayward@arm.com>
16571 David Sherwood <david.sherwood@arm.com>
16573 * coretypes.h (opt_scalar_float_mode): New typedef.
16574 * machmode.h (float_mode_for_size): New function.
16575 * emit-rtl.c (double_mode): Delete.
16576 (init_emit_once): Use float_mode_for_size.
16577 * stor-layout.c (layout_type): Likewise.
16578 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16581 Alan Hayward <alan.hayward@arm.com>
16582 David Sherwood <david.sherwood@arm.com>
16584 * output.h (assemble_real): Take a scalar_float_mode.
16585 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16586 * config/arm/arm.md (consttable_4): Likewise.
16587 (consttable_8): Likewise.
16588 (consttable_16): Likewise.
16589 * config/mips/mips.md (consttable_float): Likewise.
16590 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16591 * varasm.c (assemble_real): Take a scalar_float_mode.
16592 (output_constant_pool_2): Update accordingly.
16593 (output_constant): Likewise.
16595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16596 Alan Hayward <alan.hayward@arm.com>
16597 David Sherwood <david.sherwood@arm.com>
16599 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16600 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16601 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16602 (native_encode_real): Likewise.
16603 (native_interpret_real): Likewise.
16604 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16605 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16608 Alan Hayward <alan.hayward@arm.com>
16609 David Sherwood <david.sherwood@arm.com>
16611 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16612 <scalar_float_mode>. Simplify.
16613 (gen_extend_conv_libfunc): Likewise.
16615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16616 Alan Hayward <alan.hayward@arm.com>
16617 David Sherwood <david.sherwood@arm.com>
16619 * coretypes.h (scalar_float_mode): New type.
16620 * machmode.h (mode_traits::from_int): Use machine_mode if
16621 USE_ENUM_MODES is defined.
16622 (is_a): New function.
16624 (dyn_cast): Likewise.
16625 (scalar_float_mode): New class.
16626 (scalar_float_mode::includes_p): New function.
16627 (is_float_mode): Likewise.
16628 * gdbhooks.py (MachineModePrinter): New class.
16629 (build_pretty_printer): Use it for scalar_float_mode.
16630 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16631 (format_helper::format_helper): Turn into a template.
16632 * genmodes.c (get_mode_class): New function.
16633 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16634 or machine_mode if none.
16635 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16636 as_a <scalar_float_mode>.
16637 * dwarf2out.c (mem_loc_descriptor): Likewise.
16638 (insert_float): Likewise.
16639 (add_const_value_attribute): Likewise.
16640 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16641 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16642 (expand_unop): Update accordingly.
16643 (expand_abs_nojump): Likewise.
16644 (expand_copysign_absneg): Take a scalar_float_mode.
16645 (expand_copysign_bit): Likewise.
16646 (expand_copysign): Update accordingly.
16648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16649 Alan Hayward <alan.hayward@arm.com>
16650 David Sherwood <david.sherwood@arm.com>
16652 * coretypes.h (opt_mode): New class.
16653 * machmode.h (opt_mode): Likewise.
16654 (opt_mode::else_void): New function.
16655 (opt_mode::require): Likewise.
16656 (opt_mode::exists): Likewise.
16657 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16658 (GET_MODE_2XWIDER_MODE): Likewise.
16659 (mode_iterator::get_wider): Update accordingly.
16660 (mode_iterator::get_2xwider): Likewise.
16661 (mode_iterator::get_known_wider): Likewise, turning into a template.
16662 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16663 forcing a wider mode to exist.
16664 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16665 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16666 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16667 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16668 * lower-subreg.c (init_lower_subreg): Likewise.
16669 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16670 on the final iteration.
16671 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16672 a wider mode exists before asking for a move pattern.
16673 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16674 forcing a wider mode to exist.
16675 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16676 returning false if no such mode exists.
16677 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16678 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16679 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16680 Avoid checking for a MODE_INT if we already know the mode is not a
16682 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16683 forcing a wider mode to exist.
16684 (expmed_mult_highpart_optab): Likewise.
16685 (expmed_mult_highpart): Likewise.
16686 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16688 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16689 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16690 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16691 * internal-fn.c (expand_mul_overflow): Update use of
16692 GET_MODE_2XWIDER_MODE.
16693 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16694 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16695 GET_MODE_WIDER_MODE.
16696 (convert_plusminus_to_widen): Likewise.
16697 * tree-switch-conversion.c (array_value_type): Likewise.
16698 * var-tracking.c (emit_note_insn_var_location): Likewise.
16699 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16700 Return false inside rather than outside the loop if no wider mode
16702 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16703 and GET_MODE_2XWIDER_MODE
16704 (can_compare_p): Use else_void.
16705 * gdbhooks.py (OptMachineModePrinter): New class.
16706 (build_pretty_printer): Use it for opt_mode.
16708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16709 Alan Hayward <alan.hayward@arm.com>
16710 David Sherwood <david.sherwood@arm.com>
16712 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16713 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16716 Alan Hayward <alan.hayward@arm.com>
16717 David Sherwood <david.sherwood@arm.com>
16719 * machmode.h (mode_traits): New structure.
16720 (get_narrowest_mode): New function.
16721 (mode_iterator::start): Likewise.
16722 (mode_iterator::iterate_p): Likewise.
16723 (mode_iterator::get_wider): Likewise.
16724 (mode_iterator::get_known_wider): Likewise.
16725 (mode_iterator::get_2xwider): Likewise.
16726 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16727 (FOR_EACH_MODE): Likewise.
16728 (FOR_EACH_MODE_FROM): Likewise.
16729 (FOR_EACH_MODE_UNTIL): Likewise.
16730 (FOR_EACH_WIDER_MODE): Likewise.
16731 (FOR_EACH_2XWIDER_MODE): Likewise.
16732 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16733 * combine.c (simplify_comparison): Likewise
16734 * config/i386/i386.c (type_natural_mode): Likewise.
16735 * cse.c (cse_insn): Likewise.
16736 * dse.c (find_shift_sequence): Likewise.
16737 * emit-rtl.c (init_derived_machine_modes): Likewise.
16738 (init_emit_once): Likewise.
16739 * explow.c (hard_function_value): Likewise.
16740 * expmed.c (extract_fixed_bit_field_1): Likewise.
16741 (extract_bit_field_1): Likewise.
16742 (expand_divmod): Likewise.
16743 (emit_store_flag_1): Likewise.
16744 * expr.c (init_expr_target): Likewise.
16745 (convert_move): Likewise.
16746 (alignment_for_piecewise_move): Likewise.
16747 (widest_int_mode_for_size): Likewise.
16748 (emit_block_move_via_movmem): Likewise.
16749 (copy_blkmode_to_reg): Likewise.
16750 (set_storage_via_setmem): Likewise.
16751 (compress_float_constant): Likewise.
16752 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16753 * optabs-query.c (get_best_extraction_insn): Likewise.
16754 * optabs.c (expand_binop): Likewise.
16755 (expand_twoval_unop): Likewise.
16756 (expand_twoval_binop): Likewise.
16757 (widen_leading): Likewise.
16758 (widen_bswap): Likewise.
16759 (expand_parity): Likewise.
16760 (expand_unop): Likewise.
16761 (prepare_cmp_insn): Likewise.
16762 (prepare_float_lib_cmp): Likewise.
16763 (expand_float): Likewise.
16764 (expand_fix): Likewise.
16765 (expand_sfix_optab): Likewise.
16766 * postreload.c (move2add_use_add2_insn): Likewise.
16767 * reg-stack.c (reg_to_stack): Likewise.
16768 * reginfo.c (choose_hard_reg_mode): Likewise.
16769 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16770 * stor-layout.c (mode_for_size): Likewise.
16771 (smallest_mode_for_size): Likewise.
16772 (mode_for_vector): Likewise.
16773 (finish_bitfield_representative): Likewise.
16774 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16775 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16776 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16777 * var-tracking.c (prepare_call_arguments): Likewise.
16779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16780 Alan Hayward <alan.hayward@arm.com>
16781 David Sherwood <david.sherwood@arm.com>
16783 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16784 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16785 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16786 * machmode.h (mode_size): Move earlier in file.
16787 (mode_precision): Likewise.
16788 (mode_inner): Likewise.
16789 (mode_nunits): Likewise.
16790 (mode_unit_size): Likewise.
16791 (unit_unit_precision): Likewise.
16792 (mode_wider): Likewise.
16793 (mode_2xwider): Likewise.
16794 (machine_mode): New class.
16795 (mode_to_bytes): New function.
16796 (mode_to_bits): Likewise.
16797 (mode_to_precision): Likewise.
16798 (mode_to_inner): Likewise.
16799 (mode_to_unit_size): Likewise.
16800 (mode_to_unit_precision): Likewise.
16801 (mode_to_nunits): Likewise.
16802 (GET_MODE_SIZE): Use mode_to_bytes.
16803 (GET_MODE_BITSIZE): Use mode_to_bits.
16804 (GET_MODE_PRECISION): Use mode_to_precision.
16805 (GET_MODE_INNER): Use mode_to_inner.
16806 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16807 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16808 (GET_MODE_NUNITS): Use mode_to_nunits.
16809 * system.h (ALWAYS_INLINE): New macro.
16810 * config/powerpcspe/powerpcspe-c.c
16811 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16812 int for arg1_mode and arg2_mode.
16814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16815 Alan Hayward <alan.hayward@arm.com>
16816 David Sherwood <david.sherwood@arm.com>
16818 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16819 Prefix mode names with E_ in case statements.
16820 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16821 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16822 (aarch64_split_simd_move): Likewise.
16823 (aarch64_gen_storewb_pair): Likewise.
16824 (aarch64_gen_loadwb_pair): Likewise.
16825 (aarch64_gen_store_pair): Likewise.
16826 (aarch64_gen_load_pair): Likewise.
16827 (aarch64_get_condition_code_1): Likewise.
16828 (aarch64_constant_pool_reload_icode): Likewise.
16829 (get_rsqrte_type): Likewise.
16830 (get_rsqrts_type): Likewise.
16831 (get_recpe_type): Likewise.
16832 (get_recps_type): Likewise.
16833 (aarch64_gimplify_va_arg_expr): Likewise.
16834 (aarch64_simd_container_mode): Likewise.
16835 (aarch64_emit_load_exclusive): Likewise.
16836 (aarch64_emit_store_exclusive): Likewise.
16837 (aarch64_expand_compare_and_swap): Likewise.
16838 (aarch64_gen_atomic_cas): Likewise.
16839 (aarch64_emit_bic): Likewise.
16840 (aarch64_emit_atomic_swap): Likewise.
16841 (aarch64_emit_atomic_load_op): Likewise.
16842 (aarch64_evpc_trn): Likewise.
16843 (aarch64_evpc_uzp): Likewise.
16844 (aarch64_evpc_zip): Likewise.
16845 (aarch64_evpc_ext): Likewise.
16846 (aarch64_evpc_rev): Likewise.
16847 (aarch64_evpc_dup): Likewise.
16848 (aarch64_gen_ccmp_first): Likewise.
16849 (aarch64_gen_ccmp_next): Likewise.
16850 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16851 (alpha_emit_xfloating_libcall): Likewise.
16852 (emit_insxl): Likewise.
16853 (alpha_arg_type): Likewise.
16854 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16855 (arc_preferred_simd_mode): Likewise.
16856 (arc_secondary_reload): Likewise.
16857 (get_arc_condition_code): Likewise.
16858 (arc_print_operand): Likewise.
16859 (arc_legitimate_constant_p): Likewise.
16860 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16861 * config/arc/arc.md (casesi_load): Likewise.
16862 (casesi_compact_jump): Likewise.
16863 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16864 (cc_use_register): Likewise.
16865 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16866 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16867 (arm_init_iwmmxt_builtins): Likewise.
16868 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16869 (neon_expand_vector_init): Likewise.
16870 (arm_attr_length_move_neon): Likewise.
16871 (maybe_get_arm_condition_code): Likewise.
16872 (arm_emit_vector_const): Likewise.
16873 (arm_preferred_simd_mode): Likewise.
16874 (arm_output_iwmmxt_tinsr): Likewise.
16875 (thumb1_output_casesi): Likewise.
16876 (thumb2_output_casesi): Likewise.
16877 (arm_emit_load_exclusive): Likewise.
16878 (arm_emit_store_exclusive): Likewise.
16879 (arm_expand_compare_and_swap): Likewise.
16880 (arm_evpc_neon_vuzp): Likewise.
16881 (arm_evpc_neon_vzip): Likewise.
16882 (arm_evpc_neon_vrev): Likewise.
16883 (arm_evpc_neon_vtrn): Likewise.
16884 (arm_evpc_neon_vext): Likewise.
16885 (arm_validize_comparison): Likewise.
16886 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16887 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16888 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16889 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16890 (c6x_preferred_simd_mode): Likewise.
16891 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16892 (epiphany_rtx_costs): Likewise.
16893 * config/epiphany/predicates.md (proper_comparison_operator):
16895 * config/frv/frv.c (condexec_memory_operand): Likewise.
16896 (frv_emit_move): Likewise.
16897 (output_move_single): Likewise.
16898 (output_condmove_single): Likewise.
16899 (frv_hard_regno_mode_ok): Likewise.
16900 (frv_matching_accg_mode): Likewise.
16901 * config/h8300/h8300.c (split_adds_subs): Likewise.
16902 (h8300_rtx_costs): Likewise.
16903 (h8300_print_operand): Likewise.
16904 (compute_mov_length): Likewise.
16905 (output_logical_op): Likewise.
16906 (compute_logical_op_length): Likewise.
16907 (compute_logical_op_cc): Likewise.
16908 (h8300_shift_needs_scratch_p): Likewise.
16909 (output_a_shift): Likewise.
16910 (compute_a_shift_length): Likewise.
16911 (compute_a_shift_cc): Likewise.
16912 (expand_a_rotate): Likewise.
16913 (output_a_rotate): Likewise.
16914 * config/i386/i386.c (classify_argument): Likewise.
16915 (function_arg_advance_32): Likewise.
16916 (function_arg_32): Likewise.
16917 (function_arg_64): Likewise.
16918 (function_value_64): Likewise.
16919 (ix86_gimplify_va_arg): Likewise.
16920 (ix86_legitimate_constant_p): Likewise.
16921 (put_condition_code): Likewise.
16922 (split_double_mode): Likewise.
16923 (ix86_avx256_split_vector_move_misalign): Likewise.
16924 (ix86_expand_vector_logical_operator): Likewise.
16925 (ix86_split_idivmod): Likewise.
16926 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16927 (ix86_build_const_vector): Likewise.
16928 (ix86_build_signbit_mask): Likewise.
16929 (ix86_match_ccmode): Likewise.
16930 (ix86_cc_modes_compatible): Likewise.
16931 (ix86_expand_branch): Likewise.
16932 (ix86_expand_sse_cmp): Likewise.
16933 (ix86_expand_sse_movcc): Likewise.
16934 (ix86_expand_int_sse_cmp): Likewise.
16935 (ix86_expand_vec_perm_vpermi2): Likewise.
16936 (ix86_expand_vec_perm): Likewise.
16937 (ix86_expand_sse_unpack): Likewise.
16938 (ix86_expand_int_addcc): Likewise.
16939 (ix86_split_to_parts): Likewise.
16940 (ix86_vectorize_builtin_gather): Likewise.
16941 (ix86_vectorize_builtin_scatter): Likewise.
16942 (avx_vpermilp_parallel): Likewise.
16943 (inline_memory_move_cost): Likewise.
16944 (ix86_tieable_integer_mode_p): Likewise.
16945 (x86_maybe_negate_const_int): Likewise.
16946 (ix86_expand_vector_init_duplicate): Likewise.
16947 (ix86_expand_vector_init_one_nonzero): Likewise.
16948 (ix86_expand_vector_init_one_var): Likewise.
16949 (ix86_expand_vector_init_concat): Likewise.
16950 (ix86_expand_vector_init_interleave): Likewise.
16951 (ix86_expand_vector_init_general): Likewise.
16952 (ix86_expand_vector_set): Likewise.
16953 (ix86_expand_vector_extract): Likewise.
16954 (emit_reduc_half): Likewise.
16955 (ix86_emit_i387_round): Likewise.
16956 (ix86_mangle_type): Likewise.
16957 (ix86_expand_round_sse4): Likewise.
16958 (expand_vec_perm_blend): Likewise.
16959 (canonicalize_vector_int_perm): Likewise.
16960 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16961 (expand_vec_perm_1): Likewise.
16962 (expand_vec_perm_interleave3): Likewise.
16963 (expand_vec_perm_even_odd_pack): Likewise.
16964 (expand_vec_perm_even_odd_1): Likewise.
16965 (expand_vec_perm_broadcast_1): Likewise.
16966 (ix86_vectorize_vec_perm_const_ok): Likewise.
16967 (ix86_expand_vecop_qihi): Likewise.
16968 (ix86_expand_mul_widen_hilo): Likewise.
16969 (ix86_expand_sse2_abs): Likewise.
16970 (ix86_expand_pextr): Likewise.
16971 (ix86_expand_pinsr): Likewise.
16972 (ix86_preferred_simd_mode): Likewise.
16973 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16974 * config/i386/sse.md (*andnot<mode>3): Likewise.
16975 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16976 (*<code><mode>3): Likewise.
16977 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16978 (ia64_expand_atomic_op): Likewise.
16979 (ia64_arg_type): Likewise.
16980 (ia64_mode_to_int): Likewise.
16981 (ia64_scalar_mode_supported_p): Likewise.
16982 (ia64_vector_mode_supported_p): Likewise.
16983 (expand_vec_perm_broadcast): Likewise.
16984 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16985 (iq2000_function_arg_advance): Likewise.
16986 (iq2000_function_arg): Likewise.
16987 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16988 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16989 (m68k_libcall_value): Likewise.
16990 (m68k_function_value): Likewise.
16991 (sched_attr_op_type): Likewise.
16992 * config/mcore/mcore.c (mcore_output_move): Likewise.
16993 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16995 (microblaze_function_arg): Likewise.
16996 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16997 (mips_print_operand): Likewise.
16998 (mips_mode_ok_for_mov_fmt_p): Likewise.
16999 (mips_vector_mode_supported_p): Likewise.
17000 (mips_preferred_simd_mode): Likewise.
17001 (mips_expand_vpc_loongson_even_odd): Likewise.
17002 (mips_expand_vec_unpack): Likewise.
17003 (mips_expand_vi_broadcast): Likewise.
17004 (mips_expand_vector_init): Likewise.
17005 (mips_expand_vec_reduc): Likewise.
17006 (mips_expand_msa_cmp): Likewise.
17007 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17008 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17009 (cc_flags_for_mode): Likewise.
17010 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17011 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17012 (nds32_output_casesi_pc_relative): Likewise.
17013 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17014 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17015 (nvptx_gen_unpack): Likewise.
17016 (nvptx_gen_pack): Likewise.
17017 (nvptx_gen_shuffle): Likewise.
17018 (nvptx_gen_wcast): Likewise.
17019 (nvptx_preferred_simd_mode): Likewise.
17020 * config/pa/pa.c (pa_secondary_reload): Likewise.
17021 * config/pa/predicates.md (base14_operand): Likewise.
17022 * config/powerpcspe/powerpcspe-c.c
17023 (altivec_resolve_overloaded_builtin): Likewise.
17024 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17026 (rs6000_preferred_simd_mode): Likewise.
17027 (output_vec_const_move): Likewise.
17028 (rs6000_expand_vector_extract): Likewise.
17029 (rs6000_split_vec_extract_var): Likewise.
17030 (reg_offset_addressing_ok_p): Likewise.
17031 (rs6000_legitimate_offset_address_p): Likewise.
17032 (rs6000_legitimize_address): Likewise.
17033 (rs6000_emit_set_const): Likewise.
17034 (rs6000_const_vec): Likewise.
17035 (rs6000_emit_move): Likewise.
17036 (spe_build_register_parallel): Likewise.
17037 (rs6000_darwin64_record_arg_recurse): Likewise.
17038 (swap_selector_for_mode): Likewise.
17039 (spe_init_builtins): Likewise.
17040 (paired_init_builtins): Likewise.
17041 (altivec_init_builtins): Likewise.
17042 (do_load_for_compare): Likewise.
17043 (rs6000_generate_compare): Likewise.
17044 (rs6000_expand_float128_convert): Likewise.
17045 (emit_load_locked): Likewise.
17046 (emit_store_conditional): Likewise.
17047 (rs6000_output_function_epilogue): Likewise.
17048 (rs6000_handle_altivec_attribute): Likewise.
17049 (rs6000_function_value): Likewise.
17050 (emit_fusion_gpr_load): Likewise.
17051 (emit_fusion_p9_load): Likewise.
17052 (emit_fusion_p9_store): Likewise.
17053 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17054 (fusion_gpr_mem_load): Likewise.
17055 (fusion_addis_mem_combo_load): Likewise.
17056 (fusion_addis_mem_combo_store): Likewise.
17057 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17058 (fusion_gpr_mem_load): Likewise.
17059 (fusion_addis_mem_combo_load): Likewise.
17060 (fusion_addis_mem_combo_store): Likewise.
17061 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17063 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17064 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17065 (rs6000_preferred_simd_mode): Likewise.
17066 (output_vec_const_move): Likewise.
17067 (rs6000_expand_vector_extract): Likewise.
17068 (rs6000_split_vec_extract_var): Likewise.
17069 (reg_offset_addressing_ok_p): Likewise.
17070 (rs6000_legitimate_offset_address_p): Likewise.
17071 (rs6000_legitimize_address): Likewise.
17072 (rs6000_emit_set_const): Likewise.
17073 (rs6000_const_vec): Likewise.
17074 (rs6000_emit_move): Likewise.
17075 (rs6000_darwin64_record_arg_recurse): Likewise.
17076 (swap_selector_for_mode): Likewise.
17077 (paired_init_builtins): Likewise.
17078 (altivec_init_builtins): Likewise.
17079 (rs6000_expand_float128_convert): Likewise.
17080 (emit_load_locked): Likewise.
17081 (emit_store_conditional): Likewise.
17082 (rs6000_output_function_epilogue): Likewise.
17083 (rs6000_handle_altivec_attribute): Likewise.
17084 (rs6000_function_value): Likewise.
17085 (emit_fusion_gpr_load): Likewise.
17086 (emit_fusion_p9_load): Likewise.
17087 (emit_fusion_p9_store): Likewise.
17088 * config/rx/rx.c (rx_gen_move_template): Likewise.
17089 (flags_from_mode): Likewise.
17090 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17091 (s390_slb_comparison): Likewise.
17092 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17093 (s390_vector_mode_supported_p): Likewise.
17094 (s390_cc_modes_compatible): Likewise.
17095 (s390_match_ccmode_set): Likewise.
17096 (s390_canonicalize_comparison): Likewise.
17097 (s390_emit_compare_and_swap): Likewise.
17098 (s390_branch_condition_mask): Likewise.
17099 (s390_rtx_costs): Likewise.
17100 (s390_secondary_reload): Likewise.
17101 (__SECONDARY_RELOAD_CASE): Likewise.
17102 (s390_expand_cs): Likewise.
17103 (s390_preferred_simd_mode): Likewise.
17104 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17105 * config/sh/sh.c (sh_print_operand): Likewise.
17106 (dump_table): Likewise.
17107 (sh_secondary_reload): Likewise.
17108 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17109 * config/sh/sh.md (casesi_worker_1): Likewise.
17110 (casesi_worker_2): Likewise.
17111 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17112 (fcc_comparison_operator): Likewise.
17113 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17114 (emit_soft_tfmode_cvt): Likewise.
17115 (sparc_preferred_simd_mode): Likewise.
17116 (output_cbranch): Likewise.
17117 (sparc_print_operand): Likewise.
17118 (sparc_expand_vec_perm_bmask): Likewise.
17119 (vector_init_bshuffle): Likewise.
17120 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17121 (spu_vector_mode_supported_p): Likewise.
17122 (spu_expand_insv): Likewise.
17123 (spu_emit_branch_or_set): Likewise.
17124 (spu_handle_vector_attribute): Likewise.
17125 (spu_builtin_splats): Likewise.
17126 (spu_builtin_extract): Likewise.
17127 (spu_builtin_promote): Likewise.
17128 (spu_expand_sign_extend): Likewise.
17129 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17130 (tilegx_simd_int): Likewise.
17131 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17132 (tilepro_simd_int): Likewise.
17133 * config/v850/v850.c (const_double_split): Likewise.
17134 (v850_print_operand): Likewise.
17135 (ep_memory_offset): Likewise.
17136 * config/vax/vax.c (vax_rtx_costs): Likewise.
17137 (vax_output_int_move): Likewise.
17138 (vax_output_int_add): Likewise.
17139 (vax_output_int_subtract): Likewise.
17140 * config/visium/predicates.md (visium_branch_operator): Likewise.
17141 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17142 (visium_print_operand_address): Likewise.
17143 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17144 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17145 (xtensa_expand_conditional_branch): Likewise.
17146 (xtensa_copy_incoming_a7): Likewise.
17147 (xtensa_output_literal): Likewise.
17148 * dfp.c (decimal_real_maxval): Likewise.
17149 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17152 Alan Hayward <alan.hayward@arm.com>
17153 David Sherwood <david.sherwood@arm.com>
17155 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17156 (mode_nunits_inline): Likewise.
17157 (mode_inner_inline): Likewise.
17158 (mode_unit_size_inline): Likewise.
17159 (mode_unit_precision_inline): Likewise.
17160 (emit_insn_modes_h): Likewise. Also emit a #define of the
17162 (emit_mode_wider): Add an E_ prefix to mode names.
17163 (emit_mode_complex): Likewise.
17164 (emit_mode_inner): Likewise.
17165 (emit_mode_adjustments): Likewise.
17166 (emit_mode_int_n): Likewise.
17167 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17168 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17169 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17170 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17171 (CRC32_BUILTIN, ENTRY): Likewise.
17172 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17173 (aarch64_pop_regs): Likewise.
17174 (aarch64_process_components): Likewise.
17175 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17176 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17177 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17178 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17179 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17180 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17182 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17183 (pcmpistr): Likewise.
17184 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17185 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17186 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17188 * config/rl78/rl78.c (mduc_regs): Likewise.
17189 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17190 (htm_expand_builtin): Likewise.
17191 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17192 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17193 * config/xtensa/xtensa.c (print_operand): Likewise.
17194 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17195 (NUM_MODE_VECTOR_INT): Likewise.
17196 * genoutput.c (null_operand): Likewise.
17197 (output_operand_data): Likewise.
17198 * genrecog.c (print_parameter_value): Likewise.
17199 * lra.c (debug_operand_data): Likewise.
17201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17203 * dwarf2out.c (scompare_loc_descriptor_wide)
17204 (scompare_loc_descriptor_narrow): New functions, split out from...
17205 (scompare_loc_descriptor): ...here.
17206 * expmed.c (emit_store_flag_int): New function, split out from...
17207 (emit_store_flag): ...here.
17209 2017-08-30 Richard Biener <rguenther@suse.de>
17211 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17212 (dwarf2out_early_finish): Move setting of AT_pubnames from
17213 early debug output to early finish.
17215 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17217 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17218 and -mdata-region to the assembler.
17220 2017-08-30 Richard Biener <rguenther@suse.de>
17222 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17224 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17225 (dwarf2out_early_global_decl): Only generate a DIE for the
17226 abstract origin if it doesn't already exist or is a declaration DIE.
17227 (resolve_addr): Do not add the linkage name twice when
17228 generating a stub DIE for the DW_TAG_GNU_call_site target.
17230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17233 Use machine_mode rather than int for arg1_mode.
17235 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17238 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17239 that the second argument of the built-in functions to unpack
17240 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17241 switch statement instead a lot of if statements.
17242 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17243 Allow 64-bit values to be in Altivec registers as well as
17244 traditional floating point registers.
17245 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17247 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17249 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17250 MAX_REGS_PER_ADDRESS == 1.
17252 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17254 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17255 * config/i386/i386.c (ix86_option_override_internal): Simplify
17256 setting of opts->x_flag_entry.
17258 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17259 Jakub Jelinek <jakub@redhat.com>
17260 Richard Biener <rguenther@suse.de>
17262 PR tree-optimization/81503
17263 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17264 folded constant fits in the target type; reorder tests for clarity.
17266 2017-08-29 Martin Liska <mliska@suse.cz>
17268 * passes.def: Include pass_lower_switch.
17269 * stmt.c (dump_case_nodes): Remove and move to
17270 tree-switch-conversion.
17271 (case_values_threshold): Likewise.
17272 (expand_switch_as_decision_tree_p): Likewise.
17273 (emit_case_decision_tree): Likewise.
17274 (expand_case): Likewise.
17275 (balance_case_nodes): Likewise.
17276 (node_has_low_bound): Likewise.
17277 (node_has_high_bound): Likewise.
17278 (node_is_bounded): Likewise.
17279 (emit_case_nodes): Likewise.
17280 (struct simple_case_node): New struct.
17281 (add_case_node): Remove.
17282 (emit_case_dispatch_table): Use vector instead of case_list.
17283 (reset_out_edges_aux): Remove.
17284 (compute_cases_per_edge): Likewise.
17285 (expand_case): Build list of simple_case_node.
17286 (expand_sjlj_dispatch_table): Use it.
17287 * tree-switch-conversion.c (struct case_node): Moved from
17288 stmt.c and adjusted.
17289 (emit_case_nodes): Likewise.
17290 (node_has_low_bound): Likewise.
17291 (node_has_high_bound): Likewise.
17292 (node_is_bounded): Likewise.
17293 (case_values_threshold): Likewise.
17294 (reset_out_edges_aux): Likewise.
17295 (compute_cases_per_edge): Likewise.
17296 (add_case_node): Likewise.
17297 (dump_case_nodes): Likewise.
17298 (balance_case_nodes): Likewise.
17299 (expand_switch_as_decision_tree_p): Likewise.
17300 (emit_jump): Likewise.
17301 (emit_case_decision_tree): Likewise.
17302 (try_switch_expansion): Likewise.
17303 (do_jump_if_equal): Likewise.
17304 (emit_cmp_and_jump_insns): Likewise.
17305 (fix_phi_operands_for_edge): New function.
17306 (record_phi_operand_mapping): Likewise.
17307 (class pass_lower_switch): New pass.
17308 (pass_lower_switch::execute): New function.
17309 (make_pass_lower_switch): Likewise.
17310 (conditional_probability):
17311 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17312 * tree-pass.h: Add make_pass_lower_switch.
17314 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17317 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17320 2017-08-29 Richard Biener <rguenther@suse.de>
17322 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17323 we do not add a DW_AT_inline attribute twice.
17324 (gen_subprogram_die): Remove code setting DW_AT_inline on
17325 DECL_ABSTRACT_P nodes.
17327 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17329 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17330 calls to internal functions.
17331 (gimplify_modify_expr): Likewise.
17332 * tree-call-cdce.c (use_internal_fn): Likewise.
17333 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17334 (convert_to_divmod): Set the nothrow flag.
17335 * tree-if-conv.c (predicate_mem_writes): Likewise.
17336 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17337 (vectorizable_call): Likewise.
17338 (vectorizable_store): Likewise.
17339 (vectorizable_load): Likewise.
17340 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17341 (vect_recog_mask_conversion_pattern): Likewise.
17343 2017-08-29 Martin Liska <mliska@suse.cz>
17346 * gcc.c (driver_handle_option): Add new argument.
17347 * opts-common.c (handle_option): Pass
17348 target_option_override_hook.
17349 * opts-global.c (lang_handle_option): Add new option.
17350 (set_default_handlers): Add new argument.
17351 (decode_options): Likewise.
17352 * opts.c (target_handle_option): Likewise.
17353 (common_handle_option): Call target_option_override_hook.
17354 * opts.h (struct cl_option_handler_func): Add hook for
17355 target option override.
17356 (struct cl_option_handlers): Likewise.
17357 (set_default_handlers): Add new argument.
17358 (decode_options): Likewise.
17359 (common_handle_option): Likewise.
17360 (target_handle_option): Likewise.
17361 * toplev.c (toplev::main): Pass targetm.target_option.override
17364 2017-08-29 Richard Biener <rguenther@suse.de>
17365 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17367 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17368 life to the active subtree.
17370 2017-08-28 Jeff Law <law@redhat.com>
17372 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17373 derive_equivalences.
17374 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17375 Code moved into....
17376 (edge_info::derive_equivalences): New private member function
17378 * tree-ssa-dom.c (class edge_info): Changed from a struct
17379 to a class. Add ctor/dtor, methods and data members.
17380 (edge_info::edge_info): Renamed from allocate_edge_info.
17381 Initialize additional members.
17382 (edge_info::~edge_info): New.
17383 (free_dom_edge_info): Delete the edge info.
17384 (record_edge_info): Use new class & associated member functions.
17385 Tighten forms for testing for edge equivalences.
17386 (record_temporary_equivalences): Iterate over the simple
17387 equivalences rather than assuming there's only one per edge.
17388 (cprop_into_successor_phis): Iterate over the simple
17389 equivalences rather than assuming there's only one per edge.
17390 (optimize_stmt): Use operand_equal_p rather than pointer
17391 equality for mini-DSE code.
17393 2017-08-28 Nathan Sidwell <nathan@acm.org>
17395 * gcc.c (execute): Fold SIGPIPE handling into switch
17396 statement. Adjust internal error message.
17398 2017-08-28 Richard Biener <rguenther@suse.de>
17401 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17402 Do nothing for removed DIEs.
17404 2017-08-28 Richard Biener <rguenther@suse.de>
17406 PR tree-optimization/81977
17407 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17410 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17413 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17414 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17415 using targetm.gen_mem_thread_fence.
17417 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17420 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17421 predicate to register_operand. Reorder operands.
17422 (*btr<mode>): Ditto.
17423 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17424 (*btr<mode>_mask): Ditto.
17426 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17428 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17429 * config/rs6000/xmmintrin.h: New file.
17430 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17432 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17435 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17436 parameter and_insn and return it.
17437 (recombine_lvx_pattern): Insert a copy to ensure availability of
17438 the base register of the copied masking operation at the point of
17439 the instruction replacement.
17440 (recombine_stvx_pattern): Likewise.
17442 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17444 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17445 undocumented switches.
17446 (-mpower9-dform-vector): Likewise.
17447 (-mpower9-dform): Likewise.
17448 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17449 comments to delete references to -mpower9-dform* switches.
17450 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17451 Delete reference to -mpower9-dform* switches, test for
17452 -mpower9-vector instead.
17453 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17454 (OTHER_P9_VECTOR_MASKS): Likewise.
17455 (POWERPC_MASKS): Likewise.
17456 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17457 tests against -mpower9-dform* to -mpower9-vector. Delete code
17458 that checked for -mpower9-dform* consistancy with other options.
17459 Add test for -mpower9-misc to enable other power9 switches.
17460 (rs6000_init_hard_regno_mode_ok): Likewise.
17461 (rs6000_option_override_internal): Likewise.
17462 (rs6000_emit_prologue): Likewise.
17463 (rs6000_emit_epilogue): Likewise.
17464 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17465 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17466 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17468 (emit_fusion_p9_store): Likewise.
17469 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17470 resetting these macros if the assembler does not support ISA 3.0
17472 (TARGET_P9_DFORM_VECTOR): Likewise.
17473 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17474 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17476 2017-08-25 Alan Modra <amodra@gmail.com>
17478 PR rtl-optimization/81747
17479 * cse.c (cse_extended_basic_block): Don't attempt to record
17480 equivalences for degenerate conditional branches that branch
17481 to their fall-through.
17483 2017-08-24 Martin Sebor <msebor@redhat.com>
17485 PR middle-end/81908
17486 * gimple-fold.c (size_must_be_zero_p): New function.
17487 (gimple_fold_builtin_memory_op): Call it.
17489 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17491 * config/rs6000/mm_malloc.h: New file.
17493 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17495 PR tree-optimization/81913
17496 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17497 analysis when either IVs in condition can wrap.
17499 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17501 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17502 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17504 2017-08-24 Richard Biener <rguenther@suse.de>
17507 * targhooks.c (default_target_can_inline_p): Properly
17508 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17509 is present and always compare.
17510 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17511 infer -mfpmath=sse from TARGET_SSE_P.
17512 (ix86_can_inline_p): Properly use target_option_default_node when
17513 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17515 2017-08-24 Richard Biener <rguenther@suse.de>
17518 * dwarf2out.c (output_die): Handle flag_generate_offload like
17520 (output_comp_unit): Likewise.
17521 (gen_array_type_die): Likewise.
17522 (dwarf2out_early_finish): Likewise.
17523 (note_variable_value_in_expr): Likewise.
17524 (dwarf2out_finish): Likewise. Adjust assert.
17525 * cgraphunit.c (symbol_table::compile): Move setting of
17526 flag_generate_offload earlier ...
17527 (symbol_table::finalize_compilation_unit): ... here, before
17528 early debug finalization.
17530 2017-08-24 Richard Biener <rguenther@suse.de>
17532 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17533 and ipa-fnsummary.h.
17534 (ix86_can_inline_p): When ix86_fpmath flags do not match
17535 check whether the callee uses FP math at all.
17537 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17539 PR middle-end/81931
17540 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17541 instead of TYPE_PRECISION.
17543 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17544 Alan Hayward <alan.hayward@arm.com>
17545 David Sherwood <david.sherwood@arm.com>
17547 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17549 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17551 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17553 (ix86_frame::stack_realign_allocate): New field.
17554 (struct machine_frame_state): Modify comments.
17555 (machine_frame_state::sp_realigned_fp_end): New field.
17556 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17557 layout calculation.
17558 (sp_valid_at): Add assertion to assure no attempt to access invalid
17559 offset of a realigned stack.
17560 (fp_valid_at): Likewise.
17561 (choose_baseaddr): Modify comments.
17562 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17563 ix86_expand_prologue.
17564 (ix86_expand_prologue): Modify stack realignment and allocation.
17565 (ix86_expand_epilogue): Modify comments.
17566 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17567 avx2_runtime, avx512f, and avx512f_runtime.
17569 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17571 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17572 (mstackrealign): Do not init to -1.
17573 * config/i386/i386.c (ix86_option_override_internal):
17574 Check opts_set, not opts when setting default value of
17575 opts->x_ix86_force_align_arg_pointer.
17577 2017-08-23 Richard Biener <rguenther@suse.de>
17579 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17580 lang_hooks.decl_printable_name.
17581 * print-rtl-function.c (print_rtx_function): Likewise.
17582 * tree-pretty-print.c (dump_function_header): Likewise.
17584 2017-08-23 Richard Biener <rguenther@suse.de>
17587 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17588 -g0 at compile-time.
17590 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17592 PR middle-end/19706
17593 * doc/sourcebuild.texi (Other hardware attributes):
17596 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17598 PR middle-end/19706
17599 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17600 Fix single-use check.
17602 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17604 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17606 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17608 * doc/install.texi: Modify to add more details on running selected
17611 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17613 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17614 is combined with -mabi=ms.
17615 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17618 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17620 PR tree-optimization/81488
17621 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17622 and cached_basis fields.
17623 (MAX_SPREAD): New constant.
17624 (alloc_cand_and_find_basis): Initialize new fields.
17625 (clear_visited): New function.
17626 (create_phi_basis_1): Rename from create_phi_basis, set visited
17627 and cached_basis fields.
17628 (create_phi_basis): New wrapper function.
17629 (phi_add_costs_1): Rename from phi_add_costs, add spread
17630 parameter, set visited field, short-circuit when limits reached.
17631 (phi_add_costs): New wrapper function.
17632 (record_phi_increments_1): Rename from record_phi_increments, set
17634 (record_phi_increments): New wrapper function.
17635 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17636 (phi_incr_cost): New wrapper function.
17637 (all_phi_incrs_profitable_1): Rename from
17638 all_phi_incrs_profitable, set visited field.
17639 (all_phi_incrs_profitable): New wrapper function.
17641 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17642 Alan Hayward <alan.hayward@arm.com>
17643 David Sherwood <david.sherwood@arm.com>
17645 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17646 that takes the outer and inner modes.
17647 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17648 comparison as the canonical test for a paradoxical subreg.
17649 * combine.c (simplify_set): Use paradoxical_subreg_p.
17650 (make_extraction): Likewise.
17651 (force_to_mode): Likewise.
17652 (rtx_equal_for_field_assignment_p): Likewise.
17653 (gen_lowpart_for_combine): Likewise.
17654 (simplify_comparison): Likewise.
17655 * cse.c (equiv_constant): Likewise.
17656 * expmed.c (store_bit_field_1): Likewise.
17657 * final.c (alter_subreg): Likewise.
17658 * fwprop.c (propagate_rtx): Likewise.
17659 (forward_propagate_subreg): Likewise.
17660 * ira-conflicts.c (ira_build_conflicts): Likewise.
17661 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17662 * lra-constraints.c (curr_insn_transform): Likewise.
17663 (split_reg): Likewise.
17664 * lra-eliminations.c (move_plus_up): Likewise.
17665 (lra_eliminate_regs_1): Likewise.
17666 * recog.c (general_operand): Likewise.
17667 * ree.c (combine_reaching_defs): Likewise.
17668 * reload.c (push_reload): Likewise.
17669 (find_reloads): Likewise.
17670 * reload1.c (elimination_effects): Likewise.
17671 (compute_reload_subreg_offset): Likewise.
17672 (choose_reload_regs): Likewise.
17673 * rtlanal.c (subreg_lsb_1): Likewise.
17674 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17675 (simplify_subreg): Likewise.
17676 * var-tracking.c (track_loc_p): Likewise.
17677 * emit-rtl.c (byte_lowpart_offset): Likewise.
17678 (paradoxical_subreg_p): Delete out-of-line definition.
17680 2017-08-22 Jeff Law <law@redhat.com>
17682 PR tree-optimization/81741
17683 PR tree-optimization/71947
17684 * tree-ssa-dom.c: Include tree-inline.h.
17685 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17686 equivalences if one is more expensive to compute than the other.
17687 * tree-ssa-scopedtables.h (class const_or_copies): Make
17688 record_const_or_copy_raw method private.
17689 (class avail_exprs_stack): New method simplify_binary_operation.
17690 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17691 avail_exprs_stack::simplify_binary_operation as needed.
17692 (avail_exprs_stack::simplify_binary_operation): New function.
17694 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17696 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17697 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17698 (DOT_SYMBOLS): Likewise.
17699 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17700 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17701 (RS6000_ABI_NAME): Likewise.
17702 (TARGET_CMODEL): Likewise.
17703 (TOC_SECTION_ASM_OP): Likewise.
17704 (SET_CMODEL): New macro.
17705 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17707 2017-08-22 Richard Biener <rguenther@suse.de>
17709 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17710 to allow for free-lang-data replacements similar to verify_type_variant.
17712 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17713 Alan Hayward <alan.hayward@arm.com>
17714 David Sherwood <david.sherwood@arm.com>
17716 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17717 VOIDmode for the LABEL_REF.
17719 2017-08-22 Richard Biener <rguenther@suse.de>
17721 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17723 2017-08-22 Richard Biener <rguenther@suse.de>
17725 * common.opt (feliminate-dwarf2-dups): Ignore.
17726 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17727 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17728 same_die_p_wrap, compute_section_prefix,
17729 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17730 (comdat_symbol_id, comdat_symbol_number): Likewise.
17731 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17733 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17734 (output_die): Mark unreachable path unreachable.
17735 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17736 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17737 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17738 (dwarf2out_early_finish): Likewise.
17740 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17742 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17744 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17747 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17748 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17749 (avr_attribute_table) <io, io_low, address>: Initialize
17750 .decl_required with true.
17752 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17754 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17755 undocumented debugging options.
17756 (-mvsx-scalar-double): Likewise.
17757 (-mallow-df-permute): Likewise.
17758 (-mvectorize-builtins): Likewise.
17759 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17760 (rs6000_builtin_vectorized_function): Likewise.
17761 (rs6000_builtin_md_vectorized_function): Likewise.
17762 (rs6000_opt_vars): Likewise.
17764 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17767 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17768 (*btrq_imm): Rename from *btrq.
17769 (*btcq_imm): Rename from *btcq.
17770 (btsc): New code attribute.
17771 (*<btsc><mode>): New insn pattern.
17772 (*btr<mode>): Ditto.
17773 (*<btsc><mode>_mask): New insn_and_split pattern.
17774 (*btr<mode>_mask): Ditto.
17776 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17777 Alan Hayward <alan.hayward@arm.com>
17778 David Sherwood <david.sherwood@arm.com>
17780 * function.c (pad_below): Simplify padding calculation.
17782 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17783 Alan Hayward <alan.hayward@arm.com>
17784 David Sherwood <david.sherwood@arm.com>
17786 * target.def (function_prologue): Remove frame size argument.
17787 (function_epilogue): Likewise.
17788 * doc/tm.texi: Regenerate.
17789 * final.c (final_start_function): Update call to function_prologue.
17790 (final_end_function): Update call to function_epilogue.
17791 (default_function_pro_epilogue): Remove frame size argument.
17792 * output.h (default_function_pro_epilogue): Likewise.
17793 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17794 (arm_output_function_prologue): Likewise.
17795 * config/frv/frv.c (frv_function_prologue): Likewise.
17796 (frv_function_epilogue): Likewise.
17797 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17798 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17799 (ia64_output_function_epilogue): Likewise.
17800 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17801 (m32r_output_function_epilogue): Likewise.
17802 * config/microblaze/microblaze.c (microblaze_function_prologue)
17803 (microblaze_function_epilogue): Likewise.
17804 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17805 (mips_output_function_epilogue): Likewise.
17806 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17807 (mmix_target_asm_function_epilogue): Likewise.
17808 * config/msp430/msp430.c (msp430_start_function): Likewise.
17809 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17810 (nds32_asm_function_epilogue): Likewise.
17811 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17812 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17813 (pa_output_function_epilogue): Likewise.
17814 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17815 (rs6000_output_function_epilogue): Likewise.
17816 * config/rl78/rl78.c (rl78_start_function): Likewise.
17817 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17818 (rs6000_output_function_epilogue): Likewise.
17819 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17820 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17821 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17822 (sparc_asm_function_epilogue): Likewise.
17824 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17826 * tree.h (type_has_mode_precision_p): New function.
17827 * convert.c (convert_to_integer_1): Use it.
17828 * expr.c (expand_expr_real_2): Likewise.
17829 (expand_expr_real_1): Likewise.
17830 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17831 * match.pd: Likewise.
17832 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17833 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17834 * tree-tailcall.c (process_assignment): Likewise.
17835 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17836 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17837 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17838 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17839 (vectorizable_assignment): Likewise.
17840 (vectorizable_shift): Likewise.
17841 (vectorizable_operation): Likewise.
17842 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17844 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17846 * match.pd: Add pow (C, x) simplification.
17848 2017-08-21 Richard Biener <rguenther@suse.de>
17850 PR tree-optimization/81900
17851 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17852 for blocks with abnormal predecessors.
17853 (compute_antic): Do not set visited flag prematurely.
17855 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17858 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17860 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17862 * stor-layout.h (vector_type_mode): Move to...
17863 * tree.h (vector_type_mode): ...here.
17864 * stor-layout.c (vector_type_mode): Move to...
17865 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17867 2017-08-21 Richard Biener <rguenther@suse.de>
17869 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17870 register_external_die hooks.
17871 (debug_false_tree_charstarstar_uhwistar): Declare.
17872 (debug_nothing_tree_charstar_uhwi): Likewise.
17873 * debug.c (do_nothing_debug_hooks): Adjust.
17874 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17875 (debug_nothing_tree_charstar_uhwi): Likewise.
17876 * dbxout.c (dbx_debug_hooks): Adjust.
17877 (xcoff_debug_hooks): Likewise.
17878 * sdbout.c (sdb_debug_hooks): Likewise.
17879 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17880 * dwarf2out.c (macinfo_label_base): New global.
17881 (dwarf2out_register_external_die): New function for the
17882 register_external_die hook.
17883 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17884 (dwarf2_debug_hooks): Use them.
17885 (dwarf2_lineno_debug_hooks): Adjust.
17886 (struct die_struct): Add with_offset flag.
17887 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17888 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17889 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17890 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17891 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17892 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17893 defining section names for the early LTO debug variants.
17894 (reset_indirect_string): New helper.
17895 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17896 (print_dw_val): Add support for offsetted symbol references.
17897 (get_ultimate_context): Split out from is_cxx.
17898 (is_cxx): Use get_ultimate_context.
17899 (is_fortran): Add decl overload.
17900 (compute_comp_unit_symbol): Split out worker from
17901 compute_section_prefix.
17902 (compute_section_prefix): Call compute_comp_unit_symbol and
17903 set comdat_type_p here.
17904 (output_die): Skip DIE symbol output for the LTO added one.
17905 Handle DIE symbol references with offset.
17906 (output_comp_unit): Guard section name mangling properly.
17907 For LTO debug sections emit a symbol at the section beginning
17908 which we use to refer to its DIEs.
17909 (add_abstract_origin_attribute): For DIEs registered via
17910 dwarf2out_register_external_die directly refer to the early
17911 DIE rather than indirectly through the shadow one we created.
17912 Remove obsolete call to dwarf2out_abstract_function for
17913 non-function/block origins.
17914 (gen_array_type_die): When generating early LTO debug do
17915 not emit DW_AT_string_length.
17916 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17917 late when in LTO. As suggested place a gcc_unreachable for
17918 the DECL_ABSTRACT_P case.
17919 (gen_subprogram_die): Avoid another specification DIE
17920 for early built declarations/definitions for the late LTO case.
17921 (gen_variable_die): Add type references for late duplicated VLA dies
17923 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17924 we have the abstract instance already.
17925 (process_scope_var): Adjust decl DIE contexts in LTO which
17926 first puts them in limbo.
17927 (gen_decl_die): Do not generate type DIEs late apart from
17928 types for VLAs or for decls we do not yet have a DIE. Do not
17929 call dwarf2out_abstract_function late.
17930 (dwarf2out_early_global_decl): Make sure to create DIEs
17931 for abstract instances of a decl first.
17932 (dwarf2out_late_global_decl): Adjust comment.
17933 (output_macinfo_op): With multiple macro sections use
17934 macinfo_label_base to distinguish labels.
17935 (output_macinfo): Likewise. Update macinfo_label_base.
17936 Pass in the line info label.
17937 (note_variable_value_in_expr): When generating LTO resolve
17938 all variable values here by generating DIEs as needed.
17939 (init_sections_and_labels): Add early LTO debug flag parameter
17940 and generate different sections and names if set. Add generation
17941 counter for the labels so we can have multiple of them.
17942 (reset_dies): Helper to allow DIEs to be output multiple times.
17943 (dwarf2out_finish): When outputting DIEs to the fat part of an
17944 LTO object first reset DIEs.
17945 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17946 (modified_type_die): Check for decl_ultimate_origin being self
17948 (gen_type_die_with_usage): Likewise.
17949 (gen_typedef_die): Allow decl_ultimate_origin being self.
17950 (set_decl_abstract_flags): Remove.
17951 (set_block_abstract_flags): Likewise.
17952 (dwarf2out_abstract_function): Treat the early generated DIEs
17953 as the abstract copy and only add DW_AT_inline and
17954 DW_AT_artificial here and call set_decl_origin_self.
17955 If the DIE has an abstract origin don't do anything.
17956 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17957 if we have none yet (Go fails to build one, PR78628).
17958 (variably_modified_type_p): Prevent endless recursion for Ada
17959 cyclic pointer types.
17960 * lto-streamer-in.c: Include debug.h.
17961 (dref_queue): New global.
17962 (lto_read_tree_1): Stream in DIE references.
17963 (lto_input_tree): Register DIE references.
17964 (input_function): Stream DECL_DEBUG_ARGS.
17965 * lto-streamer-out.c: Include debug.h.
17966 (lto_write_tree_1): Output DIE references.
17967 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17968 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17969 (output_function): Stream DECL_DEBUG_ARGS.
17970 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17971 Stream DECL_ABSTRACT_ORIGIN.
17972 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17973 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17974 DECL_CONTEXT for file-scope decls.
17975 * lto-streamer.h (struct dref_entry): Declare.
17976 (dref_queue): Likewise.
17977 * cfgexpand.c (pass_expand::execute): Do not call the
17978 outlining_inline_function hook here.
17979 * lto-wrapper.c (debug_obj): New global.
17980 (tool_cleanup): Unlink it if required.
17981 (debug_objcopy): New function.
17982 (run_gcc): Handle early debug sections in the IL files by
17983 extracting them to separate files, partially linkin them and
17984 feeding the result back as result to the linker.
17985 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17986 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17987 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17988 sections into a separate segment.
17989 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17991 (darwin_asm_dwarf_section): Likewise.
17992 (darwin_asm_output_dwarf_offset): Likewise.
17993 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17995 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17996 Alan Hayward <alan.hayward@arm.com>
17997 David Sherwood <david.sherwood@arm.com>
17999 * read-md.h (md_reader::record_potential_iterator_use): Replace
18000 pointer argument with an rtx and an index.
18001 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18002 (apply_mode_iterator): Likewise.
18003 (apply_code_iterator): Likewise.
18004 (apply_int_iterator): Likewise.
18005 (apply_subst_iterator): Likewise.
18006 (record_iterator_use): Likewise.
18007 (record_attribute_use): Likewise.
18008 (md_reader::record_potential_iterator_use): Likewise. Update calls
18009 to record_iterator_use and apply_iterator.
18010 (iterator_use): Replace ptr with x and index.
18011 (attribute_use): Likewise.
18012 (apply_attribute_uses): Update calls to apply_iterator.
18013 (apply_iterators): Likewise. Update initialization of iterator_use.
18014 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18015 and record_potential_iterator_use.
18016 (rtx_reader::read_rtx_operand): Likewise.
18018 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18019 Alan Hayward <alan.hayward@arm.com>
18020 David Sherwood <david.sherwood@arm.com>
18022 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18025 2017-08-21 Richard Biener <rguenther@suse.de>
18027 PR middle-end/81884
18028 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18029 at struct end conservatively when comparing common bases.
18031 2017-08-21 Richard Biener <rguenther@suse.de>
18033 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18034 (mem_ref_in_stmt): Remove.
18035 (determine_max_movement): Use ref index to get at the reference.
18036 (invariantness_dom_walker::before_dom_children): Deal with
18037 lim data already initialized.
18038 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18040 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18042 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18043 (TARGET_ISA_ROUND): Ditto.
18044 (TARGET_ROUND): Ditto.
18045 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18046 * config/i386/i386.md: Ditto.
18047 * config/i386/sse.md: Ditto.
18048 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18049 with OPTION_MASK_ISA_SSE4_1.
18051 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18054 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18055 __builtin_ia32_lzcnt_u16.
18057 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18060 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18061 (rs6000_set_current_function): Rewrite function to use it.
18063 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18066 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18067 and TYPE_WARN_IF_NOT_ALIGN.
18068 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18069 (handle_warn_if_not_align): New.
18070 (place_union_field): Call handle_warn_if_not_align.
18071 (place_field): Call handle_warn_if_not_align.
18072 Copy TYPE_WARN_IF_NOT_ALIGN.
18073 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18074 (layout_type): Likewise.
18075 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18077 (tree_decl_common): Add warn_if_not_align.
18078 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18079 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18080 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18081 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18082 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18083 * doc/extend.texi: Document warn_if_not_aligned attribute.
18084 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18086 2017-08-17 Martin Liska <mliska@suse.cz>
18089 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18090 (get_data_dependence): Use it as pointer type.
18091 (distribute_loop): Likewise.
18093 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18095 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18096 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18097 (p8_vmrgow_<mode>_direct): New define_insn.
18098 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18099 handle endianness for vmrgew and vmrgow permute patterns.
18101 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18103 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18104 * config/rs6000/rs6000-cpus.def: Remove comment.
18105 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18106 (POWERPC_MASKS): Likewise.
18107 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18108 use of TARGET_VSX_TIMODE.
18109 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18110 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18111 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18112 (rs6000_option_override_internal): Remove dead code.
18113 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18114 (rs6000_legitimize_reload_address): Likewise.
18115 (rs6000_legitimate_address_p): Likewise.
18116 (rs6000_opt_masks): Delete "vsx-timode".
18117 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18118 from function comment.
18119 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18120 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18121 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18123 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18124 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18125 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18126 (VSX_TI): Likewise.
18128 (define_peephole2): Likewise.
18130 2017-08-17 Martin Sebor <msebor@redhat.com>
18133 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18134 past the end of an array.
18135 (test_pp_format): Add test cases.
18137 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18139 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18140 missing ECF_NOTHROW flags.
18142 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18145 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18146 operands residing in integer registers.
18147 (*vsx_le_perm_load_<mode>): Likewise.
18148 (*vsx_le_perm_store_<mode>): Likewise.
18149 (define_peephole2): Add peepholes to optimize the above.
18151 2017-08-17 Marek Polacek <polacek@redhat.com>
18153 PR middle-end/81814
18154 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18155 to mimic what shorten_compare did. Change the return type to bool.
18156 (fold_cond_expr_with_comparison): Update call to
18157 operand_equal_for_comparison_p.
18158 (fold_ternary_loc): Likewise.
18160 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18162 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18164 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18166 2017-08-17 Richard Biener <rguenther@suse.de>
18168 * tree-ssa-structalias.c (solve_graph): When propagating
18169 to successors update the graphs succ edges and avoid duplicate work.
18171 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18174 * config/i386/i386.c (ix86_option_override_internal): Save target
18175 specific options after ix86_stack_protector_guard_reg was changed.
18177 2017-08-17 Richard Biener <rguenther@suse.de>
18179 PR tree-optimization/81827
18180 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18181 (new_var_info): Initialize it conservatively.
18182 (get_call_vi): Mark register vars.
18183 (new_scalar_tmp_constraint_exp): Likewise.
18184 (handle_rhs_call): Likewise.
18185 (handle_const_call): Likewise.
18186 (create_function_info_for): Likewise.
18187 (solve_constraints): Sort varinfos to separate register from
18188 non-register vars to pack points-to solution bitmaps during
18191 2017-08-17 Marek Polacek <polacek@redhat.com>
18193 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18195 2017-08-17 Richard Biener <rguenther@suse.de>
18197 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18198 to true when overflow is undefined and we saturated the result.
18200 2017-08-17 Alan Modra <amodra@gmail.com>
18203 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18204 Don't use store multiple if only one reg needs saving.
18205 (interesting_frame_related_regno): New function.
18206 (rs6000_frame_related): Don't emit frame info for regs that
18208 (rs6000_emit_epilogue): Likewise.
18210 2017-08-16 Nathan Sidwell <nathan@acm.org>
18212 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18213 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18214 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18215 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18216 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18217 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18218 (verify_type): Adjust for TYPE_BINFO move.
18219 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18220 process TYPE_BINFO directly.
18221 (hash_tree): Likewise.
18222 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18224 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18227 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18229 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18231 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18234 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18235 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18236 Add ix86_binary_operator_ok to insn constraint.
18237 (*iordi_1_bts): Ditto.
18238 (*xordi_1_btc): Ditto.
18239 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18240 Update corresponding peephole2 pattern.
18244 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18246 PR tree-optimization/81832
18247 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18248 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18250 2017-08-16 Marek Polacek <polacek@redhat.com>
18253 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18254 regarding size_zero_node.
18256 2017-08-16 Martin Liska <mliska@suse.cz>
18259 * config.gcc: Respect previously set extra_objs in case
18262 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18264 PR tree-optimization/81835
18265 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18266 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18267 not depend on the phi.
18269 2017-08-16 Alan Modra <amodra@gmail.com>
18271 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18274 2017-08-16 Alan Modra <amodra@gmail.com>
18276 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18277 (save_reg_p): ..into this. Update all callers.
18278 (first_reg_to_save): Simplify.
18280 2017-08-16 Alan Modra <amodra@gmail.com>
18282 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18285 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18289 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18290 constant count if that count is less than 32.
18292 2017-08-15 Nathan Sidwell <nathan@acm.org>
18294 * gcc.c (execute): Emit friendlier message if inferior is killed
18295 by an external cause.
18297 2017-08-15 Richard Biener <rguenther@suse.de>
18299 PR tree-optimization/81790
18300 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18301 CONSTRUCTORs from simplifying and VN.
18303 2017-08-14 Martin Sebor <msebor@redhat.com>
18305 * builtin-attrs.def: Add comments.
18307 2017-08-14 Martin Sebor <msebor@redhat.com>
18310 * doc/extend.texi (attribute nonstring): Document new attribute.
18312 2017-08-14 Martin Sebor <msebor@redhat.com>
18315 * tree-diagnostic.c (default_tree_printer): Handle %G.
18316 * gimple-pretty-print.h (percent_G_format): Declare new function.
18317 * gimple-pretty-print.c (percent_G_format): Define.
18318 * tree-pretty-print.c (percent_K_format): Add argument.
18320 2017-08-14 Martin Sebor <msebor@redhat.com>
18322 PR translation/79998
18323 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18324 Remove a stray space.
18326 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18329 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18330 (*iordi_1_bts): Ditto.
18331 (*xordi_1_btc): Ditto.
18333 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18336 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18339 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18340 quoted strings, and make more translator-friendly.
18341 (darwin_rs6000_override_options): Likewise.
18342 (rs6000_option_override_internal): Likewise.
18343 (rs6000_return_in_memory): Fix overlong line.
18344 (init_cmulative_args): Use quoted strings, and make more
18345 translator-friendly.
18346 (rs6000_pass_by_reference): Fix overlong line.
18347 (def_builtin): Use quoted strings.
18348 (altivec_expand_predicate_builtin): Use quoted strings, and make
18349 more translator-friendly.
18350 (htm_expand_builtin): Use quoted strings.
18351 (cpu_expand_builtin): Use quoted strings, and make more
18352 translator-friendly.
18353 (altivec_expand_builtin): Likewise.
18354 (paired_expand_predicate_builtin): Likewise.
18355 (rs6000_invalid_builtin): Likewise.
18356 (builtin_function_type): Use quoted strings.
18357 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18358 more translator-friendly.
18359 (rs6000_trampoline_init): Likewise.
18360 (rs6000_handle_altivec_attribute): Likewise.
18361 (rs6000_inner_target_options): Use quoted strings.
18362 (rs6000_disable_incompatible_switches): Likewise.
18363 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18364 strings, and make more translator-friendly.
18365 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18367 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18369 PR tree-optimization/81799
18370 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18371 cond_expr to simple gimple operand.
18373 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18375 PR middle-end/46932
18376 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18378 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18382 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18383 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18384 TARGET_GASISR_PROLOGUES.
18385 * config/avr/avr.c (avr_option_override): Same.
18386 (avr_pass_pre_proep::execute): Same.
18388 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18391 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18392 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18394 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18396 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18397 define_insn_and_split. Split to a memory load from 0 in
18398 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18399 using PTR mode iterator.
18400 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18401 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18402 (*add_tp_<mode>): Redefine as define_insn_and_split.
18403 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18404 address space. Merge with *add_tp_x32 using PTR mode iterator.
18405 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18406 Split to an add with a memory load from 0 in
18407 DEFAULT_TLS_SEG_REG address space.
18409 2017-08-12 Andrew Pinski <apinski@cavium.com>
18411 * config/aarch64/aarch64-option-extensions.def (rdma):
18412 Fix feature string to what Linux prints out in /proc/cpuinfo.
18414 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18417 * dwarf2out.c (modified_type_die): For C typedef types that have
18418 an ultimate origin, process the ultimate origin instead of the
18420 (gen_typedef_die): Assert that input DECLs have no ultimate
18422 (gen_type_die_with_usage): For typedef variants that have an
18423 ultimate origin, just call gen_decl_die on the original DECL.
18424 (process_scope_var): Avoid creating DIEs for local typedefs and
18425 concrete static variables.
18427 2017-08-12 Alan Modra <amodra@gmail.com>
18431 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18432 match gnu-user.h startfile.
18433 (ENDFILE_LINUX_SPEC): Similarly.
18435 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18438 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18440 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18442 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18443 * config/aarch64/aarch64.md (mov<mode>): Change.
18444 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18445 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18446 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18448 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18450 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18451 for storage order barriers.
18453 2017-08-11 Martin Liska <mliska@suse.cz>
18456 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18457 variables of void type.
18459 2017-08-11 Martin Liska <mliska@suse.cz>
18461 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18462 TARGET_SUPPORTS_ALIASES.
18463 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18464 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18465 (optimize_weakref): Likewise.
18466 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18467 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18468 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18470 2017-08-11 Martin Liska <mliska@suse.cz>
18473 * config/i386/i386.c (make_resolver_func): Do complete
18474 refactoring of the function.
18476 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18479 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18480 * config/i386/i386.c (ix86_stack_protect_guard): Use
18481 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18482 * doc/invoke.texi (x86 Options): Document
18483 -mstack-protector-guard-symbol= option.
18485 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18487 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18488 * config/i386/i386.c (ix86_split_stack_guard): New function.
18489 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18490 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18491 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18492 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18493 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18494 (split_stack_space_check): Call ix86_split_stack_guard.
18496 2017-08-10 Martin Sebor <msebor@redhat.com>
18498 * print-tree.c (print_node): Print location using the established
18500 Replace spaces with colons.
18501 (debug_raw, debug): Ditto.
18503 2017-08-10 Martin Sebor <msebor@redhat.com>
18506 * pretty-print.c (pp_format): Correct the handling of %s precision.
18508 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18511 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18513 (ix86_finalize_stack_frame_flags): This. Also clear
18514 frame_pointer_needed if -fno-omit-frame-pointer is used without
18516 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18517 with ix86_finalize_stack_frame_flags.
18518 (ix86_expand_epilogue): Likewise.
18519 (ix86_expand_split_stack_prologue): Likewise.
18520 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18522 2017-08-10 Martin Liska <mliska@suse.cz>
18525 * c-attribs.c (handle_target_attribute):
18526 Report warning for an empty string argument of target attribute.
18528 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18531 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18533 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18534 or DECL_NONLOCAL labels.
18535 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18536 or DECL_NONLOCAL labels here.
18538 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18541 to indicate when early gimple folding has been disabled.
18542 (rs6000_gimple_fold_builtin): Add debug content.
18543 (rs6000_invalid_builtin): Fix whitespace.
18544 (rs6000_expand_builtin): Fix whitespace.
18545 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18547 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18550 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18551 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18553 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18555 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18556 (qdf24xx): Likewise.
18557 * config/aarch64/aarch64.md: Include falkor.md.
18558 * config/aarch64/falkor.md: New.
18560 2017-08-09 Marek Polacek <polacek@redhat.com>
18563 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18564 * diagnostic.c (emit_diagnostic): Add a comment.
18565 (emit_diagnostic_valist): New function.
18567 2017-08-09 Marek Polacek <polacek@redhat.com>
18570 * input.c (make_location): New overload.
18571 * input.h (make_location): Declare.
18573 2017-08-08 Alan Modra <amodra@gmail.com>
18574 H.J. Lu <hongjiu.lu@intel.com>
18577 * gcc.c (NO_PIE_SPEC): Delete.
18578 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18580 (LINK_PIE_SPEC): ..to here.
18581 (LINK_COMMAND_SPEC): Support -no-pie.
18582 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18583 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18585 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18586 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18587 (ENDFILE_CRTEND_SPEC): Similarly.
18589 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18592 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18593 (mstack-protector-guard-offset=): Ditto.
18594 * config/i386/i386.c (ix86_option_override): Handle
18595 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18597 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18598 ix86_stack_protect_guard_offset variables.
18599 (TARGET_STACK_PROTECT_GUARD): Always define.
18600 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18601 and -mstack-protector-guard-offset= options.
18603 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18605 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18606 boundary case for the last candidate.
18608 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18610 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18611 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18613 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18615 PR middle-end/19706
18616 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18617 * config/aarch64/aarch64-builtins.c
18618 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18619 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18620 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18622 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18623 Andrew Pinski <pinskia@gmail.com>
18625 PR middle-end/19706
18626 * internal-fn.def (XORSIGN): New.
18627 * optabs.def (xorsign_optab): New.
18628 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18629 (convert_expand_mult_copysign): New.
18630 (pass_optimize_widening_mul::execute): Call
18631 convert_expand_mult_copysign.
18633 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18635 PR tree-optimization/81354
18636 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18637 Insert on edges rather than explicitly creating landing pads.
18638 (analyze_candidates_and_replace): Commit edge inserts.
18640 2017-08-08 Richard Biener <rguenther@suse.de>
18642 PR middle-end/81719
18643 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18644 (expand_simple_operations): Also look through ADDR_EXPRs with
18645 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18647 2017-08-08 Richard Biener <rguenther@suse.de>
18649 PR tree-optimization/81723
18650 * tree-vect-slp.c (struct bst_traits): New hash traits.
18651 (bst_fail): New global.
18652 (vect_build_slp_tree_2): New worker, split out from ...
18653 (vect_build_slp_tree): ... this now wrapping it with using
18654 bst_fail set to cache SLP tree build fails. Properly handle
18656 (vect_analyze_slp_instance): Allocate and free bst_fail.
18658 2017-08-08 Martin Liska <mliska@suse.cz>
18661 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18662 LABEL_DECLs that can be from a different function.
18664 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18666 PR tree-optimization/81744
18667 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18668 loop's number of iterations.
18670 2017-08-08 Martin Liska <mliska@suse.cz>
18672 * asan.c: Include header files.
18673 * attribs.c (build_decl_attribute_variant): New function moved
18675 (build_type_attribute_qual_variant): Likewise.
18676 (cmp_attrib_identifiers): Likewise.
18677 (simple_cst_list_equal): Likewise.
18678 (omp_declare_simd_clauses_equal): Likewise.
18679 (attribute_value_equal): Likewise.
18680 (comp_type_attributes): Likewise.
18681 (build_type_attribute_variant): Likewise.
18682 (lookup_ident_attribute): Likewise.
18683 (remove_attribute): Likewise.
18684 (merge_attributes): Likewise.
18685 (merge_type_attributes): Likewise.
18686 (merge_decl_attributes): Likewise.
18687 (merge_dllimport_decl_attributes): Likewise.
18688 (handle_dll_attribute): Likewise.
18689 (attribute_list_equal): Likewise.
18690 (attribute_list_contained): Likewise.
18691 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18692 (lookup_attribute_by_prefix): Likewise.
18693 * bb-reorder.c: Include header files.
18694 * builtins.c: Likewise.
18695 * calls.c: Likewise.
18696 * cfgexpand.c: Likewise.
18697 * cgraph.c: Likewise.
18698 * cgraphunit.c: Likewise.
18699 * convert.c: Likewise.
18700 * dwarf2out.c: Likewise.
18701 * final.c: Likewise.
18702 * fold-const.c: Likewise.
18703 * function.c: Likewise.
18704 * gimple-expr.c: Likewise.
18705 * gimple-fold.c: Likewise.
18706 * gimple-pretty-print.c: Likewise.
18707 * gimple.c: Likewise.
18708 * gimplify.c: Likewise.
18709 * hsa-common.c: Likewise.
18710 * hsa-gen.c: Likewise.
18711 * internal-fn.c: Likewise.
18712 * ipa-chkp.c: Likewise.
18713 * ipa-cp.c: Likewise.
18714 * ipa-devirt.c: Likewise.
18715 * ipa-fnsummary.c: Likewise.
18716 * ipa-inline.c: Likewise.
18717 * ipa-visibility.c: Likewise.
18719 * lto-cgraph.c: Likewise.
18720 * omp-expand.c: Likewise.
18721 * omp-general.c: Likewise.
18722 * omp-low.c: Likewise.
18723 * omp-offload.c: Likewise.
18724 * omp-simd-clone.c: Likewise.
18725 * opts-global.c: Likewise.
18726 * passes.c: Likewise.
18727 * predict.c: Likewise.
18728 * sancov.c: Likewise.
18729 * sanopt.c: Likewise.
18730 * symtab.c: Likewise.
18731 * toplev.c: Likewise.
18732 * trans-mem.c: Likewise.
18733 * tree-chkp.c: Likewise.
18734 * tree-eh.c: Likewise.
18735 * tree-into-ssa.c: Likewise.
18736 * tree-object-size.c: Likewise.
18737 * tree-parloops.c: Likewise.
18738 * tree-profile.c: Likewise.
18739 * tree-ssa-ccp.c: Likewise.
18740 * tree-ssa-live.c: Likewise.
18741 * tree-ssa-loop.c: Likewise.
18742 * tree-ssa-sccvn.c: Likewise.
18743 * tree-ssa-structalias.c: Likewise.
18744 * tree-ssa.c: Likewise.
18745 * tree-streamer-in.c: Likewise.
18746 * tree-vectorizer.c: Likewise.
18747 * tree-vrp.c: Likewise.
18748 * tsan.c: Likewise.
18749 * ubsan.c: Likewise.
18750 * varasm.c: Likewise.
18751 * varpool.c: Likewise.
18752 * tree.c: Remove functions moved to attribs.[ch].
18753 * tree.h: Likewise.
18754 * config/aarch64/aarch64.c: Add attrs.h header file.
18755 * config/alpha/alpha.c: Likewise.
18756 * config/arc/arc.c: Likewise.
18757 * config/arm/arm.c: Likewise.
18758 * config/avr/avr.c: Likewise.
18759 * config/bfin/bfin.c: Likewise.
18760 * config/c6x/c6x.c: Likewise.
18761 * config/cr16/cr16.c: Likewise.
18762 * config/cris/cris.c: Likewise.
18763 * config/darwin.c: Likewise.
18764 * config/epiphany/epiphany.c: Likewise.
18765 * config/fr30/fr30.c: Likewise.
18766 * config/frv/frv.c: Likewise.
18767 * config/ft32/ft32.c: Likewise.
18768 * config/h8300/h8300.c: Likewise.
18769 * config/i386/winnt.c: Likewise.
18770 * config/ia64/ia64.c: Likewise.
18771 * config/iq2000/iq2000.c: Likewise.
18772 * config/lm32/lm32.c: Likewise.
18773 * config/m32c/m32c.c: Likewise.
18774 * config/m32r/m32r.c: Likewise.
18775 * config/m68k/m68k.c: Likewise.
18776 * config/mcore/mcore.c: Likewise.
18777 * config/microblaze/microblaze.c: Likewise.
18778 * config/mips/mips.c: Likewise.
18779 * config/mmix/mmix.c: Likewise.
18780 * config/mn10300/mn10300.c: Likewise.
18781 * config/moxie/moxie.c: Likewise.
18782 * config/msp430/msp430.c: Likewise.
18783 * config/nds32/nds32-isr.c: Likewise.
18784 * config/nds32/nds32.c: Likewise.
18785 * config/nios2/nios2.c: Likewise.
18786 * config/nvptx/nvptx.c: Likewise.
18787 * config/pa/pa.c: Likewise.
18788 * config/pdp11/pdp11.c: Likewise.
18789 * config/powerpcspe/powerpcspe.c: Likewise.
18790 * config/riscv/riscv.c: Likewise.
18791 * config/rl78/rl78.c: Likewise.
18792 * config/rx/rx.c: Likewise.
18793 * config/s390/s390.c: Likewise.
18794 * config/sh/sh.c: Likewise.
18795 * config/sol2.c: Likewise.
18796 * config/sparc/sparc.c: Likewise.
18797 * config/spu/spu.c: Likewise.
18798 * config/stormy16/stormy16.c: Likewise.
18799 * config/tilegx/tilegx.c: Likewise.
18800 * config/tilepro/tilepro.c: Likewise.
18801 * config/v850/v850.c: Likewise.
18802 * config/vax/vax.c: Likewise.
18803 * config/visium/visium.c: Likewise.
18804 * config/xtensa/xtensa.c: Likewise.
18806 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18809 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18810 constraints since the -mupper-regs-* switches have been
18812 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18813 into a vector from a double word element that was extracted from
18814 another vector, and eliminate extra XXPERMDI instructions.
18815 (vsx_concat_<mode>_2): Likewise.
18816 (vsx_concat_<mode>_3): Likewise.
18817 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18818 concat to allow optimizing inserts from previous extracts.
18820 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18822 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18823 memory reference to a SSP offset in TLS address space.
18824 (ix86_print_operand) <case '@'>: Remove.
18825 (ix86_print_operand_punct_valid_p): Remove '@' code.
18826 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18827 UNSPEC_SP_TLS_TEST.
18828 (stack_tls_protect_set_<mode>): Remove.
18829 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18830 (stack_tls_protect_test_<mode>): Remove.
18831 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18833 2017-08-07 Olivier Hainque <hainque@adacore.com>
18836 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18838 2017-08-07 Douglas Rupp <rupp@adacore.com>
18840 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18841 variable was referenced as multidir in command.
18843 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18846 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18849 2017-08-07 Martin Liska <mliska@suse.cz>
18851 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18852 * config/rl78/rl78.c: Add include of attribs.h.
18853 * config/sh/sh.c: Likewise.
18854 * config/v850/v850.c: Likewise.
18856 2017-08-07 Tom de Vries <tom@codesourcery.com>
18858 PR middle-end/78266
18859 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18861 2017-08-07 Martin Liska <mliska@suse.cz>
18863 * config/mips/mips.c: Include attribs.h.
18865 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18868 * doc/invoke.texi: Document change in behvaior for -Ofast for
18871 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18873 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18875 (aarch64_pop_regs): Likewise.
18876 (aarch64_gen_load_pair): Likewise.
18877 (aarch64_save_callee_saves): Likewise.
18878 (aarch64_restore_callee_saves): Likewise.
18880 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18882 * config/i386/i386.c: Revert the last change.
18884 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18887 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18889 (ix86_finalize_stack_frame_flags): This. Also clear
18890 frame_pointer_needed if -fno-omit-frame-pointer is used without
18892 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18893 with ix86_finalize_stack_frame_flags.
18894 (ix86_expand_epilogue): Likewise.
18895 (ix86_expand_split_stack_prologue): Likewise.
18897 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18900 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18901 to P_AES for Westmere.
18903 2017-08-07 Jonathan Yong <10walls@gmail.com>
18905 * config/i386/mingw.opt (fset-stack-executable): Removed.
18906 * config/i386/cygming.opt (fset-stack-executable): Moved
18908 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18910 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18912 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18914 2017-08-07 Marek Polacek <polacek@redhat.com>
18916 PR middle-end/81737
18917 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18919 2017-08-07 Martin Liska <mliska@suse.cz>
18921 * attribs.h (canonicalize_attr_name): New function.
18922 (cmp_attribs): Move from c-format.c and adjusted.
18923 (is_attribute_p): Moved from tree.h.
18924 * tree-inline.c: Add new includes.
18925 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18926 (private_is_attribute_p): Remove.
18927 (private_lookup_attribute): Likewise.
18928 (private_lookup_attribute_by_prefix): Simplify.
18929 (remove_attribute): Use is_attribute_p.
18930 * tree.h: Remove removed declarations.
18932 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18934 PR middle-end/81698
18935 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18936 instead of computing it in the function. Formatting fix.
18937 (expand_case): Don't rely on default_edge being the first edge,
18938 clear it if removing it, pass default_edge to
18939 emit_case_dispatch_table.
18940 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18943 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18945 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18946 insn in the function, emit NOP after the insn.
18948 2017-08-06 Tom de Vries <tom@codesourcery.com>
18950 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18953 2017-08-06 Tom de Vries <tom@codesourcery.com>
18955 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18958 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18960 PR tree-optimization/57371
18961 * match.pd: New pattern.
18963 2017-08-04 Marek Polacek <polacek@redhat.com>
18965 PR middle-end/81695
18966 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18967 perform the computation in offset_int.
18969 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18971 PR tree-optimization/81136
18972 * tree-vectorizer.h: Include tree-hash-traits.h.
18973 (vec_base_alignments): New typedef.
18974 (vec_info): Add a base_alignments field.
18975 (vect_record_base_alignments): Declare.
18976 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18978 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18979 (create_data_ref): Add an is_conditional_in_stmt argument.
18980 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18981 the is_conditional_in_stmt field.
18982 (data_ref_loc): Add an is_conditional_in_stmt field.
18983 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18984 (find_data_references_in_stmt): Update call to create_data_ref.
18985 (graphite_find_data_references_in_stmt): Likewise.
18986 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18987 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18988 (vect_record_base_alignment): New function.
18989 (vect_record_base_alignments): Likewise.
18990 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18991 for nested statements even if we fail to compute a misalignment.
18992 Use pooled base alignments for unconditional references.
18993 (vect_find_same_alignment_drs): Compare base addresses instead
18995 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18996 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18998 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19000 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19001 Add an explicit name for the enum. Use auto_vec for slp_instances
19002 and grouped_stores.
19003 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19005 (_bb_vec_info): Add a constructor and destructor.
19006 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19007 (destroy_loop_vec_info): Delete.
19008 (vect_destroy_datarefs): Likewise.
19009 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19010 (vec_info::vec_info): New function.
19011 (vec_info::~vec_info): Likewise.
19012 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19013 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19014 destroy_loop_vec_info.
19015 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19016 (_loop_vec_info::_loop_vec_info): ...this.
19017 (destroy_loop_vec_info): Replace with...
19018 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19019 the stmt_vec_infos. Leave handling of vec_info information to its
19020 destructor. Remove explicit vector releases.
19021 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19022 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19023 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19024 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19025 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19026 (destroy_bb_vec_info): Replace with...
19027 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19028 information to its destructor.
19029 (vect_slp_analyze_bb_1): Use new and delete instead of
19030 new_bb_vec_info and destroy_bb_vec_info.
19031 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19034 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19036 * tree-data-ref.h (subscript): Add access_fn field.
19037 (data_dependence_relation): Add could_be_independent_p.
19038 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19039 (same_access_functions): Move to tree-data-ref.c.
19040 * tree-data-ref.c (ref_contains_union_access_p): New function.
19041 (access_fn_component_p): Likewise.
19042 (access_fn_components_comparable_p): Likewise.
19043 (dr_analyze_indices): Add a reference to access_fn_component_p.
19044 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19046 (constant_access_functions): Likewise.
19047 (add_other_self_distances): Likewise.
19048 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19049 (initialize_data_dependence_relation): Use XCNEW and remove
19050 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19051 of access functions that have the same type. Allow the
19052 subsequence to end with different bases in some circumstances.
19053 Record the chosen access functions in SUB_ACCESS_FN.
19054 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19055 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19056 (subscript_dependence_tester_1): Likewise dra and drb.
19057 (build_classic_dist_vector): Update calls accordingly.
19058 (subscript_dependence_tester): Likewise.
19059 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19060 DDR_COULD_BE_INDEPENDENT_P.
19061 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19062 comp_alias_ddrs instead of may_alias_ddrs.
19063 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19065 (vect_analyze_data_ref_dependence): Use it if
19066 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19067 distance vectors if that fails.
19068 (dependence_distance_ge_vf): New function.
19069 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19070 LOOP_VINFO_MAY_ALIAS_DDRS.
19072 2017-08-04 Richard Biener <rguenther@suse.de>
19074 PR middle-end/81705
19075 * fold-const.c (fold_binary_loc): Properly restrict
19076 minus_var0 && minus_var1 case when associating undefined overflow
19079 2017-08-04 Tom de Vries <tom@codesourcery.com>
19081 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19083 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19085 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19086 Don't start diagnostic messages with a capital letter.
19087 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19089 (rs6000_invalid_builtin): Likewise.
19090 (rs6000_trampoline_init): Likewise.
19092 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19095 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19096 after setting changeable df flags.
19098 2017-08-03 Richard Biener <rguenther@suse.de>
19100 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19101 up if the use is in USE - X.
19103 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19105 * toplev.c (dumpfile.h): New include.
19106 (internal_error_reentered): New static function. Use it...
19107 (internal_error_function): ...here to handle reentered internal_error.
19109 2017-08-03 Richard Biener <rguenther@suse.de>
19111 PR middle-end/81148
19112 * fold-const.c (split_tree): Add minus_var and minus_con
19113 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19114 here but always use minus_*.
19115 (associate_trees): Assert we never associate with MINUS_EXPR
19116 and NULL first operand. Do not recurse for PLUS_EXPR operands
19117 when associating as MINUS_EXPR either.
19118 (fold_binary_loc): Track minus_var and minus_con.
19120 2017-08-03 Tom de Vries <tom@codesourcery.com>
19123 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19124 ACCEL_COMPILER, apply finish_options on
19125 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19127 2017-08-03 Tom de Vries <tom@codesourcery.com>
19130 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19131 function_entry_patch_area_size > 0.
19133 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19136 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19137 instead of 10??LU, perform unit multiplication in wide_int,
19138 don't change alloc_object_size_limit if the limit is larger
19141 PR tree-optimization/81655
19142 PR tree-optimization/81588
19143 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19144 the case when ranges[i].low and high are 1 for unsigned type with
19147 PR middle-end/81052
19148 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19149 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19151 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19153 * tree-vrp.h: Add include guard.
19155 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19158 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19159 (ud2): New insn pattern.
19160 * config/i386/i386.c (ix86_expand_epilogue):
19161 For naked functions, generate ud2 instead of trap insn.
19163 2017-08-02 Marek Polacek <polacek@redhat.com>
19166 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19168 2017-08-02 Tom de Vries <tom@codesourcery.com>
19169 Cesar Philippidis <cesar@codesourcery.com>
19171 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19172 Add missing edge probabilities.
19174 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19176 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19177 Correct endianness.
19179 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19181 PR middle-end/79499
19182 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19183 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19184 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19185 prologue_seq sequences - if any.
19187 2017-08-02 Richard Biener <rguenther@suse.de>
19189 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19190 via vectors if supported, integer extracts via punning if supported
19191 or otherwise vector extracts.
19193 2017-08-02 Richard Biener <rguenther@suse.de>
19195 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19197 (bitmap_insert_into_set): ... this.
19199 2017-08-02 Richard Biener <rguenther@suse.de>
19201 PR tree-optimization/81633
19203 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19205 PR tree-optimization/71752
19206 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19208 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19210 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19211 (machine_function::call_ms2sysv_pad_out): Remove field.
19212 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19213 (ix86_compute_frame_layout): Likewise.
19215 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19218 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19219 attribute with interrupt attribute.
19221 2017-08-01 Andrew Pinski <apinski@cavium.com>
19223 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19224 BIT_INSERT_EXPR's operand 1
19225 to see if the types precision matches.
19227 2017-08-01 Martin Liska <mliska@suse.cz>
19229 PR middle-end/70140
19230 * builtins.c (expand_builtin_memcpy_args): Remove.
19231 (expand_builtin_memcpy): Call newly added function
19232 expand_builtin_memory_copy_args.
19233 (expand_builtin_memcpy_with_bounds): Likewise.
19234 (expand_builtin_mempcpy): Remove last argument.
19235 (expand_builtin_mempcpy_with_bounds): Likewise.
19236 (expand_builtin_memory_copy_args): New function created from
19237 expand_builtin_mempcpy_args with small modifications.
19238 (expand_builtin_mempcpy_args): Remove.
19239 (expand_builtin_stpcpy): Remove unused argument.
19240 (expand_builtin): Likewise.
19241 (expand_builtin_with_bounds): Likewise.
19243 2017-08-01 Martin Liska <mliska@suse.cz>
19246 Make mempcpy more optimal (PR middle-end/70140).
19248 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19251 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19252 __builtin_vec_cmpne verify both arguments are compatible vectors
19253 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19254 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19255 move computation of aligned to after checking the argument types.
19259 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19260 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19263 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19264 Doug Rupp <rupp@adacore.com>
19265 Olivier Hainque <hainque@adacore.com>
19267 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19268 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19270 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19271 for TARGET_OS_CPP_BUILTIN.
19272 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19273 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19275 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19276 passing required abi options to the assembler for EABI configurations.
19277 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19278 of .text.hot and .text.unlikely sections for kernel modules when
19279 using ARM style exceptions.
19280 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19281 options. Add EXTRA_CC1_SPEC.
19282 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19284 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19286 (ARM_TARGET2_DWARF_FORMAT): Define.
19287 * config/arm/t-vxworks: Adjust multilib control to removal of the
19288 Diab command line options.
19290 2017-08-01 Martin Liska <mliska@suse.cz>
19292 PR gcov-profile/81561
19293 * gcov.c (unblock): Make unblocking safe as we need to preserve
19294 index correspondence of blocks and block_lists.
19296 2017-08-01 Richard Biener <rguenther@suse.de>
19298 PR tree-optimization/81181
19299 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19300 (compute_antic): ... end of iteration here.
19302 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19304 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19305 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19306 (ftree-slp-vectorize): Likewise.
19307 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19308 can no longer be set independent of flag_tree_loop_vectorize.
19309 * omp-general.c (emp_max_vf): Likewise.
19310 * opts.c (enable_fdo_optimizations): Remove references to
19311 flag_tree_vectorize, these are now implicit.
19312 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19313 and leave it for the options machinery.
19315 2017-08-01 Martin Liska <mliska@suse.cz>
19317 PR middle-end/70140
19318 * builtins.c (expand_builtin_memcpy_args): Remove.
19319 (expand_builtin_memcpy): Call newly added function
19320 expand_builtin_memory_copy_args.
19321 (expand_builtin_memcpy_with_bounds): Likewise.
19322 (expand_builtin_mempcpy): Remove last argument.
19323 (expand_builtin_mempcpy_with_bounds): Likewise.
19324 (expand_builtin_memory_copy_args): New function created from
19325 expand_builtin_mempcpy_args with small modifications.
19326 (expand_builtin_mempcpy_args): Remove.
19327 (expand_builtin_stpcpy): Remove unused argument.
19328 (expand_builtin): Likewise.
19329 (expand_builtin_with_bounds): Likewise.
19331 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19334 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19335 print "ds:" only for immediates in generic address space.
19337 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19340 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19341 (ix86_function_ok_for_sibcall): Return false for naked functions.
19343 2017-08-01 Richard Biener <rguenther@suse.de>
19345 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19346 (compute_antic): Seed worklist with exit block predecessors.
19347 * cfganal.c (dfs_find_deadend): For a cycle return the source
19348 of the edge closing it.
19350 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19352 * config/aarch64/aarch64.c
19353 (aarch64_can_const_movi_rtx_p): Move 0 check.
19355 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19357 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19358 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19360 * match.pd: Ditto in address comparison pattern.
19362 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19364 PR tree-optimization/81627
19365 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19366 closed ssa form for store-store chain.
19368 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19370 PR tree-optimization/81620
19371 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19372 for store-store chain.
19374 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19376 PR tree-optimization/81588
19377 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19378 ranges[i].in_p, invert comparison code ccode. For >/>=,
19379 swap rhs1 and rhs2 and comparison code unconditionally,
19380 for </<= don't do that. Don't swap rhs1/rhs2 again if
19381 ranges[i].in_p, instead invert comparison code ccode if
19382 opcode or oe->rank is BIT_IOR_EXPR.
19385 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19386 a direct optab to conversion optab.
19387 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19388 with GET_MODE_INNER as last argument instead of optab_handler.
19389 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19390 vector extraction if possible and optab is available.
19391 * expr.c (store_constructor): Use convert_optab_handler instead
19392 of optab_handler. Use vector initialization from smaller
19393 vectors if possible and optab is available.
19394 * tree-vect-stmts.c (vectorizable_load): Likewise.
19395 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19396 now have two modes.
19397 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19398 of vec_init from half-sized vectors with the same element mode.
19399 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19400 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19401 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19402 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19403 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19404 after mode in gen_vec_extract* calls.
19405 (vec_extract<mode>): Renamed to ...
19406 (vec_extract<mode><ssescalarmodelower>): ... this.
19407 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19408 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19409 element mode after mode in gen_vec_init* calls.
19410 (VEC_INIT_HALF_MODE): New mode iterator.
19411 (vec_init<mode>): Renamed to ...
19412 (vec_init<mode><ssescalarmodelower>): ... this.
19413 (vec_init<mode><ssehalfvecmodelower>): New expander.
19414 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19415 (vec_extractv2sfsf): ... this.
19416 (vec_initv2sf): Renamed to ...
19417 (vec_initv2sfsf): ... this.
19418 (vec_extractv2si): Renamed to ...
19419 (vec_extractv2sisi): ... this.
19420 (vec_initv2si): Renamed to ...
19421 (vec_initv2sisi): ... this.
19422 (vec_extractv4hi): Renamed to ...
19423 (vec_extractv4hihi): ... this.
19424 (vec_initv4hi): Renamed to ...
19425 (vec_initv4hihi): ... this.
19426 (vec_extractv8qi): Renamed to ...
19427 (vec_extractv8qiqi): ... this.
19428 (vec_initv8qi): Renamed to ...
19429 (vec_initv8qiqi): ... this.
19430 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19431 (vec_init<mode>): Renamed to ...
19432 (vec_init<mode><VEC_base_l>): ... this.
19433 (vec_extract<mode>): Renamed to ...
19434 (vec_extract<mode><VEC_base_l>): ... this.
19435 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19436 (vec_initv2sfsf): ... this.
19437 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19438 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19439 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19440 element mode after mode in gen_vec_init* calls.
19441 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19442 (vec_init<mode><Vel>): ... this.
19443 (vec_extract<mode>): Renamed to ...
19444 (vec_extract<mode><Vel>): ... this.
19445 * config/aarch64/iterators.md (Vel): New mode attribute.
19446 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19447 Add element mode after mode in gen_vec_extract* calls.
19448 * config/s390/vector.md (non_vec_l): New mode attribute.
19449 (vec_extract<mode>): Renamed to ...
19450 (vec_extract<mode><non_vec_l>): ... this.
19451 (vec_init<mode>): Renamed to ...
19452 (vec_init<mode><non_vec_l>): ... this.
19453 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19454 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19456 * config/arm/iterators.md (V_elem_l): New mode attribute.
19457 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19458 (vec_extract<mode><V_elem_l>): ... this.
19459 (vec_extractv2di): Renamed to ...
19460 (vec_extractv2didi): ... this.
19461 (vec_init<mode>): Renamed to ...
19462 (vec_init<mode><V_elem_l>): ... this.
19463 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19464 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19465 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19466 Add element mode after gen_vec_extract* calls.
19467 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19468 (vec_init<mode><unitmode>): ... this.
19469 (vec_extract<mode>): Renamed to ...
19470 (vec_extract<mode><unitmode>): ... this.
19471 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19472 (vec_init<mode><unitmode>): ... this.
19473 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19474 (vec_initv2sfsf): ... this.
19475 (vec_extractv2sf): Renamed to ...
19476 (vec_extractv2sfsf): ... this.
19477 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19478 Add element mode after gen_vec_extract* calls.
19479 * config/mips/mips.md (unitmode): New mode iterator.
19480 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19481 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19482 * config/spu/spu.md (inner_l): New mode attribute.
19483 (vec_init<mode>): Renamed to ...
19484 (vec_init<mode><inner_l>): ... this.
19485 (vec_extract<mode>): Renamed to ...
19486 (vec_extract<mode><inner_l>): ... this.
19487 * config/sparc/sparc.md (veltmode): New mode iterator.
19488 (vec_init<VMALL:mode>): Renamed to ...
19489 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19490 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19491 (vec_initv2sisi): ... this.
19492 (vec_initv2sf): Renamed to ...
19493 (vec_initv2sfsf): ... this.
19494 (vec_extractv2sf): Renamed to ...
19495 (vec_extractv2sfsf): ... this.
19496 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19497 (vec_init<mode>): Renamed to ...
19498 (vec_init<mode><VEC_base_l>): ... this.
19499 (vec_extract<mode>): Renamed to ...
19500 (vec_extract<mode><VEC_base_l>): ... this.
19501 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19502 (vec_initv2sfsf): ... this.
19503 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19504 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19505 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19506 gen_vec_init* calls.
19508 2017-08-01 Richard Biener <rguenther@suse.de>
19510 PR tree-optimization/81297
19511 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19512 TREE_OVERFLOW from INTEGER_CSTs.
19514 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19516 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19518 2017-07-31 Carl Love <cel@us.ibm.com>
19520 * config/rs6000/rs6000-c: Add support for built-in functions
19521 vector signed char vec_xl_be (signed long long, signed char *);
19522 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19523 vector signed int vec_xl_be (signed long long, signed int *);
19524 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19525 vector signed long long vec_xl_be (signed long long, signed long long *);
19526 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19527 vector signed short vec_xl_be (signed long long, signed short *);
19528 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19529 vector double vec_xl_be (signed long long, double *);
19530 vector float vec_xl_be (signed long long, float *);
19531 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19532 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19533 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19535 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19536 (altivec_expand_builtin): Add switch statement to call
19537 altivec_expand_xl_be for each builtin.
19538 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19539 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19540 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19541 __builtin_vsx_le_be_v16qi.
19542 * doc/extend.texi: Update the built-in documentation file for the
19543 new built-in functions.
19545 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19548 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19550 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19552 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19554 * config.gcc: Add z14.
19555 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19556 CPU model numbers for z13s and z14.
19557 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19559 * config/s390/s390-opts.h (enum processor_type): Rename
19560 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19561 * config/s390/s390.c (processor_table): Add field for CPU name to
19562 be passed to Binutils.
19563 (s390_asm_output_machine_for_arch): Use the new field in
19564 processor_table for Binutils.
19565 (s390_expand_builtin): Replace arch12 with z14.
19566 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19567 (s390_get_sched_attrmask): Likewise.
19568 (s390_get_unit_mask): Likewise.
19569 * config/s390/s390.opt: Add z14 to processor_type enum.
19571 2017-07-31 Martin Jambor <mjambor@suse.cz>
19574 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19575 regardless of optimization level.
19577 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19578 Martin Liska <mliska@suse.cz>
19580 * predict.def: Remove old comment and adjust probability.
19581 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19582 PREDICT statements.
19584 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19587 * config/i386/i386.c (ix86_function_naked): New function.
19588 (ix86_can_use_return_insn_p): Return false for naked functions.
19589 (ix86_expand_prologue): Skip prologue for naked functions.
19590 (ix86_expand_epilogue): Skip epilogue for naked functions
19591 and emit trap instruction.
19592 (ix86_warn_func_return): New function.
19593 (ix86_attribute_table): Add "naked" attribute specification.
19594 (TARGET_WARN_FUNC_RETURN): Define.
19595 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19597 2017-07-31 Martin Liska <mliska@suse.cz>
19599 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19600 (dump_gimple_bb_header): Always dump BB info.
19601 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19603 2017-07-31 Martin Liska <mliska@suse.cz>
19606 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19607 also with current_function_decl non-null equality.
19609 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19612 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19613 change type to the element type, instead add eltype variable and
19614 use it where we are interested in the element type.
19616 PR tree-optimization/81603
19617 * ipa-polymorphic-call.c
19618 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19619 offset arithmetic in offset_int, bail out if the resulting bit offset
19620 doesn't fit into shwi.
19622 2017-07-31 Martin Liska <mliska@suse.cz>
19624 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19625 (gimplify_save_expr): Fix comment.
19627 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19630 * config/i386/i386.c (ix86_function_arg): Update arguments for
19632 (ix86_compute_frame_layout): Set the initial stack offset to
19633 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19634 INCOMING_FRAME_SP_OFFSET.
19635 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19636 stack before exception handler returns.
19637 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19638 the 'ERROR_CODE' for exception handler.
19640 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19642 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19643 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19644 (ASM_OUTPUT_REG_POP): Ditto.
19645 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19646 instead of asm_fprintf to output pure string.
19648 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19650 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19651 to imported_module_or_decl hook.
19652 (debug_nothing_tree_tree_tree_bool): Remove.
19653 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19654 * debug.c (do_nothing_debug_hooks): Use
19655 debug_nothing_tree_tree_tree_bool_bool instead of
19656 debug_nothing_tree_tree_tree_bool.
19657 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19658 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19659 * sdbout.c (sdb_debug_hooks): Likewise.
19660 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19661 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19663 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19664 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19665 attribute, don't add anything.
19667 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19669 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19670 (fold_build2_stat_loc): Likewise.
19671 (fold_build3_stat_loc): Likewise.
19672 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19673 (fold_build1_loc): Remove macro.
19674 (fold_build2_loc): Likewise.
19675 (fold_build3_loc): Likewise.
19677 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19679 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19680 (gimple_build_debug_bind_source_stat): Likewise.
19681 * gimple.h (gimple_build_debug_bind): Remove macro.
19682 (gimple_build_debug_bind_source): Likewise.
19684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19686 * bitmap.c (bitmap_alloc): Adjust.
19687 (bitmap_gc_alloc): Likewise.
19688 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19692 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19693 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19694 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19695 (bitmap_gc_alloc_stat): Likewise.
19696 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19698 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19700 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19701 * rtl.h (shallow_copy_rtx): Remove macro.
19703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19705 * emit-rtl.c (gen_raw_REG): Adjust.
19706 * gengenrtl.c (gendef): Likewise.
19707 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19708 * rtl.h (rtx_alloc): Remove macro.
19710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19712 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19713 (build_tree_list_stat): Likewise.
19714 * tree.h (build_tree_list): Remove macro.
19715 (build_tree_list_vec): Likewise.
19717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19719 * tree.c (make_vector_stat): Remove _stat from name.
19720 (build_vector_stat): Likewise.
19721 * tree.h (make_vector_stat): Remove macro.
19722 (build_vector_stat): Likewise.
19724 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19726 * tree.h (build_var_debug_value): Remove prototype.
19728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19730 * tree.c (tree_cons_stat): Remove _stat from name.
19731 * tree.h (tree_cons): Remove macro.
19733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19735 * tree.c (build_vl_exp_stat): Remove _stat from name.
19736 * tree.h (build_vl_exp): Remove macro.
19738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19740 * tree.c (build_decl_stat): Remove _stat from name.
19741 * tree.h (build_decl): Remove macro.
19743 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19745 * gimple.c (gimple_build_with_ops_stat): Adjust.
19746 (gimple_alloc_stat): Remove _stat from name.
19747 * gimple.h (gimple_alloc): Remove macro.
19749 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19751 * tree.c (make_tree_vec_stat): Remove _stat from name.
19752 (grow_tree_vec_stat): Likewise.
19753 * tree.h (make_tree_vec_stat): Adjust prototype.
19754 (grow_tree_vec_stat): Likewise.
19755 (make_tree_vec): Remove macro.
19756 (grow_tree_vec): Likewise.
19758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19760 * fold-const.c (fold_build1_stat_loc): Adjust.
19761 (fold_build2_stat_loc): Likewise.
19762 (fold_build3_stat_loc): Likewise.
19763 * tree.c (build0_stat): Remove _stat from name.
19764 (build1_stat): Likewise.
19765 (build2_stat): Likewise.
19766 (build3_stat): Likewise.
19767 (build4_stat): Likewise.
19768 (build5_stat): Likewise.
19769 * tree.h (build1_loc): Remove macro, and rename _stat function
19771 (build2_loc): Likewise.
19772 (build3_loc): Likewise.
19773 (build4_loc): Likewise.
19774 (build5_loc): Likewise.
19776 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19778 * tree.c (make_int_cst_stat): Remove _stat from name.
19779 * tree.h (make_int_cst_stat): Adjust prototype.
19780 (make_int_cst): Remove macro.
19782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19784 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19785 * tree.h (make_tree_binfo_stat): Adjust prototype.
19786 (make_tree_binfo): Remove.
19788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19790 * tree.c (copy_node_stat): Rename to copy_node.
19791 (build_distinct_type_copy): Adjust.
19792 * tree.h (copy_node_stat): Adjust prototype.
19793 (copy_node): Remove macro.
19795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19797 * tree.c (make_node_stat): rename to make_node.
19798 (build_tree_list_stat): Adjust.
19799 (build0_stat): Likewise.
19800 (build2_stat): Likewise.
19801 (build3_stat): Likewise.
19802 (build4_stat): Likewise.
19803 (build5_stat): Likewise.
19804 (build_decl_stat): Likewise.
19805 * tree.h (make_node_stat): Adjust prototype.
19806 (make_node): remove macro.
19808 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19810 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19811 (PPC_FEATURE2_SCV): Likewise.
19812 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19814 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19816 * config/aarch64/aarch64.c
19817 (aarch64_internal_mov_immediate): Add new special pattern.
19818 * config/aarch64/aarch64.md (*movdi_aarch64):
19819 Add reg/32bit const mov case.
19821 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19822 Richard Sandiford <richard.sandiford@linaro.org>
19824 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19825 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19826 Add integer and movi cases.
19827 (movi-split-hf-df-sf split, fp16): New.
19828 (enabled): Added TARGET_FP_F16INST.
19829 * config/aarch64/iterators.md (GPF_HF): New.
19830 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19832 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19834 * config/aarch64/aarch64.c
19835 (aarch64_simd_container_mode): Add prototype.
19836 (aarch64_expand_mov_immediate): Add HI support.
19837 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19838 (aarch64_can_const_movi_rtx_p): New.
19839 (aarch64_preferred_reload_class):
19840 Remove restrictions of using FP registers for certain SIMD operations.
19841 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19842 (aarch64_valid_floating_const): Add integer move validation.
19843 (aarch64_simd_imm_scalar_p): Remove.
19844 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19845 (aarch64_legitimate_constant_p): Expand list of supported cases.
19846 * config/aarch64/aarch64-protos.h
19847 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19848 (aarch64_reinterpret_float_as_int): New.
19849 (aarch64_simd_imm_scalar_p): Remove.
19850 * config/aarch64/constraints.md (Uvi): New.
19851 (Dd): Split into Ds and new Dd.
19852 * config/aarch64/aarch64.md (*movsi_aarch64):
19854 (*movdi_aarch64): Add SIMD mov case.
19856 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19858 * tree-predcom.c: (struct chain): Handle store-store chain in which
19859 stores for elimination only store loop invariant values.
19860 (execute_pred_commoning_chain): Ditto.
19861 (prepare_initializers_chain_store_elim): Ditto.
19862 (prepare_finalizers): Ditto.
19863 (is_inv_store_elimination_chain): New function.
19864 (initialize_root_vars_store_elim_1): New function.
19866 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19868 * tree-predcom.c: Revise general description of the pass.
19869 (enum chain_type): New enum type for store elimination.
19870 (struct chain): New field supporting store elimination.
19871 (struct component): Ditto.
19872 (dump_chain): Dump store-stores chain.
19873 (release_chain): Release resources.
19874 (split_data_refs_to_components): Compute and create component
19875 contains only stores for elimination.
19876 (get_chain_last_ref_at): New function.
19877 (make_invariant_chain): Initialization.
19878 (make_rooted_chain): Specify chain type in parameter and record it.
19879 (add_looparound_copies): Skip for store-stores chain.
19880 (determine_roots_comp): Compute type of chain and pass it to
19882 (initialize_root_vars_store_elim_2): New function.
19883 (finalize_eliminated_stores): New function.
19884 (remove_stmt): Handle store for elimination.
19885 (execute_pred_commoning_chain): Execute predictive commoning on
19886 store-store chains.
19887 (determine_unroll_factor): Skip unroll for store-stores chain.
19888 (prepare_initializers_chain_store_elim): New function.
19889 (prepare_initializers_chain): Hanlde store-store chain.
19890 (prepare_finalizers_chain, prepare_finalizers): New function.
19891 (tree_predictive_commoning_loop): Return integer value indicating
19892 if loop is unrolled or lcssa form is corrupted.
19893 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19895 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19897 * tree-predcom.c (initialize_root): Delete.
19898 (execute_pred_commoning_chain): Initialize root vars and replace
19899 reference of non-combined chain directly, rather than call above
19902 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19904 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19905 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19907 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19909 * tree-predcom.c (struct chain): New field init_seq.
19910 (release_chain): Release init_seq.
19911 (prepare_initializers_chain): Record intialization stmts in above
19913 (insert_init_seqs): New function.
19914 (tree_predictive_commoning_loop): Call insert_init_seqs.
19916 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19918 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19920 2017-07-28 Richard Biener <rguenther@suse.de>
19922 * match.pd: Remove superfluous :c.
19923 * genmatch.c (simplify::id): Add member.
19924 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19926 (current_id): New global.
19927 (dt_node::parent): Move from ...
19928 (dt_operand::parent): ... here. Add for_id member.
19929 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19930 (decision_tree::find_node): Relax order requirement when
19931 merging DT_TRUE nodes to ones inbetween the current simplify
19932 and the one we try to merge with. Add diagnostic whenever
19933 we need to enforce pattern order by not merging.
19934 (decision_tree::insert): Set current_id.
19935 (decision_tree::print_node): Dump parent node and for_id.
19936 (parser::last_id): Add member.
19937 (parser::push_simplify): Assign unique id.
19938 (parser::parser): Initialize last_id.
19940 2017-07-28 Martin Liska <mliska@suse.cz>
19943 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19944 gimple_build_debug_bind.
19946 2017-07-28 Richard Biener <rguenther@suse.de>
19948 PR tree-optimization/81502
19949 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19951 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19953 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19954 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19955 for BIT_FIELD_REF args.
19956 * fold-const.c (make_bit_field_ref): Likewise.
19957 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19959 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19962 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19963 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19964 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19965 Or it into SANITIZER_UNDEFINED.
19966 * ubsan.c: Include gimple-fold.h and varasm.h.
19967 (ubsan_expand_ptr_ifn): New function.
19968 (instrument_pointer_overflow): New function.
19969 (maybe_instrument_pointer_overflow): New function.
19970 (instrument_object_size): Formatting fix.
19971 (pass_ubsan::execute): Call instrument_pointer_overflow
19972 and maybe_instrument_pointer_overflow.
19973 * internal-fn.c (expand_UBSAN_PTR): New function.
19974 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19975 * sanitizer.def (__ubsan_handle_pointer_overflow,
19976 __ubsan_handle_pointer_overflow_abort): New builtins.
19977 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19978 * internal-fn.def (UBSAN_PTR): New internal function.
19979 * opts.c (sanitizer_opts): Add pointer-overflow.
19980 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19981 * fold-const.c (build_range_check): Compute pointer range check in
19982 integral type if pointer arithmetics would be needed. Formatting
19985 2017-07-28 Martin Liska <mliska@suse.cz>
19988 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19989 parameters that are of a variable-length.
19991 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19993 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19995 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19996 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19997 (CRT_CALL_STATIC_FUNCTION): Likewise.
19998 (ASM_DEFAULT_SPEC): New define.
19999 (ASM_SPEC32): Likewise.
20000 (ASM_SPEC64): Likewise.
20001 (ASM_SPEC_COMMON): Likewise.
20002 (ASM_SPEC): Likewise.
20003 (INVALID_64BIT): Likewise.
20004 (LINK_OS_DEFAULT_SPEC): Likewise.
20005 (LINK_OS_SPEC32): Likewise.
20006 (LINK_OS_SPEC64): Likewise.
20007 (POWERPC_LINUX): Likewise.
20008 (PTRDIFF_TYPE): Likewise.
20009 (RESTORE_FP_PREFIX): Likewise.
20010 (RESTORE_FP_SUFFIX): Likewise.
20011 (SAVE_FP_PREFIX): Likewise.
20012 (SAVE_FP_SUFFIX): Likewise.
20013 (SIZE_TYPE): Likewise.
20014 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20015 (TARGET_64BIT): Likewise.
20016 (TARGET_64BIT): Likewise.
20017 (TARGET_AIX): Likewise.
20018 (WCHAR_TYPE_SIZE): Likewise.
20019 (WCHAR_TYPE): Undefine.
20020 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20021 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20022 (CPP_OS_RTEMS_SPEC): Delete.
20023 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20024 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20026 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20028 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20030 PR tree-optimization/81578
20031 * tree-parloops.c (build_new_reduction): Bail out if
20032 reduction_code isn't one of the standard OpenMP reductions.
20033 Move the details printing after that decision.
20035 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20037 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20038 related to reload_in_progress.
20039 (splat_input_operand): Likewise.
20040 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20042 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20044 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20045 (TARGET_INSTANTIATE_DECLS): Likewise.
20046 (legitimate_indexed_address_p): Delete reload_in_progress code.
20047 (rs6000_debug_legitimate_address_p): Likewise.
20048 (rs6000_eliminate_indexed_memrefs): Likewise.
20049 (rs6000_emit_le_vsx_store): Likewise.
20050 (rs6000_emit_move_si_sf_subreg): Likewise.
20051 (rs6000_emit_move): Likewise.
20052 (register_to_reg_type): Likewise.
20053 (rs6000_pre_atomic_barrier): Likewise.
20054 (rs6000_machopic_legitimize_pic_address): Likewise.
20055 (rs6000_allocate_stack_temp): Likewise.
20056 (rs6000_address_for_fpconvert): Likewise.
20057 (rs6000_address_for_altivec): Likewise.
20058 (rs6000_secondary_memory_needed_rtx): Delete function.
20059 (rs6000_check_sdmode): Likewise.
20060 (rs6000_alloc_sdmode_stack_slot): Likewise.
20061 (rs6000_instantiate_decls): Likewise.
20062 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20063 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20064 Delete reload_in_progress.
20065 (*vec_reload_and_plus_<mptrsize>): Likewise.
20066 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20067 (vsx_div_v2di): Likewise.
20068 (vsx_udiv_v2di): Likewise.
20070 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20072 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20073 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20074 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20075 (rs6000_debug_reg_global): Delete print of LRA status.
20076 (rs6000_option_override_internal): Delete dead LRA related code.
20077 (rs6000_lra_p): Delete function.
20078 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20080 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20082 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20083 * config/riscv/rtems.h: New file.
20085 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20086 Sudakshina Das <sudi.das@arm.com>
20088 * config/aarch64/aarch64.md
20089 (define_split for and<mode>3nr_compare): Move
20090 non aarch64_logical_operand to a register.
20091 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20092 register immediate operand to a register.
20093 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20095 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20097 PR middle-end/81564
20098 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20100 2017-07-27 Richard Biener <rguenther@suse.de>
20102 PR tree-optimization/81573
20103 PR tree-optimization/81494
20104 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20105 multi defuse cycle case.
20107 2017-07-27 Richard Biener <rguenther@suse.de>
20109 PR tree-optimization/81571
20110 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20113 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20115 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20116 earlier and only if MASK_FPU is set. Adjust formatting.
20118 2017-07-27 Martin Liska <mliska@suse.cz>
20120 * opt-functions.awk: Add validation of value of Init.
20121 * optc-gen.awk: Pass new argument.
20123 2017-07-27 Martin Liska <mliska@suse.cz>
20125 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20126 Fix wrong condition.
20128 2017-07-27 Martin Liska <mliska@suse.cz>
20130 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20131 BBs and edges seen by autoFDO.
20133 2017-07-27 Richard Biener <rguenther@suse.de>
20135 PR tree-optimization/81502
20136 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20137 with incompatible but same sized type.
20138 (execute_update_addresses_taken): Likewise.
20140 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20142 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20143 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20145 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20148 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20149 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20150 Change s_operand to memory_operand.
20152 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20154 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20155 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20156 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20157 Emit instructions rather than returning an expression. Handle TFmode
20158 and KFmode by casting to TImode.
20159 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20160 (rs6000_emit_le_vsx_store): Likewise.
20161 * config/rs6000/vsx.md (VSX_TI): New iterator.
20162 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20163 (*vsx_le_undo_permute_<mode>): Likewise.
20164 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20165 emit the split sequence.
20166 (*vsx_le_perm_store_<mode>): Likewise.
20168 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20170 PR tree-optimization/81555
20171 PR tree-optimization/81556
20172 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20173 if true, force CHANGED for the recursive invocation.
20174 (reassociate_bb): Remember original length of ops array, pass
20175 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20177 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20178 attributes for noipa attribute. For naked attribute use
20179 lookup_attribute first before lookup_attribute_spec.
20180 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20182 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20183 for functions with noipa attribute.
20184 (cgraph_externally_visible_p): Return true for functions with noipa
20186 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20187 for functions with noipa attribute.
20188 * doc/extend.texi: Document noipa function attribute.
20189 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20190 also for functions with noipa attribute.
20191 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20193 2017-07-26 Andrew Pinski <apinski@cavium.com>
20195 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20196 vec_unalign_load_cost and vec_unalign_store_cost.
20198 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20200 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20201 -mvsx-small-integer option.
20202 (ISA_3_0_MASKS_IEEE): Likewise.
20203 (OTHER_VSX_VECTOR_MASKS): Likewise.
20204 (POWERPC_MASKS): Likewise.
20205 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20206 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20207 code, only testing for DImode being allowed in non-VSX floating
20209 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20210 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20212 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20213 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20214 TARGET_P8_VECTOR test.
20215 (rs6000_secondary_reload_simple_move): Likewise.
20216 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20217 since TARGET_P9_VECTOR was already tested.
20218 (rs6000_opt_masks): Remove -mvsx-small-integer.
20219 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20220 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20222 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20223 test for TARGET_VEXTRACTUB was used, and that uses
20225 (p9 extract splitter): Likewise.
20226 (vsx_extract_<mode>_di_p9): Likewise.
20227 (vsx_extract_<mode>_store_p9): Likewise.
20228 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20229 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20230 the elimination of TARGET_VSX_SMALL_INTEGER.
20231 (vsx_extract_<mode>_p8): Likewise.
20232 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20233 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20234 (vsx_set_<mode>_p9): Likewise.
20235 (vsx_set_v4sf_p9): Likewise.
20236 (vsx_set_v4sf_p9_zero): Likewise.
20237 (vsx_insert_extract_v4sf_p9): Likewise.
20238 (vsx_insert_extract_v4sf_p9_2): Likewise.
20239 * config/rs6000/rs6000.md (sign extend splitter): Change
20240 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20241 (floatsi<mode>2_lfiwax_mem): Likewise.
20242 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20243 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20244 since a test for TARGET_P9_VECTOR was used.
20245 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20246 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20247 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20248 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20249 TARGET_P8_VECTOR test.
20250 (fix_trunc<mode>si2_stfiwx): Likewise.
20251 (fix_trunc<mode>si2_internal): Likewise.
20252 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20253 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20255 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20256 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20257 TARGET_P8_VECTOR test.
20258 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20259 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20260 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20262 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20263 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20264 since a test for TARGET_P9_VECTOR was used.
20265 (splitter for loading small constants): Likewise.
20267 2017-07-26 Andrew Pinski <apinski@cavium.com>
20269 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20272 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20275 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20276 (fp_valid_at): Likewise.
20278 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20280 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20281 (qdf24xx_addrcost_table): Likewise.
20282 (cortexa57_tunings): Update to use generic_branch_cost.
20283 (cortexa72_tunings): Likewise.
20284 (cortexa73_tunings): Likewise.
20285 (qdf24xx_tunings): Likewise.
20287 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20289 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20290 (thunderx2t99_branch_cost): Likewise.
20291 (cortexa35_tunings): Update to use generic_branch_cost.
20292 (cortexa53_tunings): Likewise.
20293 (cortexa57_tunings): Likewise.
20294 (cortexa72_tunings): Likewise.
20295 (cortexa73_tunings): Likewise.
20296 (thunderx2t99_tunings): Likewise.
20298 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20300 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20301 (sparc_option_override): Honour MASK_FSMULD.
20302 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20303 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20304 * config/sparc/sparc.opt (mfsmuld): New option.
20305 * doc/invoke.texi (mfsmuld): Document option.
20307 2017-07-26 Marek Polacek <polacek@redhat.com>
20309 PR middle-end/70992
20310 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20312 2017-07-26 Richard Biener <rguenther@suse.de>
20314 * gimple-match-head.c (do_valueize): Return OP if valueize
20316 (get_def): New helper to get at the def stmt of a SSA name
20317 if valueize allows.
20318 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20319 do_valueize to get at the def stmt.
20320 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20322 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20324 PR middle-end/46932
20325 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20327 2017-07-26 Martin Liska <mliska@suse.cz>
20330 * function.c (expand_function_start): Make expansion of
20331 nonlocal_goto_save_area after parm_birth_insn.
20333 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20335 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20336 from all CPU target flags enable members.
20338 2017-07-26 Richard Biener <rguenther@suse.de>
20340 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20341 (decision_tree::gen): Make 'type' const.
20342 (write_predicate): Likewise.
20344 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20346 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20347 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20348 (rs6000_option_override_internal): Likewise.
20349 (rs6000_expand_vector_set): Likewise.
20350 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20351 (TARGET_UPPER_REGS_SF): Likewise.
20352 (TARGET_UPPER_REGS_DI): Likewise.
20353 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20354 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20355 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20356 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20357 (Splitters for DI constants in Altivec registers): Likewise.
20358 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20359 (vsx_set_v4sf_p9): Likewise.
20360 (vsx_set_v4sf_p9_zero): Likewise.
20361 (vsx_insert_extract_v4sf_p9): Likewise.
20362 (vsx_insert_extract_v4sf_p9_2): Likewise.
20364 2017-07-25 Carl Love <cel@us.ibm.com>
20366 * doc/extend.texi: Update the built-in documentation file for the
20367 existing built-in functions
20368 vector signed char vec_cnttz (vector signed char);
20369 vector unsigned char vec_cnttz (vector unsigned char);
20370 vector signed short vec_cnttz (vector signed short);
20371 vector unsigned short vec_cnttz (vector unsigned short);
20372 vector signed int vec_cnttz (vector signed int);
20373 vector unsigned int vec_cnttz (vector unsigned int);
20374 vector signed long long vec_cnttz (vector signed long long);
20375 vector unsigned long long vec_cnttz (vector unsigned long long);
20377 2017-07-25 Andrew Pinski <apinski@cavium.com>
20379 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20380 accesses where the use is for the first operand of a BIT_INSERT.
20382 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20385 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20386 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20388 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20390 * config/i386/gstabs.h: Delete.
20391 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20393 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20395 * config/i386/i386.c (ix86_decompose_address): Do not check for
20396 register RTX when looking at index_reg or base_reg.
20397 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20399 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20401 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20402 to update EH info here.
20404 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20406 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20408 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20410 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20412 2017-07-25 Torsten Duwe <duwe@suse.de>
20414 * common.opt: Introduce -fpatchable-function-entry
20415 command line option, and its variables function_entry_patch_area_size
20416 and function_entry_patch_area_start.
20417 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20418 including a two-value parser.
20419 * target.def (print_patchable_function_entry): New target hook.
20420 * targhooks.h (default_print_patchable_function_entry): New function.
20421 * targhooks.c (default_print_patchable_function_entry): Likewise.
20422 * toplev.c (process_options): Switch off IPA-RA if
20423 patchable function entries are being generated.
20424 * varasm.c (assemble_start_function): Look at the
20425 patchable-function-entry command line switch and current
20426 function attributes and maybe generate NOP instructions by
20427 calling the print_patchable_function_entry hook.
20428 * doc/extend.texi: Document patchable_function_entry attribute.
20429 * doc/invoke.texi: Document -fpatchable_function_entry
20430 command line option.
20431 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20433 * doc/tm.texi: Re-generate.
20435 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20438 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20439 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20441 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20443 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20445 2017-07-25 Richard Biener <rguenther@suse.de>
20447 PR tree-optimization/81455
20448 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20449 not walk in cycles when looking for guards.
20451 2017-07-25 Richard Biener <rguenther@suse.de>
20453 PR tree-optimization/81529
20454 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20455 when optimizing backedge uses.
20457 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20459 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20461 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20462 to dl_section_ref. On AIX, append an expression to subtract
20463 the size of the section length to dl_section_ref.
20465 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20467 * configure.ac: If any of the config.* scripts fail, exit 1.
20468 * configure: Regenerate.
20470 2017-07-25 Richard Biener <rguenther@suse.de>
20472 PR middle-end/81546
20473 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20474 of immediate uses, be more verbose on errors.
20476 2017-07-25 Richard Biener <rguenther@suse.de>
20478 PR tree-optimization/81510
20479 * tree-vect-loop.c (vect_is_simple_reduction): When the
20480 reduction stmt is not inside the loop bail out.
20482 2017-07-25 Richard Biener <rguenther@suse.de>
20484 PR tree-optimization/81303
20485 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20486 profitability check against LOOP_VINFO_NITERSM1.
20488 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20490 * domwalk.c (cmp_bb_postorder): Simplify.
20491 (sort_bbs_postorder): New function. Use it...
20492 (dom_walker::walk): ...here to optimize common cases.
20494 2017-07-25 Martin Liska <mliska@suse.cz>
20497 * ipa-visibility.c (function_and_variable_visibility): Make the
20498 redirection just on target that supports aliasing.
20499 Fix GNU coding style.
20501 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20504 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20506 * config/arm/rtems.h: Likewise.
20507 * config/bfin/rtems.h: Likewise.
20508 * config/i386/rtemself.h: Likewise.
20509 * config/lm32/rtems.h: Likewise.
20510 * config/m32c/rtems.h: Likewise.
20511 * config/m68k/rtemself.h: Likewise.
20512 * config/microblaze/rtems.h: Likewise.
20513 * config/mips/rtems.h: Likewise.
20514 * config/moxie/rtems.h: Likewise.
20515 * config/nios2/rtems.h: Likewise.
20516 * config/powerpcspe/rtems.h: Likewise.
20517 * config/rs6000/rtems.h: Likewise.
20518 * config/rtems.h: Likewise.
20519 * config/sh/rtems.h: Likewise.
20520 * config/sh/rtemself.h: Likewise.
20521 * config/sparc/rtemself.h: Likewise.
20523 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20526 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20527 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20528 * tree-ssa-structalias.c (alias_get_name): Same.
20530 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20533 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20534 instructions if no du chain is found.
20536 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20538 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20540 2017-07-25 Richard Biener <rguenther@suse.de>
20542 PR middle-end/81505
20543 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20546 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20548 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20549 upper-regs options.
20550 (ISA_2_7_MASKS_SERVER): Likewise.
20551 (ISA_3_0_MASKS_IEEE): Likewise.
20552 (OTHER_P8_VECTOR_MASKS): Likewise.
20553 (OTHER_VSX_VECTOR_MASKS): Likewise.
20554 (POWERPC_MASKS): Likewise.
20555 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20556 duplicate list of options.
20557 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20558 explicit -mupper-regs options.
20559 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20560 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20561 alias for -mupper-regs-df.
20562 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20563 (rs6000_init_hard_regno_mode_ok): Likewise.
20564 (rs6000_option_override_internal): Likewise.
20565 (rs6000_opt_masks): Likewise.
20566 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20567 options in terms of whether -mvsx or -mpower8-vector was used.
20568 (TARGET_UPPER_REGS_DI): Likewise.
20569 (TARGET_UPPER_REGS_SF): Likewise.
20570 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20571 -mupper-regs-* options.
20573 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20575 * passes.c (emergency_dump_function): Print some empty lines and a
20576 header before the RTL dump.
20578 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20580 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20582 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20585 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20586 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20588 2017-07-24 Carl Love <cel@us.ibm.com>
20590 * config/rs6000/rs6000-c.c: Add support for built-in functions
20591 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20592 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20593 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20594 vec_extract_fp_from_shortl): Add defines for the two builtins.
20595 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20596 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20598 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20599 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20600 * doc/extend.texi: Update the built-in documentation file for the
20601 new built-in function.
20603 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20606 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20608 * doc/generic.texi: Likewise.
20609 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20610 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20612 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20614 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20615 (aarch64_mls_elt_merge<mode>): Likewise.
20617 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20619 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20620 having __cxa_atexit.
20622 2017-07-23 Michael Collison <michael.collison@arm.com>
20624 * config/arm/arm.c (arm_option_override): Deprecate
20625 use of -mstructure-size-boundary.
20626 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20627 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20629 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20632 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20633 Reduce cost estimate for direct moves.
20635 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20638 * config/i386/i386.c (ix86_option_override_internal): Disable
20639 BMI, BMI2 and TBM instructions for -m16.
20641 2017-07-21 Carl Love <cel@us.ibm.com>
20643 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20644 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20645 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20646 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20647 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20648 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20649 VMULOSW): New enum "unspec" values.
20650 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20651 altivec_vmulosw): New patterns.
20652 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20653 VMULOSW): Add definitions.
20655 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20657 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20658 (qdf24xx): Likewise.
20659 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20660 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20661 (AARCH64_FL_V8_1): Renumber.
20662 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20663 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20664 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20665 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20666 rdma to feature modifiers list.
20668 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20670 PR middle-end/56727
20671 * ipa-visibility (function_and_variable_visibility): Convert
20672 recursive PLT call to direct call if appropriate.
20674 2017-07-21 Andrew Pinski <apinski@cavium.com>
20676 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20677 operand 1 to see if the types precision matches.
20678 * fold-const.c (operand_equal_p): Likewise.
20680 2017-07-21 Richard Biener <rguenther@suse.de>
20682 PR tree-optimization/81303
20683 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20684 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20685 (vect_peeling_hash_get_lowest_cost): Adjust.
20686 (vect_enhance_data_refs_alignment): Likewise. Use
20687 vect_get_peeling_costs_all_drs to compute the penalty for no
20688 peeling to match up costs.
20690 2017-07-21 Richard Biener <rguenther@suse.de>
20692 PR tree-optimization/81500
20693 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20694 we didn't identify a reduction path.
20696 2017-07-21 Tom de Vries <tom@codesourcery.com>
20697 Cesar Philippidis <cesar@codesourcery.com>
20699 PR gcov-profile/81442
20700 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20703 2017-07-21 Tom de Vries <tom@codesourcery.com>
20706 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20708 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20709 nvptx_override_options_after_change.
20711 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20713 * dwarf2out.c (output_file_names): Avoid double testing for
20714 dwarf_version >= 5.
20716 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20718 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20720 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20722 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20724 (try_crossjump_to_edge): Do not punt on partitioned functions.
20726 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20728 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20729 Put all BBs reachable only via paths crossing cold region to cold
20731 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20733 2016-07-21 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/81303
20736 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20737 into account prologue and epilogue iterations when raising
20738 min_profitable_iters to sth at least covering one vector iteration.
20740 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20742 * config/arm/arm.c (arm_test_cpu_arch_dat):
20745 2017-07-20 Nathan Sidwell <nathan@acm.org>
20747 Remove TYPE_METHODS.
20748 * tree.h (TYPE_METHODS): Delete.
20749 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20750 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20751 (dbxout_type_methods): Scan TYPE_FIELDS.
20752 (dbxout_type): Don't check TYPE_METHODS here.
20753 * function.c (use_register_for_decl): Always ignore register for
20754 class types when not optimizing.
20755 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20756 * tree.c (free_lang_data_in_type): Stitch out member functions and
20757 templates from TYPE_FIELDS.
20758 (build_distinct_type_copy, verify_type_variant,
20759 verify_type): Member fns are on TYPE_FIELDS.
20760 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20761 * tree-pretty-print.c (dump_generic_node): Likewise.
20763 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20766 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20767 V2TImode and V4TImode.
20768 (ix86_expand_vector_extract): Likewise.
20769 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20770 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20771 (ssescalarmode): Handle V4TImode and V2TImode.
20772 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20773 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20774 (VEXTRACTI128_MODE): New mode iterator.
20775 (splitter for *vec_extractv?ti first element): New.
20776 (VEC_INIT_MODE): New mode iterator.
20777 (vec_init<mode>): Consolidate 3 expanders into one using
20778 VEC_INIT_MODE mode iterator.
20780 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20782 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20783 non_spilled_static_chain_regno_p.
20785 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20787 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20789 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20791 * bb-reorder.c (connect_traces): Allow copying of blocks within
20794 2017-07-20 Richard Biener <rguenther@suse.de>
20796 * gimple.h (gimple_phi_result): Add gphi * overload.
20797 (gimple_phi_result_ptr): Likewise.
20798 (gimple_phi_arg): Likewise. Adjust index assert to only
20799 allow actual argument accesses rather than all slots available
20801 (gimple_phi_arg_def): Add gphi * overload.
20802 * tree-phinodes.c (make_phi_node): Initialize only actual
20804 (resize_phi_node): Clear memory not covered by old node,
20805 do not initialize excess argument slots.
20806 (reserve_phi_args_for_new_edge): Initialize new argument slot
20809 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20811 PR tree-optimization/81388
20813 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20815 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20816 by removing computation of may_be_zero.
20818 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20819 Tom de Vries <tom@codesourcery.com>
20821 PR middle-end/81030
20822 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20823 when gimple level profile disagrees with what RTL expander did.
20825 2017-07-20 Richard Biener <rguenther@suse.de>
20827 PR tree-optimization/61171
20828 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20829 (vect_analyze_stmt): Add slp instance parameter.
20830 (vectorizable_reduction): Likewise.
20831 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20832 (vect_is_simple_reduction): Deal with chains not detected
20833 as SLP reduction chain, specifically not properly associated
20834 chains containing a mix of plus/minus.
20835 (get_reduction_op): Remove.
20836 (get_initial_defs_for_reduction): Simplify, pass in whether
20837 this is a reduction chain, pass in the SLP node for the PHIs.
20838 (vect_create_epilog_for_reduction): Get the SLP instance as
20840 (vectorizable_reduction): Get the SLP instance as arg.
20841 During analysis remember the SLP node with the PHIs in the
20842 instance. Simplify getting at the vectorized reduction PHIs.
20843 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20844 through SLP instance.
20845 (vect_slp_analyze_operations): Likewise.
20846 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20847 (vect_transform_stmt): Likewise.
20849 2017-07-20 Tom de Vries <tom@codesourcery.com>
20851 PR tree-optimization/81489
20852 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20853 read of phi arg location to before loop that modifies phi.
20855 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20857 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20860 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20862 PR middle-end/81331
20863 * except.c (execute): Fix ordering issue.
20865 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20867 PR rtl-optimization/81423
20868 * combine.c (make_compound_operation_int): Don't try to optimize
20869 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20871 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20873 PR rtl-optimization/81423
20874 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20875 with a constant that is -1 in the truncated to mode.
20877 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20879 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20880 (determine_unlikely_bbs): ... here.
20881 * predict.h (propagate_unlikely_bbs_forward): Declare.
20882 * cfgexpand.c (pass_expand::execute): Use it.
20883 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20885 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20886 propagate_unlikely_bbs_forward.
20888 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20890 PR middle-end/81331
20891 * except.c (maybe_add_nop_after_section_switch): New function.
20894 2017-07-19 Tom de Vries <tom@codesourcery.com>
20896 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20898 2017-07-19 Tom de Vries <tom@codesourcery.com>
20900 * gimple.h (gimple_phi_arg): Make assert more strict.
20902 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20904 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20905 * config/rs6000/mmintrin.h: New file.
20906 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20908 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20910 PR tree-optimization/81346
20911 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20913 2017-07-19 Tom de Vries <tom@codesourcery.com>
20915 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20917 2017-07-19 Tom de Vries <tom@codesourcery.com>
20919 * config/nvptx/nvptx-modes.def: Add V2DImode.
20920 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20921 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20922 (nvptx_output_mov_insn): Handle lack of mov.b128.
20923 (nvptx_print_operand): Handle 'H' and 'L' codes.
20924 (nvptx_vector_mode_supported): Allow V2DImode.
20925 (nvptx_preferred_simd_mode): New function.
20926 (nvptx_data_alignment): New function.
20927 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20928 nvptx_preferred_simd_mode.
20929 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20931 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20933 2017-07-19 Tom de Vries <tom@codesourcery.com>
20935 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20936 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20937 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20938 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20939 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20940 (mov<VECIM>_insn): New define_insn.
20941 (define_expand "mov<VECIM>): New define_expand.
20943 2017-07-19 Tom de Vries <tom@codesourcery.com>
20945 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20947 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20949 PR tree-optimization/81346
20950 * fold-const.h (fold_div_compare, range_check_type): Declare.
20951 * fold-const.c (range_check_type): New function.
20952 (build_range_check): Use range_check_type.
20953 (fold_div_compare): No longer static, rewritten into
20954 a match.pd helper function.
20955 (fold_comparison): Don't call fold_div_compare here.
20956 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20957 as helper function.
20959 2017-07-19 Nathan Sidwell <nathan@acm.org>
20961 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20962 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20963 * tree.c (find_decls_types_r, verify_type): Use
20964 TYPE_{MIN,MAX}_VALUE_RAW.
20965 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20966 (hash_tree): Likewise.
20967 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20969 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20972 2017-07-18 Tom de Vries <tom@codesourcery.com>
20974 PR middle-end/81464
20975 * omp-expand.c (expand_omp_for_static_chunk): Handle
20976 equal-argument loop exit phi.
20978 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20981 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20982 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20983 operand 2 predicate.
20984 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20985 operand 2 predicate.
20986 (ror,rol -> rorx splitters): Use const_int_operand as
20987 operand 2 predicate.
20989 2017-06-18 Richard Biener <rguenther@suse.de>
20991 PR tree-optimization/81410
20992 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20993 the gap in the ! slp_perm SLP case after each group.
20995 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20997 PR middle-end/81463
20998 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21001 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21003 PR middle-end/81462
21004 * predict.c (set_even_probabilities): Cleanup; do not affect
21005 probabilities that are already known.
21006 (combine_predictions_for_bb): Call even when count is set.
21008 2017-07-18 Nathan Sidwell <nathan@acm.org>
21010 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21013 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21016 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21017 optimization for loop niter analysis.
21019 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21022 * config/avr/avr.c (avr_optimize_casesi): Don't use
21023 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21025 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21027 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21028 body_cost_vec from _vect_peel_extended_info.
21029 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21030 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21033 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21035 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21037 2017-07-18 Richard Biener <rguenther@suse.de>
21039 PR tree-optimization/80620
21040 PR tree-optimization/81403
21041 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21042 info when re-using a VN table entry.
21044 2017-07-18 Richard Biener <rguenther@suse.de>
21046 PR tree-optimization/81418
21047 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21048 vectype_in. Verify that with lane-reducing reduction operations
21049 we have a single def-use cycle.
21051 2017-07-17 Carl Love <cel@us.ibm.com>
21053 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21055 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21056 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21057 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21059 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21060 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21061 VMULOSW): New enum "unspec" values.
21062 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21063 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21064 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21065 altivec_vmulosw): New patterns.
21066 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21067 VMULOSW): Add definitions.
21069 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21071 * config/alpha/alpha.c: Include predict.h.
21073 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21075 * tree-vrp.c (compare_assert_loc): Fix comparison function
21076 to return predictable results.
21078 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21080 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21082 (subdi3): Likewise.
21083 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21084 * doc/invoke.texi (mexpand-adddi): Update text.
21086 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21088 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21089 that also clobbers the CC register. The old expand code is moved
21091 (*arc_clzsi2): ... here.
21092 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21093 the CC register. The old expand code is moved to ...
21094 (arc_ctzsi2): ... here.
21096 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21098 * config/arc/arc.opt (mindexed-loads): Use initial value
21099 TARGET_INDEXED_LOADS_DEFAULT.
21100 (mauto-modify-reg): Use initial value
21101 TARGET_AUTO_MODIFY_REG_DEFAULT.
21102 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21103 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21104 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21105 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21107 2017-07-17 Martin Liska <mliska@suse.cz>
21110 * opts.c (finish_options): Do not allow -fgnu-tm
21111 w/ -fsanitize={kernel-,}address. Say sorry.
21113 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21116 * tree-loop-distribution.c (classify_partition): Only assert on
21117 numer of iterations.
21118 (merge_dep_scc_partitions): Delete prameter. Update function call.
21119 (distribute_loop): Remove code handling loop with unknown niters.
21120 (pass_loop_distribution::execute): Skip loop with unknown niters.
21122 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21125 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21126 function sort_partitions_by_post_order.
21128 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21130 PR tree-optimization/81374
21131 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21132 the max index of basic blocks, rather than number of basic blocks.
21134 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21136 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21138 (arc_legitimate_pic_operand_p): Likewise.
21139 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21141 (arc_needs_pcl_p): Likewise.
21142 (arc_legitimate_pc_offset_p): Likewise.
21143 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21144 function is also used in constrains.md.
21145 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21146 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21147 PLUS. Only return true/false in known cases, otherwise assert.
21148 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21149 is already called in arc_legitimate_constant_p.
21150 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21152 (LEGITIMATE_PIC_OPERAND_P): Use
21153 arc_raw_symbolic_reference_mentioned_p function.
21154 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21159 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21160 Andrew Burgess <andrew.burgess@embecosm.com>
21162 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21163 (arc_return_address_register): New function.
21164 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21165 (arc_handle_fndecl_attribute): Add naked attribute.
21166 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21167 (TARGET_WARN_FUNC_RETURN): Likewise.
21168 (arc_allocate_stack_slots_for_args): New function.
21169 (arc_warn_func_return): Likewise.
21170 (machine_function): Change type fn_type.
21171 (arc_compute_function_type): Consider new naked function type,
21172 change function return type.
21173 (arc_must_save_register): Adapt to handle new
21174 arc_compute_function_type's return type.
21175 (arc_expand_prologue): Likewise.
21176 (arc_expand_epilogue): Likewise.
21177 (arc_return_address_regs): Delete.
21178 (arc_return_address_register): New function.
21179 (arc_epilogue_uses): Use above function.
21180 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21181 (arc_function_type): Change encoding, add naked type.
21182 (ARC_INTERRUPT_P): Change to handle the new encoding.
21183 (ARC_FAST_INTERRUPT_P): Likewise.
21184 (ARC_NORMAL_P): Define.
21185 (ARC_NAKED_P): Likewise.
21186 (arc_compute_function_type): Delete prototype.
21187 * config/arc/arc.md (in_ret_delay_slot): Use
21188 arc_return_address_register function.
21189 (simple_return): Likewise.
21190 (p_return_i): Likewise.
21192 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21194 PR tree-optimization/81428
21195 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21196 can't be built for those types.
21198 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21200 Remove stuff dead since r239246.
21202 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21203 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21204 (avr_inform_devices): Remove dead stuff.
21206 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21208 * config/arm/arm_neon.h: Fix softp typo.
21210 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21212 PR tree-optimization/81365
21213 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21214 aggregate moves onto bb predecessor edges, make sure there are no
21215 loads that could alias the lhs in between the start of bb and the
21218 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21221 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21222 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21223 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21225 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21227 PR tree-optimization/81396
21228 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21229 (init_symbolic_number): Initialize it to 1.
21230 (perform_symbolic_merge): Add n_ops from both operands into the new
21232 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21233 without base_addr as useless if they need more than one operation.
21234 (bswap_replace): Handle !bswap case for NULL base_addr.
21236 2017-07-17 Tom de Vries <tom@codesourcery.com>
21239 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21242 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21244 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21245 conditional builtin define __FIX_LEON3FT_B2BST.
21247 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21249 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21250 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21253 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21255 PR rtl-optimization/81424
21256 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21257 to remove potential trapping from operands if -fnon-call-exceptions.
21259 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21261 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21262 profile_proability for scalling.
21263 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21265 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21267 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21269 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21271 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21272 fixpoint arithmetics.
21274 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21276 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21277 fixpoint arithmetics.
21279 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21281 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21282 fixpoint arithmetics.
21284 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21286 * profile-count.h (profile_probability::from_reg_br_prob_note,
21287 profile_probability::to_reg_br_prob_note): New functions.
21288 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21289 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21290 * predict.c (probability_reliable_p): Update.
21291 (edge_probability_reliable_p): Update.
21292 (br_prob_note_reliable_p): Update.
21293 (invert_br_probabilities): Update.
21294 (add_reg_br_prob_note): New function.
21295 (combine_predictions_for_insn): Update.
21296 * asan.c (asan_clear_shadow): Update.
21297 * cfgbuild.c (compute_outgoing_frequencies): Update.
21298 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21299 (update_br_prob_note): Update.
21300 (rtl_verify_edges): Update.
21301 (purge_dead_edges): Update.
21302 (fixup_reorder_chain): Update.
21303 * emit-rtl.c (try_split): Update.
21304 * ifcvt.c (cond_exec_process_insns): Update.
21305 (cond_exec_process_if_block): Update.
21306 (dead_or_predicable): Update.
21307 * internal-fn.c (expand_addsub_overflow): Update.
21308 (expand_neg_overflow): Update.
21309 (expand_mul_overflow): Update.
21310 * loop-doloop.c (doloop_modify): Update.
21311 * loop-unroll.c (compare_and_jump_seq): Update.
21312 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21313 * predict.h: Update.
21314 * reorg.c (mostly_true_jump): Update.
21316 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21317 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21318 * config/arc/arc.c: (emit_unlikely_jump): Update.
21319 * config/arm/arm.c: (emit_unlikely_jump): Update.
21320 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21321 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21322 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21323 (ix86_print_operand): Update.
21324 (ix86_split_fp_branch): Update.
21325 (predict_jump): Update.
21326 * config/ia64/ia64.c (ia64_print_operand): Update.
21327 * config/mmix/mmix.c (mmix_print_operand): Update.
21328 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21329 (rs6000_expand_split_stack_prologue): Update.
21330 * config/rs6000/rs6000.c: Update.
21331 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21332 (s390_expand_vec_movstr): Update.
21333 (s390_expand_cs_tdsi): Update.
21334 (s390_expand_split_stack_prologue): Update.
21335 * config/sh/sh.c (sh_print_operand): Update.
21336 (expand_cbranchsi4): Update.
21337 (expand_cbranchdi4): Update.
21338 * config/sparc/sparc.c (output_v9branch): Update.
21339 * config/spu/spu.c (get_branch_target): Update.
21340 (ea_load_store_inline): Update.
21341 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21342 * config/tilepro/tilepro.c: Update.
21344 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21346 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21347 (gimplify_save_expr): Likewise.
21349 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21351 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21353 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21355 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21357 * timevar.def (TV_IPA_FNSUMMARY): Define.
21359 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21361 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21362 to back store errata sensitive sequence from being generated.
21363 (sqrtdf2_fix): Likewise.
21365 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21367 * tree-ssa-threadupdate.c (compute_path_counts,
21368 update_joiner_offpath_counts): Use profile_probability.
21370 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21373 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21375 * config/arm/arm-c.c (arm_cpu_builtins): Define
21376 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21378 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21380 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21381 array entries to represent __ieee128 versions of the
21382 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21383 scalar_extract_sig, and scalar_insert_exp built-in functions.
21384 (altivec_resolve_overloaded_builtin): Add special case handling
21385 for the __builtin_scalar_insert_exp function, as represented by
21386 the P9V_BUILTIN_VEC_VSIEDP constant.
21387 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21388 exponent support for __ieee128 argument.
21389 (VSESQP): Add scalar extract signature support for __ieee128
21391 (VSTDCNQP): Add scalar test negative support for __ieee128
21393 (VSIEQP): Add scalar insert exponent support for __int128 argument
21394 with __ieee128 result.
21395 (VSIEQPF): Add scalar insert exponent support for __ieee128
21396 argument with __ieee128 result.
21397 (VSTDCQP): Add scalar test data class support for __ieee128
21399 (VSTDCNQP): Add overload support for scalar test negative with
21400 __ieee128 argument.
21401 (VSTDCQP): Add overload support for scalar test data class
21402 __ieee128 argument.
21403 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21404 UNSPEC_VSX_SXSIGDP.
21405 (UNSPEC_VSX_SIEXPQP): New constant.
21406 (xsxexpqp): New insn for VSX scalar extract exponent quad
21408 (xsxsigqp): New insn for VSX scalar extract significand quad
21410 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21411 precision with floating point argument.
21412 (xststdcqp): New expand for VSX scalar test data class quad
21414 (xststdcnegqp): New expand for VSX scalar test negative quad
21416 (xststdcqp): New insn to match expansions for VSX scalar test data
21417 class quad precision and VSX scalar test negative quad precision.
21418 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21419 special case operand checking to enforce that second operand of
21420 VSX scalar test data class with quad precision argument is a 7-bit
21422 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21423 prototypes and descriptions of __ieee128 versions of
21424 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21425 scalar_test_data_class, and scalar_test_neg built-in functions.
21427 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21429 PR tree-optimization/81162
21430 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21431 replace a negate with an add.
21433 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21435 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21437 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21439 * config/arm/arm-c.c (arm_cpu_builtins): Define
21440 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21442 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21444 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21445 (armv8-r): Set ARM Cortex-R52 as default CPU.
21446 * config/arm/arm-tables.opt: Regenerate.
21447 * config/arm/arm-tune.md: Regenerate.
21448 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21450 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21451 extension for -mcpu=cortex-r52.
21453 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21455 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21456 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21457 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21458 (fp-armv8): Define it as FP_ARMv8 only.
21459 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21460 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21462 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21463 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21464 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21465 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21466 than TARGET_FPU_ARMV8.
21467 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21468 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21469 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21471 * config/arm/neon.md (neon_vrint): Likewise.
21472 (neon_vcvt): Likewise.
21473 (neon_<fmaxmin_op><mode>): Likewise.
21474 (<fmaxmin><mode>3): Likewise.
21475 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21476 * config/arm/predicates.md (arm_cond_move_operator): Check against
21477 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21479 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21481 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21482 to top of function.
21484 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21486 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21487 loop in comment with memset.
21489 2017-07-14 Martin Liska <mliska@suse.cz>
21491 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21492 * dwarf2out.c (is_java): Remove the function.
21493 (output_pubname): Remove usage of the function.
21494 (lower_bound_default): Remove usage of DW_LANG_Java.
21495 (gen_compile_unit_die): Likewise.
21496 * gcc.c: Remove compiler defaults for .java and .zip files.
21497 * gimple-expr.c (remove_suffix): Change as there's no longer
21498 extension than 4-letter one.
21499 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21500 (gimplify_save_expr): Likewise.
21501 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21502 as it's possible even for other languages than Java.
21503 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21504 * lto-opts.c (lto_write_options): Remove reference to Java.
21505 * opts.c (strip_off_ending): Update file extension handling.
21506 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21507 * tree-eh.c (lower_resx): Likewise.
21508 * tree.c (free_lang_data_in_type): Remove dead code.
21509 (find_decls_types_r): Likewise.
21510 (build_common_builtin_nodes): Remove Java from a comment.
21511 (verify_type): Remove dead code.
21512 * varasm.c (assemble_external): Remove Java from a comment.
21514 2017-07-14 Martin Liska <mliska@suse.cz>
21516 * opts.c (finish_options): Add quotes.
21517 (common_handle_option): Likewise.
21519 2017-07-14 Martin Liska <mliska@suse.cz>
21521 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21522 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21523 Remove N_SO_PASCAL.
21524 * dwarf2out.c (lower_bound_default): Do not handle
21526 (gen_compile_unit_die): Likewise.
21527 * gcc.c: Remove default extension binding for GNU Pascal.
21528 * stmt.c: Remove Pascal language from a comment.
21529 * xcoffout.c: Likewise.
21531 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21534 * diagnostic-show-locus.c (fixit_cmp): New function.
21535 (layout::layout): Sort m_fixit_hints.
21536 (column_range::column_range): Assert that the values are valid.
21537 (struct char_span): New struct.
21538 (correction::overwrite): New method.
21539 (struct source_line): New struct.
21540 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21541 calls in terms of classes source_line and char_span, and
21542 correction::overwrite.
21543 (selftest::test_overlapped_fixit_printing_2): New function.
21544 (selftest::diagnostic_show_locus_c_tests): Call it.
21546 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21549 early if there is no lhs.
21551 2017-07-13 Martin Liska <mliska@suse.cz>
21553 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21554 (gen_reference_type_die): Likewise.
21555 * stor-layout.c: Remove Pascal-related comment.
21557 2017-07-13 Martin Liska <mliska@suse.cz>
21559 * opts.c (finish_options): Add quotes to error messages.
21560 (parse_sanitizer_options): Likewise.
21562 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21564 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21566 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21568 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21570 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21572 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21574 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21576 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21579 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21580 provides the hardware capability bits, define the macro
21581 __BUILTIN_CPU_SUPPORTS__.
21582 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21583 if GLIBC does not provide the hardware capability bits. Add a
21584 gcc_unreachable call if the built-in cpu function is neither
21585 __builtin_cpu_is nor __builtin_cpu_supports.
21586 (rs6000_get_function_versions_dispatcher): Change the warning
21587 that an old GLIBC is used which does not export the capability
21588 bits to be an error.
21589 * doc/extend.texi (target_clones attribute): Document the
21590 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21591 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21592 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21593 the macros defined by GCC if the newer GLIBC is available.
21595 2017-07-12 Jeff Law <law@redhat.com>
21597 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21598 remaining includes slightly.
21599 * config/riscv/riscv-builtins.c: Include profile-count.h.
21601 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21604 * config/avr/avr.c (avr_set_current_function): In diagnostic
21605 messages: Quote keywords and (parts of) identifiers.
21606 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21609 2017-07-12 Carl Love <cel@us.ibm.com>
21611 * config/rs6000/rs6000-c.c: Add support for built-in functions
21612 vector bool char vec_revb (vector bool char);
21613 vector bool short vec_revb (vector short char);
21614 vector bool int vec_revb (vector bool int);
21615 vector bool long long vec_revb (vector bool long long);
21616 * doc/extend.texi: Update the built-in documentation file for the
21617 new built-in functions.
21619 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21621 * config/s390/s390.md: Remove movcc splitter.
21623 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21625 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21626 load/store on condition.
21628 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21631 * config/avr/avr.c (avr_encode_section_info)
21632 [progmem && !TREE_READONLY]: Error if progmem object needs
21635 2017-07-11 Michael Collison <michael.collison@arm.com>
21637 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21640 2017-07-11 Carl Love <cel@us.ibm.com>
21642 * config/rs6000/rs6000-c.c: Add support for builtins
21643 vector unsigned int vec_parity_lsbb (vector signed int);
21644 vector unsigned int vec_parity_lsbb (vector unsigned int);
21645 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21646 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21647 vector unsigned long long vec_parity_lsbb (vector signed long long);
21648 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21649 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21650 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21651 * doc/extend.texi: Update the built-in documentation file for the
21652 new built-in functions.
21654 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21656 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21657 (layout::m_primary_loc): New field.
21658 (layout::layout): Initialize new field. Move location filtering
21659 logic from here to...
21660 (layout::maybe_add_location_range): ...this new method. Add
21661 support for filtering to just the lines already specified by other
21663 (layout::will_show_line_p): New method.
21664 (gcc_rich_location::add_location_if_nearby): New method.
21665 (selftest::test_add_location_if_nearby): New test function.
21666 (selftest::diagnostic_show_locus_c_tests): Call it.
21667 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21670 2017-07-11 Tom de Vries <tom@codesourcery.com>
21672 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21673 (bb_first_real_insn): New function.
21674 (nvptx_single): Add extra initialization of broadcasted condition
21677 2017-07-11 Nathan Sidwell <nathan@acm.org>
21679 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21681 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21683 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21684 Binutils doc as TEXI will mess them up.
21685 * doc/invoke.texi (AVR Options): Same here.
21687 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21689 * config/sparc/sparc.opt (mfix-ut700): New option.
21690 (mfix-gr712rc): Likewise.
21691 (sparc_fix_b2bst): New variable.
21692 * doc/invoke.texi (SPARC options): Document them.
21693 (ARM options): Fix warnings.
21694 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21695 instructions to prevent sequences that can trigger the store-store
21696 errata for certain LEON3FT processors.
21697 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21698 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21699 * config/sparc/sparc.md (fix_b2bst): New attribute.
21700 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21702 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21705 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21707 (*rsqrtsf2_sse): Ditto.
21709 (div<mode>3): Macroize insn from divdf3 and divsf3
21710 using MODEF mode iterator.
21712 2017-07-10 Martin Sebor <msebor@redhat.com>
21714 PR tree-optimization/80397
21715 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21716 instead of testing for equality to INTEGER_TYPE.
21718 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21720 * config.gcc: Remove uclibc from arc target spec.
21722 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21724 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21726 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21729 * lto-wrapper.c (remove_option): New function.
21730 (merge_and_complain): Merge PIC/PIE options more realistically.
21732 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21734 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21738 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21739 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21740 * config.in: Regenerate.
21741 * configure: Regenerate.
21742 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21743 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21744 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21745 (TARGET_GASISR_PROLOGUES): ...target mask.
21746 * common/config/avr/avr-common.c
21747 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21748 Set -mgas-isr-prologues.
21749 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21750 INSERT_PASS_BEFORE for it.
21751 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21752 * config/avr/avr.c (avr_option_override)
21753 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21754 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21755 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21756 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21757 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21759 (make_avr_pass_pre_proep): New function.
21760 (emit_push_sfr) <treg>: Add argument to function and use it
21761 instead of TMP_REG.
21762 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21763 and set machine->gasisr.yes.
21764 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21765 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21766 __gcc_isr.n_pushed to .L__stack_usage.
21767 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21768 (avr_asm_final_postscan_insn): ...this new static function.
21769 * config/avr/avr.h (machine_function)
21770 <is_no_gccisr, use_L__stack_usage>: New fields.
21771 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21772 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21773 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21774 (gasisr, *gasisr): New expander and insn.
21775 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21776 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21777 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21779 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21781 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21784 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21786 Move jump-tables out of .text again.
21789 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21790 (ASM_OUTPUT_ADDR_VEC): New function.
21791 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21792 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21793 INSN_ADDRESSes as asm comment.
21794 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21795 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21796 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21797 * config/avr/avr.md (*tablejump): Adjust comment.
21798 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21799 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21801 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21802 (avr_output_addr_vec): New proto.
21803 (avr_log_t) <insn_addresses>: New field.
21805 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21808 * config/i386/i386.c (ix86_function_arg_advance): Set
21809 outgoing_args_on_stack to true if there are outgoing arguments
21811 (ix86_function_arg): Likewise.
21812 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21813 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21814 * config/i386/i386.h (machine_function): Add
21815 outgoing_args_on_stack.
21817 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21819 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21820 supporting pthreds.
21821 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21823 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21825 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21826 (REAL_H): Remove $(MACHMODE_H).
21827 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21829 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21830 $(MACHMODE_H) and double-int.h.
21831 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21833 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21836 2017-07-07 Andrew Pinski <apinski@cavium.com>
21838 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21839 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21841 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21843 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21844 Add warning if GCC was not configured to link against a GLIBC that
21845 exports the hardware capability bits.
21846 (make_resolver_func): Make resolver function private and not a
21847 COMDAT function. Create the name with clone_function_name instead
21848 of make_unique_name.
21851 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21852 correct operand in doing the split.
21854 2017-07-07 Carl Love <cel@us.ibm.com>
21856 * config/rs6000/rs6000-c: Add support for built-in function
21857 vector unsigned short vec_pack_to_short_fp32 (vector float,
21859 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21860 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21861 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21862 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21863 (convert_4f32_8i16): Add define_expand.
21864 * doc/extend.texi: Update the built-in documentation file for the
21865 new built-in function.
21867 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21869 * config/sparc/m8.md: New file.
21870 * config/sparc/sparc.md: Include m8.md.
21872 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21874 * config/sparc/sparc.opt: New option -mvis4b.
21875 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21876 (sparc_option_override): Handle VIS4B.
21877 (enum sparc_builtins): Define
21878 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21879 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21880 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21881 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21882 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21883 (check_constant_argument): New function.
21884 (sparc_vis_init_builtins): Define builtins
21885 __builtin_vis_dictunpack{8,16,32},
21886 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21887 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21888 __builtin_vis_fpcmpde{8,16,32}shl and
21889 __builtin_vis_fpcmpur{8,16,32}shl.
21890 (sparc_expand_builtin): Check that the constant operands to
21891 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21892 constant and in range.
21893 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21895 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21896 (SPARC_IMM5_P): Likewise.
21897 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21898 (enabled): Handle vis4b.
21899 (UNSPEC_DICTUNPACK): New unspec.
21900 (UNSPEC_FPCMPSHL): Likewise.
21901 (UNSPEC_FPUCMPSHL): Likewise.
21902 (UNSPEC_FPCMPDESHL): Likewise.
21903 (UNSPEC_FPCMPURSHL): Likewise.
21904 (cpu_feature): New CPU feature `vis4b'.
21905 (dictunpack{8,16,32}): New insns.
21906 (FPCSMODE): New mode iterator.
21907 (fpcscond): New code iterator.
21908 (fpcsucond): Likewise.
21909 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21910 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21911 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21912 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21913 * config/sparc/constraints.md: Define constraints `q' for unsigned
21914 2-bit integer constants and `t' for unsigned 5-bit integer
21916 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21918 (imm5_operand_dictunpack16): Likewise.
21919 (imm5_operand_dictunpack32): Likewise.
21920 (imm2_operand): Likewise.
21921 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21922 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21923 ditunpack* and fpcmp*shl builtins.
21925 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21927 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21928 * config.in: Add HAVE_AS_SPARC6 define.
21929 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21931 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21933 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21934 (CPP_CPU_SPEC): Handle m8.
21935 (ASM_CPU_SPEC): Likewise.
21936 * config/sparc/sparc-opts.h (enum processor_type): Add
21938 * config/sparc/sparc.c (m8_costs): New struct.
21939 (sparc_option_override): Handle TARGET_CPU_m8.
21940 (sparc32_initialize_trampoline): Likewise.
21941 (sparc64_initialize_trampoline): Likewise.
21942 (sparc_issue_rate): Likewise.
21943 (sparc_register_move_cost): Likewise.
21944 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21945 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21946 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21947 (CPP_CPU_SPEC): Handle M8.
21948 (ASM_CPU_SPEC): Likewise.
21949 (AS_M8_FLAG): Define.
21950 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21951 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21952 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21954 * configure: Regenerate.
21955 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21958 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21960 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21962 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21963 ("*movdi_insn_sp32"): Do not set v3pipe.
21964 ("*movsi_insn"): Likewise.
21965 ("*movdi_insn_sp64"): Likewise.
21966 ("*movsf_insn"): Likewise.
21967 ("*movdf_insn_sp32"): Likewise.
21968 ("*movdf_insn_sp64"): Likewise.
21969 ("*zero_extendsidi2_insn_sp64"): Likewise.
21970 ("*sign_extendsidi2_insn"): Likewise.
21971 ("*mov<VM32:mode>_insn"): Likewise.
21972 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21973 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21974 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21975 ("<vlop:code><VL:mode>3"): Likewise.
21976 ("*not_<vlop:code><VL:mode>3"): Likewise.
21977 ("*nand<VL:mode>_vis"): Likewise.
21978 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21979 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21980 ("one_cmpl<VL:mode>2"): Likewise.
21981 ("faligndata<VM64:mode>_vis"): Likewise.
21982 ("alignaddrsi_vis"): Likewise.
21983 ("alignaddrdi_vis"): Likweise.
21984 ("alignaddrlsi_vis"): Likewise.
21985 ("alignaddrldi_vis"): Likewise.
21986 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21987 ("bmaskdi_vis"): Likewise.
21988 ("bmasksi_vis"): Likewise.
21989 ("bshuffle<VM64:mode>_vis"): Likewise.
21990 ("cmask8<P:mode>_vis"): Likewise.
21991 ("cmask16<P:mode>_vis"): Likewise.
21992 ("cmask32<P:mode>_vis"): Likewise.
21993 ("pdistn<P:mode>_vis"): Likewise.
21994 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21998 * config/sparc/sparc.md ("subtype"): New insn attribute.
21999 ("*wrgsr_sp64"): Set insn subtype.
22000 ("*rdgsr_sp64"): Likewise.
22001 ("alignaddrsi_vis"): Likewise.
22002 ("alignaddrdi_vis"): Likewise.
22003 ("alignaddrlsi_vis"): Likewise.
22004 ("alignaddrldi_vis"): Likewise.
22005 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22006 ("fexpand_vis"): Likewise.
22007 ("fpmerge_vis"): Likewise.
22008 ("faligndata<VM64:mode>_vis"): Likewise.
22009 ("bshuffle<VM64:mode>_vis"): Likewise.
22010 ("cmask8<P:mode>_vis"): Likewise.
22011 ("cmask16<P:mode>_vis"): Likewise.
22012 ("cmask32<P:mode>_vis"): Likewise.
22013 ("fchksm16_vis"): Likewise.
22014 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22015 ("fmean16_vis"): Likewise.
22016 ("fp<plusminus_insn>64_vis"): Likewise.
22017 ("<plusminus_insn>v8qi3"): Likewise.
22018 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22019 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22020 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22021 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22022 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22023 ("*movqi_insn"): Likewise.
22024 ("*movhi_insn"): Likewise.
22025 ("*movsi_insn"): Likewise.
22026 ("movsi_pic_gotdata_op"): Likewise.
22027 ("*movdi_insn_sp32"): Likewise.
22028 ("*movdi_insn_sp64"): Likewise.
22029 ("movdi_pic_gotdata_op"): Likewise.
22030 ("*movsf_insn"): Likewise.
22031 ("*movdf_insn_sp32"): Likewise.
22032 ("*movdf_insn_sp64"): Likewise.
22033 ("*zero_extendhisi2_insn"): Likewise.
22034 ("*zero_extendqihi2_insn"): Likewise.
22035 ("*zero_extendqisi2_insn"): Likewise.
22036 ("*zero_extendqidi2_insn"): Likewise.
22037 ("*zero_extendhidi2_insn"): Likewise.
22038 ("*zero_extendsidi2_insn_sp64"): Likewise.
22039 ("ldfsr"): Likewise.
22040 ("prefetch_64"): Likewise.
22041 ("prefetch_32"): Likewise.
22042 ("tie_ld32"): Likewise.
22043 ("tie_ld64"): Likewise.
22044 ("*tldo_ldub_sp32"): Likewise.
22045 ("*tldo_ldub1_sp32"): Likewise.
22046 ("*tldo_ldub2_sp32"): Likewise.
22047 ("*tldo_ldub_sp64"): Likewise.
22048 ("*tldo_ldub1_sp64"): Likewise.
22049 ("*tldo_ldub2_sp64"): Likewise.
22050 ("*tldo_ldub3_sp64"): Likewise.
22051 ("*tldo_lduh_sp32"): Likewise.
22052 ("*tldo_lduh1_sp32"): Likewise.
22053 ("*tldo_lduh_sp64"): Likewise.
22054 ("*tldo_lduh1_sp64"): Likewise.
22055 ("*tldo_lduh2_sp64"): Likewise.
22056 ("*tldo_lduw_sp32"): Likewise.
22057 ("*tldo_lduw_sp64"): Likewise.
22058 ("*tldo_lduw1_sp64"): Likewise.
22059 ("*tldo_ldx_sp64"): Likewise.
22060 ("*mov<VM32:mode>_insn"): Likewise.
22061 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22062 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22064 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22066 * config/sparc/sparc.md ("type"): New insn type viscmp.
22067 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22069 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22070 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22071 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22072 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22074 ("n7_vis_logical_11cycle"): Likewise.
22075 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22076 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22077 * config/sparc/niagara.md ("niag_vis"): Likewise.
22078 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22079 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22081 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22083 * config/sparc/sparc.md: New instruction type `bmask'.
22084 (bmaskdi_vis): Use the `bmask' type.
22085 (bmasksi_vis): Likewise.
22086 * config/sparc/ultra3.md (us3_array): Likewise.
22087 * config/sparc/niagara7.md (n7_array): Likewise.
22088 * config/sparc/niagara4.md (n4_array): Likewise.
22089 * config/sparc/niagara2.md (niag2_vis): Likewise.
22090 (niag3_vis): Likewise.
22091 * config/sparc/niagara.md (niag_vis): Likewise.
22093 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22095 * ipa-comdats.c: Remove optimize check from gate.
22096 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22097 for functions not optimized.
22098 (ipa_fn_summary_read): Skip optimize check.
22099 (ipa_fn_summary_write): Likewise.
22100 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22102 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22104 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22105 (check_callers): Give up on caller which is not optimized.
22106 (inline_small_functions): Likewise.
22107 (ipa_inline): Do not give up when not optimizing.
22108 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22109 away unoptimizes cdtors.
22110 (whole_program_function_and_variable_visibility): Do
22111 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22112 * ipa.c (process_references): Do not check optimize.
22113 (symbol_table::remove_unreachable_nodes): Update optimize check.
22114 (set_writeonly_bit): Update optimize check.
22115 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22116 (pass_ipa_single_use::gate): Remove.
22118 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22120 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22121 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22122 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22123 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22124 permute_load, permute_store, adjust_extract, adjust_splat,
22125 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22126 replace_swap_with_copy, dump_swap_insn_table,
22127 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22128 recombine_lvx_pattern, recombine_stvx_pattern,
22129 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22130 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22131 to file rs6000-p8swap.c.
22132 * config/rs6000/rs6000-p8swap.c: New file.
22133 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22134 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22135 and rs6000*-*-* targets.
22137 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22139 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22141 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22143 * lto-wrapper.c (merge_and_complain): Do not merge
22144 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22145 fsigned_zeros, ftrapping_math, fwrapv.
22146 (append_compiler_options): Do not track these options.
22147 (append_linker_options): Likewie
22149 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22151 * cgraphunit.c (cgraph_node::finalize_function): When
22152 !flag_toplevel_reorde set no_reorder flag.
22153 (varpool_node::finalize_decl): Likewise.
22154 (symbol_table::compile): Drop no toplevel reorder path.
22156 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22158 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22159 edges; zero probability is not better than uninitialized.
22161 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22163 * asan.h (asan_sanitize_allocas_p): Declare.
22164 * asan.c (asan_sanitize_allocas_p): New function.
22165 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22166 (handle_builtin_alloca): Likewise.
22167 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22168 if !asan_sanitize_allocas_p.
22169 * params.def (asan-instrument-allocas): Add new option.
22170 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22171 * opts.c (common_handle_option): Disable allocas sanitization for
22174 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22176 * asan.c: Include gimple-fold.h.
22177 (get_last_alloca_addr): New function.
22178 (handle_builtin_stackrestore): Likewise.
22179 (handle_builtin_alloca): Likewise.
22180 (asan_emit_allocas_unpoison): Likewise.
22181 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22182 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22183 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22184 (instrument_builtin_call): Pass gimple iterator to
22185 get_mem_refs_of_builtin_call.
22186 (last_alloca_addr): New global.
22187 * asan.h (asan_emit_allocas_unpoison): Declare.
22188 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22189 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22190 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22191 if function calls alloca.
22192 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22193 * gimple-fold.h (replace_call_with_value): Declare.
22194 * internal-fn.c: Include asan.h.
22195 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22196 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22198 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22200 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22201 (C_SELFTEST_FLAGS): New.
22202 (CPP_SELFTEST_FLAGS): New.
22203 (SELFTEST_DEPS): New, from deps of s-selftest.
22204 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22205 (CPP_SELFTEST_DEPS): New.
22206 (selftest): Add dependency on s-selftest-c++.
22207 (s-selftest): Rename to...
22208 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22209 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22210 than SELFTEST_FLAGS.
22211 (selftest-gdb): Rename to...
22212 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22214 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22215 (selftest-valgrind): Rename to...
22216 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22218 (selftest-valgrind): Reintroduce as an alias for
22219 selftest-c-valgrind.
22220 (s-selftest-c++): New.
22221 (selftest-c++-gdb): New.
22222 (selftest-c++-valgrind): New.
22224 2017-07-06 Olivier Hainque <hainque@adacore.com>
22226 * gcc.c (process_command): When deciding if undefined variables
22227 should be ignored when processing specs, accept "gcc -v" as well.
22229 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22231 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22232 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22234 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22236 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22237 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22238 * config/arm/arm-tables.opt: Regenerate.
22239 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22241 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22243 2017-07-06 Carl Love <cel@us.ibm.com>
22245 * ChangeLog: Clean up from mid air collision
22247 2017-07-06 Carl Love <cel@us.ibm.com>
22249 * config/rs6000/rs6000-c.c: Add support for built-in functions
22250 vector signed int vec_subc (vector signed int, vector signed int);
22251 vector signed __int128 vec_subc (vector signed __int128,
22252 vector signed __int128);
22253 vector unsigned __int128 vec_subc (vector unsigned __int128,
22254 vector unsigned __int128);
22255 vector signed int vec_sube (vector signed int, vector signed int,
22256 vector signed int);
22257 vector unsigned int vec_sube (vector unsigned int,
22258 vector unsigned int,
22259 vector unsigned int);
22260 vector signed __int128 vec_sube (vector signed __int128,
22261 vector signed __int128,
22262 vector signed__int128);
22263 vector unsigned __int128 vec_sube (vector unsigned __int128,
22264 vector unsigned __int128,
22265 vector unsigned __int128);
22266 vector signed int vec_subec (vector signed int, vector signed int,
22267 vector signed int);
22268 vector unsigned int vec_subec (vector unsigned int,
22269 vector unsigned int,
22270 vector unsigned int);
22271 vector signed __int128 vec_subec (vector signed __int128,
22272 vector signed __int128,
22273 vector signed__int128);
22274 vector unsigned __int128 vec_subec (vector unsigned __int128,
22275 vector unsigned __int128,
22276 vector unsigned __int128);
22277 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22278 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22279 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22280 BU_ALTIVEC_OVERLOAD_X definitions.
22281 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22282 * doc/extend.texi: Update the built-in documentation file for the new
22283 built-in functions.
22285 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22288 * diagnostic-show-locus.c (layout::layout): Use start and finish
22289 spelling location for the start and finish of each range.
22290 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22291 Add unused aspect param.
22292 * input.c (expand_location_1): Add "aspect" param, and use it
22293 to access the correct part of the location.
22294 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22296 (expand_location_to_spelling_point): Likewise.
22297 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22298 param, and pass it to expand_location_1.
22300 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22302 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22303 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22304 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22305 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22306 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22307 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22308 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22309 _mm_maskz_getmant_ss): New intrinsics.
22310 (__builtin_ia32_getexpss128_mask): Changed to ...
22311 __builtin_ia32_getexpss128_round ... this.
22312 (__builtin_ia32_getexpsd128_mask): Changed to ...
22313 __builtin_ia32_getexpsd128_round ... this.
22314 * config/i386/i386-builtin-types.def
22315 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22316 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22317 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22318 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22319 __builtin_ia32_getmantss_mask_round): New builtins.
22320 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22321 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22322 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22323 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22324 * config/i386/sse.md
22325 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22326 avx512f_sgetexp<mode><mask_scalar_name>
22327 <round_saeonly_scalar_name> ... this.
22328 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22329 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22330 vgetexp<ssescalarmodesuffix>
22331 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22332 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22333 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22334 avx512f_vgetmant<mode><mask_scalar_name>
22335 <round_saeonly_scalar_name> ... this.
22336 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22337 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22338 vgetmant<ssescalarmodesuffix>
22339 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22340 %0<mask_scalar_operand4>, %1, %2
22341 <round_saeonly_scalar_mask_op4>, %3} ... this.
22342 * config/i386/subst.md (mask_scalar_operand4,
22343 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22344 round_saeonly_scalar_nimm_predicate): New subst attributes.
22346 2017-07-06 Julia Koval <julia.koval@intel.com>
22348 * config/i386/i386.c (ix86_erase_embedded_rounding):
22349 Remove code for old rounding pattern.
22351 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22353 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22355 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22357 * doc/sourcebuild.texi (Test Directives, Variants of
22358 dg-require-support): Add documentation for dg-require-stack-check.
22360 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22362 * config/i386/subst.md (mask_scalar, round_scalar,
22363 round_saeonly_scalar): New meta-templates.
22364 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22365 round_scalar_mask_operand3, round_scalar_mask_op3,
22366 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22367 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22368 round_saeonly_scalar_constraint,
22369 round_saeonly_scalar_prefix): New subst attribute.
22370 * config/i386/sse.md
22371 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22372 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22373 <round_scalar_name> ... this.
22374 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22375 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22376 <round_scalar_name> ... this.
22377 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22378 <sse>_vm<code><mode>3<mask_scalar_name>
22379 <round_saeonly_scalar_name> ... this.
22380 (v<plusminus_mnemonic><ssescalarmodesuffix>
22381 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22382 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22383 v<plusminus_mnemonic><ssescalarmodesuffix>
22384 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22385 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22386 (v<multdiv_mnemonic><ssescalarmodesuffix>
22387 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22388 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22389 v<multdiv_mnemonic><ssescalarmodesuffix>
22390 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22391 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22392 (v<maxmin_float><ssescalarmodesuffix>
22393 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22394 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22395 v<maxmin_float><ssescalarmodesuffix>
22396 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22397 %0<mask_scalar_operand3>, %1, %<iptr>2
22398 <round_saeonly_scalar_mask_op3>} ... this.
22400 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22402 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22403 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22405 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22406 Alan Hayward <alan.hayward@arm.com>
22407 David Sherwood <david.sherwood@arm.com>
22409 * combine.c (simplify_if_then_else): Remove "enum" before
22411 * compare-elim.c (can_eliminate_compare): Likewise.
22412 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22414 (aarch64_lookup_simd_builtin_type): Likewise.
22415 (aarch64_simd_builtin_type): Likewise.
22416 (aarch64_init_simd_builtin_types): Likewise.
22417 (aarch64_simd_expand_args): Likewise.
22418 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22420 (aarch64_reverse_mask): Likewise.
22421 (aarch64_simd_emit_reg_reg_move): Likewise.
22422 (aarch64_gen_adjusted_ldpstp): Likewise.
22423 (aarch64_ccmp_mode_to_code): Likewise.
22424 (aarch64_operands_ok_for_ldpstp): Likewise.
22425 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22426 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22428 (aarch64_min_divisions_for_recip_mul): Likewise.
22429 (aarch64_reassociation_width): Likewise.
22430 (aarch64_get_condition_code_1): Likewise.
22431 (aarch64_simd_emit_reg_reg_move): Likewise.
22432 (aarch64_simd_attr_length_rglist): Likewise.
22433 (aarch64_reverse_mask): Likewise.
22434 (aarch64_operands_ok_for_ldpstp): Likewise.
22435 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22436 (aarch64_gen_adjusted_ldpstp): Likewise.
22437 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22439 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22440 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22441 (arm_lookup_simd_builtin_type): Likewise.
22442 (arm_simd_builtin_type): Likewise.
22443 (arm_init_simd_builtin_types): Likewise.
22444 (arm_expand_builtin_args): Likewise.
22445 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22446 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22447 (ft32_setup_incoming_varargs): Likewise.
22448 (ft32_function_arg): Likewise.
22449 (ft32_function_arg_advance): Likewise.
22450 (ft32_pass_by_reference): Likewise.
22451 (ft32_arg_partial_bytes): Likewise.
22452 (ft32_valid_pointer_mode): Likewise.
22453 (ft32_addr_space_pointer_mode): Likewise.
22454 (ft32_addr_space_legitimate_address_p): Likewise.
22455 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22457 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22458 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22459 (iamcu_alignment): Likewise.
22460 (canonicalize_vector_int_perm): Likewise.
22461 (ix86_noce_conversion_profitable_p): Likewise.
22462 (ix86_mpx_bound_mode): Likewise.
22463 (ix86_operands_ok_for_move_multiple): Likewise.
22464 * config/microblaze/microblaze-protos.h
22465 (microblaze_expand_conditional_branch_reg): Likewise.
22466 * config/microblaze/microblaze.c
22467 (microblaze_expand_conditional_branch_reg): Likewise.
22468 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22470 (rs6000_reassociation_width): Likewise.
22471 (rs6000_invalid_binary_op): Likewise.
22472 (fusion_p9_p): Likewise.
22473 (emit_fusion_p9_load): Likewise.
22474 (emit_fusion_p9_store): Likewise.
22475 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22477 (riscv_hard_regno_mode_ok_p): Likewise.
22478 (riscv_address_insns): Likewise.
22479 (riscv_split_symbol): Likewise.
22480 (riscv_legitimize_move): Likewise.
22481 (riscv_function_value): Likewise.
22482 (riscv_hard_regno_nregs): Likewise.
22483 (riscv_expand_builtin): Likewise.
22484 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22485 (riscv_build_integer): Likewise.
22486 (riscv_split_integer): Likewise.
22487 (riscv_legitimate_constant_p): Likewise.
22488 (riscv_cannot_force_const_mem): Likewise.
22489 (riscv_regno_mode_ok_for_base_p): Likewise.
22490 (riscv_valid_base_register_p): Likewise.
22491 (riscv_valid_offset_p): Likewise.
22492 (riscv_valid_lo_sum_p): Likewise.
22493 (riscv_classify_address): Likewise.
22494 (riscv_legitimate_address_p): Likewise.
22495 (riscv_address_insns): Likewise.
22496 (riscv_load_store_insns): Likewise.
22497 (riscv_force_binary): Likewise.
22498 (riscv_split_symbol): Likewise.
22499 (riscv_force_address): Likewise.
22500 (riscv_legitimize_address): Likewise.
22501 (riscv_move_integer): Likewise.
22502 (riscv_legitimize_const_move): Likewise.
22503 (riscv_legitimize_move): Likewise.
22504 (riscv_address_cost): Likewise.
22505 (riscv_subword): Likewise.
22506 (riscv_output_move): Likewise.
22507 (riscv_canonicalize_int_order_test): Likewise.
22508 (riscv_emit_int_order_test): Likewise.
22509 (riscv_function_arg_boundary): Likewise.
22510 (riscv_pass_mode_in_fpr_p): Likewise.
22511 (riscv_pass_fpr_single): Likewise.
22512 (riscv_pass_fpr_pair): Likewise.
22513 (riscv_get_arg_info): Likewise.
22514 (riscv_function_arg): Likewise.
22515 (riscv_function_arg_advance): Likewise.
22516 (riscv_arg_partial_bytes): Likewise.
22517 (riscv_function_value): Likewise.
22518 (riscv_pass_by_reference): Likewise.
22519 (riscv_setup_incoming_varargs): Likewise.
22520 (riscv_print_operand): Likewise.
22521 (riscv_elf_select_rtx_section): Likewise.
22522 (riscv_save_restore_reg): Likewise.
22523 (riscv_for_each_saved_reg): Likewise.
22524 (riscv_register_move_cost): Likewise.
22525 (riscv_hard_regno_mode_ok_p): Likewise.
22526 (riscv_hard_regno_nregs): Likewise.
22527 (riscv_class_max_nregs): Likewise.
22528 (riscv_memory_move_cost): Likewise.
22529 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22530 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22531 (rl78_addr_space_address_mode): Likewise.
22532 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22534 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22535 (rs6000_reassociation_width): Likewise.
22536 (rs6000_invalid_binary_op): Likewise.
22537 (fusion_p9_p): Likewise.
22538 (emit_fusion_p9_load): Likewise.
22539 (emit_fusion_p9_store): Likewise.
22540 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22541 (ok_for_simple_move_operands): Likewise.
22542 (ok_for_simple_move_strict_operands): Likewise.
22543 (ok_for_simple_arith_logic_operands): Likewise.
22544 (visium_legitimize_reload_address): Likewise.
22545 (visium_select_cc_mode): Likewise.
22546 (output_cbranch): Likewise.
22547 (visium_split_double_move): Likewise.
22548 (visium_expand_copysign): Likewise.
22549 (visium_expand_int_cstore): Likewise.
22550 (visium_expand_fp_cstore): Likewise.
22551 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22552 (visium_function_arg): Likewise.
22553 (visium_function_arg_advance): Likewise.
22554 (visium_libcall_value): Likewise.
22555 (visium_setup_incoming_varargs): Likewise.
22556 (visium_legitimate_constant_p): Likewise.
22557 (visium_legitimate_address_p): Likewise.
22558 (visium_legitimize_address): Likewise.
22559 (visium_secondary_reload): Likewise.
22560 (visium_register_move_cost): Likewise.
22561 (visium_memory_move_cost): Likewise.
22562 (prepare_move_operands): Likewise.
22563 (ok_for_simple_move_operands): Likewise.
22564 (ok_for_simple_move_strict_operands): Likewise.
22565 (ok_for_simple_arith_logic_operands): Likewise.
22566 (visium_function_value_1): Likewise.
22567 (rtx_ok_for_offset_p): Likewise.
22568 (visium_legitimize_reload_address): Likewise.
22569 (visium_split_double_move): Likewise.
22570 (visium_expand_copysign): Likewise.
22571 (visium_expand_int_cstore): Likewise.
22572 (visium_expand_fp_cstore): Likewise.
22573 (visium_split_cstore): Likewise.
22574 (visium_select_cc_mode): Likewise.
22575 (visium_split_cbranch): Likewise.
22576 (output_cbranch): Likewise.
22577 (visium_print_operand_address): Likewise.
22578 * expmed.c (flip_storage_order): Likewise.
22579 * expmed.h (emit_cstore): Likewise.
22580 (flip_storage_order): Likewise.
22581 * genrecog.c (validate_pattern): Likewise.
22582 * hsa-gen.c (gen_hsa_addr): Likewise.
22583 * internal-fn.c (expand_arith_overflow): Likewise.
22584 * ira-color.c (allocno_copy_cost_saving): Likewise.
22585 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22586 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22587 (process_invariant_for_inheritance): Likewise.
22588 * lra-eliminations.c (move_plus_up): Likewise.
22589 * omp-low.c (lower_oacc_reductions): Likewise.
22590 * simplify-rtx.c (simplify_subreg): Likewise.
22591 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22592 (TARGET_CHKP_BOUND_MODE): Likewise..
22593 * targhooks.c (default_chkp_bound_mode): Likewise.
22594 (default_setup_incoming_vararg_bounds): Likewise.
22595 * targhooks.h (default_chkp_bound_mode): Likewise.
22596 (default_setup_incoming_vararg_bounds): Likewise.
22597 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22598 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22599 (have_whole_vector_shift): Likewise.
22600 * tree-vect-stmts.c (vectorizable_load): Likewise.
22601 * doc/tm.texi: Regenerate.
22603 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22605 Graceful degrade if Binutils PR21472 is not available.
22608 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22609 .rodata in flash test fails.
22610 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22611 * confgure: Regenerate.
22612 * config.in: Regenerate.
22613 * config/avr/avr.c (avr_asm_named_section)
22614 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22615 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22616 (avr_asm_init_sections): Same.
22618 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22620 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22621 (fma<VH:mode>4_intrinsic): Likewise.
22622 (*fmsub<VCVTF:mode>4): Likewise.
22623 (*fmsub<VH:mode>4_intrinsic): Likewise.
22625 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22628 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22629 Don't depend on "optimize > 0".
22630 (out_movhi_r_mr, out_movqi_mr_r): Same.
22631 (out_movhi_mr_r, out_movqi_r_mr): Same.
22632 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22633 io_address_operand on "optimize > 0".
22635 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22637 * tree-loop-distribution.c: Add general explanantion on the pass.
22638 (generate_loops_for_partition): Mark distributed loop.
22639 (pg_add_dependence_edges): New parameter. Handle alias data
22640 dependence specially and record it in the parameter if asked.
22641 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22642 (init_partition_graph_vertices, add_partition_graph_edge): New.
22643 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22644 (free_partition_graph_vdata, build_partition_graph): New.
22645 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22646 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22647 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22648 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22649 (version_for_distribution_p, finalize_partitions): New.
22650 (distribute_loop): Handle alias data dependence specially. Factor
22651 out loop fusion code as functions and call these functions.
22653 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22655 * tree-loop-distribution.c (classify_partition): New parameter and
22656 better handle reduction statement.
22657 (rdg_build_partitions): Revise comment.
22658 (distribute_loop): Compute statements in all partitions and pass it
22659 to classify_partition.
22661 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22663 * tree-loop-distribution.c (enum partition_type): New.
22664 (struct partition): New field type.
22665 (partition_merge_into): Add parameter. Update partition type.
22666 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22667 (build_rdg_partition_for_vertex): Compute partition type.
22668 (rdg_build_partitions): Dump partition type.
22669 (distribute_loop): Update calls to partition_merge_into.
22671 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22673 * tree-loop-distribution.c (struct ddr_hasher): New.
22674 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22676 (classify_partition): Call get_data_dependence.
22677 (pg_add_dependence_edges): Ditto.
22678 (distribute_loop): Release data dependence hash table.
22680 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22682 * tree-loop-distribution.c (ref_base_address): Delete.
22683 (similar_memory_accesses): Rename ...
22684 (share_memory_accesses): ... to this. Check if partitions access
22685 the same memory reference.
22686 (distribute_loop): Call share_memory_accesses.
22688 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22690 * tree-loop-distribution.c (struct partition): New field recording
22691 its data reference.
22692 (partition_alloc, partition_free): Init and release data refs.
22693 (partition_merge_into): Merge data refs.
22694 (build_rdg_partition_for_vertex): Collect data refs for partition.
22695 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22697 (distribute_loop): Remve data refs from vertice data of partition
22700 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22702 * tree-loop-distribution.c (params.h): Include header file.
22703 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22704 (datarefs_vec): New global var.
22705 (create_rdg_vertices): Use datarefs_vec directly.
22706 (free_rdg): Don't free data references.
22707 (build_rdg): Update use. Don't free data references.
22708 (distribute_loop): Compute global variable for data references.
22709 Bail out if there are too many data references.
22711 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22713 * tree-loop-distribution.c (loop_nest): New global var.
22714 (build_rdg): Use loop directly, rather than loop nest.
22715 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22717 (distribute_loop): Compute global variable loop nest. Update use.
22719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22721 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22722 (partition_merge_into): New parameter. Dump reason for fusion.
22723 (distribute_loop): Update use of partition_merge_into.
22725 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22727 * tree-loop-distribution.c (bb_top_order_index): New.
22728 (bb_top_order_index_size, bb_top_order_cmp): New.
22729 (stmts_from_loop): Use topological order.
22730 (pass_loop_distribution::execute): Compute and release topological
22731 order for basic blocks.
22733 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22735 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22740 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22741 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22742 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22743 * internal-fn.def (LOOP_DIST_ALIAS): New.
22744 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22745 (fold_loop_internal_call): ... this.
22746 (vect_loop_dist_alias_call): New function.
22747 (set_uid_loop_bbs): Call fold_loop_internal_call.
22748 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22751 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22754 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22755 Require dead FLAGS_REG at the beginning of a peephole.
22757 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22760 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22761 arguments in the call to __builtin_ia32_sbb_u32.
22762 (_subborrow_u64): Swap _X and _Y arguments in the call to
22763 __builtin_ia32_sbb_u64.
22765 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22768 * tree-vrp.c (compare_assert_loc): Turn into a function template
22769 with stable template parameter. Only test if a->e is NULL,
22770 !a->e == !b->e has been verified already. Use e == NULL or
22771 e != NULL instead of e or ! e tests. If stable is true, don't use
22772 iterative_hash_expr, on the other side allow a or b or both NULL
22773 and sort the NULLs last.
22774 (process_assert_insertions): Sort using compare_assert_loc<false>
22775 instead of compare_assert_loc, later sort using
22776 compare_assert_loc<true> before calling process_assert_insertions_for
22777 in a loop. Use break instead of continue once seen NULL pointer.
22779 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22781 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22782 Cortex-R7 and Cortex-R8 processors.
22784 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22786 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22787 uninitialized while src is not.
22789 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22791 * common/config/arm/arm-common.c: Adjust include path for
22793 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22794 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22795 (arm-cpu-data.h): Likewise.
22796 (arm-cpu-cdata.h): Likewise.
22797 * config/arm/arm-cpu.h: Delete.
22798 * config/arm/arm-cpu-cdata.h: Delete.
22799 * config/arm/arm-cpu-data.h: Delete.
22801 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22803 * config/arm/arm-cpus.in (cortex-a55): New.
22804 (cortex-a75): Likewise.
22805 (cortex-a75.cortex-a55): Likewise.
22806 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22808 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22809 * config/arm/arm-cpu-cdata.h: Regenerate.
22810 * config/arm/arm-cpu-data.h: Regenerate.
22811 * config/arm/arm-cpu.h: Regenerate.
22812 * config/arm/arm-tables.opt: Regenerate.
22813 * config/arm/arm-tune.md: Regenerate.
22815 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22817 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22819 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22821 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22824 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22826 PR tree-optimization/81292
22827 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22828 full_string_p, also call adjust_related_strinfos if the adjustment
22829 is simple, otherwise invalidate related strinfos.
22831 2017-07-04 Martin Liska <mliska@suse.cz>
22834 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22835 newly created variable as DECL_IGNORED_P.
22837 2017-07-04 Martin Liska <mliska@suse.cz>
22840 * ipa-inline.c (inline_small_functions):
22841 Use xstrdup_for_dump.
22843 2017-07-04 Tom de Vries <tom@codesourcery.com>
22845 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22847 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22850 * config/darwin.c (darwin_function_switched_text_sections):
22853 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22855 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22857 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22859 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22861 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22863 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22864 min_profitable_iters, and th as inclusive lower bounds.
22865 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22866 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22867 for min_profitable_iters and min_profitable_estimate.
22868 (vect_transform_loop): Treat th as an inclusive lower bound.
22869 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22871 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22874 * config/darwin.c (darwin_function_switched_text_sections):
22875 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22876 in two pieces, and suppress the use of buf.
22878 2017-07-03 Nathan Sidwell <nathan@acm.org>
22880 * hash-table.h (hash_table_mod1): Fix indentation.
22882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22884 PR middle-end/81290
22885 * predict.c (force_edge_cold): Be more careful about propagation
22887 * profile-count.h (profile_probability::guessed,
22888 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22890 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22892 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22894 * doc/invoke.texi (rcpc architecture extension): Document it.
22896 2017-07-03 Richard Biener <rguenther@suse.de>
22898 PR tree-optimization/60510
22899 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22900 the scalar reduction PHI and use it.
22901 (vectorizable_reduction): Properly guard the single_defuse_cycle
22902 path for non-SLP reduction chains where we cannot use it.
22903 Rework reduc_def/index and vector type deduction. Rework
22904 vector operand gathering during reduction op code-gen.
22905 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22906 chains dissolve the chain and leave it to non-SLP reduction
22909 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22911 * tree-data-ref.h (dr_alignment): Declare.
22912 * tree-data-ref.c (dr_alignment): New function.
22913 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22914 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22916 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22920 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22921 and base_misalignment fields.
22922 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22923 * tree-data-ref.c: Include builtins.h.
22924 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22925 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22926 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22927 * tree-vect-data-refs.c: Include tree-cfg.h.
22928 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22929 fields instead of calculating an alignment here.
22930 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22931 innermost_loop_behavior fields.
22933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22935 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22937 (DR_STEP_ALIGNMENT): New macro.
22938 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22939 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22940 (create_data_ref): Print it.
22941 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22942 to tell whether the step preserves vector (mis)alignment.
22943 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22944 Move the check for an integer step and generalise to all INTEGER_CST.
22945 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22946 Print the outer step alignment.
22948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22950 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22951 with offset_alignment.
22952 (DR_ALIGNED_TO): Delete.
22953 (DR_OFFSET_ALIGNMENT): New macro.
22954 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22955 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22956 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22957 (create_data_ref): Likewise.
22958 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22959 (vect_analyze_data_refs): Likewise.
22960 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22961 creating dummy innermost behavior.
22963 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22965 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22966 with a "innermost_loop_behavior *" and refeence tree.
22967 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22968 (create_data_ref): Update call accordingly.
22969 * tree-predcom.c (find_looparound_phi): Likewise.
22971 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22973 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22974 fields with dr_wrt_vec_loop.
22975 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22976 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22977 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22978 (vect_dr_behavior): New function.
22979 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22980 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22981 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22982 track whether the step preserves the misalignment.
22983 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22984 Use vect_dr_behavior.
22985 (vect_setup_realignment): Update call accordingly.
22986 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22987 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22988 call to vect_create_addr_base_for_vector_ref.
22989 (vect_create_cond_for_align_checks): Likewise.
22990 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22991 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22992 (vect_recog_mask_conversion_pattern): Likewise.
22993 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22994 (new_stmt_vec_info): Remove redundant zeroing.
22996 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22998 * common/config/arm/arm-common.c (arm_be8_option): New function.
22999 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23000 (ISA_ARMv6): Add isa_bit_be8.
23001 * config/arm/arm.h (arm_be8_option): Add prototype.
23002 (BE8_SPEC_FUNCTION): New define.
23003 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23004 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23005 (mlittle-endian): Similarly.
23006 (mbe8, mbe32): New options.
23007 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23008 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23010 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23012 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23014 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23016 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23017 (cleanup_tree_cfg_bb): Use it.
23018 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23020 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23022 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23025 * loop-doloop.c (add_test): Update profile.
23027 2017-07-03 Martin Liska <mliska@suse.cz>
23030 * sanopt.c (rewrite_usage_of_param): New function.
23031 (sanitize_rewrite_addressable_params): Likewise.
23032 (pass_sanopt::execute): Call rewrite_usage_of_param.
23034 2017-07-03 Richard Biener <rguenther@suse.de>
23036 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23037 back to using VIEW_CONVERT_EXPR.
23039 2017-07-03 Martin Liska <mliska@suse.cz>
23042 * doc/extend.texi: Document when a resolver function is
23043 generated for target_clones.
23045 2017-07-03 Martin Liska <mliska@suse.cz>
23047 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23048 and shadow memory of auto variables which are subject of
23049 use-after-scope sanitization.
23050 (asan_expand_mark_ifn): Add do set only when is_poison.
23052 2016-07-03 Richard Biener <rguenther@suse.de>
23054 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23056 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23057 (vectorizable_reduction): Perform reduction PHI creation when
23058 visiting a reduction PHI and adjust and simplify code generation
23059 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23060 (vect_transform_loop): Visit reduction PHIs.
23061 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23062 defs into the SLP tree.
23063 (vect_build_slp_tree): Reduction defs terminate the recursion.
23064 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23066 (vect_get_vec_defs_for_stmt_copy): Export.
23067 (vect_get_vec_defs): Likewise.
23068 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23070 (vect_get_vec_defs_for_stmt_copy): Declare.
23071 (vect_get_vec_defs): Likewise.
23073 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23075 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23076 parameter with a "loop" parameter and use it instead of the
23077 loop containing DR_STMT. Don't check simple_iv when doing
23078 BB analysis. Describe the two analysis modes in the comment.
23080 2017-07-03 Tom de Vries <tom@codesourcery.com>
23082 PR tree-optimization/69468
23083 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23084 (find_same_succ_bb): Handle ignore_edge_flags.
23086 2017-07-03 Tom de Vries <tom@codesourcery.com>
23088 PR tree-optimization/81192
23089 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23091 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23093 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23095 2017-07-03 Tom de Vries <tom@codesourcery.com>
23097 PR tree-optimization/81192
23098 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23099 BB_SAME_SUCC (bb) == NULL.
23101 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23103 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23106 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23108 * dumpfile.c: Include profile-count.h
23109 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23111 (insert_cond_bb): Update profile.
23112 * tree-cfg.h (insert_cond_bb): Update prototype.
23113 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23114 * tree-dump.c: Do not include tree-cfg.
23116 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23118 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23120 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23122 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23125 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23127 * tree-complex.c (expand_complex_div_wide): update profile.
23129 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23130 Alan Hayward <alan.hayward@arm.com>
23131 David Sherwood <david.sherwood@arm.com>
23133 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23134 (CORETYPES_H): New define.
23135 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23136 (insn-modes-inline.h, s-modes-inline-h): New rules.
23137 (generated_files): Add insn-modes-inline.h.
23138 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23139 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23140 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23141 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23142 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23143 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23144 (build/gencodes.o, build/genconditions.o): Likewise.
23145 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23146 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23147 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23148 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23149 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23150 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23151 * coretypes.h: Include everything up to real.h for generators.
23152 Include insn-modes.h first. Include wide-int-print.h after
23153 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23154 * machmode.h: Don't include insn-modes.h here.
23155 * function-tests.c: Remove includes of signop.h, machmode.h,
23156 double-int.h and wide-int.h.
23158 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23160 * optc-save-gen.awk: Likewise.
23161 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23162 * godump.c: Remove include of wide-int-print.h.
23163 * pretty-print.h: Likewise.
23164 * wide-int-print.cc: Likewise.
23165 * wide-int.cc: Likewise.
23166 * hash-map-tests.c: Remove include of signop.h.
23167 * hash-set-tests.c: Likewise.
23168 * rtl-tests.c: Likewise.
23169 * mkconfig.sh: Remove include of machmode.h.
23170 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23172 (emit_insn_modes_inline_h): ...this new function. Emit the code
23173 into an insn-modes-inline.h header file, adding appropriate
23174 include guards and end comments.
23175 (emit_insn_modes_c_header): Remove include of machmode.h.
23176 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23178 (main): Handle -i flag and call emit_insn_modes_inline_h when
23181 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23183 * tree-ssa-strlen.c (strinfo): Rename the length field to
23184 nonzero_chars. Add a full_string_p field.
23185 (compare_nonzero_chars, zero_length_string_p): New functions.
23186 (get_addr_stridx): Add an offset_out parameter.
23187 Use compare_nonzero_chars.
23188 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23189 (new_strinfo): Update after above changes to strinfo.
23190 (set_endptr_and_length): Set full_string_p.
23191 (get_string_length): Update after above changes to strinfo.
23192 (unshare_strinfo): Update call to new_strinfo.
23193 (maybe_invalidate): Likewise.
23194 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23195 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23196 as a uhwi instead of an shwi. Update after above changes to
23197 strinfo and new_strinfo.
23198 (zero_length_string): Assert that chainsi contains full strings.
23199 Use zero_length_string_p. Update call to new_strinfo.
23200 (adjust_related_strinfos): Update after above changes to strinfo.
23201 Copy full_string_p from origsi.
23202 (adjust_last_stmt): Use zero_length_string_p.
23203 (handle_builtin_strlen): Update after above changes to strinfo and
23204 new_strinfo. Install the lhs as the string length if the previous
23205 entry didn't describe a full string.
23206 (handle_builtin_strchr): Update after above changes to strinfo
23208 (handle_builtin_strcpy): Likewise.
23209 (handle_builtin_strcat): Likewise.
23210 (handle_builtin_malloc): Likewise.
23211 (handle_pointer_plus): Likewise.
23212 (handle_builtin_memcpy): Likewise. Track nonzero characters
23213 that aren't necessarily followed by a nul terminator.
23214 (handle_char_store): Likewise.
23216 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23218 PR tree-optimization/80769
23219 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23220 for malloc and calloc. Document the new invariant that all related
23221 strinfos have delayed lengths or none do.
23222 (verify_related_strinfos): Move earlier in file.
23223 (set_endptr_and_length): New function, split out from...
23224 (get_string_length): ...here. Also set the lengths of related
23226 (zero_length_string): Assert that chainsi has known (rather than
23228 (adjust_related_strinfos): Likewise.
23230 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23232 PR tree-optimization/81136
23233 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23234 assert that two references with the same misalignment have the same
23235 compile-time misalignment if those compile-time misalignments
23238 2017-07-01 Andi Kleen <ak@linux.intel.com>
23240 * print-tree.c (print_node): Print all attributes.
23242 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23244 * cfg.c (scale_bbs_frequencies): New function.
23245 * cfg.h (scale_bbs_frequencies): Declare it.
23246 * cfgloopanal.c (single_likely_exit): Cleanup.
23247 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23249 (scale_loop_profile): Likewise.
23250 (loop_version): Likewise.
23251 (create_empty_loop_on_edge): Update.
23252 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23253 scale_loop_frequencies, scale_loop_profile, loopify,
23254 loop_version): Update prototypes.
23255 * modulo-sched.c (sms_schedule): Update.
23256 * predict.c (unlikely_executed_edge_p): Also check probability.
23257 (probably_never_executed_edge_p): Fix typo.
23258 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23259 * tree-parloops.c (gen_parallel_loop): Update.
23260 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23261 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23262 * tree-ssa-loop-split.c (split_loop): Update.
23263 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23264 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23265 (vect_loop_versioning): Update.
23266 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23268 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23270 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23272 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23274 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23275 to keep profile consistent.
23277 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23279 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23280 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23281 * profile-count.h (max_safe_multiplier): Make unsigned.
23282 (profile_count::guessed_zero): New.
23284 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23286 * bb-reorder.c (fix_up_crossing_landing_pad,
23287 fix_crossing_conditional_branches): Use make_single_succ_edge
23288 to keep profile consistent.
23290 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23292 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23295 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23298 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23299 the right scopes, make sure cond_jump isn't preserved between multiple
23300 iterations. Search for fallthru edge whenever there are 3+ edges and
23301 use find_fallthru_edge for it.
23303 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23305 Patch by Alexander Monakov <amonakov@ispras.ru>
23306 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23307 probabilities consistently.
23309 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23311 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23312 emit_cmp_and_jump_insns.
23314 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23317 * tree-inline.c (expand_call_inline): Combine profile statuses.
23319 2017-06-30 Andrew Pinski <apinski@cavium.com>
23321 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23322 fold_stmt returned true.
23324 2017-06-30 Nathan Sidwell <nathan@acm.org>
23326 * ggc.h (empty_string): Delete.
23327 * cfgexpand.c (expand_asm_stmt): Use plain "".
23328 * optabs.c (expand_asm_memory_barrier): Likewise.
23329 * stringpool.c (empty_string): Delete.
23330 (digit_vector, digit_string): Delete.
23331 (ggc_alloc_string): Use plain "", don't optimize single digit
23332 strings. Use ggc_alloc_atomic.
23334 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23336 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23337 comparison set and one other set, use the cost of the non-comparison
23340 2017-06-30 Nathan Sidwell <nathan@acm.org>
23342 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23345 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23347 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23348 loops. Remove now unneeded calls to gimple_switch_set_label() that
23349 just set removed labels to NULL_TREE.
23351 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23353 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23354 (set_range_info): ...here. Only call set_range_info_raw if domain
23356 (set_nonzero_bits): Call set_range_info_raw.
23357 * tree-ssanames.h (set_range_info_raw): New.
23359 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23362 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23363 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23364 of nonimmediate_operand and <store_mask_constraint> instead of m
23365 for the input operand. For V8FI iterator, always split if input
23366 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23367 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23368 <store_mask_predicate> instead of register_operand and
23369 <store_mask_constraint> instead of v for the input operand. Make
23370 sure both operands aren't MEMs for if not <mask_applied>.
23372 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23374 * lto-wrapper.c (copy_file) Close both file descriptors before
23377 2017-06-30 Martin Liska <mliska@suse.cz>
23380 * multiple_target.c (create_dispatcher_calls): Make ifunc
23381 also for function that don't have calls or are not referenced.
23383 2017-06-30 Richard Biener <rguenther@suse.de>
23385 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23386 analyze the first scalar stmt. Move vector type computation
23387 for the BB case here from ...
23388 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23389 live operation processing in the SLP case properly.
23391 2017-06-30 Richard Biener <rguenther@suse.de>
23393 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23395 2017-06-30 Martin Liska <mliska@suse.cz>
23398 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23399 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23401 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23403 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23406 2017-06-30 Richard Biener <rguenther@suse.de>
23408 PR tree-optimization/81249
23409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23410 condition reduction result to original scalar type.
23412 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23414 * profile-count.h (enum profile_quality): Fix typos and whitespace
23417 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23419 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23420 type for branch probabilities.
23422 2017-06-29 Julian Brown <julian@codesourcery.com>
23423 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23425 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23426 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23427 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23428 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23430 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23432 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23433 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23434 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23435 CC usage from generic code to here.
23436 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23437 CC usage into the target macros.
23439 2017-06-29 Maya Rashish <coypu@sdf.org>
23441 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23444 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23446 * arm/arm-builtins.c: Include profile-count.h
23447 * except.c (sjlj_emit_function_enter): Use
23448 profile_probability::unlikely.
23450 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23452 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23453 and tocrel_offset be pointer args rather than implicitly using
23455 (legitimate_constant_pool_address_p, rs6000_emit_move,
23456 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23457 tocrel_offset and use in toc_relative_expr_p call.
23458 (print_operand, print_operand_address): Use static tocrel_base_oac
23459 and tocrel_offset_oac.
23460 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23463 2017-06-29 Maya Rashish <coypu@sdf.org>
23465 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23467 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23469 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23470 objects, take into account only the alignment of 'op0' and 'mode1' if
23473 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23475 * ccmp.c (ccmp_tree_comparison_p): New function.
23476 (ccmp_candidate_p): Update to use above function.
23477 (get_compare_parts): New function.
23478 (expand_ccmp_next): Update to use new functions.
23479 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23481 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23482 take mode as argument.
23483 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23484 * expr.c (expand_expr_real_1): Pass mode as argument.
23486 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23488 * combine.c (combine_instructions): Print insns to dump_file, together
23491 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23493 * asan.c (asan_emit_stack_protection): Update.
23494 (create_cond_insert_point): Update.
23495 * auto-profile.c (afdo_propagate_circuit): Update.
23496 * basic-block.h (struct edge_def): Turn probability to
23497 profile_probability.
23498 (EDGE_FREQUENCY): Update.
23499 * bb-reorder.c (find_traces_1_round): Update.
23500 (better_edge_p): Update.
23501 (sanitize_hot_paths): Update.
23502 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23503 (make_single_succ_edge): Update.
23504 (check_bb_profile): Update.
23505 (dump_edge_info): Update.
23506 (update_bb_profile_for_threading): Update.
23507 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23508 probabilitycount to 0.
23509 * cfgbuild.c (compute_outgoing_frequencies): Update.
23510 * cfgcleanup.c (try_forward_edges): Update.
23511 (outgoing_edges_match): Update.
23512 (try_crossjump_to_edge): Update.
23513 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23514 (expand_gimple_tailcall): Update.
23515 (construct_init_block): Use make_single_succ_edge.
23516 (construct_exit_block): Use make_single_succ_edge.
23517 * cfghooks.c (verify_flow_info): Update.
23518 (redirect_edge_succ_nodup): Update.
23519 (split_edge): Update.
23520 (account_profile_record): Update.
23521 * cfgloopanal.c (single_likely_exit): Update.
23522 * cfgloopmanip.c (scale_loop_profile): Update.
23523 (set_zero_probability): Remove.
23524 (duplicate_loop_to_header_edge): Update.
23525 * cfgloopmanip.h (loop_version): Update prototype.
23526 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23527 (force_nonfallthru_and_redirect): Update.
23528 (update_br_prob_note): Update.
23529 (rtl_verify_edges): Update.
23530 (purge_dead_edges): Update.
23531 (rtl_lv_add_condition_to_bb): Update.
23532 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23533 * cgraphunit.c (init_lowered_empty_function): Update.
23534 (cgraph_node::expand_thunk): Update.
23535 * cilk-common.c: Include profile-count.h
23536 * dojump.c (inv): Remove.
23537 (jumpifnot): Update.
23538 (jumpifnot_1): Update.
23539 (do_jump_1): Update.
23541 (do_jump_by_parts_greater_rtx): Update.
23542 (do_compare_rtx_and_jump): Update.
23543 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23544 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23545 * dwarf2cfi.c: Include profile-count.h
23546 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23547 (sjlj_emit_dispatch_table): Likewise.
23548 * explow.c: Include profile-count.h
23549 * expmed.c (emit_store_flag_force): Update.
23550 (do_cmp_and_jump): Update.
23551 * expr.c (compare_by_pieces_d::generate): Update.
23552 (compare_by_pieces_d::finish_mode): Update.
23553 (emit_block_move_via_loop): Update.
23554 (store_expr_with_bounds): Update.
23555 (store_constructor): Update.
23556 (expand_expr_real_2): Update.
23557 (expand_expr_real_1): Update.
23558 * expr.h (try_casesi, try_tablejump): Update prototypes.
23559 * gimple-pretty-print.c (dump_probability): Update.
23560 (dump_profile): New.
23561 (dump_gimple_label): Update.
23562 (dump_gimple_bb_header): Update.
23563 * graph.c (draw_cfg_node_succ_edges): Update.
23564 * hsa-gen.c (convert_switch_statements): Update.
23565 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23566 (find_if_case_1): Update.
23567 (find_if_case_2): Update.
23568 * internal-fn.c (expand_arith_overflow_result_store): Update.
23569 (expand_addsub_overflow): Update.
23570 (expand_neg_overflow): Update.
23571 (expand_mul_overflow): Update.
23572 (expand_vector_ubsan_overflow): Update.
23573 * ipa-cp.c (good_cloning_opportunity_p): Update.
23574 * ipa-split.c (split_function): Use make_single_succ_edge.
23575 * ipa-utils.c (ipa_merge_profiles): Update.
23576 * loop-doloop.c (add_test): Update.
23577 (doloop_modify): Update.
23578 * loop-unroll.c (compare_and_jump_seq): Update.
23579 (unroll_loop_runtime_iterations): Update.
23580 * lra-constraints.c (lra_inheritance): Update.
23581 * lto-streamer-in.c (input_cfg): Update.
23582 * lto-streamer-out.c (output_cfg): Update.
23583 * mcf.c (adjust_cfg_counts): Update.
23584 * modulo-sched.c (sms_schedule): Update.
23585 * omp-expand.c (expand_omp_for_init_counts): Update.
23586 (extract_omp_for_update_vars): Update.
23587 (expand_omp_ordered_sink): Update.
23588 (expand_omp_for_ordered_loops): Update.
23589 (expand_omp_for_generic): Update.
23590 (expand_omp_for_static_nochunk): Update.
23591 (expand_omp_for_static_chunk): Update.
23592 (expand_cilk_for): Update.
23593 (expand_omp_simd): Update.
23594 (expand_omp_taskloop_for_outer): Update.
23595 (expand_omp_taskloop_for_inner): Update.
23596 * omp-simd-clone.c (simd_clone_adjust): Update.
23597 * optabs.c (expand_doubleword_shift): Update.
23598 (expand_abs): Update.
23599 (emit_cmp_and_jump_insn_1): Update.
23600 (expand_compare_and_swap_loop): Update.
23601 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23602 * predict.c (predictable_edge_p): Update.
23603 (edge_probability_reliable_p): Update.
23604 (set_even_probabilities): Update.
23605 (combine_predictions_for_insn): Update.
23606 (combine_predictions_for_bb): Update.
23607 (propagate_freq): Update.
23608 (estimate_bb_frequencies): Update.
23609 (force_edge_cold): Update.
23610 * profile-count.c (profile_count::dump): Add missing space into dump.
23611 (profile_count::debug): Add newline.
23612 (profile_count::differs_from_p): Explicitly convert to unsigned.
23613 (profile_count::stream_in): Update.
23614 (profile_probability::dump): New member function.
23615 (profile_probability::debug): New member function.
23616 (profile_probability::differs_from_p): New member function.
23617 (profile_probability::differs_lot_from_p): New member function.
23618 (profile_probability::stream_in): New member function.
23619 (profile_probability::stream_out): New member function.
23620 * profile-count.h (profile_count_quality): Rename to ...
23621 (profile_quality): ... this one.
23622 (profile_probability): New.
23623 (profile_count): Update.
23624 * profile.c (compute_branch_probabilities): Update.
23625 * recog.c (peep2_attempt): Update.
23626 * sched-ebb.c (schedule_ebbs): Update.
23627 * sched-rgn.c (find_single_block_region): Update.
23628 (compute_dom_prob_ps): Update.
23629 (schedule_region): Update.
23630 * sel-sched-ir.c (compute_succs_info): Update.
23631 * stmt.c (struct case_node): Update.
23632 (do_jump_if_equal): Update.
23633 (get_outgoing_edge_probs): Update.
23634 (conditional_probability): Update.
23635 (emit_case_dispatch_table): Update.
23636 (expand_case): Update.
23637 (expand_sjlj_dispatch_table): Update.
23638 (emit_case_nodes): Update.
23639 * targhooks.c: Update.
23640 * tracer.c (better_p): Update.
23641 (find_best_successor): Update.
23642 * trans-mem.c (expand_transaction): Update.
23643 * tree-call-cdce.c: Update.
23644 * tree-cfg.c (gimple_split_edge): Upate.
23645 (move_sese_region_to_fn): Upate.
23646 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23647 * tree-eh.c (lower_resx): Upate.
23648 (cleanup_empty_eh_move_lp): Upate.
23649 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23650 * tree-inline.c (copy_edges_for_bb): Update.
23651 (copy_cfg_body): Update.
23652 * tree-parloops.c (gen_parallel_loop): Update.
23653 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23654 (gimple_gen_time_profiler): Update.
23655 * tree-ssa-dce.c (remove_dead_stmt): Update.
23656 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23657 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23658 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23659 (unloop_loops): Update.
23660 (try_peel_loop): Update.
23661 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23662 * tree-ssa-loop-split.c (connect_loops): Update.
23663 (split_loop): Update.
23664 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23665 (hoist_guard): Update.
23666 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23667 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23668 (value_replacement): Update.
23669 * tree-ssa-reassoc.c (branch_fixup): Update.
23670 * tree-ssa-tail-merge.c (replace_block_by): Update.
23671 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23672 (create_edge_and_update_destination_phis): Update.
23673 (compute_path_counts): Update.
23674 (recompute_probabilities): Update.
23675 (update_joiner_offpath_counts): Update.
23676 (freqs_to_counts_path): Update.
23677 (duplicate_thread_path): Update.
23678 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23679 (struct switch_conv_info): Update.
23680 (gen_inbound_check): Update.
23681 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23682 (vect_do_peeling): Update.
23683 (vect_loop_versioning): Update.
23684 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23685 (optimize_mask_stores): Update.
23686 * ubsan.c (ubsan_expand_null_ifn): Update.
23687 * value-prof.c (gimple_divmod_fixed_value): Update.
23688 (gimple_divmod_fixed_value_transform): Update.
23689 (gimple_mod_pow2): Update.
23690 (gimple_mod_pow2_value_transform): Update.
23691 (gimple_mod_subtract): Update.
23692 (gimple_mod_subtract_transform): Update.
23693 (gimple_ic): Update.
23694 (gimple_stringop_fixed_value): Update.
23695 (gimple_stringops_transform): Update.
23696 * value-prof.h: Update.
23698 2017-06-29 Carl Love <cel@us.ibm.com>
23700 * config/rs6000/rs6000-c.c: Add support for built-in functions
23701 vector signed int vec_signed (vector float);
23702 vector signed long long vec_signed (vector double);
23703 vector signed int vec_signed2 (vector double, vector double);
23704 vector signed int vec_signede (vector double);
23705 vector signed int vec_signedo (vector double);
23706 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23707 instruction generator.
23708 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23709 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23710 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23712 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23713 vunsignede_v2df): Add define_expands.
23714 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23715 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23716 VEC_UNSIGNEDO): Add definitions.
23717 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23718 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23719 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23720 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23721 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23722 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23723 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23724 vec_unsignede, vec_unsignedo): Add builtin defines.
23725 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23727 * doc/extend.texi: Update the built-in documentation file for the
23728 new built-in functions.
23730 2017-06-29 Richard Biener <rguenther@suse.de>
23732 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23733 reduction chains to LOOP_VINFO_REDUCTIONS.
23734 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23735 SLP reductions after processing reduction chains.
23737 2017-06-29 Nathan Sidwell <nathan@acm.org>
23739 * builtins.c (fold_builtin_FUNCTION): Use
23740 lang_hooks.decl_printable_name.
23742 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23744 PR middle-end/81194
23745 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23746 with only one label.
23747 * stmt.c (expand_case): Assert NCASES is greater than one.
23749 2017-06-29 Richard Biener <rguenther@suse.de>
23751 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23753 (group_case_labels): Likewise.
23754 (find_taken_edge): Push sanity checking on val to workers...
23755 (find_taken_edge_cond_expr): ... here
23756 (find_taken_edge_switch_expr): ... and here, handle cases
23757 with just a default label.
23758 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23759 (group_case_labels): Likewise.
23760 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23761 group_case_labels does anything cleanup the CFG again.
23763 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23765 PR tree-optimization/81196
23766 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23767 exit condition comparing two IVs.
23769 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23771 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23772 profile to the dummy entry at the end of the list of architectures.
23773 * config/arm/arm-cpu-cdata.h: Regenerated.
23775 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23776 Michael Collison <michael.collison@arm.com>
23779 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23781 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23782 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23783 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23784 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23785 of shift when the shift amount is masked with constant equal to
23786 the size of the mode.
23787 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23790 2017-06-29 Martin Liska <mliska@suse.cz>
23792 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23794 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23797 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23800 (edge_growth_cache_entry::edge_growth_cache_entry): New
23802 (reset_edge_growth_cache): Update to use constructor.
23804 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23806 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23807 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23808 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23810 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23812 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23813 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23815 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23817 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23818 (*-linux-uclibc*): Add t-uclibc tmake_file.
23819 * config/t-musl: New.
23820 * config/t-uclibc: New.
23822 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23824 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23826 (gen_comm_data): Emit architectural setting of arch_prof.
23827 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23829 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23830 (armv8-m.base, armv8-m.main): Likewise.
23831 * arm-protos.h (arm_build_target): Add profile field.
23832 (arch_option): Likewise.
23833 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23835 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23836 arm_active_target.profile.
23838 2017-06-28 Richard Biener <rguenther@suse.de>
23840 PR middle-end/81227
23841 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23842 TYPE_OVERFLOW_WRAPS.
23843 * match.pd (negate_expr_p): Likewise.
23844 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23845 fold_build2, not fold_binary.
23847 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23849 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23850 Convert memory address to Pmode.
23851 (aarch64_print_operand): Assert MEM operands are always Pmode.
23853 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23856 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23857 Remove redundant if.
23858 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23859 * config/arm/aarch-common-protos.h
23860 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23861 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23863 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23866 * multiple_target.c (create_dispatcher_calls): Set the default
23867 clone to be static, not public.
23869 2017-06-28 Richard Biener <rguenther@suse.de>
23871 * tree-vect-loop.c (vectorizable_reduction): Move special
23872 cond reduction IV var creation ...
23873 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23874 parameter. Use STMT_VINFO_VECTYPE.
23875 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23878 2017-06-28 Martin Liska <mliska@suse.cz>
23881 * ipa-visibility.c (non_local_p): Handle visibility.
23883 2017-06-28 Martin Liska <mliska@suse.cz>
23886 * common.opt: Add IntegerRange to various options.
23887 * opt-functions.awk (integer_range_info): New function.
23888 * optc-gen.awk: Add integer_range_info to cl_options struct.
23889 * opts-common.c (decode_cmdline_option): Handle
23890 CL_ERR_INT_RANGE_ARG.
23891 (cmdline_handle_error): Likewise.
23892 * opts.c (print_filtered_help): Show valid interval in
23893 when --help is provided.
23894 * opts.h (struct cl_option): Add range_min and range_max fields.
23895 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23897 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23899 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23900 (x * C EQ/NE y * C): New transformation.
23902 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23904 * genmultilib (combination_space): Accept '+' in option names.
23906 2017-06-28 Martin Liska <mliska@suse.cz>
23909 * asan.c (instrument_derefs): Bail out inner references
23910 that are hard register variables.
23912 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23915 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23916 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23918 2017-06-28 Richard Biener <rguenther@suse.de>
23920 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23921 (vect_get_slp_defs): Adjust.
23922 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23924 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23926 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23927 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23928 (vectorizable_reduction): Adjust.
23929 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23931 (vect_get_slp_defs): Likewise.
23932 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23933 (vectorizable_bswap): Adjust.
23934 (vectorizable_call): Likewise.
23935 (vectorizable_conversion): Likewise.
23936 (vectorizable_assignment): Likewise.
23937 (vectorizable_shift): Likewise.
23938 (vectorizable_operation): Likewise.
23939 (vectorizable_store): Likewise.
23940 (vectorizable_condition): Likewise.
23941 (vectorizable_comparison): Likewise.
23943 2017-06-28 Michael Collison <michael.collison@arm.com>
23946 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23948 (arm_gen_movmemqi): Removed unused variable 'i'.
23949 Convert 'for' loop into 'while' loop.
23950 (arm_expand_prologue): Remove last unnecessary set of insn.
23951 (thumb_pop): Remove unused variable 'pushed_words'.
23952 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23954 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23956 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23957 * config/s390/s390-protos.h: Add prototype of
23958 s390_rel_address_ok_p.
23959 * config/s390/s390.c (s390_got_symbol): New function.
23960 (s390_rel_address_ok_p): New function.
23961 (legitimize_pic_address): Use s390_rel_address_ok_p.
23962 (s390_load_got): Use s390_got_symbol.
23963 (s390_option_override): Issue error if
23964 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23965 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23967 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23969 2017-06-27 Andrew Pinski <apinski@cavium.com>
23971 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23972 (X * copysign (1.0, X)): New pattern.
23973 (X * copysign (1.0, -X)): New pattern.
23974 (copysign (-1.0, CST)): New pattern.
23976 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23978 * genmultilib (combination_space): Remove variable.
23979 Validate reuse rules against regular expression for any sequence
23980 of multilib options in any order.
23982 2017-06-27 Michael Collison <michael.collison@arm.com>
23984 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23985 call aarch64_split_simd_combine.
23986 * (aarch64_combine_internal<mode>): Delete pattern.
23987 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23988 Allow register and subreg operands.
23990 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23992 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23993 specific need, just fallback on defaults.
23994 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23996 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23997 Olivier Hainque <hainque@adacore.com>
23999 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24001 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24002 targets. Pick a default if no particular attempt applied.
24003 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24006 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24008 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24009 (x86_64-wrs-vxworks7): Likewise.
24011 2017-06-27 Marek Polacek <polacek@redhat.com>
24014 * ubsan.c (instrument_null): Check get_base_address's result for null.
24016 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24018 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24020 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24022 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24023 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24024 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24025 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24026 New function types.
24027 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24028 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24029 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24030 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24031 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24032 BUILT_IN_FEUPDATEENV): New builtins.
24033 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24034 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24035 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24036 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24038 (builtin_structptr_types): Adjust size.
24039 * tree.c (builtin_structptr_types): Add four entries.
24041 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24042 Olivier Hainque <hainque@adacore.com>
24044 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24045 (TLS_SYM): New local macro, forcing reference to __tls__ on
24046 link command lines for VxWorks 7 RTPs, triggering initialization
24048 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24049 OS features TLS support, true for RTPs on VxWorks 7.
24050 * config/vxworks.c (vxworks_override_options): Setup emutls
24053 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24055 * predict.c (test_prediction_value_range): Use -1U instead of -1
24056 to avoid narrowing conversion warning.
24057 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24058 to avoid narrowing conversion warning.
24059 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24061 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24063 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24065 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24066 64bit configurations.
24067 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24068 (SIZE_TYPE): Likewise.
24069 * config/vxworks.c (vxworks_emutls_var_fields): Use
24070 long_unsigned_type_node instead of unsigned_type_node as the offset
24071 field type, which is "pointer" mode in emutls.c.
24073 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24076 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24078 PR middle-end/81207
24079 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24080 gimple_vuse copying separately from gimple_vdef copying.
24082 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24084 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24087 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24088 Olivier Hainque <hainque@adacore.com>
24090 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24091 for all vxworks7 targets.
24092 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24093 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24094 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24095 variations for VX6/VX7 and 32/64bits later on in ...
24096 (VXWORKS_LIB_SPEC): Leverage new macros.
24097 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24098 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24100 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24102 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24103 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24105 2017-06-26 Carl Love <cel@us.ibm.com>
24107 * config/rs6000/rs6000-c.c: Add support for built-in functions
24108 vector bool char vec_reve (vector bool char);
24109 vector signed char vec_reve (vector signed char);
24110 vector unsigned char vec_reve (vector unsigned char);
24111 vector bool int vec_reve (vector bool int);
24112 vector signed int vec_reve (vector signed int);
24113 vector unsigned int vec_reve (vector unsigned int);
24114 vector bool long long vec_reve (vector bool long long);
24115 vector signed long long vec_reve (vector signed long long);
24116 vector unsigned long long vec_reve (vector unsigned long long);
24117 vector bool short vec_reve (vector bool short);
24118 vector signed short vec_reve (vector signed short);
24119 vector double vec_reve (vector double);
24120 vector float vec_reve (vector float);
24121 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24122 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24123 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24124 (altivec_vreve): New pattern.
24125 * config/rs6000/altivec.h (vec_reve): New define.
24126 * doc/extend.texi (vec_rev): Update the built-in documentation file
24127 for the new built-in functions.
24129 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24131 PR tree-optimization/71815
24132 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24134 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24136 (slsr_process_phi): Likewise.
24137 (replace_uncond_cands_and_profitable_phis): Don't replace a
24138 multiply candidate with a stride of 1 (copy or cast).
24139 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24141 (lowest_cost_path): Likewise.
24142 (total_savings): Likewise.
24144 2017-06-26 Richard Biener <rguenther@suse.de>
24147 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24148 Use def_builtin_pure for all gather builtins.
24150 2017-06-26 Richard Biener <rguenther@suse.de>
24152 PR tree-optimization/81203
24153 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24154 non-dominating BBs.
24156 2017-06-26 Marek Polacek <polacek@redhat.com>
24159 * doc/invoke.texi: Document -Wmultistatement-macros.
24161 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24163 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24164 arm_neon_ok_no_float_abi effective target.
24166 2017-06-26 Richard Biener <rguenther@suse.de>
24168 PR tree-optimization/80928
24169 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24170 (copy_bbs): Set BB_DUPLICATED flag early.
24171 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24173 (execute_on_shrinking_pred): Likewise.
24174 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24175 BB_DUPLICATED blocks.
24176 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24177 iterate over all PHIs considering removal of *gsi.
24179 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24181 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24184 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24186 * config/rs6000/rs6000-string.c: (expand_block_clear,
24187 do_load_for_compare, select_block_compare_mode,
24188 compute_current_alignment, expand_block_compare,
24189 expand_strncmp_align_check, expand_strn_compare,
24190 expand_block_move, rs6000_output_load_multiple)
24191 Move functions related to string/block move/compare
24192 to a separate file.
24193 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24194 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24195 for this function which is now used in two files.
24196 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24197 * config.gcc: Add rs6000-string.o to extra_objs for
24198 targets powerpc*-*-* and rs6000*-*-*.
24200 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24203 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24204 32-bit, since indexed is not valid for DImode.
24205 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24206 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24207 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24208 (define_peephole2 for Altivec d-form store): Likewise.
24211 * multiple_target.c (create_dispatcher_calls): Only create the
24212 dispatcher call if the function is the default clone of a
24213 versioned function.
24215 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24217 PR middle-end/80902
24218 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24219 a call, force the call to not be a tail call.
24221 2017-06-23 Jeff Law <law@redhat.com>
24223 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24226 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24228 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24229 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24230 handling for early expansion of vector shifts (sl,sr,sra,rl).
24231 (builtin_function_type): Add vector shift right instructions
24232 to the unsigned argument list.
24234 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24236 rtl-optimizatoin/79286
24237 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24238 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24239 trap. PIC register plus a const unspec without offset can never trap.
24241 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24243 * tree.h (builtin_structptr_type): New type.
24244 (builtin_structptr_types): Declare new array.
24245 * tree.c (builtin_structptr_types): New array.
24246 (free_lang_data, build_common_tree_nodes): Use it.
24248 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24251 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24254 2017-06-22 Matt Turner <mattst88@gmail.com>
24256 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24257 Lake models to skylake case. Assume skylake for unknown
24258 models with clflushopt.
24260 2017-06-22 Jeff Law <law@redhat.com>
24262 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24263 frame sizes that do not satisfy aarch64_uimm12_shift.
24265 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24267 * profile-count.h (apply_probability,
24268 apply_scale, probability_in): Fix checks for zero.
24270 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24272 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24273 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24275 2016-06-22 Richard Biener <rguenther@suse.de>
24277 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24278 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24279 REDUC_MAX_EXPR support.
24280 (vectorizable_reduction): Likewise.
24281 (vect_create_epilog_for_reduction): Likewise.
24283 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24285 * match.pd (A / (1 << B) -> A >> B): New.
24286 * generic-match-head.c: Include optabs-tree.h.
24287 * gimple-match-head.c: Likewise.
24288 * optabs-tree.h (target_supports_op_p): New.
24289 * optabs-tree.c (target_supports_op_p): New.
24291 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24293 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24294 $gcc_cv_ld --help output.
24295 (gcc_cv_ld_demangle): Likewise.
24296 (gcc_cv_ld_eh_frame_hdr): Likewise.
24297 (gcc_cv_ld_pie): Likewise.
24298 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24299 (gcc_cv_ld_buildid): Likewise.
24300 (gcc_cv_ld_sysroot): Likewise.
24301 (ld_bndplt_support): Likewise.
24302 (ld_pushpopstate_support): Likewise.
24303 * configure: Regenerate.
24304 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24306 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24309 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24310 operands indexes to avoid gap between operands and match_dups.
24312 2017-06-21 Andrew Pinski <apinski@cavium.com>
24314 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24315 Increment Arith_shift and Arith_shift_reg by 1.
24316 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24318 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24319 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24320 (aarch64_strip_extend): Add new argument and test for it.
24321 (aarch64_cheap_mult_shift_p): New function.
24322 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24323 add a cost if it is true.
24324 Update calls to aarch64_strip_extend.
24325 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24327 2017-06-21 Andrew Pinski <apinski@cavium.com>
24329 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24331 (thunderxt88): Likewise.
24332 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24333 (thunderx_prefetch_tune): New variable.
24334 (thunderx2t99_prefetch_tune): Update for the correct values.
24335 (thunderxt88_tunings): New variable.
24336 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24337 generic_prefetch_tune.
24338 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24340 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24342 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24343 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24344 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24345 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24346 (aarch64_atomic_cas<mode>, GPI): Likewise.
24348 2017-06-21 Martin Liska <mliska@suse.cz>
24350 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24351 statements on cold and hot labels.
24352 * predict.c (tree_estimate_probability_bb): Remove the
24353 prediction from this place.
24355 2017-06-21 Martin Liska <mliska@suse.cz>
24357 PR tree-optimization/79489
24358 * gimplify.c (maybe_add_early_return_predict_stmt): New
24360 (gimplify_return_expr): Call the function.
24361 * predict.c (tree_estimate_probability_bb): Remove handling
24363 * predict.def: Update comment about early return predictor.
24364 * gimple-predict.h (is_gimple_predict): New function.
24365 * predict.def: Change default value of early return to 66.
24366 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24368 * passes.def: Put pass_strip_predict_hints to the beginning of
24371 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24373 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24374 FUNCTION_DECL declarations.
24375 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24377 (dwaf2out_decl): Likewise.
24378 * godump.c (go_early_global_decl): Skip call to the real debug hook
24379 for FUNCTION_DECL declarations.
24380 * passes.c (rest_of_decl_compilation): Skip call to the
24381 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24382 -fdump-go-spec is passed.
24384 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24386 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24387 Reorder for compactness.
24388 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24389 (def_builtin_pure, def_builtin_pure2): New functions.
24390 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24392 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24394 * match.pd (nop_convert): New predicate.
24395 ((A +- CST1) +- CST2): Allow some NOP conversions.
24397 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24400 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24401 with ctors/dtors if GOVD_SHARED is set.
24403 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24405 * config/aarch64/aarch64.md (movti_aarch64):
24406 Emit mov rather than orr.
24407 (movtf_aarch64): Likewise.
24408 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24409 Emit mov rather than orr.
24411 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24413 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24414 Swap alternatives, make integer dup more expensive.
24416 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24418 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24419 Return true for non-tls symbols.
24421 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24423 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24424 (cortex-a75): Likewise.
24425 (cortex-a75.cortex-a55): Likewise.
24426 * config/aarch64/aarch64-tune.md: Regenerate.
24427 * doc/invoke.texi (-mtune): Document new values for -mtune.
24429 2017-06-21 Tom de Vries <tom@codesourcery.com>
24431 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24432 stack_size feature.
24433 (Effective-Target Keywords, Other attributes): Suggest using
24434 dg-add-options stack_size feature to get stack limit in stack_size
24435 effective target documentation.
24437 2017-06-21 Julian Brown <julian@codesourcery.com>
24438 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24440 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24441 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24442 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24444 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24445 attribute type list for neon_multiply.
24446 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24447 attribute type list for neon_multiply.
24448 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24449 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24450 attribute type list for neon_multiply.
24451 * config/arm/types.md (crypto_pmull): Add.
24452 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24453 attribute type list.
24455 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24457 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24460 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24462 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24463 to make sure not to dereference a NULL cost_classes_ptr pointer.
24465 2017-06-20 Carl Love <cel@us.ibm.com>
24467 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24468 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24469 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24471 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24472 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24473 VMULOSW): New enum "unspec" values.
24474 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24475 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24476 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24477 altivec_vmulosw): New patterns.
24478 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24479 VMULOSW): Add definitions.
24481 2017-06-20 Julia Koval <julia.koval@intel.com>
24483 * config/i386/i386.c: Fix rounding expand for new pattern.
24484 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24486 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24488 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24489 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24491 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24493 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24496 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24498 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24499 architecture, then by implementer ID.
24500 * config/aarch64/aarch64-tune.md: Regenerate.
24502 2017-06-20 Richard Biener <rguenther@suse.de>
24504 PR middle-end/81097
24505 * fold-const.c (split_tree): Fold to type before negating.
24507 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24509 * diagnostic-show-locus.c
24510 (selftest::test_fixit_deletion_affecting_newline): New function.
24511 (selftest::diagnostic_show_locus_c_tests): Call it.
24513 2017-06-20 Andreas Schwab <schwab@suse.de>
24516 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24519 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24521 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24522 __ARM_FEATURE_COPROC according to support.
24524 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24526 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24527 Rewritten to avoid overflow for > 32-bit pointers.
24530 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24531 by removing enum keyword.
24532 (ubsan_type_descriptor): Likewise. Formatting fix.
24535 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24536 splitter): Require TARGET_SSE2 in the condition.
24538 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24541 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24542 for doing vector set of SFmode on ISA 3.0.
24543 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24544 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24546 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24547 SFmode value into a V4SF variable that was extracted from another
24548 V4SF variable without converting the element to double precision
24549 and back to single precision vector format.
24550 (vsx_insert_extract_v4sf_p9_2): Likewise.
24552 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24554 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24555 in UWHI to avoid undefined overflow.
24558 * ubsan.h (enum ubsan_encode_value_phase): New.
24559 (ubsan_encode_value): Change second argument to
24560 enum ubsan_encode_value_phase with default value of
24561 UBSAN_ENCODE_VALUE_GENERIC.
24562 * ubsan.c (ubsan_encode_value): Change second argument to
24563 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24564 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24565 create_tmp_var_raw instead of create_tmp_var and use a
24567 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24568 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24569 ubsan_encode_value callers.
24572 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24573 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24574 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24576 2017-06-19 Richard Biener <rguenther@suse.de>
24578 PR middle-end/81118
24579 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24580 estimates if we changed anything.
24582 2017-06-19 Richard Biener <rguenther@suse.de>
24584 PR tree-optimization/80887
24585 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24586 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24587 simplified lookups, then reset mprts_hook.
24588 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24590 (try_to_simplify): Likewise.
24592 2017-06-19 Martin Liska <mliska@suse.cz>
24595 * gimplify.c (gimplify_switch_expr):
24596 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24598 2017-06-19 Martin Liska <mliska@suse.cz>
24600 * doc/install.texi: Document that PGO runs in 4 stages.
24602 2017-06-19 Martin Liska <mliska@suse.cz>
24605 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24606 to dispatcher function name.
24607 * multiple_target.c (replace_function_decl): New function.
24608 (create_dispatcher_calls): Redirect both edges and references.
24610 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24612 * profile-count.c (profile_count::dump): Dump quality.
24613 (profile_count::differs_from_p): Update for unsigned val.
24614 * profile-count.h (profile_count_quality): New enum.
24615 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24617 2017-06-19 Richard Biener <rguenther@suse.de>
24619 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24620 struct function as arg.
24621 (estimate_numbers_of_iterations): Export overload with loop arg.
24622 (free_numbers_of_iterations_estimates_loop): Use an overload of
24623 free_numbers_of_iterations_estimates instead.
24624 * tree-cfg.c (remove_bb): Adjust.
24625 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24626 * tree-parloops.c (gen_parallel_loop): Likewise.
24627 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24629 (tree_unroll_loops_completely): Likewise.
24630 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24631 Use an overload instead and export.
24632 (estimated_loop_iterations): Adjust.
24633 (max_loop_iterations): Likewise.
24634 (likely_max_loop_iterations): Likewise.
24635 (estimate_numbers_of_iterations): Take struct function as arg
24637 (loop_exits_before_overflow): Adjust.
24638 (free_numbers_of_iterations_estimates_loop): Use an overload.
24639 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24640 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24642 2017-06-19 Richard Biener <rguenther@suse.de>
24645 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24646 RANGE_EXPR conservatively.
24648 2017-06-16 Carl Love <cel@us.ibm.com>
24650 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24651 definitions for vec_float, vec_float2, vec_floato,
24652 vec_floate built-ins.
24653 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24654 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24656 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24657 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24658 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24659 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24660 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24661 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24662 vec_floato): Add builtin defines.
24663 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24664 Update the built-in documentation file for the new built-in
24667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24669 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24670 (mthumb): Mark as the negative of -marm.
24672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24674 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24676 (ARM Options, -mtune): Document that this accepts the same
24677 extension options as -mcpu.
24678 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24682 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24683 permitted extensions.
24685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24687 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24688 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24689 (armv8-m.main): Add option +nodsp.
24690 * config/arm/arm-cpu-cdata.h: Regenerated.
24692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24694 * config/arm/t-fuchsia: New file.
24695 * config.gcc (arm*-*-fuchsia*): Use it.
24697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24699 * config/arm/t-symbian: Rewrite for new option infrastructure.
24701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24703 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24704 (MULTILIB_REQUIRED): Likewise.
24706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24708 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24709 (MULTILIB_RESUE): Likewise.
24710 (MULTILIB_MATCHES): Likewise.
24711 (MULTLIB_REQUIRED): Likewise.
24713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24715 * config/arm/t-rtems: Rewrite for new option framework.
24717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24719 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24720 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24721 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24722 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24723 * config/arm/t-multilib: ... here.
24724 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24725 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24726 armv7-a and armv8*-a when A-profile libraries have not been built.
24727 * config/arm/t-rmprofile: Rewrite.
24729 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24731 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24732 with a backslash. Remove the backslash after substituting unescaped
24734 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24736 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24738 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24739 the driver to pass through the default setting of -mfloat-abi.
24740 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24742 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24743 (all_feat_combs): New rule.
24744 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24746 * config/arm/t-aprofile: Rewrite.
24748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24750 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24751 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24752 fpu is not specified by the user/command-line.
24753 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24754 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24755 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24756 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24757 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24758 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24760 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24762 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24763 * config/arm/t-arm-elf: Rewritten.
24765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24767 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24768 have some floating-point instructions.
24769 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24770 (TARGET_MAYBE_HARD_FLOAT): New macro.
24771 * config/arm/arm-builtins.c (arm_init_builtins): Use
24772 TARGET_MAYBE_HARD_FLOAT.
24773 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24777 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24778 (configargs.h): Include it.
24779 (arm_print_hint_for_fpu_option): New function.
24780 (arm_parse_fpu_option): New function.
24781 (candidate_extension): New class.
24782 (arm_canon_for_multilib): New function.
24783 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24784 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24785 (ARCH_CANONICAL_SPECS): New macro.
24786 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24790 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24791 are set after handling multilib fragments. Set target_cpu_default2
24794 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24796 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24798 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24799 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24801 (check_arch): Likewise.
24802 * config/arm/arm.c (arm_configure_build_target): Handle
24803 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24804 options in the default.
24806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24808 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24809 when an option is an alias of another.
24810 * config/arm/parsecpu.awk (optalias): New parser token.
24811 (gen_comm_data): Mark non-alias options as such. Emit entries
24812 for extension aliases.
24813 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24814 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24815 (armv6kz, armv6zk, armv6t2): Likewise.
24816 (armv7): Make vfpv3-d16 an alias.
24817 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24819 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24820 Sort in canonical order.
24821 (armv8-a): Sort in canonical order.
24822 (armv8.1-a, armv8.2-a): Likewise.
24823 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24825 (cortex-a9): Sort in canonical order.
24826 * config/arm/arm.c (selftests.h): Include it.
24827 (arm_test_cpu_arch_data): New function.
24828 (arm_run_self_tests): New function.
24829 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24830 (targetm): Move declaration to the end of the file.
24831 * arm-cpu-cdata.h: Regenerated.
24833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24835 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24836 call to target_mode_check describing the type of option passed.
24837 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24838 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24839 arm_parse_cpu_option_name to match parameters against list of
24841 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24842 arm_arch_core_flags data structure.
24843 * config/arm/arm-cpu_cdata.h: Regenerated.
24845 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24847 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24849 (arm_print_hint_for_cpu_option): Likewise.
24850 (arm_print_hint_for_arch_option): Likewise.
24851 (arm_parse_cpu_option_name): Likewise.
24852 (arm_parse_arch_option_name): Likewise.
24853 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24854 of entries in the all_fpus list.
24855 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24856 (arm_parse_cpu_option_name): Declare.
24857 (arm_parse_arch_option_name): Declare.
24858 (arm_parse_option_features): Declare.
24859 (arm_intialize_isa): Declare.
24860 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24862 (gen_comm_data): ... here. Make definitions non-static.
24863 * config/arm/arm-cpu-data.h: Regenerated.
24864 * config/arm/arm-cpu-cdata.h: Regenerated.
24866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24868 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24869 (cpu_arch_extension): New structure.
24870 (cpu_arch_option, arch_option, cpu_option): New structures.
24871 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24872 architecture types.
24873 (gen_data): Generate new format data tables.
24874 * config/arm/arm.c (cpu_tune): New structure.
24875 (cpu_option, processors): Delete.
24876 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24877 (arm_print_hint_for_cpu_option): ... this and ...
24878 (arm_print_hint_for_arch_option): ... this.
24879 (arm_parse_arch_cpu_name): Delete. Replace with ...
24880 (arm_parse_cpu_option_name): ... this and ...
24881 (arm_parse_arch_option_name): ... this.
24882 (arm_unrecognized_feature): Change type of target parameter to
24884 (arm_parse_arch_cpu_features): Delete. Replace with ...
24885 (arm_parse_option_features): ... this.
24886 (arm_configure_build_target): Rework to use new configuration data
24888 (arm_print_tune_info): Rework for new configuration data tables.
24889 * config/arm/arm-cpu-data.h: Regenerated.
24890 * config/arm/arm-cpu.h: Regenerated.
24892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24894 * Makefile.in (OBJS): Move sbitmap.o from here ...
24895 (OBJS-libcommon): ... to here.
24897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24899 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24900 (ISA_ALL_CRYPTO): New macro.
24901 (ISA_ALL_SIMD): New macro
24902 (ISA_ALL_FP): New macro.
24903 * config/arm/arm.c (fpu_bitlist): Update initializer.
24904 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24906 (arm9e): Add fpu. Add option for nofp
24907 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24908 (arm926ej-s, arm1026ej-s): Likewise.
24909 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24910 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24911 neon-fp16, neon-vfpv4, nofp and nosimd.
24912 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24913 (cortex-a8): Add fpu. Add option for nofp.
24914 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24915 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24916 (cortex-r4f): Add fpu.
24917 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24918 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24920 (cortex-r8): Likewise.
24921 (cortex-m4): Add fpu. Add option for nofp.
24922 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24923 (cortex-a17.cortex-a7): Likewise.
24924 (cortex-a32): Add fpu. Add options for crypto and nofp.
24925 (cortex-a35, cortex-a53): Likewise.
24926 (cortex-a57): Add fpu. Add option for crypto.
24927 (cortex-a72, cortex-a73): Likewise.
24928 (exynos-m1): Likewise.
24929 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24930 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24931 (cortex-m33): Add fpu. Add option for nofp.
24932 * config/arm/arm-cpu-cdata.h: Regenerated
24933 * config/arm/arm-cpu-data.h: Regenerated.
24935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24937 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24938 (armv5te, armv5tej): Likewise.
24939 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24940 (armv7): Add options fp and vfpv3-d16.
24941 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24942 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24944 (armv7ve): Likewise.
24945 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24946 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24947 (armv8-a): Add nocrypto option.
24948 (armv8.1-a, armv8.2-a): Likewise.
24949 (armv8-m.main): add options fp, fp.dp and nofp.
24951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24953 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24955 (armv8-a+crc): Delete.
24956 (armv8.1-a): Add options simd, crypto and nofp.
24957 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24958 (armv8.2-a+fp16): Delete.
24959 (armv8-m.main): Add option dsp.
24960 (armv8-m.main+dsp): Delete.
24961 (cortex-a8): Add fpu. Add nofp option.
24962 (cortex-a9): Add fpu. Add nofp and nosimd options.
24963 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24964 link to main cpu and architecture data structures.
24965 (gen_comm_data): Only put isa attributes from the main architecture
24967 (option): New statement for architecture and CPU entries.
24968 * arm.c (struct cpu_option): New structure.
24969 (struct processors): Add entry for options.
24970 (arm_unrecognized_feature): New function.
24971 (arm_parse_arch_cpu_name): Ignore any characters after the first
24973 (arm_parse_arch_cpu_feature): New function.
24974 (arm_configure_build_target): Separate out any CPU and architecture
24975 features and parse separately. Don't error out if -mfpu=auto is
24976 used with only an architecture string.
24977 (arm_print_asm_arch_directives): New function.
24978 (arm_file_start): Call it.
24979 * config/arm/arm-cpu-cdata.h: Regenerated.
24980 * config/arm/arm-cpu-data.h: Likewise.
24981 * config/arm/arm-tables.opt: Likewise.
24983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24985 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24986 assembler when it is not -mfpu=auto.
24988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24990 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24991 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24992 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24993 (ASM_CPU_SPEC): Rewrite.
24994 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24995 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24996 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24997 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24998 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24999 copied string is NUL-terminated. Also strip any characters prefixed
25001 (arm_rewrite_selected_arch): New function.
25002 (arm_rewrite_march): New function.
25004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25006 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25007 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25008 (march, mcpu, mtune): Convert to string-based options.
25009 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25010 (arm_parse_arch_cpu_name): New function.
25011 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25012 identify selected architecture or CPU.
25013 (arm_option_save): New function.
25014 (TARGET_OPTION_SAVE): Redefine.
25015 (arm_option_restore): Restore string options.
25016 (arm_option_print): Print string options.
25018 2017-06-16 Martin Sebor <msebor@redhat.com>
25020 PR tree-optimization/80933
25021 PR tree-optimization/80934
25022 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25023 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25024 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25025 (gimple_fold_builtin): Call them.
25027 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25029 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25030 as unlikely; update profile.
25032 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25034 * predict.c (force_edge_cold): Handle declaring edges impossible
25037 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25039 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25041 (try_unroll_loop_completely): Fix reporting.
25043 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25045 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25047 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25050 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25051 if given a non-constant argument for an intrinsic which requires a
25054 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25056 * profile.c (compare_freqs): New function.
25057 (branch_prob): Sort edge list.
25058 (find_spanning_tree): Assume that the list is priority sorted.
25060 2017-06-16 Richard Biener <rguenther@suse.de>
25062 PR tree-optimization/81090
25063 * passes.def (pass_record_bounds): Remove.
25064 * tree-pass.h (make_pass_record_bounds): Likewise.
25065 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25066 make_pass_record_bounds): Likewise.
25067 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25068 not free niter estimates at the beginning but at the end.
25069 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25071 2017-06-16 Richard Biener <rguenther@suse.de>
25073 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25074 initializer to workaround ICE in host GCC 4.8.
25076 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25078 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25080 (clone_inlined_nodes): Update.
25082 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25084 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25085 prefetch settings, and enable prefetching by default at -O3.
25087 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25089 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25090 Set flag_prefetch_loop_arrays according to tuning data.
25092 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25094 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25095 New tune structure.
25096 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25097 [Unrelated to main purpose of the patch] Place the pointer field last
25098 to enable type checking errors when tune structure are wrongly merged.
25099 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25100 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25101 (thunderx2t99_prefetch_tune): New tune constants.
25102 (tune_params *_tunings): Update all tunings (no functional change).
25103 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25104 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25105 from tunings structures.
25107 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25110 * ubsan.c (instrument_null): Add T argument, use it instead
25111 of computing it based on IS_LHS.
25112 (instrument_object_size): Likewise.
25113 (pass_ubsan::execute): Adjust instrument_null and
25114 instrument_object_size callers to pass gimple_get_lhs or
25115 gimple_assign_rhs1 result to it. Use instrument_null instead of
25116 calling get_base_address and instrument_mem_ref. Handle
25117 aggregate call arguments for object-size sanitization.
25119 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25121 PR tree-optimization/81089
25122 * tree-vrp.c (is_masked_range_test): Validate operands of
25125 2017-06-15 Martin Sebor <msebor@redhat.com>
25128 * dumpfile.c (dump_register): Avoid calling memset to initialize
25129 a class with a default ctor.
25130 * gcc.c (struct compiler): Remove const qualification.
25131 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25132 * hash-table.h: Ditto.
25133 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25135 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25136 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25138 * params.h (struct param_info): Make struct members non-const.
25139 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25140 with default initialization.
25141 * vec.h (vec_copy_construct, vec_default_construct): New helper
25143 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25144 with vec_copy_construct.
25145 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25146 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25147 * doc/invoke.texi (-Wclass-memaccess): Document.
25149 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25151 * emit-rtl.h (is_leaf): Update comment about local
25152 register allocator.
25154 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25157 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25158 for a variable to have a section before checking if the section has a
25160 Set section to.persistent if persistent attribute is set.
25161 Warn if .persistent attribute is used on an automatic variable.
25163 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25165 PR rtl-optimization/80474
25166 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25168 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25170 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25173 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25175 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25176 (casesi): Emit bounds checking as RTL.
25177 (casesi_internal_mips16_<mode>): Remove bounds checking.
25179 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25181 * config/xtensa/xtensa.c (xtensa_option_override): Append
25182 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25183 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25184 xtensa_doloop_hooks): Define unconditionally.
25185 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25187 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25188 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25189 for it in xtensa_option_override.
25190 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25191 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25193 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25195 * doc/cppopts.texi: Document '-' special value to -MF.
25197 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25199 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25200 (cortex_a53_fconst): Likewise.
25201 (cortex_a53_fpmul): Likewise.
25202 (cortex_a53_f_load_64): Likewise.
25203 (cortex_a53_f_load_many): Likewise.
25204 (cortex_a53_advsimd_alu): Likewise.
25205 (cortex_a53_advsimd_alu_q): Likewise.
25206 (cortex_a53_advsimd_mul): Likewise.
25207 (cortex_a53_advsimd_mul_q): Likewise.
25208 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25209 Add missing fmul, r2f_cvt and fconst cases.
25211 2017-06-14 Richard Biener <rguenther@suse.de>
25213 PR middle-end/81088
25214 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25216 (fold_binary_loc): When associating do not treat pre-existing
25217 TREE_OVERFLOW on literal constants as a reason to allow
25218 TREE_OVERFLOW on associated literal constants.
25220 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25222 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25223 (MASK_FEATURES): New macro.
25224 * config/sparc/sparc.c (sparc_option_override): Remove the special
25225 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25227 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25229 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25230 a division of 0 if non-call exceptions are enabled.
25232 2017-06-14 Andrew Pinski <apinski@cavium.com>
25233 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25236 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25237 Improve vector initialization code gen for only variable case.
25239 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25241 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25243 2017-06-14 Richard Biener <rguenther@suse.de>
25245 PR tree-optimization/81083
25246 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25249 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25251 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25252 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25253 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25254 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25255 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25256 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25258 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25260 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25261 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25263 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25265 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25267 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25269 * config/rs6000/t-rtems: Don't handle SPE.
25271 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25273 * config/rs6000/t-linux: Don't handle SPE.
25275 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25277 * config/rs6000/eabispe.h: Delete file.
25279 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25281 * config/rs6000/t-spe: Delete file.
25283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25285 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25286 (rs6000_legitimate_offset_address_p): Return false for anything in
25287 V2SImode or V2SFmode.
25289 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25291 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25292 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25293 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25295 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25296 (rs6000_legitimate_offset_address_p): Ditto.
25297 (rs6000_emit_move): Ditto.
25298 (rs6000_init_builtins): Remove V4HI_type_node.
25300 2017-06-13 Martin Liska <mliska@suse.cz>
25303 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25304 (gate_asan): Likewise.
25305 * asan.h (asan_no_sanitize_address_p): Remove the function.
25306 (sanitize_flags_p): New function.
25307 * builtins.def: Fix coding style.
25308 * common.opt: Use renamed enum value.
25309 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25310 * doc/extend.texi: Document no_sanitize attribute.
25311 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25312 to SANITIZE_UNDEFINED_NONDEFAULT.
25313 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25314 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25315 Use sanitize_flags_p.
25316 * gimplify.c (gimplify_function_tree): Likewise.
25317 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25318 * opts.c (parse_no_sanitize_attribute): New function.
25319 (common_handle_option): Use renamed enum value.
25320 * opts.h (parse_no_sanitize_attribute): Declare.
25321 * tree.c (sanitize_flags_p): New function.
25322 * tree.h: Declared here.
25323 * tsan.c: Use sanitize_flags_p.
25324 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25325 (instrument_mem_ref): Likewise.
25326 (instrument_bool_enum_load): Likewise.
25327 (do_ubsan_in_current_function): Remove the function.
25328 (pass_ubsan::execute): Use sanitize_flags_p.
25329 * ubsan.h: Remove do_ubsan_in_current_function
25330 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25331 (dump_function_to_file): Use it here.
25333 2017-06-13 Martin Jambor <mjambor@suse.cz>
25335 PR tree-optimization/80803
25336 PR tree-optimization/81063
25337 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25338 (propagate_subaccesses_across_link): Enqueue subtree whenever
25339 necessary instead of relying on the caller.
25341 2017-06-13 Martin Jambor <mjambor@suse.cz>
25343 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25344 that have a first_link.
25345 (sort_and_splice_var_accesses): Do not check first_link before
25347 (subtree_mark_written_and_enqueue): Likewise.
25348 (propagate_all_subaccesses): Likewise and do not stop at first
25349 parent with a first_link.
25351 2017-06-13 Martin Jambor <mjambor@suse.cz>
25353 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25356 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25358 * match.pd: New pattern.
25360 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25362 * tree-vrp.c (is_masked_range_test): New function.
25363 (register_edge_assert_for): Determine ranges for
25366 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25368 PR tree-optimization/67328
25369 * fold-const.c (maskable_range_p): New function.
25370 (build_range_check): Generate bittests if possible.
25372 2017-06-13 Martin Liska <mliska@suse.cz>
25374 * gimple-pretty-print.c (dump_probability): Add new argument.
25375 (dump_edge_probability): Dump both probability and count.
25376 (dump_gimple_label): Likewise.
25377 (dump_gimple_bb_header): Likewise.
25379 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25382 * config/avr/avr-devices.c: Fix indentation.
25383 * config/avr/gen-avr-mmcu-specs.c: Dito.
25385 2017-06-13 Richard Biener <rguenther@suse.de>
25387 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25388 instead get vector type from stmt_info.
25389 (vectorizable_reduction): Adjust. Remove dead code.
25391 2017-06-13 Richard Biener <rguenther@suse.de>
25393 PR middle-end/81065
25394 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25395 case of C * (x * C2 + C3).
25396 (fold_addr_of_array_ref_difference): Properly fold index difference.
25398 2017-06-12 David S. Miller <davem@davemloft.net>
25401 * config/sparc/sparc.md (return expander): Emit frame blockage if
25402 function uses alloca.
25404 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25406 * combine.c (make_field_assignment): Check len rather than the mode
25407 precision when calling force_to_mode.
25409 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25411 Support multilibs and devices that see flash in RAM address range.
25414 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25415 (avr_mcu_t) <flash_pm_offset>: New field.
25416 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25417 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25418 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25419 (AVR_TINY_PM_OFFSET): Remove macro.
25420 * config/avr/avr.opt (-mshort-calls): New option.
25421 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25422 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25423 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25424 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25425 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25426 instead of avr_arch->have_jmp_call.
25427 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25428 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25429 avr_arch->flash_pm_offset to define.
25430 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25431 new field flash_pm_offset. Add entry for avrxmega3.
25432 (avr_texinfo): Add entry for avrxmega3.
25433 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25434 attiny212, attiny214,
25435 attiny412, attiny414, attiny416, attiny417,
25436 attiny814, attiny816, attiny817,
25437 attiny1614, attiny1616, attiny1617,
25438 attiny3214, attiny3216, attiny3217.
25439 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25440 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25441 (avr_print_operand_address) [AVR_TINY]: Same.
25442 (avr_asm_init_sections) <readonly_data_section>: Only patch
25443 callback if avr_arch->flash_pm_offset = 0.
25444 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25445 for rodata if avr_arch->flash_pm_offset != 0.
25446 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25447 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25448 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25449 (m_options): Append opt_rcall.
25450 (m_dirnames): Append dir_rcall.
25451 * config/avr/t-multilib: Regenerate.
25453 * configure.ac [target=avr]: Check whether avrxmega3 default
25454 linker description file works as needed.
25455 * configure: Regenerate.
25456 * doc/avr-mmcu.texi: Regenerate.
25457 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25458 <__AVR_ARCH__>: Document avrxmega3 and 103.
25459 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25460 <__AVR_SHORT_CALLS__>: Document it.
25461 <__AVR_PM_BASE_ADDRESS__>: Document it.
25462 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25463 (AVR Variable Attributes) <progmem>: Document this is
25464 not needed for avrxmega3.
25465 (AVR Named Address Spaces) <__flash>: Dito.
25467 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25469 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25471 2017-06-12 Doug Rupp <rupp@adacore.com>
25473 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25474 Append vxworks-stdint.h to the tm_file list.
25475 * config/vxworks-stdint.h: New file.
25477 2017-06-12 Martin Liska <mliska@suse.cz>
25479 PR tree-optimization/81041
25480 * tree-profile.c (gimple_gen_ic_func_profiler):
25481 Create an extra BB in profile-generate
25482 (gimple_gen_time_profiler): Likewise.
25484 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25486 PR tree-optimization/81003
25487 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25488 (update_range_test): Use it instead of force_gimple_operand_gsi.
25490 2017-06-12 Richard Biener <rguenther@suse.de>
25492 PR tree-optimization/81053
25493 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25494 with backedge value not defined in loop. Simplify def stmt
25497 2017-06-11 Tom de Vries <tom@codesourcery.com>
25500 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25502 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25503 nvptx_cannot_force_const_mem.
25505 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25507 * opts.c (finish_options): Move test for flag_split_stack after
25508 it has been initialized.
25510 2017-06-11 Jason Merrill <jason@redhat.com>
25512 * tree.h (id_equal): New.
25513 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25514 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25515 instead of strcmp of IDENTIFIER_POINTER.
25517 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25519 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25520 (mark_all_inlined_calls_cdtor): Fix formating.
25521 (inline_transform): Rescale profile before inlining.
25523 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25525 * cgraph.h (cgraph_edge::clone): Update prototype.
25526 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25527 (cgraph_node::create_clone): Update.
25528 (cgraph_node::create_version_clone): Update.
25529 * tree-inline.c (copy_bb): Update.
25530 (expand_call_inline): Update.
25532 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25534 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25535 factored out from ...
25536 (rs6000_emit_prologue): ... here.
25538 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25540 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25541 factored out from ...
25542 (rs6000_emit_prologue): ... here.
25544 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25546 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25548 (handle_missing_profiles): Fix computation of tp_first_run.
25549 (counts_to_freqs): Do not touch freqs when count is 0.
25551 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25553 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25556 2017-06-10 Tom de Vries <tom@codesourcery.com>
25558 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25559 attributes): Document signal effective target.
25561 2017-06-10 Tom de Vries <tom@codesourcery.com>
25563 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25564 Document effective target stack_size.
25566 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25568 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25569 to the edit_context if they can be auto-applied.
25571 2017-06-9 Ian Lance Taylor <iant@golang.org>
25573 * opts.c (finish_options): If -fsplit-stack, disable implicit
25574 -forder-blocks-and-partition.
25575 * doc/invoke.texi (Optimize Options): Document that when using
25576 -fsplit-stack -forder-blocks-and-partition is not implicitly
25579 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25581 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25582 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25583 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25584 * builtins.def (abort, trap, unreachable): Declare cold.
25585 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25586 * tree-core.h (ECF_COLD): New.
25587 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25588 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25590 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25592 * predict.c (unlikely_executed_stmt_p): Cleanup.
25594 2017-06-09 Richard Biener <rguenther@suse.de>
25596 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25597 model if the ref is always written to.
25599 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25601 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25603 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25605 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25608 2017-06-09 Tom de Vries <tom@codesourcery.com>
25611 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25612 "target cannot support label values" when encountering LABEL_REF.
25614 2017-06-09 Martin Liska <mliska@suse.cz>
25616 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25617 (gimple_gen_ic_func_profiler): Emit direct comparison
25618 of __gcov_indirect_call_callee with NULL.
25619 (gimple_gen_time_profiler): Change probability from
25620 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25622 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25624 * profile.c (edge_gcov_counts): Turn to pointer.
25625 (compute_branch_probabilities, compute_branch_probabilities): Update.
25626 (branch_prob): Do not clear edge_gcov_count.
25627 * profile.h (edge_gcov_counts): Turn to pointer.
25628 (edge_gcov_count): Update.
25630 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25632 * gimple.h (gimple_check_failed): Mark cold.
25634 2017-06-09 Richard Biener <rguenther@suse.de>
25636 PR tree-optimization/66623
25637 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25638 refactor check_reduction into two parts, properly computing
25639 whether we have to check reduction validity for outer loop
25642 2017-06-09 Richard Biener <rguenther@suse.de>
25644 PR tree-optimization/79483
25645 * graphite-scop-detection.c (order): New global.
25646 (get_order): Compute bb to order mapping that satisfies code
25647 generation constraints.
25648 (cmp_pbbs): New helper.
25649 (build_scops): Start domwalk at entry block, sort generated
25652 2017-06-09 Richard Biener <rguenther@suse.de>
25654 PR middle-end/81007
25655 * ipa-polymorphic-call.c
25656 (ipa_polymorphic_call_context::restrict_to_inner_class):
25657 Skip FIELD_DECLs with error_mark_node type.
25658 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25661 2017-06-09 Martin Liska <mliska@suse.cz>
25663 * predict.c (struct branch_predictor): New struct.
25664 (test_prediction_value_range): New test.
25665 (predict_c_tests): New function.
25666 * selftest-run-tests.c (selftest::run_tests): Run the function.
25667 * selftest.h: Declare new tests.
25669 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25672 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25673 gen_add3_insn did not fail.
25674 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25675 r0, construct that number in a temporary reg and add that reg to r0.
25676 If asked to put the result in r0 as well, fail.
25678 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25680 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25681 for early expansion of vec_eqv.
25683 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25685 PR middle-end/81005
25686 * ubsan.c (instrument_null): Avoid pointless code temporary.
25687 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25689 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25693 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25696 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25699 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25701 * predict.c (unlikely_executed_edge_p): Move ahead.
25702 (probably_never_executed_edge_p): Use it.
25704 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25706 PR middle-end/79988
25707 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25708 gimple_call_builtin_p call.
25710 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25712 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25713 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25714 rtl_check_failed_type2, rtl_check_failed_code1,
25715 rtl_check_failed_code2, rtl_check_failed_code_mode,
25716 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25717 rtvec_check_failed_bounds, rtl_check_failed_flag,
25718 _fatal_insn_not_found, _fatal_insn): Likewise.
25719 * tree.h (tree_contains_struct_check_failed,
25720 tree_check_failed, tree_not_check_failed,
25721 tree_class_check_failed, tree_range_check_failed,
25722 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25723 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25724 tree_operand_check_failed, omp_clause_check_failed,
25725 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25728 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25730 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25731 flag_branch_probabilities.
25732 * ipa-inline.c (edge_badness): Likewise.
25733 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25734 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25735 * predict.c (maybe_hot_frequency_p): Likewise.
25736 (probably_never_executed): Likewise.
25737 * sched-ebb.c (schedule_ebbs): Likewise.
25738 * sched-rgn.c (find_single_block_region): Likewise.
25739 * tracer.c (tail_duplicate): Likewise.
25741 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25743 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25744 longer requires x_flag_profile_use.
25746 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25748 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25749 instead of flag_reorder_blocks_and_partition.
25750 * dbxout.c (dbxout_function_end): Likewise.
25751 * dwarf2out.c (gen_subprogram_die): Likewise.
25752 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25753 * hw-doloop.c (reorg_loops): Likewise.
25754 * varasm.c (assemble_start_function,
25755 assemble_end_function): Likewise.
25756 (decide_function_section): Do not check for
25757 flag_reorder_blocks_and_partition.
25759 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25761 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25763 (chkp_get_hard_register_fake_addr_expr): Ditto.
25764 (chkp_build_addr_expr): Add check for hard reg case.
25765 (chkp_parse_array_and_component_ref): Ditto.
25766 (chkp_find_bounds_1): Ditto.
25767 (chkp_process_stmt): Don't generate bounds store for
25770 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25772 * predict.c (maybe_hot_bb_p): Do not check profile status.
25773 (maybe_hot_edge_p): Likewise.
25774 (probably_never_executed): Check for zero counts even if profile
25776 (unlikely_executed_edge_p): New function.
25777 (unlikely_executed_stmt_p): New function.
25778 (unlikely_executed_bb_p): New function.
25779 (set_even_probabilities): Use unlikely predicates.
25780 (combine_predictions_for_bb): Likewise.
25781 (predict_paths_for_bb): Likewise.
25782 (predict_paths_leading_to_edge): Likewise.
25783 (determine_unlikely_bbs): New function.
25784 (estimate_bb_frequencies): Use it.
25785 (compute_function_frequency): Use zero counts even if profile is
25787 * profile-count.h: Fix typo.
25789 2017-08-08 Julia Koval <julia.koval@intel.com>
25791 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25792 _mm512_mask_cvtsepi16_storeu_epi8,
25793 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25794 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25795 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25796 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25797 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25798 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25799 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25800 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25801 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25802 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25803 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25804 __builtin_ia32_pmovuswb256mem_mask,
25805 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25806 __builtin_ia32_pmovwb512mem_mask): New builtins.
25808 2017-08-08 Julia Koval <julia.koval@intel.com>
25810 PR target/73350,80862
25811 * config/i386/subst.md (round): Fix round pattern.
25812 * config/i386/i386.c (ix86_erase_embedded_rounding):
25813 Fix erasing rounding for the fixed pattern.
25815 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25817 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25819 2017-06-08 Martin Liska <mliska@suse.cz>
25821 PR gcov-profile/80911
25822 * gcov.c (block_info::block_info): New constructor.
25824 2017-06-07 Carl Love <cel@us.ibm.com>
25826 * config/rs6000/rs6000-c: The return type of the following
25827 built-in functions was implemented as int not long long. Fix sign
25828 of return value for the unsigned version of vec_mulo and vec_mule.
25829 vector unsigned long long vec_bperm (vector unsigned long long,
25830 vector unsigned char)
25831 vector signed long long vec_mule (vector signed int,
25833 vector unsigned long long vec_mule (vector unsigned int,
25834 vector unsigned int)
25835 vector signed long long vec_mulo (vector signed int,
25837 vector unsigned long long vec_mulo (vector unsigned int,
25838 vector unsigned int)
25839 * doc/extend.texi: Fix the documentation for the built-in
25842 2017-06-07 Carl Love <cel@us.ibm.com>
25845 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25848 2017-06-07 Carl Love <cel@us.ibm.com>
25850 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25851 support, Generate doublehv for signed int/float for BE case only.
25853 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25855 * doc/invoke.texi (mcx16): Rewrite.
25857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25859 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25860 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25861 *mov<mode>_softfloat, and an anonymous splitter): Use
25862 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25866 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25868 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25869 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25870 (rs6000_debug_reg_global): Adjust.
25871 (rs6000_init_hard_regno_mode_ok): Adjust.
25872 (rs6000_dbx_register_number): Adjust.
25873 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25874 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25875 Remove SPE_ACC and SPEFSCR.
25876 (REG_ALLOC_ORDER): Ditto.
25877 (FRAME_POINTER_REGNUM): Change to 111.
25878 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25879 (REG_CLASS_NAMES): Ditto.
25880 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25881 (REGISTER_NAMES): Ditto.
25882 (ADDITIONAL_REG_NAMES): Ditto.
25883 (rs6000_reg_names): Ditto.
25884 * config/rs6000/rs6000.md: Renumber some register number
25887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25889 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25891 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25892 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25894 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25895 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25896 Delete the SPE high registers.
25897 (REG_ALLOC_ORDER): Ditto.
25898 (enum reg_class): Remove SPE_HIGH_REGS.
25899 (REG_CLASS_NAMES): Ditto.
25900 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25901 (REGISTER_NAMES): Ditto.
25902 (rs6000_reg_names): Ditto.
25903 * doc/tm.texi.in: Remove SPE as example.
25904 * doc/tm.texi: Regenerate.
25906 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25908 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25909 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25910 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25911 * config/rs6000/rs6000.md (type): Remove "brinc".
25913 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25915 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25916 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25917 * config/rs6000/linuxspe.h: Delete file.
25918 * config/rs6000/rs6000.md: Don't include spe.md.
25919 * config/rs6000/spe.h: Delete file.
25920 * config/rs6000/spe.md: Delete file.
25921 * config/rs6000/t-rs6000: Remove spe.md.
25923 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25925 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25926 (reg_or_none500mem_operand): Delete.
25927 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25928 instead of reg_or_none500mem_operand.
25930 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25932 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25933 handling of SPE flags.
25934 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25936 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25938 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25940 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25941 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25942 paired_divv2sf3): Similar.
25943 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25944 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25945 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25947 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25948 Rename the paired_* instruction patterns.
25949 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25951 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25952 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25953 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25954 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25955 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25956 PAIRED_VECTOR_MODE.
25957 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25958 (spe_func_has_64bit_regs_p): Delete.
25959 (spe_expand_predicate_builtin): Delete.
25960 (spe_expand_evsel_builtin): Delete.
25961 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25962 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25963 (invalid_e500_subreg): Delete.
25964 (rs6000_legitimize_address): Always force_reg op2 as well, for
25965 paired single memory accesses.
25966 (rs6000_member_type_forces_blk): Delete.
25967 (rs6000_spe_function_arg): Delete.
25968 (rs6000_expand_unop_builtin): Delete SPE handling.
25969 (rs6000_expand_binop_builtin): Ditto.
25970 (spe_expand_stv_builtin): Delete.
25971 (bdesc_2arg_spe): Delete.
25972 (spe_expand_builtin): Delete.
25973 (spe_expand_predicate_builtin): Delete.
25974 (spe_expand_evsel_builtin): Delete.
25975 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25976 (spe_init_builtins): Delete.
25977 (spe_func_has_64bit_regs_p): Delete.
25978 (savres_routine_name): Delete "info" parameter. Adjust callers.
25979 (rs6000_emit_stack_reset): Ditto.
25980 (rs6000_dwarf_register_span): Delete.
25981 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25982 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25983 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25985 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25987 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25988 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25989 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25990 mulv2sf3, divv2sf3): Delete expanders.
25992 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25994 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25998 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25999 * config/rs6000/rs6000.c: Ditto.
26001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26003 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26004 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26005 comparison_operator.
26007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26009 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26010 * config/rs6000/rs6000.opt: Ditto.
26011 * config/rs6000/t-rtems: Ditto.
26013 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26015 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26016 TARGET_E500_SINGLE by 0, simplify.
26017 * config/rs6000/rs6000.c: Ditto.
26018 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26019 (spe_build_register_parallel): Delete.
26020 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26021 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26022 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26023 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26024 (E500_CONVERT): Delete.
26025 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26027 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26029 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26030 * config/rs6000/dfp.md: Ditto.
26031 (negdd2, *negdd2_fpr): Merge.
26032 (absdd2, *absdd2_fpr): Merge.
26033 (negtd2, *negtd2_fpr): Merge.
26034 (abstd2, *abstd2_fpr): Merge.
26035 * config/rs6000/e500.h: Delete file.
26036 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26037 TARGET_FPRS by 1 and simplify.
26038 * config/rs6000/rs6000-c.c: Ditto.
26039 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26040 TARGET_DF_SPE by 0.
26041 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26043 * config/rs6000/rs6000.md: Ditto.
26044 (floatdidf2, *floatdidf2_fpr): Merge.
26045 (move_from_CR_gt_bit): Delete.
26046 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26047 (E500_CR_IOR_COMPARE): Delete.
26048 (All patterns that require !TARGET_FPRS): Delete.
26049 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26051 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26053 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26055 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26057 * graphds.c (add_edge): Intitialize edge's attached data.
26058 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26059 pointer parameter. Call pointed function on each edge during
26060 graph traversing. Skip traversing the edge when the function
26062 (graphds_dfs, graphds_scc): Ditto.
26063 (for_each_edge): New parameter. Pass the new parameter to callback
26065 * graphds.h (skip_edge_callback): New function pointer type.
26066 (graphds_dfs, graphds_scc): New function pointer parameter.
26067 (graphds_edge_callback, for_each_edge): New parameter.
26069 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26071 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26072 out code checking if runtime alias check is possible to below ...
26073 Call the new function.
26074 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26075 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26077 2017-06-07 Marek Polacek <polacek@redhat.com>
26080 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26081 TYPE_OVERFLOW_WRAPS check.
26083 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26085 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26086 if versioning is required.
26087 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26088 peeling with the check for versioning.
26090 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26092 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26093 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26094 Set true to new parameter if new ssa variable is defined.
26095 (vect_gen_vector_loop_niters): Refactor. Set range information
26096 for the new vector loop bound variable.
26097 (vect_do_peeling): Ditto.
26099 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26101 * tree-affine.c (ssa.h): Include header file.
26102 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26103 has wrapping overflow behavior.
26105 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26107 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26109 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26111 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26112 (tree_to_aff_combination): ... here.
26114 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26116 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26117 reg_pressure model function.
26118 (ivopts_global_cost_for_size): Delete.
26119 (determine_set_costs, iv_ca_recount_cost): Call new model function
26120 ivopts_estimate_reg_pressure.
26122 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26124 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26125 expensive than udiv. Remove floating point cases from mod.
26127 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26129 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26130 Increase idiv cost.
26132 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26134 * config/aarch64/aarch64.md
26135 (copysignsf3): Fix mask generation.
26137 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26139 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26141 (class dump_manager): Add register_dumps method.
26142 * dumpfile.c: Include langhooks.h.
26143 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26144 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26145 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26146 (dump_manager::dump_register): Start with 512 entries instead of 32.
26147 (dump_manager::register_dumps): New method.
26148 * toplev.c (general_init): Instead of invoking register_dumps
26149 langhook, invoke register_dumps method on the dump manager.
26150 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26153 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26155 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26156 Say that # requires an associated define_split to exist, and that
26157 the define_split must be suitable for use after register allocation.
26159 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26161 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26162 (compute_outgoing_frequencies): Also initialize zero counts.
26163 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26164 around loops; preserve more of profile when nothing changes.
26166 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26168 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26170 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26171 * config/arm/arm-cpu-cdata.h: Regenerate.
26172 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26173 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26174 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26175 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26176 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26178 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26180 * config/arm/t-rmprofile: Likewise.
26181 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26183 2017-06-06 David S. Miller <davem@davemloft.net>
26186 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26187 blockage if function uses alloca.
26189 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26191 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26192 New "uid" fields to hold pretty-print IDs of group and ref.
26193 Memory references are now identified as <group_id>:<ref_id>
26194 instead of using [random] addresses.
26195 (dump_mem_details): Simplify, no functional change.
26196 (dump_mem_ref): Simplify and make output more concise.
26197 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26198 (find_or_create_group): Initialize group uid.
26199 (record_ref): Initialize ref uid. Improve debug output.
26200 (prune_group_by_reuse, should_issue_prefetch_p,)
26201 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26202 (mark_nontemporal_store, determine_loop_nest_reuse):
26203 Improve debug output.
26205 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26207 * dbgcnt.def (prefetch): New debug counter.
26208 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26209 (schedule_prefetches): Stop issueing prefetches if debug counter
26212 2017-06-06 Tom de Vries <tom@codesourcery.com>
26214 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26215 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26217 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26219 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26220 Use aarch64_reg_or_zero predicate for operand 4.
26221 (aarch64_compare_and_swap<mode> define_insn_and_split):
26222 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26223 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26225 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26227 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26228 (arm_compute_save_core_reg_mask): This.
26229 (thumb1_compute_save_reg_mask): Rename into ...
26230 (thumb1_compute_save_core_reg_mask): This.
26231 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26232 (arm_compute_frame_layout): Likewise.
26234 2017-06-06 Richard Biener <rguenther@suse.de>
26236 PR tree-optimization/80974
26237 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26238 keep or clear leaders SSA info.
26240 2017-06-06 Tom de Vries <tom@codesourcery.com>
26242 * config/nvptx/nvptx.c (split_mode_p): New function.
26243 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26245 2017-06-06 Tom de Vries <tom@codesourcery.com>
26247 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26249 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26252 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26255 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26257 * shrink-wrap.c (handle_simple_exit): Update profile.
26258 (try_shrink_wrapping): Upate profile.
26260 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26262 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26263 (tree_guess_outgoing_edge_probabilities): New.
26264 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26265 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26267 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26269 * ipa-split.c (split_function): Initialize return bb profile.
26271 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26273 * profile.c (compute_branch_probabilities): Also initialize
26274 EXIT_BLOCK profile.
26276 2017-06-06 Richard Biener <rguenther@suse.de>
26278 PR tree-optimization/80928
26279 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26280 (vect_analyze_loop_operations): Properly guard analysis for
26282 (vect_transform_loop): Likewise.
26283 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26284 (vect_model_induction_cost): Do not cost for pure SLP.
26285 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26286 of induction in inner loop vectorization.
26287 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26288 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26289 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26291 (vect_analyze_slp_cost_1): Cost induction.
26292 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26293 (vect_get_slp_vect_defs): Likewise.
26294 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26295 (vect_transform_stmt): Handle SLP reductions.
26296 * tree-vectorizer.h (vectorizable_induction): Adjust.
26298 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26300 * config/rs6000/rs6000.c (make_resolver_func): Update
26301 init_lowered_empty_function call.
26303 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26305 * doc/invoke.texi: Document the -fprofile-abs-path option.
26306 * common.opt (fprofile-abs-path): New option.
26307 * gcov-io.h (gcov_write_filename): Declare.
26308 * gcov-io.c (gcov_write_filename): New function.
26309 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26310 * profile.c (output_location): Likewise.
26312 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26314 * shring-wrap.c: Revert accidental commit.
26316 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26318 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26322 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26324 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26325 profile in callgraph edge.
26326 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26327 (apply_scale): Likewise.
26328 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26329 Also scale profile when inlining function with zero profile.
26330 (initialize_cfun): Update exit block profile even when it is zero.
26331 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26332 when profile is read.
26334 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26336 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26337 (CLONE_*): New constants to define the processors we can generate
26338 code for with the target_clone attribute.
26339 (rs6000_clone_map): New array to identify which clone processors
26340 the current program is running on.
26341 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26342 target_clone attribute.
26343 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26344 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26345 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26346 (cpu_expand_builtin): Add support for target_clone attribute.
26347 (rs6000_valid_attribute_p): Allow "default" attribute.
26348 (get_decl_name): New debug function to simplify printing the
26349 current function name in debugging statements.
26350 (rs6000_clone_priority): New functions to support the target_clone
26351 attribute, and be able to generate code to switch between ISA 2.05
26352 through ISA 3.0 (power6 through power9).
26353 (rs6000_compare_version_priority): Likewise.
26354 (rs6000_get_function_versions_dispatcher): Likewise.
26355 (make_resolver_func): Likewise.
26356 (add_condition_to_bb): Likewise.
26357 (dispatch_function_versions): Likewise.
26358 (rs6000_generate_version_dispatcher_body): Likewise.
26359 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26360 (fusion_gpr_load_p): Fix a spacing issue.
26361 * doc/extend.texi (Common Function Attributes): Document that the
26362 PowerPC supports the target_clone attribute.
26364 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26366 * config/arm/arm.h: explain F symbol found in description of ARM
26367 register allocation in its legend.
26369 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26371 * config/mips/frame-header-opt.c: Include profile-count.h.
26372 * config/riscv/riscv.c: Include profile-count.h
26374 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26376 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26378 (sm_set_flag_if_changed): Add bbs field.
26379 (execute_sm_if_changed_flag_set): Pass BBS.
26380 (execute_sm): Update.
26382 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26384 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26387 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26389 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26390 (peephole2): New peephole2 to emit the above.
26391 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26393 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26395 * config/aarch64/aarch64.c (define_peephole2 above
26396 *sub_<shift>_<mode>): New peephole.
26398 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26400 * config/i386/i386.c (make_resolver_func): Update.
26401 * Makefile.in: Add profile-count.h and profile-count.o
26402 * auto-profile.c (afdo_indirect_call): Update to new API.
26403 (afdo_set_bb_count): Update.
26404 (afdo_propagate_edge): Update.
26405 (afdo_propagate_circuit): Update.
26406 (afdo_calculate_branch_prob): Update.
26407 (afdo_annotate_cfg): Update.
26408 * basic-block.h: Include profile-count.h
26409 (struct edge_def): Turn count to profile_count.
26410 (struct basic_block_def): Likewie.
26411 (REG_BR_PROB_BASE): Move to profile-count.h
26412 (RDIV): Move to profile-count.h
26413 * bb-reorder.c (max_entry_count): Turn to profile_count.
26414 (find_traces): Update.
26415 (rotate_loop):Update.
26416 (connect_traces):Update.
26417 (sanitize_hot_paths):Update.
26418 * bt-load.c (migrate_btr_defs): Update.
26419 * cfg.c (RDIV): Remove.
26420 (init_flow): Use alloc_block.
26421 (alloc_block): Uninitialize count.
26422 (unchecked_make_edge): Uninitialize count.
26423 (check_bb_profile): Update.
26424 (dump_edge_info): Update.
26425 (dump_bb_info): Update.
26426 (update_bb_profile_for_threading): Update.
26427 (scale_bbs_frequencies_int): Update.
26428 (scale_bbs_frequencies_gcov_type): Update.
26429 (scale_bbs_frequencies_profile_count): New.
26430 * cfg.h (update_bb_profile_for_threading): Update.
26431 (scale_bbs_frequencies_profile_count): Declare.
26432 * cfgbuild.c (compute_outgoing_frequencies): Update.
26433 (find_many_sub_basic_blocks): Update.
26434 * cfgcleanup.c (try_forward_edges): Update.
26435 (try_crossjump_to_edge): Update.
26436 * cfgexpand.c (expand_gimple_tailcall): Update.
26437 (construct_exit_block): Update.
26438 * cfghooks.c (verify_flow_info): Update.
26439 (dump_bb_for_graph): Update.
26440 (split_edge): Update.
26441 (make_forwarder_block): Update.
26442 (duplicate_block): Update.
26443 (account_profile_record): Update.
26444 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26445 (get_estimated_loop_iterations): Update.
26446 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26447 (single_likely_exit): Update.
26448 * cfgloopmanip.c (scale_loop_profile): Update.
26450 (set_zero_probability): Update.
26451 (lv_adjust_loop_entry_edge): Update.
26452 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26453 (purge_dead_edges): Update.
26454 (rtl_account_profile_record): Update.
26455 * cgraph.c (cgraph_node::create): Uninitialize count.
26456 (symbol_table::create_edge): Uninitialize count.
26457 (cgraph_update_edges_for_call_stmt_node): Update.
26458 (cgraph_edge::dump_edge_flags): Update.
26459 (cgraph_node::dump): Update.
26460 (cgraph_edge::maybe_hot_p): Update.
26461 * cgraph.h: Include profile-count.h
26462 (create_clone), create_edge, create_indirect_edge): Update.
26463 (cgraph_node): Turn count to profile_count.
26464 (cgraph_edge0: Likewise.
26465 (make_speculative, clone): Update.
26466 (create_edge): Update.
26467 (init_lowered_empty_function): Update.
26468 * cgraphclones.c (cgraph_edge::clone): Update.
26469 (duplicate_thunk_for_node): Update.
26470 (cgraph_node::create_clone): Update.
26471 * cgraphunit.c (cgraph_node::analyze): Update.
26472 (cgraph_node::expand_thunk): Update.
26473 * final.c (dump_basic_block_info): Update.
26474 * gimple-streamer-in.c (input_bb): Update.
26475 * gimple-streamer-out.c (output_bb): Update.
26476 * graphite.c (print_global_statistics): Update.
26477 (print_graphite_scop_statistics): Update.
26478 * hsa-brig.c: Include basic-block.h.
26479 * hsa-dump.c: Include basic-block.h.
26480 * hsa-gen.c (T sum_slice): Update.
26481 (convert_switch_statements):Update.
26482 * hsa-regalloc.c: Include basic-block.h.
26483 * ipa-chkp.c (chkp_produce_thunks): Update.
26484 * ipa-cp.c (struct caller_statistics): Update.
26485 (init_caller_stats): Update.
26486 (gather_caller_stats): Update.
26487 (ipcp_cloning_candidate_p): Update.
26488 (good_cloning_opportunity_p): Update.
26489 (get_info_about_necessary_edges): Update.
26490 (dump_profile_updates): Update.
26491 (update_profiling_info): Update.
26492 (update_specialized_profile): Update.
26493 (perhaps_add_new_callers): Update.
26494 (decide_about_value): Update.
26495 (ipa_cp_c_finalize): Update.
26496 * ipa-devirt.c (struct odr_type_warn_count): Update.
26497 (struct decl_warn_count): Update.
26498 (struct final_warning_record): Update.
26499 (possible_polymorphic_call_targets): Update.
26500 (ipa_devirt): Update.
26501 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26502 * ipa-icf.c (sem_function::merge): Update.
26503 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26504 * ipa-inline.c (compute_uninlined_call_time): Update.
26505 (compute_inlined_call_time): Update.
26506 (want_inline_small_function_p): Update.
26507 (want_inline_self_recursive_call_p): Update.
26508 (edge_badness): Update.
26509 (lookup_recursive_calls): Update.
26510 (recursive_inlining): Update.
26511 (inline_small_functions): Update.
26512 (dump_overall_stats): Update.
26513 (dump_inline_stats): Update.
26514 * ipa-profile.c (ipa_profile_generate_summary): Update.
26515 (ipa_propagate_frequency): Update.
26516 (ipa_profile): Update.
26517 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26518 * ipa-utils.c (ipa_merge_profiles): Update.
26519 * loop-doloop.c (doloop_modify): Update.
26520 * loop-unroll.c (report_unroll): Update.
26521 (unroll_loop_runtime_iterations): Update.
26522 * lto-cgraph.c (lto_output_edge): Update.
26523 (lto_output_node): Update.
26524 (input_node): Update.
26525 (input_edge): Update.
26526 (merge_profile_summaries): Update.
26527 * lto-streamer-in.c (input_cfg): Update.
26528 * lto-streamer-out.c (output_cfg): Update.
26529 * mcf.c (create_fixup_graph): Update.
26530 (adjust_cfg_counts): Update.
26531 (sum_edge_counts): Update.
26532 * modulo-sched.c (sms_schedule): Update.
26533 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26534 * predict.c (maybe_hot_count_p): Update.
26535 (probably_never_executed): Update.
26536 (dump_prediction): Update.
26537 (combine_predictions_for_bb): Update.
26538 (propagate_freq): Update.
26539 (handle_missing_profiles): Update.
26540 (counts_to_freqs): Update.
26541 (rebuild_frequencies): Update.
26542 (force_edge_cold): Update.
26543 * predict.h: Include profile-count.h
26544 (maybe_hot_count_p, counts_to_freqs): UPdate.
26545 * print-rtl-function.c: Do not include cfg.h
26546 * print-rtl.c: Include basic-block.h
26547 * profile-count.c: New file.
26548 * profile-count.h: New file.
26549 * profile.c (is_edge_inconsistent): Update.
26550 (correct_negative_edge_counts): Update.
26551 (is_inconsistent): Update.
26552 (set_bb_counts): Update.
26553 (read_profile_edge_counts): Update.
26554 (compute_frequency_overlap): Update.
26555 (compute_branch_probabilities): Update; Initialize and deinitialize
26557 (branch_prob): Update.
26558 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26559 (edge_gcov_count): New.
26560 (bb_gcov_count): New.
26561 * shrink-wrap.c (try_shrink_wrapping): Update.
26562 * tracer.c (better_p): Update.
26563 * trans-mem.c (expand_transaction): Update.
26564 (ipa_tm_insert_irr_call): Update.
26565 (ipa_tm_insert_gettmclone_call): Update.
26566 * tree-call-cdce.c: Update.
26567 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26568 (gimple_duplicate_sese_tail): Update.
26569 (gimple_account_profile_record): Update.
26570 (execute_fixup_cfg): Update.
26571 * tree-inline.c (copy_bb): Update.
26572 (copy_edges_for_bb): Update.
26573 (initialize_cfun): Update.
26574 (freqs_to_counts): Update.
26575 (copy_cfg_body): Update.
26576 (expand_call_inline): Update.
26577 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26578 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26579 (try_unroll_loop_completely): Update.
26580 (try_peel_loop): Update.
26581 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26582 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26583 * tree-ssa-loop-split.c (connect_loops): Update.
26584 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26585 * tree-ssa-reassoc.c (branch_fixup): Update.
26586 * tree-ssa-tail-merge.c (replace_block_by): Update.
26587 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26588 (compute_path_counts): Update.
26589 (update_profile): Update.
26590 (recompute_probabilities): Update.
26591 (update_joiner_offpath_counts): Update.
26592 (estimated_freqs_path): Update.
26593 (freqs_to_counts_path): Update.
26594 (clear_counts_path): Update.
26595 (ssa_fix_duplicate_block_edges): Update.
26596 (duplicate_thread_path): Update.
26597 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26598 (struct switch_conv_info): Update.
26599 * tree-tailcall.c (decrease_profile): Update.
26600 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26601 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26602 * value-prof.c (check_counter): Update.
26603 (gimple_divmod_fixed_value): Update.
26604 (gimple_mod_pow2): Update.
26605 (gimple_mod_subtract): Update.
26606 (gimple_ic_transform): Update.
26607 (gimple_stringop_fixed_value): Update.
26608 * value-prof.h (gimple_ic): Update.
26610 2017-06-02 Carl Love <cel@us.ibm.com>
26612 * config/rs6000/rs6000-c: Add support for built-in functions
26613 vector double vec_doublee (vector signed int);
26614 vector double vec_doublee (vector unsigned int);
26615 vector double vec_doublee (vector float);
26616 vector double vec_doubleh (vector signed int);
26617 vector double vec_doubleh (vector unsigned int);
26618 vector double vec_doubleh (vector float);
26619 vector double vec_doublel (vector signed int);
26620 vector double vec_doublel (vector unsigned int);
26621 vector double vec_doublel (vector float);
26622 vector double vec_doubleo (vector signed int);
26623 vector double vec_doubleo (vector unsigned int);
26624 vector double vec_doubleo (vector float);.
26625 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26626 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26628 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26629 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26630 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26632 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26633 vec_doublel, vec_doubleh.
26634 * doc/extend.texi: Update the built-in documentation file for the
26635 new built-in functions.
26637 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26640 * ipa-inline-analysis.c (free_growth_caches): Set
26641 edge_removal_hook_holder to NULL after removing it.
26643 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26645 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26646 comparision with zero.
26648 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26649 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26650 for early expansion of vec_min and vec_max builtins.
26651 (builtin_function_type): Add min/max unsigned variants to those
26652 identified as having unsigned arguments.
26654 2017-06-02 Olivier Hainque <hainque@adacore.com>
26656 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26658 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26660 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26661 Use VALL_F16 iterator rather than VALL.
26663 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26665 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26666 Emit CBNZ inside loop when doing a strong exchange and comparing
26667 against zero. Generate the CC flags after the loop.
26669 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26671 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26672 (dl_section_ref): New.
26673 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26674 On AIX, append an expression to subtract the size of the
26675 section length to dl_section_ref.
26677 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26679 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26680 for early expansion of vector absolute builtins.
26682 2017-06-02 Richard Biener <rguenther@suse.de>
26684 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26685 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26687 2017-06-02 Richard Biener <rguenther@suse.de>
26689 PR tree-optimization/80948
26690 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26691 stmt order as well.
26693 2017-06-02 Richard Biener <rguenther@suse.de>
26695 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26697 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26698 for inductions as relevant.
26700 2017-06-02 Richard Biener <rguenther@suse.de>
26702 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26703 (vectorizable_induction): ... this. Remove dead code.
26705 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26707 * builtins. (expand_builtin_alloca): Remove second parameter and
26708 infer its value from the first parameter instead.
26709 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26711 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26713 PR rtl-optimization/80903
26714 * loop-doloop.c (add_test): Unshare sequence.
26716 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26718 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26720 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26722 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26724 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26725 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26726 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26727 (xlogue_layout::get_stub_rtx): Make static.
26728 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26729 (xlogue_layout::compute_stub_managed_regs): Rename to...
26730 (xlogue_layout::count_stub_managed_regs): ...this.
26731 (xlogue_layout::is_stub_managed_reg): New function.
26732 (xlogue_layout::m_stub_names): Rename to...
26733 (xlogue_layout::s_stub_names): ...this, make static.
26734 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26735 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26736 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26737 xlogue_layout::s_stub_names): Instantiate statics.
26738 (stub_managed_regs): Remove.
26739 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26740 (disable_call_ms2sysv_xlogues): Rename to...
26741 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26742 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26744 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26745 change after reload_completed.
26746 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26748 (ix86_expand_prologue): Likewise.
26749 (ix86_expand_epilogue): Likewise.
26750 (ix86_expand_split_stack_prologue): Likewise.
26751 (ix86_compute_frame_layout): Remove frame parameter ...
26752 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26753 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26755 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26756 (ix86_frame): Move from here ...
26757 * config/i386/i386.h (ix86_frame): ... to here.
26758 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26759 complete ix86_frame data structure instead. Remove some_ld_name.
26761 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26763 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26764 symbols that hold a DECL_VALUE_EXPR.
26766 2017-06-01 Martin Jambor <mjambor@suse.cz>
26768 PR tree-optimization/80898
26769 * tree-sra.c (process_subtree_disqualification): Removed.
26770 (disqualify_candidate): Do not acll
26771 process_subtree_disqualification.
26772 (subtree_mark_written_and_enqueue): New function.
26773 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26774 RHS has been disqualified and re-queue LHS if necessary. Apart
26775 from that, ignore disqualified RHS.
26777 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26779 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26780 address fetch for z10 or later.
26782 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26784 * config/arc/arc.md (tst_movb): Add guard when splitting.
26786 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26788 * config/arc/arc.c (arc_can_eliminate): Test against
26789 arc_frame_pointer_needed.
26791 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26793 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26794 to prevent store reordering.
26795 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26796 (type): Add block type.
26797 (stack_tie): Define special instruction to be used in
26800 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26802 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26803 constraint. It is not valid for the pattern.
26804 (noncommutative_binary_comparison): Likewise.
26806 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26808 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26811 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26813 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26814 be used by the reg-alloc.
26816 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26818 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26819 reg-alloc when having mul64 or mul32x16 instructions.
26820 (mulsidi3): Likewise.
26821 (umulsidi3): Likewise.
26822 (mulsi32x16): New pattern.
26823 (mulsi64): Likewise.
26824 (mulsidi64): Likewise.
26825 (umulsidi64): Likewise.
26826 (MUL32x16_REG): Define.
26827 (mul64_600): Use MUL32x16_REG.
26828 (mac64_600): Likewise.
26829 (umul64_600): Likewise.
26830 (umac64_600): Likewise.
26832 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26834 * config/arc/arc.md (mulsi3_700): Make it commutative.
26836 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26838 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26840 (*sign_extendsidi2_insn): Likewise for movstosw.
26842 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26844 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26845 the type of the input discriminant value. Convert the
26846 discriminant value of signedness vary.
26848 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26850 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26851 Add to -Wall section.
26853 2017-06-01 Richard Biener <rguenther@suse.de>
26855 PR middle-end/66313
26856 * fold-const.c (fold_plusminus_mult_expr): If the factored
26857 factor may be zero use a wrapping type for the inner operation.
26858 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26859 and handle moved defs.
26860 (process_assignment): Properly guard the unary op case. Return a
26861 tri-state indicating that moving the stmt before the call may allow
26862 to continue. Pass through to_move.
26863 (find_tail_calls): Handle moving unrelated defs before
26866 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26869 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26870 splitter result in the canonical way.
26872 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26874 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26875 also for 32bit target. Update insn attributes.
26876 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26878 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26880 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26881 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26882 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26883 (_mm_maskz_min_ss): New intrinsics.
26885 2017-05-31 Martin Liska <mliska@suse.cz>
26887 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26888 Change comment style to one we normally use.
26889 (vectorizable_reduction): Likewise.
26890 (vectorizable_induction): Likewise.
26891 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26892 (vectorizable_call): Likewise.
26893 (vectorizable_simd_clone_call): Likewise.
26894 (vectorizable_conversion): Likewise.
26895 (vectorizable_assignment): Likewise.
26896 (vectorizable_shift): Likewise.
26897 (vectorizable_operation): Likewise.
26898 (vectorizable_store): Likewise.
26899 (vectorizable_load): Likewise.
26900 * tree-vectorizer.h: Likewise.
26902 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26904 * passes.c (emergency_dump_function): New.
26905 * tree-pass.h (emergency_dump_function): Declare.
26906 * plugin.c (plugins_internal_error_function): Remove.
26907 * plugin.h (plugins_internal_error_function): Remove declaration.
26908 * toplev.c (internal_error_function): New static function. Use it...
26909 (general_init): ...here.
26911 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26913 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26914 (arc_rtx_costs): Add costs for new patterns.
26915 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26916 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26918 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26920 * tree-ssa-strlen.c (get_next_strinfo): New function.
26921 (get_stridx_plus_constant): Use it.
26922 (zero_length_string): Likewise.
26923 (adjust_related_strinfos): Likewise.
26924 (adjust_last_stmt): Likewise.
26926 2017-05-31 Richard Biener <rguenther@suse.de>
26929 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26930 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26932 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26934 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26935 loop_vinfo argument and use of dependence distance vectors.
26936 Check instead whether the two references differ only in their
26937 initial value and assume that they have the same alignment if the
26938 difference is a multiple of the vector alignment.
26939 (vect_analyze_data_refs_alignment): Update call accordingly.
26941 2017-05-31 Martin Liska <mliska@suse.cz>
26944 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26946 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26948 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26949 (create_intersect_range_checks): Move from ...
26950 * tree-data-ref.c (create_intersect_range_checks_index)
26951 (create_intersect_range_checks): ... to here.
26952 (create_runtime_alias_checks): New function factored from ...
26953 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26954 here. Call above function.
26955 * tree-data-ref.h (create_runtime_alias_checks): New function.
26957 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26959 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26960 segment length for dr_b and compute it in wide_int.
26962 2017-05-31 Richard Biener <rguenther@suse.de>
26964 PR tree-optimization/80906
26965 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26966 and pass through iv_map.
26967 (copy_bb_and_scalar_dependences): Adjust.
26968 (translate_pending_phi_nodes): Likewise.
26969 (copy_loop_close_phi_args): Handle code-generating IVs instead
26972 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26974 * diagnostic-color.c (color_dict): Add "type-diff".
26975 (parse_gcc_colors): Update comment.
26976 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26977 -fdiagnostics-show-template-tree and -fno-elide-type.
26978 (GCC_COLORS): Add type-diff to example.
26980 (-fdiagnostics-show-template-tree): New.
26981 (-fno-elide-type): New.
26982 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26983 the pp_format_decoder callback. Call any m_format_postprocessor's
26985 (pretty_printer::pretty_printer): Initialize
26986 m_format_postprocessor.
26987 (pretty_printer::~pretty_printer): Delete any
26988 m_format_postprocessor.
26989 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26990 (class format_postprocessor): New class.
26991 (struct pretty_printer::format_decoder): Document the new parameters.
26992 (struct pretty_printer::m_format_postprocessor): New field.
26993 * tree-diagnostic.c (default_tree_printer): Update for new
26994 bool and const char ** params.
26995 * tree-diagnostic.h (default_tree_printer): Likewise.
26997 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26999 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27000 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27001 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27002 rs6000_gen_cell_microcode code.
27003 (rs6000_final_prescan_insn): Delete.
27004 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27005 "warn-cell-microcode" entries.
27006 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27007 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27008 throughout. Change cc_reg_not_micro_cr0_operand to
27009 cc_reg_not_cr0_operand throughout.
27010 (*extendhi<mode>2_noload): Delete.
27011 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27012 (mwarn-cell-microcode): Delete.
27013 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27014 -mgen-cell-microcode and -mwarn-cell-microcode.
27016 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27019 * config/i386/constraints.md (Yd): New constraint.
27021 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27022 and (?Yd, r) alternatives. Update insn attributes.
27023 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27024 and (?*Yd, r) alternatives. Update insn attributes.
27025 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27027 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27029 * gimplify.c (gimplify_modify_expr): Don't create a
27030 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27033 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27035 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27037 2017-05-30 Richard Biener <rguenther@suse.de>
27039 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27040 and reduc_def fields.
27041 (STMT_VINFO_REDUC_TYPE): New define.
27042 (STMT_VINFO_REDUC_DEF): Likewise.
27043 (vect_force_simple_reduction): Adjust prototype.
27044 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27045 (vect_is_simple_reduction): Remove check_reduction argument.
27046 (vect_force_simple_reduction): Adjust and set
27047 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27048 (vectorizable_reduction): Do not re-do reduction analysis
27049 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27050 * tree-parloops.c (gather_scalar_reductions): Adjust.
27052 2017-05-30 Richard Biener <rguenther@suse.de>
27054 PR middle-end/80901
27055 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27058 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27060 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27061 Introduce unknown_misalignment parameter and remove vf.
27062 (vect_peeling_hash_get_lowest_cost):
27063 Pass unknown_misalignment parameter.
27064 (vect_enhance_data_refs_alignment):
27065 Fix unsupportable data ref treatment.
27067 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27069 * tree-vect-data-refs.c (vect_get_data_access_cost):
27070 Workaround for SLP handling.
27071 (vect_enhance_data_refs_alignment):
27072 Compute costs for doing no peeling at all, compare to the best
27073 peeling costs so far and avoid peeling if cheaper.
27075 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27077 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27078 Return peeling info and set costs to zero for unlimited cost
27080 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27081 with unknown misalignment. Compute and costs for unknown
27082 misalignment, compare them to the costs for known misalignment
27083 and choose the cheapest for peeling.
27085 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27087 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27088 (vect_get_peeling_costs_all_drs): Create function.
27089 (vect_peeling_hash_get_lowest_cost):
27090 Use vect_get_peeling_costs_all_drs.
27091 (vect_peeling_supportable): Create function.
27092 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27094 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27096 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27097 DR_HAS_NEGATIVE_STEP.
27098 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27099 (vect_enhance_data_refs_alignment): Use.
27100 (vect_duplicate_ssa_name_ptr_info): Use.
27101 * tree-vectorizer.h (dr_misalignment): Use.
27102 (known_alignment_for_access_p): Use.
27104 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27107 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27109 (has_section_name): New function.
27111 2017-05-30 Martin Liska <mliska@suse.cz>
27114 * auto-profile.c (get_function_decl_from_block): Fix
27117 2017-05-30 Richard Biener <rguenther@suse.de>
27119 PR middle-end/80876
27120 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27122 2017-05-30 Martin Liska <mliska@suse.cz>
27124 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27125 * dumpfile.h (struct dump_file_info): Remove ctors.
27127 2017-05-30 Martin Liska <mliska@suse.cz>
27129 * predict.def: Fix GNU coding style.
27131 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27133 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27134 Mark 'to' argument with ATTRIBUTE_UNUSED.
27136 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27138 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27139 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27140 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27143 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27145 * doc/install.texi (Options specification): Restore entry of
27146 --enable-sjlj-exceptions.
27148 2017-05-27 Michael Eager <eager@eagercon.com>
27151 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27153 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27155 * config/microblaze/microblaze.h
27156 (FIXED_REGISTERS): Update in macro.
27157 (CALL_USED_REGISTERS): Update in macro.
27159 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27161 * doc/install.texi: Add links to macOS binary distributions.
27163 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27167 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27169 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27171 2017-05-26 Martin Liska <mliska@suse.cz>
27173 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27175 2017-05-26 Martin Liska <mliska@suse.cz>
27177 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27178 always leading ';; '.
27179 (dump_bb_info): Likewise.
27180 (brief_dump_cfg): Likewise.
27181 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27182 * dumpfile.c: Remove usage of TDF_VERBOSE.
27183 * dumpfile.h (enum dump_kind): Likewise.
27184 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27185 * print-tree.c (debug_verbose): Remove.
27186 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27187 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27188 * tree-diagnostic.c (default_tree_printer): Replace
27189 TDF_DIAGNOSTIC with TDF_SLIM.
27191 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27193 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27194 in parameter loop, rather than loop_vinfo.
27195 (create_intersect_range_checks): Ditto.
27196 (vect_create_cond_for_alias_checks): Update call to above functions.
27198 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27200 PR tree-optimization/80815
27201 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27202 for merging runtime alias checks. Handle negative DR_STEPs.
27204 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27206 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27208 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27209 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27210 out code pruning runtime alias checks.
27211 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27212 factored out from above.
27213 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27215 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27217 (prune_runtime_alias_test_list): New decalaration.
27219 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27221 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27222 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27223 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27224 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27225 (operator==, comp_dr_with_seg_len_pair): Ditto.
27226 (vect_prune_runtime_alias_test_list): Ditto.
27228 2017-05-26 Martin Liska <mliska@suse.cz>
27231 * params.def: Bound partial-inlining-entry-probability param.
27233 2017-05-26 Marek Polacek <polacek@redhat.com>
27236 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27239 2017-05-26 Richard Biener <rguenther@suse.de>
27241 PR tree-optimization/80842
27242 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27245 2017-05-26 Richard Biener <rguenther@suse.de>
27247 PR tree-optimization/80844
27248 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27250 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27252 * doc/md.texi (Machine Constraints): Update x86 family
27253 machine constraints section to match 'config/i386/constraints.md'.
27255 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27257 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27259 2017-05-25 Nathan Sidwell <nathan@acm.org>
27261 * doc/invoke.texi (--enable-languages): Update documentation.
27263 2017-05-25 Martin Liska <mliska@suse.cz>
27265 * dumpfile.c: Add TDF_FOLDING.
27266 * dumpfile.h (enum dump_kind): Likewise.
27267 * genmatch.c (dt_simplify::gen_1): Use it.
27269 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27271 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27273 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27275 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27276 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27278 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27280 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27281 * match.pd (X == C): Rewrite it here.
27282 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27283 with_certain_nonzero_bits2): New predicates.
27284 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27286 2017-05-24 Nathan Sidwell <nathan@acm.org>
27288 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27291 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27294 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27296 * config/powerpcspe: New port. Files are copied from the rs6000
27297 port, with "rs6000" in filenames replaced by "powerpcspe".
27299 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27301 PR rtl-optimization/80754
27302 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27304 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27306 * config/sparc/sparc.md (length): Return the correct value for -mflat
27307 sibcalls to match output_sibcall.
27309 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27313 * config/rs6000/rs6000.c (struct machine_function): Add new field
27315 (rs6000_get_separate_components): Init that field, use it.
27316 (rs6000_components_for_bb): Use the field.
27318 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27320 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27322 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27324 PR middle-end/80823
27325 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27327 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27330 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27331 address against address_operand predicate.
27332 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27334 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27336 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27337 parameters passed indirectly.
27339 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27341 * config/i386/i386.md (*movdi_internal): Remove SSE4
27342 alternative 18 (?r, *v). Update insn attributes.
27343 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27344 Update insn attributes.
27345 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27346 Update insn attributes.
27347 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27348 alternative 1 (r, v). Remove isa attribute.
27349 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27350 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27351 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27353 2017-05-23 Tom de Vries <tom@codesourcery.com>
27355 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27358 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27360 * cgraphunit.c (symbol_table::process_new_functions): Update.
27361 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27362 (inline_generate_summary): Rename to ...
27363 (ipa_fn_summary_generate): ... this one.
27364 (inline_read_summary): Rename to ...
27365 (ipa_fn_summary_read): ... this one.
27366 (inline_write_summary): Rename to ...
27367 (ipa_fn_summary_write): ... this one.
27368 (inline_free_summary): Rename to ...
27369 (ipa_free_fn_summary): ... this one.
27370 (pass_data_local_fn_summary, pass_local_fn_summary,
27371 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27372 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27373 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27374 make_pass_ipa_fn_summary): New.
27375 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27376 inline_write_summary, inline_free_summary): Remove.
27377 (ipa_free_fn_summary) : New.
27378 * ipa-inline.c (ipa_inline): Update.
27379 (pass_ipa_inline): Do not generate summaries.
27380 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27382 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27383 and add pass_ipa_fn_summary.
27384 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27386 (make_pass_inline_parameters): Remove.
27388 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27390 * omp-low.c (struct omp_context): Remove "default_kind" member.
27393 * omp-offload.c (execute_oacc_device_lower): Remove the
27394 parallelism dimensions function attributes for unparallelized
27395 OpenACC kernels constructs.
27397 2017-05-23 Martin Liska <mliska@suse.cz>
27399 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27401 (cgraph_edge::make_speculative): Likewise.
27402 (cgraph_edge::resolve_speculation): Likewise.
27403 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27404 (cgraph_node::dump): Likewise.
27405 * cgraph.h: Likewise.
27406 * cgraphunit.c (analyze_functions): Likewise.
27407 (symbol_table::compile): Likewise.
27408 * ipa-cp.c (print_all_lattices): Likewise.
27409 (determine_versionability): Likewise.
27410 (initialize_node_lattices): Likewise.
27411 (ipcp_verify_propagated_values): Likewise.
27412 (estimate_local_effects): Likewise.
27413 (update_profiling_info): Likewise.
27414 (create_specialized_node): Likewise.
27415 (perhaps_add_new_callers): Likewise.
27416 (decide_about_value): Likewise.
27417 (decide_whether_version_node): Likewise.
27418 (identify_dead_nodes): Likewise.
27419 (ipcp_store_bits_results): Likewise.
27420 * ipa-devirt.c (dump_targets): Likewise.
27421 (ipa_devirt): Likewise.
27422 * ipa-icf.c (sem_item::dump): Likewise.
27423 (sem_function::equals): Likewise.
27424 (sem_variable::equals): Likewise.
27425 (sem_item_optimizer::read_section): Likewise.
27426 (sem_item_optimizer::execute): Likewise.
27427 (congruence_class::dump): Likewise.
27428 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27429 (dump_inline_summary): Likewise.
27430 (estimate_node_size_and_time): Likewise.
27431 (inline_analyze_function): Likewise.
27432 * ipa-inline-transform.c (inline_call): Likewise.
27433 * ipa-inline.c (report_inline_failed_reason): Likewise.
27434 (want_early_inline_function_p): Likewise.
27435 (edge_badness): Likewise.
27436 (update_edge_key): Likewise.
27437 (inline_small_functions): Likewise.
27438 * ipa-profile.c (ipa_profile): Likewise.
27439 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27440 (ipa_make_edge_direct_to_target): Likewise.
27441 (remove_described_reference): Likewise.
27442 (ipa_impossible_devirt_target): Likewise.
27443 (propagate_controlled_uses): Likewise.
27444 (ipa_print_node_params): Likewise.
27445 (ipcp_transform_function): Likewise.
27446 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27447 (propagate_pure_const): Likewise.
27448 * ipa-reference.c (generate_summary): Likewise.
27449 (read_write_all_from_decl): Likewise.
27450 (propagate): Likewise.
27451 (ipa_reference_read_optimization_summary): Likewise.
27452 * ipa-utils.c (ipa_merge_profiles): Likewise.
27453 * ipa.c (walk_polymorphic_call_targets): Likewise.
27454 (symbol_table::remove_unreachable_nodes): Likewise.
27455 (ipa_single_use): Likewise.
27456 * passes.c (execute_todo): Likewise.
27457 * predict.c (drop_profile): Likewise.
27458 * symtab.c (symtab_node::get_dump_name): New function.
27459 (symtab_node::dump_name): Likewise.
27460 (symtab_node::dump_asm_name): Likewise.
27461 (symtab_node::dump_references): Likewise.
27462 (symtab_node::dump_referring): Likewise.
27463 (symtab_node::dump_base): Likewise.
27464 (symtab_node::debug_symtab): Likewise.
27465 * tree-sra.c (convert_callers_for_node): Likewise.
27466 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27467 * value-prof.c (init_node_map): Likewise.
27469 2017-05-23 Martin Liska <mliska@suse.cz>
27471 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27472 and symtab_node::debug_symtab to symbol_table::debug.
27473 * cgraphunit.c (analyze_functions): Use the renamed function.
27474 (symbol_table::compile): Likewise.
27475 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27476 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27477 * passes.c (execute_todo): Likewise.
27478 * symtab.c (symbol_table::dump): New function.
27479 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27481 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27483 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27484 that nonconst implies exec.
27486 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27488 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27489 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27490 (inline_edge_summary_vec): Turn into ...
27491 (ipa_call_summaries): ... this one.
27492 (redirect_to_unreachable, edge_set_predicate,
27493 evaluate_properties_for_edge, inline_summary_alloc,
27494 reset_ipa_call_summary, reset_inline_summary,
27495 inline_summary_t::duplicate): Update.
27496 (inline_edge_duplication_hook): Turn to ...
27497 (ipa_call_summary_t::duplicate): ... this one.
27498 (inline_edge_removal_hook): Turn to ...
27499 (ipa_call_summary_t::remove): ... this one.
27500 (dump_inline_edge_summary): Turn to ...
27501 (dump_ipa_call_summary): ... this one.
27502 (estimate_function_body_sizes): Update.
27503 (inline_update_callee_summaries): Update.
27504 (remap_edge_change_prob): Update.
27505 (remap_edge_summaries): Update.
27506 (inline_merge_summary): Update.
27507 (do_estimate_edge_time): Update.
27508 (inline_generate_summary): Update.
27509 (inline_read_section): Update.
27510 (inline_read_summary): Update.
27511 (inline_free_summary): Update.
27512 * ipa-inline.c (can_inline_edge_p): Update.
27513 (compute_inlined_call_time): Update.
27514 (want_inline_small_function_p): Update.
27515 (edge_badness): Update.
27516 (early_inliner): Update.
27517 * ipa-inline.h (inline_edge_summary): Turn to ...
27518 (ipa_call_summary): ... this one.
27519 (ipa_call_summary_t): New class.
27520 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27521 (ipa_call_summaries): New.
27522 (inline_edge_summary): Remove.
27523 (estimate_edge_growth): Update.
27524 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27525 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27526 * ipa-split.c (execute_split_functions): Update.
27527 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27529 2017-05-23 Tom de Vries <tom@codesourcery.com>
27531 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27532 attributes): Document rdrand effective target.
27534 2017-05-23 Tom de Vries <tom@codesourcery.com>
27536 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27537 attributes): Sort alphabetically.
27539 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27541 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27543 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27546 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27547 V2DF/V2DI splat into two separate patterns, one that handles
27548 registers, and the other that only handles memory. Drop support
27549 for splatting from a GPR on ISA 2.07 and then splitting the
27550 splat into direct move and splat.
27551 (vsx_splat_<mode>_reg): Likewise.
27552 (vsx_splat_<mode>_mem): Likewise.
27554 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27556 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27558 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27560 PR middle-end/80809
27561 * omp-low.c (finish_taskreg_remap): New function.
27562 (finish_taskreg_scan): If unit size of ctx->record_type
27563 is non-constant, unshare the size expression and replace
27564 decls in it with possible outer var refs.
27566 PR middle-end/80809
27567 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27568 GOVD_SHARED rather than GOVD_PRIVATE with it.
27569 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27570 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27572 PR middle-end/80853
27573 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27574 as last argument to build_outer_var_ref for pointer bases of array
27575 section reductions.
27577 2017-05-19 Martin Sebor <msebor@redhat.com>
27579 * print-tree.c (print_node): Print DECL_READ_P flag.
27581 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27583 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27584 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27585 * cgraph.c: Likewise.
27586 * cgraphunit.c: Likewise.
27587 * gengtype.c: Likewise.
27588 * ipa-cp.c: Likewise.
27589 * ipa-devirt.c: Likewise.
27590 * ipa-icf.c: Likewise.
27591 * ipa-predicate.c: Likewise.
27592 * ipa-profile.c: Likewise.
27593 * ipa-prop.c: Likewise.
27594 * ipa-split.c: Likewise.
27596 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27597 edge_predicate_pool, dump_inline_hints,
27598 inline_summary::account_size_time, redirect_to_unreachable,
27599 edge_set_predicate, set_hint_predicate,
27600 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27601 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27602 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27603 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27604 ipa_call_summary_t::remove, initialize_growth_caches,
27605 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27606 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27607 mark_modified, unmodified_parm_1, unmodified_parm,
27608 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27609 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27610 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27611 will_be_nonconstant_predicate, record_modified_bb_info,
27612 get_minimal_bb, record_modified, param_change_prob,
27613 phi_result_unknown_predicate, predicate_for_phi_result,
27614 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27615 estimate_function_body_sizes, compute_inline_parameters,
27616 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27617 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27618 inline_update_callee_summaries, remap_edge_change_prob,
27619 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27620 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27621 inline_analyze_function, inline_summary_t::insert,
27622 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27623 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27624 inline_free_summary): Move to ipa-fnsummary.h
27625 (predicate_t): Remove.
27626 * ipa-fnsummary.c: New file.
27627 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27628 (enum inline_hints_vals, inline_hints, agg_position_info,
27629 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27630 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27631 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27632 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27633 inline_read_summary, inline_write_summary, inline_free_summary,
27634 inline_analyze_function, initialize_inline_failed,
27635 inline_merge_summary, inline_update_overall_summary,
27636 compute_inline_parameters): Move to ipa-fnsummary.h
27637 * ipa-fnsummary.h: New file.
27638 * ipa-inline-transform.h: Include ipa-inline.h.
27639 * ipa-inline.c: LIkewise.
27641 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27643 * ipa-inline.c (edge_badness): Use inlined_time instead of
27644 inline_summaries->get.
27646 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27648 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27650 2017-05-22 Nathan Sidwell <nathan@acm.org>
27652 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27653 (fdump-lang): Document 'raw' option.
27654 * dumpfile.h (TDI_tu): Delete.
27655 * dumpfile.c (dump_files): Remove translation-unit.
27656 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27658 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27660 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27661 command option from $(AWK) call.
27662 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27663 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27664 [FORMAT]: Remove handling of variable.
27665 * config/avr/t-multilib: Regenerate.
27667 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27669 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27671 (dump_inline_summary): Do not print self_time.
27672 (estimate_function_body_sizes): Do not set self_time.
27673 (compute_inline_parameters): Likewise.
27674 (inline_read_section, inline_write_summary): Do not stream self_time.
27675 * ipa-inline.h (inline_summary): Drop self_time.
27677 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27679 * ipa-inline-analysis.c (account_size_time): Rename to ...
27680 (inline_summary::account_size_time): ... this one.
27681 (reset_ipa_call_summary): Turn to ...
27682 (ipa_call_summary::reset): ... this one.
27683 (reset_inline_summary): Turn to ...
27684 (inline_summary::reset): ... this one.
27685 (inline_summary_t::remove): Update.
27686 (inline_summary_t::duplicate): Update.
27687 (ipa_call_summary_t::remove): Update.
27688 (dump_inline_summary): Update.
27689 (estimate_function_body_sizes): Update.
27690 (compute_inline_parameters): Update.
27691 (estimate_node_size_and_time): Update.
27692 (inline_merge_summary): Update.
27693 (inline_update_overall_summary): Update.
27694 (inline_read_section): Update.
27695 (inline_write_summary): Update.
27696 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27697 add account_size_time and reset member functions.
27698 (ipa_call_summary): Add reset function.
27699 * ipa-predicate.h (predicate::operator &): Constify.
27701 2017-05-22 Richard Biener <rguenther@suse.de>
27703 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27705 2017-05-19 Jason Merrill <jason@redhat.com>
27707 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27709 2017-05-19 Marek Polacek <polacek@redhat.com>
27712 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27713 TYPE_OVERFLOW_WRAPS checks.
27715 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27717 * tree-core.h (enum omp_clause_default_kind): Add
27718 "OMP_CLAUSE_DEFAULT_PRESENT".
27719 * tree-pretty-print.c (dump_omp_clause): Handle it.
27720 * gimplify.c (enum gimplify_omp_var_data): Add
27721 "GOVD_MAP_FORCE_PRESENT".
27722 (gimplify_adjust_omp_clauses_1): Map it to
27723 "GOMP_MAP_FORCE_PRESENT".
27724 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27726 * gimplify.c (oacc_default_clause): Clarify.
27728 2017-05-19 Nathan Sidwell <nathan@acm.org>
27730 LANG_HOOK_REGISTER_DUMPS
27731 * toplev.c (general_init): Call register dump lang hook.
27732 * doc/invoke.texi: Document -fdump-lang option family.
27733 * dumpfile.c (dump_files): Remove class dump here.
27734 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27735 * dumpfile.h (tree_dump_index): Remove TDI_class.
27736 * langhooks-def.h (lhd_register_dumps): Declare.
27737 (LANG_HOOKS_REGISTER_DUMPS): Define.
27738 (LANG_HOOKS_INITIALIZER): Add it.
27739 * langhooks.c (lhd_register_dumps): Define.
27740 * langhooks.h (struct lang_hooks): Add register_dumps.
27742 2017-05-19 Nathan Sidwell <nathan@acm.org>
27744 * context.h (context::set_passes): New.
27745 * context.c (context::context): Do not create pass manager.
27746 * toplev.c (general_init): Create pass manager here.
27748 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27750 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27751 use this splitter if two add or or instructions would also work for
27752 the constant we want to generate.
27754 2017-05-19 Richard Biener <rguenther@suse.de>
27757 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27758 predicate evaluation.
27760 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27762 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27764 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27767 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27769 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27770 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27771 (inline_edge_summary_vec): Turn into ...
27772 (ipa_call_summaries): ... this one.
27773 (redirect_to_unreachable, edge_set_predicate,
27774 evaluate_properties_for_edge, inline_summary_alloc,
27775 reset_ipa_call_summary, reset_inline_summary,
27776 inline_summary_t::duplicate): Update.
27777 (inline_edge_duplication_hook): Turn to ...
27778 (ipa_call_summary_t::duplicate): ... this one.
27779 (inline_edge_removal_hook): Turn to ...
27780 (ipa_call_summary_t::remove): ... this one.
27781 (dump_inline_edge_summary): Turn to ...
27782 (dump_ipa_call_summary): ... this one.
27783 (estimate_function_body_sizes): Update.
27784 (inline_update_callee_summaries): Update.
27785 (remap_edge_change_prob): Update.
27786 (remap_edge_summaries): Update.
27787 (inline_merge_summary): Update.
27788 (do_estimate_edge_time): Update.
27789 (inline_generate_summary): Update.
27790 (inline_read_section): Update.
27791 (inline_read_summary): Update.
27792 (inline_free_summary): Update.
27793 * ipa-inline.c (can_inline_edge_p): Update.
27794 (compute_inlined_call_time): Update.
27795 (want_inline_small_function_p): Update.
27796 (edge_badness): Update.
27797 (early_inliner): Update.
27798 * ipa-inline.h (inline_edge_summary): Turn to ...
27799 (ipa_call_summary): ... this one.
27800 (ipa_call_summary_t): New class.
27801 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27802 (ipa_call_summaries): New.
27803 (inline_edge_summary): Remove.
27804 (estimate_edge_growth): Update.
27805 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27806 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27807 * ipa-split.c (execute_split_functions): Update.
27808 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27810 2017-05-19 Richard Biener <rguenther@suse.de>
27812 PR middle-end/80764
27813 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27815 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27817 * config/rs6000/rs6000.c (struct machine_function): Add field
27818 fpr_is_wrapped_separately.
27819 (rs6000_get_separate_components): Use 64 components. Handle the
27820 new FPR components.
27821 (rs6000_components_for_bb): Handle the FPR components.
27822 (rs6000_emit_prologue_components): Handle the FPR components.
27823 (rs6000_emit_epilogue_components): Handle the FPR components.
27824 (rs6000_set_handled_components): Handle the FPR components.
27825 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27826 that are already separately shrink-wrapped.
27827 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27828 that are already separately shrink-wrapped.
27830 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27833 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27836 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27837 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27838 cases where the register allocator uses a move and an offsettable
27839 memory operation to/from a FPR register on ISA 2.06/2.07.
27840 (define_peephole2 for Altivec d-form store): Likewise.
27842 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27845 * config/i386/mmx.md (*mov<mode>_internal): Enable
27846 alternatives 11, 12, 13 and 14 also for 32bit targets.
27847 Remove alternatives 15, 16, 17 and 18.
27848 * config/i386/sse.md (vec_concatv2di): Change
27849 alternative (!x, *y) to (x, ?!*Yn).
27851 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27853 * dumpfile.h (enum dump_kind): Remove stray comma.
27855 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27857 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27858 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27859 predicate::num_conditions
27860 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27861 (CHANGED): turn into predicate::changed.
27862 (agg_position_info): Move to ipa-predicate.h
27863 (add_condition, predicate::add_clause, predicate::operator &=,
27864 predicate::or_with, predicate::evaluate, predicate::probability,
27865 dump_condition, dump_clause, predicate::dump,
27866 predicate::remap_after_duplication, predicate::remap_after_inlining,
27867 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27868 (evaluate_conditions_for_known_args): Update.
27869 (set_cond_stmt_execution_predicate): Update.
27870 * ipa-inline.h: Include ipa-predicate.h
27871 (condition, inline_param_summary, conditions, agg_position_info,
27872 predicate): Move to ipa-predicate.h
27873 * ipa-predicate.c: New file.
27874 * ipa-predicate.h: New file.
27876 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27878 * final.c (leaf_function_p): Check we are not in a sequence.
27880 2017-05-18 Martin Liska <mliska@suse.cz>
27882 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27883 * dumpfile.c (dump_register): Use new enum dump_kind.
27884 (get_dump_file_name): Likewise.
27885 (dump_enable_all): Likewise.
27886 (dump_switch_p_1): Likewise.
27887 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27888 * dumpfile.h (enum dump_kind): New enum type.
27889 (struct dump_file_info): Create constructor and
27890 format fields and comments.
27891 * passes.c (pass_manager::register_one_dump_file):
27893 * statistics.c (statistics_early_init): Likewise.
27894 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27895 TDF_TREE with TDF_SLIM.
27896 (gather_memory_references_ref): Likewise.
27898 2017-05-18 Martin Liska <mliska@suse.cz>
27900 * vec.h (struct vnull): Use it.
27902 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27904 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27905 (true_predicate, false_predicate, true_predicate_p,
27906 false_predicate_p): Remove.
27907 (single_cond_predicate, not_inlined_predicate): Turn to member function
27909 (add_condition): Update.
27910 (add_clause): Turn to...
27911 (predicate::add_clause): ... this one; update; allow passing NULL
27913 (and_predicates): Turn to ...
27914 (predicate::operator &=): ... this one.
27915 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27916 (or_predicates): Turn to ...
27917 (predicate::or_with): ... this one.
27918 (evaluate_predicate): Turn to ...
27919 (predicate::evaluate): ... this one.
27920 (predicate_probability): Turn to ...
27921 (predicate::probability): ... this one.
27922 (dump_condition): Update.
27923 (dump_predicate): Turn to ...
27924 (predicate::dump): ... this one.
27925 (account_size_time): Update.
27926 (edge_set_predicate): Update.
27927 (set_hint_predicate): UPdate.
27928 (evaluate_conditions_for_known_args): Update.
27929 (evaluate_properties_for_edge): Update.
27930 (remap_predicate_after_duplication): Turn to...
27931 (predicate::remap_after_duplication): ... this one.
27932 (remap_hint_predicate_after_duplication): Update.
27933 (inline_summary_t::duplicate): UPdate.
27934 (dump_inline_edge_summary): Update.
27935 (dump_inline_summary): Update.
27936 (set_cond_stmt_execution_predicate): Update.
27937 (set_switch_stmt_execution_predicate): Update.
27938 (compute_bb_predicates): Update.
27939 (will_be_nonconstant_expr_predicate): Update.
27940 (will_be_nonconstant_predicate): Update.
27941 (phi_result_unknown_predicate): Update.
27942 (predicate_for_phi_result): Update.
27943 (array_index_predicate): Update.
27944 (estimate_function_body_sizes): Update.
27945 (estimate_node_size_and_time): Update.
27946 (estimate_ipcp_clone_size_and_time): Update.
27947 (remap_predicate): Rename to ...
27948 (predicate::remap_after_inlining): ... this one.
27949 (remap_hint_predicate): Update.
27950 (inline_merge_summary): Update.
27951 (inline_update_overall_summary): Update.
27952 (estimate_size_after_inlining): Update.
27953 (read_predicate): Rename to ...
27954 (predicate::stream_in): ... this one.
27955 (read_inline_edge_summary): Update.
27956 (write_predicate): Rename to ...
27957 (predicate::stream_out): ... this one.
27958 (write_inline_edge_summary): Update.
27959 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27960 (clause_t): Turn to uint32_t
27961 (predicate): Turn to class; implement constructor and operators
27963 (size_time_entry): Update.
27964 (inline_summary): Update.
27965 (inline_edge_summary): Update.
27967 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27969 * fold-const.c (fold_binary_loc): Move transformation...
27970 * match.pd (C - X CMP X): ... here.
27972 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27974 * config/sparc/sparc.c (sparc_option_override): Set function
27975 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27976 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27978 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27980 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27982 2017-05-18 Marek Polacek <polacek@redhat.com>
27985 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27986 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27987 gimple_assign_load_p.
27989 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27991 PR middle-end/80692
27992 * real.c (do_compare): Give decimal_do_compare preference over
27993 comparing just the signs.
27995 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27997 * doc/md.texi (Canonicalization of Instructions): Describe the
27998 canonical form of instructions that inherently set a condition
28001 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28003 PR middle-end/80775
28004 * tree-cfg.c: Move deletion of unreachable case statements to after
28005 the merging of consecutive case labels.
28007 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28009 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28010 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28011 restoring of callee-saved registers.
28013 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28015 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28016 * config/visium/visium.c (single_set_and_flags): Likewise.
28017 * config/visium/visium.md (Substitutions): Likewise.
28019 2017-05-17 Martin Liska <mliska@suse.cz>
28021 * cfg.c: Introduce dump_flags_t type and
28022 use it instead of int type.
28024 * cfghooks.c: Likewise.
28025 * cfghooks.h (struct cfg_hooks): Likewise.
28026 * cfgrtl.c: Likewise.
28027 * cfgrtl.h: Likewise.
28028 * cgraph.c (cgraph_node::get_body): Likewise.
28029 * coretypes.h: Likewise.
28030 * domwalk.c: Likewise.
28031 * domwalk.h: Likewise.
28032 * dumpfile.c (struct dump_option_value_info): Likewise.
28033 (dump_enable_all): Likewise.
28034 (dump_switch_p_1): Likewise.
28035 (opt_info_switch_p): Likewise.
28036 * dumpfile.h (enum tree_dump_index): Likewise.
28037 (struct dump_file_info): Likewise.
28038 * genemit.c: Likewise.
28039 * generic-match-head.c: Likewise.
28040 * gengtype.c (open_base_files): Likewise.
28041 * gimple-pretty-print.c: Likewise.
28042 * gimple-pretty-print.h: Likewise.
28043 * graph.c (print_graph_cfg): Likewise.
28044 * graphite-scop-detection.c (dot_all_sese): Likewise.
28045 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28046 * loop-unroll.c (report_unroll): Likewise.
28047 * passes.c (pass_manager::register_one_dump_file): Likewise.
28048 * print-tree.c: Likewise.
28049 * statistics.c: Likewise.
28050 * tree-cfg.c: Likewise.
28051 * tree-cfg.h: Likewise.
28052 * tree-dfa.c: Likewise.
28053 * tree-dfa.h: Likewise.
28054 * tree-dump.c (dump_function): Likewise.
28055 * tree-dump.h (struct dump_info): Likewise.
28056 * tree-pretty-print.c: Likewise.
28057 * tree-pretty-print.h: Likewise.
28058 * tree-ssa-live.c: Likewise.
28059 * tree-ssa-live.h: Likewise.
28060 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28061 * tree-vect-loop.c: Likewise.
28062 * tree-vect-slp.c: Likewise.
28064 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28065 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28067 PR tree-optimization/80457
28068 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28069 of all arguments to a statement as scalar_to_vec operations.
28070 (vectorizable_call): Adjust call to vect_model_simple_cost for
28072 (vectorizable_conversion): Likewise.
28073 (vectorizable_assignment): Likewise.
28074 (vectorizable_shift): Likewise.
28075 (vectorizable_operation): Likewise.
28076 (vectorizable_comparison): Likewise.
28077 (vect_is_simple_cond): Record the def types for operands.
28078 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28079 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28080 for statement argument count.
28082 2017-05-16 Carl Love <cel@us.ibm.com>
28084 * config/rs6000/rs6000-c: Add support for built-in functions
28085 vector unsigned long long vec_bperm (vector unsigned long long,
28086 vector unsigned char)
28087 vector signed long long vec_mule (vector signed int,
28089 vector unsigned long long vec_mule (vector unsigned int,
28090 vector unsigned int)
28091 vector signed long long vec_mulo (vector signed int,
28093 vector unsigned long long vec_mulo (vector unsigned int,
28094 vector unsigned int)
28095 vector signed char vec_sldw (vector signed char,
28096 vector signed char,
28098 vector unsigned char vec_sldw (vector unsigned char,
28099 vector unsigned char,
28101 vector signed short vec_sldw (vector signed short,
28102 vector signed short,
28104 vector unsigned short vec_sldw (vector unsigned short,
28105 vector unsigned short,
28107 vector signed int vec_sldw (vector signed int,
28110 vector unsigned int vec_sldw (vector unsigned int,
28111 vector unsigned int,
28113 vector signed long long vec_sldw (vector signed long long,
28114 vector signed long long,
28116 vector unsigned long long vec_sldw (vector unsigned long long,
28117 vector unsigned long long,
28119 * config/rs6000/rs6000-c: Add support for built-in functions
28120 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28121 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28122 * doc/extend.texi: Update the built-in documentation for the
28123 new built-in functions.
28125 2017-05-16 Marek Polacek <polacek@redhat.com>
28129 * tree.c (save_expr): Don't fold the expression.
28131 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28133 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28134 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28135 and (?*y,m). Update insn attributes.
28137 2017-05-16 Martin Liska <mliska@suse.cz>
28139 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28140 flags argument of print_gimple_stmt, print_gimple_expr,
28141 print_generic_stmt and print_generic_expr.
28142 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28143 * coretypes.h: Likewise.
28144 * except.c (dump_eh_tree): Likewise.
28145 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28146 * gimple-pretty-print.h: Likewise.
28147 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28148 (backprop::push_to_worklist): Likewise.
28149 (backprop::pop_from_worklist): Likewise.
28150 (backprop::process_use): Likewise.
28151 (backprop::intersect_uses): Likewise.
28152 (note_replacement): Likewise.
28153 * gimple-ssa-store-merging.c
28154 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28155 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28156 (pass_store_merging::execute): Likewise.
28157 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28158 (ssa_base_cand_dump_callback): Likewise.
28159 (dump_incr_vec): Likewise.
28160 (replace_refs): Likewise.
28161 (replace_mult_candidate): Likewise.
28162 (create_add_on_incoming_edge): Likewise.
28163 (create_phi_basis): Likewise.
28164 (insert_initializers): Likewise.
28165 (all_phi_incrs_profitable): Likewise.
28166 (introduce_cast_before_cand): Likewise.
28167 (replace_one_candidate): Likewise.
28168 * gimplify.c (gimplify_expr): Likewise.
28169 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28170 (set_rename): Likewise.
28171 (rename_uses): Likewise.
28172 (copy_loop_phi_nodes): Likewise.
28173 (add_close_phis_to_merge_points): Likewise.
28174 (copy_loop_close_phi_args): Likewise.
28175 (copy_cond_phi_args): Likewise.
28176 (graphite_copy_stmts_from_block): Likewise.
28177 (translate_pending_phi_nodes): Likewise.
28178 * graphite-poly.c (print_pdr): Likewise.
28179 (dump_gbb_cases): Likewise.
28180 (dump_gbb_conditions): Likewise.
28181 (print_scop_params): Likewise.
28182 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28183 (build_cross_bb_scalars_use): Likewise.
28184 (gather_bbs::before_dom_children): Likewise.
28185 * hsa-dump.c (dump_hsa_immed): Likewise.
28186 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28187 (get_replacement_map): Likewise.
28188 * ipa-inline-analysis.c (dump_condition): Likewise.
28189 (estimate_function_body_sizes): Likewise.
28190 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28191 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28192 * ipa-prop.c (ipa_dump_param): Likewise.
28193 (ipa_print_node_jump_functions_for_edge): Likewise.
28194 (ipa_modify_call_arguments): Likewise.
28195 (ipa_modify_expr): Likewise.
28196 (ipa_dump_param_adjustments): Likewise.
28197 (ipa_dump_agg_replacement_values): Likewise.
28198 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28199 * ipa-pure-const.c (check_stmt): Likewise.
28200 (pass_nothrow::execute): Likewise.
28201 * ipa-split.c (execute_split_functions): Likewise.
28202 * omp-offload.c (dump_oacc_loop_part): Likewise.
28203 (dump_oacc_loop): Likewise.
28204 * trans-mem.c (tm_log_emit): Likewise.
28205 (tm_memopt_accumulate_memops): Likewise.
28206 (dump_tm_memopt_set): Likewise.
28207 (dump_tm_memopt_transform): Likewise.
28208 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28209 (print_loop): Likewise.
28210 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28211 (chkp_gather_checks_info): Likewise.
28212 (chkp_get_check_result): Likewise.
28213 (chkp_remove_check_if_pass): Likewise.
28214 (chkp_use_outer_bounds_if_possible): Likewise.
28215 (chkp_reduce_bounds_lifetime): Likewise.
28216 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28217 (chkp_mark_completed_bounds): Likewise.
28218 (chkp_register_incomplete_bounds): Likewise.
28219 (chkp_mark_invalid_bounds): Likewise.
28220 (chkp_maybe_copy_and_register_bounds): Likewise.
28221 (chkp_build_returned_bound): Likewise.
28222 (chkp_get_bound_for_parm): Likewise.
28223 (chkp_build_bndldx): Likewise.
28224 (chkp_get_bounds_by_definition): Likewise.
28225 (chkp_generate_extern_var_bounds): Likewise.
28226 (chkp_get_bounds_for_decl_addr): Likewise.
28227 * tree-chrec.c (chrec_apply): Likewise.
28228 * tree-data-ref.c (dump_data_reference): Likewise.
28229 (dump_subscript): Likewise.
28230 (dump_data_dependence_relation): Likewise.
28231 (analyze_overlapping_iterations): Likewise.
28232 * tree-inline.c (expand_call_inline): Likewise.
28233 (tree_function_versioning): Likewise.
28234 * tree-into-ssa.c (dump_defs_stack): Likewise.
28235 (dump_currdefs): Likewise.
28236 (dump_names_replaced_by): Likewise.
28237 (dump_update_ssa): Likewise.
28238 (update_ssa): Likewise.
28239 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28240 * tree-parloops.c (build_new_reduction): Likewise.
28241 (try_create_reduction_list): Likewise.
28242 (ref_conflicts_with_region): Likewise.
28243 (oacc_entry_exit_ok_1): Likewise.
28244 (oacc_entry_exit_single_gang): Likewise.
28245 * tree-pretty-print.h: Likewise.
28246 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28247 (get_scalar_evolution): Likewise.
28248 (add_to_evolution): Likewise.
28249 (get_loop_exit_condition): Likewise.
28250 (analyze_evolution_in_loop): Likewise.
28251 (analyze_initial_condition): Likewise.
28252 (analyze_scalar_evolution): Likewise.
28253 (instantiate_scev): Likewise.
28254 (number_of_latch_executions): Likewise.
28255 (gather_chrec_stats): Likewise.
28256 (final_value_replacement_loop): Likewise.
28257 (scev_const_prop): Likewise.
28258 * tree-sra.c (dump_access): Likewise.
28259 (disqualify_candidate): Likewise.
28260 (create_access): Likewise.
28261 (reject): Likewise.
28262 (maybe_add_sra_candidate): Likewise.
28263 (create_access_replacement): Likewise.
28264 (analyze_access_subtree): Likewise.
28265 (analyze_all_variable_accesses): Likewise.
28266 (sra_modify_assign): Likewise.
28267 (initialize_constant_pool_replacements): Likewise.
28268 (find_param_candidates): Likewise.
28269 (decide_one_param_reduction): Likewise.
28270 (replace_removed_params_ssa_names): Likewise.
28271 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28272 * tree-ssa-copy.c (dump_copy_of): Likewise.
28273 (copy_prop_visit_cond_stmt): Likewise.
28274 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28275 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28276 (record_equivalences_from_stmt): Likewise.
28277 * tree-ssa-dse.c (compute_trims): Likewise.
28278 (delete_dead_call): Likewise.
28279 (delete_dead_assignment): Likewise.
28280 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28281 (forward_propagate_into_cond): Likewise.
28282 (pass_forwprop::execute): Likewise.
28283 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28284 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28286 (move_computations_worker): Likewise.
28287 (execute_sm): Likewise.
28288 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28289 (remove_exits_and_undefined_stmts): Likewise.
28290 (remove_redundant_iv_tests): Likewise.
28291 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28292 (adjust_iv_update_pos): Likewise.
28293 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28294 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28295 (value_replacement): Likewise.
28296 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28297 * tree-ssa-pre.c (print_pre_expr): Likewise.
28298 (get_representative_for): Likewise.
28299 (create_expression_by_pieces): Likewise.
28300 (insert_into_preds_of_block): Likewise.
28301 (eliminate_insert): Likewise.
28302 (eliminate_dom_walker::before_dom_children): Likewise.
28303 (eliminate): Likewise.
28304 (remove_dead_inserted_code): Likewise.
28305 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28306 * tree-ssa-reassoc.c (get_rank): Likewise.
28307 (eliminate_duplicate_pair): Likewise.
28308 (eliminate_plus_minus_pair): Likewise.
28309 (eliminate_not_pairs): Likewise.
28310 (undistribute_ops_list): Likewise.
28311 (eliminate_redundant_comparison): Likewise.
28312 (update_range_test): Likewise.
28313 (optimize_range_tests_var_bound): Likewise.
28314 (optimize_vec_cond_expr): Likewise.
28315 (rewrite_expr_tree): Likewise.
28316 (rewrite_expr_tree_parallel): Likewise.
28317 (linearize_expr): Likewise.
28318 (break_up_subtract): Likewise.
28319 (linearize_expr_tree): Likewise.
28320 (attempt_builtin_powi): Likewise.
28321 (attempt_builtin_copysign): Likewise.
28322 (transform_stmt_to_copy): Likewise.
28323 (transform_stmt_to_multiply): Likewise.
28324 (dump_ops_vector): Likewise.
28325 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28326 (print_scc): Likewise.
28327 (set_ssa_val_to): Likewise.
28328 (visit_reference_op_store): Likewise.
28329 (visit_use): Likewise.
28330 (sccvn_dom_walker::before_dom_children): Likewise.
28331 (run_scc_vn): Likewise.
28332 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28334 (expr_hash_elt::print): Likewise.
28335 (const_and_copies::pop_to_marker): Likewise.
28336 (const_and_copies::record_const_or_copy_raw): Likewise.
28337 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28338 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28339 (dump_predicates): Likewise.
28340 (find_uninit_use): Likewise.
28341 (warn_uninitialized_phi): Likewise.
28342 (pass_late_warn_uninitialized::execute): Likewise.
28343 * tree-ssa.c (verify_vssa): Likewise.
28344 (verify_ssa): Likewise.
28345 (maybe_optimize_var): Likewise.
28346 * tree-vrp.c (dump_value_range): Likewise.
28347 (dump_all_value_ranges): Likewise.
28348 (dump_asserts_for): Likewise.
28349 (register_edge_assert_for_2): Likewise.
28350 (vrp_visit_cond_stmt): Likewise.
28351 (vrp_visit_switch_stmt): Likewise.
28352 (vrp_visit_stmt): Likewise.
28353 (vrp_visit_phi_node): Likewise.
28354 (simplify_cond_using_ranges_1): Likewise.
28355 (fold_predicate_in): Likewise.
28356 (evrp_dom_walker::before_dom_children): Likewise.
28357 (evrp_dom_walker::push_value_range): Likewise.
28358 (evrp_dom_walker::pop_value_range): Likewise.
28359 (execute_early_vrp): Likewise.
28361 2017-05-16 Richard Biener <rguenther@suse.de>
28363 * dwarf2out.c (loc_list_from_tree_1): Do not create
28364 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28366 2017-05-16 Richard Biener <rguenther@suse.de>
28368 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28370 (note_variable_value_in_expr): If we resolved the decl ref
28371 do not push to the stack.
28373 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28375 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28376 operations in fast-math mode.
28377 (vaddq_f16): Likewise.
28378 (vmul_f16): Likewise.
28379 (vmulq_f16): Likewise.
28380 (vsub_f16): Likewise.
28381 (vsubq_f16): Likewise.
28382 * config/arm/neon.md (add<mode>3): New.
28384 (fma:<VH:mode>3): New. Also remove outdated comment.
28387 2017-05-16 Martin Liska <mliska@suse.cz>
28391 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28392 (odr_types_equivalent_p): Likewise.
28394 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28396 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28398 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28401 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28402 non-interunit SSE move alternatives with '?'.
28403 (zero-extendsidi peephole2): New peephole to skip intermediate
28404 general register in SSE zero-extend sequence.
28406 2017-05-15 Jeff Law <law@redhat.com>
28408 * reorg.c (relax_delay_slots): Create a new variable to hold
28409 the temporary target rather than clobbering TARGET_LABEL.
28411 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28412 missing argument to extract_bit_field call.
28413 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28415 2017-05-15 Martin Liska <mliska@suse.cz>
28418 * gcc.c (process_command): Do not allow empty argument of -o option.
28420 2017-05-15 Renlin Li <renlin.li@arm.com>
28422 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28423 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28424 * config/aarch64/constraints.md (Usf): Add long call check.
28425 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28426 (call_value): Likewise.
28427 (sibcall): Likewise.
28428 (sibcall_value): Likewise.
28430 (call_value_insn): New.
28431 (sibcall_insn): Update rtx pattern.
28432 (sibcall_value_insn): Likewise.
28433 (call_internal): Remove.
28434 (call_value_internal): Likewise.
28435 (sibcall_internal): Likewise.
28436 (sibcall_value_internal): Likewise.
28437 (call_reg): Likewise.
28438 (call_symbol): Likewise.
28439 (call_value_reg): Likewise.
28440 (call_value_symbol): Likewise.
28442 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28445 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28447 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28449 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28450 compatible with CCGOCmode and with CCZmode.
28452 2017-05-14 Martin Sebor <msebor@redhat.com>
28454 PR middle-end/77671
28455 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28456 (gimple_fold_builtin_snprintf): Same.
28457 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28458 (gimple_fold_builtin_snprintf): Same.
28459 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28460 of character types.
28461 (is_call_safe): New function.
28462 (try_substitute_return_value): Call it.
28463 (try_simplify_call): New function.
28464 (pass_sprintf_length::handle_gimple_call): Call it.
28466 2017-05-14 Martin Sebor <msebor@redhat.com>
28468 PR middle-end/80669
28469 * builtins.c (expand_builtin_stpncpy): Simplify.
28471 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28473 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28474 * config/i386/i386.h
28475 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28476 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28477 (struct machine_function): Add new members call_ms2sysv,
28478 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28479 (struct machine_frame_state): New fields sp_realigned and
28480 sp_realigned_offset.
28481 * config/i386/i386.c
28482 (enum xlogue_stub): New enum.
28483 (enum xlogue_stub_sets): New enum.
28484 (class xlogue_layout): New class.
28485 (struct ix86_frame): New fields stack_realign_allocate_offset,
28486 stack_realign_offset and outlined_save_offset. Modify comments to
28487 detail stack layout when using out-of-line stubs.
28488 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28489 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28490 -mcall-ms2sysv-xlogues.
28491 (stub_managed_regs): New static variable.
28492 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28493 registers managed by out-of-line stub.
28494 (disable_call_ms2sysv_xlogues): New function.
28495 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28496 m->call_ms2sysv when appropriate and compute frame layout for
28498 (sp_valid_at, fp_valid_at): New inline functions.
28499 (choose_basereg): New function.
28500 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28502 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28503 Use align parameter of choose_baseaddr to generated aligned SSE movs
28505 (pro_epilogue_adjust_stack): Modify to track
28506 machine_frame_state::sp_realigned.
28507 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28508 (ix86_nsaved_sseregs): Likewise.
28509 (ix86_emit_save_regs): Likewise.
28510 (ix86_emit_save_regs_using_mov): Likewise.
28511 (ix86_emit_save_sse_regs_using_mov): Likewise.
28512 (get_scratch_register_on_entry): Likewise.
28513 (gen_frame_set): New function.
28514 (gen_frame_load): Likewise.
28515 (gen_frame_store): Likewise.
28516 (emit_outlined_ms2sysv_save): Likewise.
28517 (emit_outlined_ms2sysv_restore): Likewise.
28518 (ix86_expand_prologue): Modify stack re-alignment code and call
28519 emit_outlined_ms2sysv_save when appropriate.
28520 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28521 parameter rtx_insn *insn, which allows the function to be used to only
28522 generate the notes.
28523 (ix86_expand_epilogue): Modify validity checks of frame and stack
28524 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28525 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28526 * config/i386/predicates.md
28527 (save_multiple): New predicate.
28528 (restore_multiple): Likewise.
28529 * config/i386/sse.md
28530 (save_multiple<mode>): New pattern.
28531 (save_multiple_realign<mode>): Likewise.
28532 (restore_multiple<mode>): Likewise.
28533 (restore_multiple_and_return<mode>): Likewise.
28534 (restore_multiple_leave_return<mode>): Likewise.
28535 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28537 2017-05-14 Julia Koval <julia.koval@intel.com>
28539 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28540 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28541 (__builtin_ia32_xsetbv): New builtins.
28542 * config/i386/i386.c (ix86_expand_special_args_builtin):
28544 (ix86_expand_builtin): Special expand for new intrinsics.
28545 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28546 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28547 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28549 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28551 * cfganal.c (inverted_post_order_compute): Change argument type
28553 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28554 * df-core.c (rest_of_handle_df_initialize): Adjust.
28555 (rest_of_handle_df_finish): Likewise.
28556 (df_analyze_1): Likewise.
28557 (df_analyze): Likewise.
28558 (loop_inverted_post_order_compute): Change argument to be a vec *.
28559 (df_analyze_loop): Adjust.
28560 (df_get_n_blocks): Likewise.
28561 (df_get_postorder): Likewise.
28562 * df.h (struct df_d): Change field to be a vec.
28563 * lcm.c (compute_laterin): Adjust.
28564 (compute_available): Likewise.
28565 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28566 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28567 * tree-ssa-pre.c (compute_antic): Likewise.
28569 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28571 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28572 (depth_first_search::depth_first_search): Change structure init
28573 function to this constructor.
28574 (depth_first_search::add_bb): Rename function to this member.
28575 (depth_first_search::execute): Likewise.
28576 (flow_dfs_compute_reverse_finish): Adjust.
28578 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28580 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28581 (longest_simple_path): Likewise.
28582 * shrink-wrap.c (spread_components): Likewise.
28583 (disqualify_problematic_components): Likewise.
28584 (emit_common_heads_for_components): Likewise.
28585 (emit_common_tails_for_components): Likewise.
28586 (insert_prologue_epilogue_for_components): Likewise.
28588 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28590 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28593 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28595 * df-core.c (df_set_blocks): Start using auto_bitmap.
28596 (df_compact_blocks): Likewise.
28597 * df-problems.c (df_rd_confluence_n): Likewise.
28598 * df-scan.c (df_insn_rescan_all): Likewise.
28599 (df_process_deferred_rescans): Likewise.
28600 (df_update_entry_block_defs): Likewise.
28601 (df_update_exit_block_uses): Likewise.
28602 (df_entry_block_bitmap_verify): Likewise.
28603 (df_exit_block_bitmap_verify): Likewise.
28604 (df_scan_verify): Likewise.
28605 * lra-constraints.c (lra_constraints): Likewise.
28606 (undo_optional_reloads): Likewise.
28607 (lra_undo_inheritance): Likewise.
28608 * lra-remat.c (calculate_gen_cands): Likewise.
28609 (do_remat): Likewise.
28610 * lra-spills.c (assign_spill_hard_regs): Likewise.
28611 (spill_pseudos): Likewise.
28612 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28613 (bitmap_set_subtract_values): Likewise.
28615 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28617 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28618 management with auto_bitmap.
28619 (fix_inter_tick): Likewise.
28620 (fix_recovery_deps): Likewise.
28621 * ira.c (add_store_equivs): Likewise.
28622 (find_moveable_pseudos): Likewise.
28623 (split_live_ranges_for_shrink_wrap): Likewise.
28624 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28625 (rtx_reuse_manager::seen_def_p): Likewise.
28626 (rtx_reuse_manager::set_seen_def): Likewise.
28627 * print-rtl.h (class rtx_reuse_manager): Likewise.
28629 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28631 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28633 (migrate_btr_def): Likewise.
28634 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28635 * df-core.c (loop_post_order_compute): Likewise.
28636 (loop_inverted_post_order_compute): Likewise.
28637 * hsa-common.h: Likewise.
28638 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28639 * init-regs.c (initialize_uninitialized_regs): Likewise.
28640 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28641 (inline_small_functions): Likewise.
28642 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28643 * ira.c (combine_and_move_insns): Likewise.
28644 (build_insn_chain): Likewise.
28645 * loop-invariant.c (find_invariants): Likewise.
28646 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28647 * predict.c (tree_predict_by_opcode): Likewise.
28648 (predict_paths_leading_to): Likewise.
28649 (predict_paths_leading_to_edge): Likewise.
28650 (estimate_loops_at_level): Likewise.
28651 (estimate_loops): Likewise.
28652 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28653 (spread_components): Likewise.
28654 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28655 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28656 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28657 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28658 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28659 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28660 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28661 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28662 (mark_threaded_blocks): Likewise.
28663 (thread_through_all_blocks): Likewise.
28664 * tree-ssa.c (verify_ssa): Likewise.
28665 (execute_update_addresses_taken): Likewise.
28666 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28668 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28670 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28672 (post_order_compute): Likewise.
28673 (inverted_post_order_compute): Likewise.
28674 (pre_and_rev_post_order_compute_fn): Likewise.
28676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28678 * genrecog.c (int_set::int_set): Explicitly construct our
28679 auto_vec base class.
28680 * vec.h (auto_vec::auto_vec): New constructor.
28682 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28684 * bitmap.h (class auto_bitmap): New constructor taking
28685 bitmap_obstack * argument.
28687 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28689 * bitmap.h (class auto_bitmap): Change type of m_bits to
28690 bitmap_head, and adjust ctor / dtor and member operators.
28692 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28694 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28695 when returned register mode doesn't match original mode.
28697 2017-05-12 Jeff Law <law@redhat.com>
28698 Jakub Jelinek <jakub@redhat.com>
28700 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28701 we look for cc setter after the compare-elim changes.
28702 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28703 within the vector to match what compare-elim now expects.
28704 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28705 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28707 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28708 after the compare-elim changes.
28709 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28710 the vector to match what compare-elim now expects.
28711 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28712 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28713 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28714 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28715 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28717 * config/visium/visium.c (single_set_and_flags): Fix where
28718 we look for cc setter after the compare-elim changes.
28719 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28720 with the vector to match what compare-elim now expects.
28721 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28722 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28723 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28724 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28725 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28726 (neg<mode>2_insn_set_overflow): Likewise.
28728 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28730 PR middle-end/79794
28731 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28732 maybe_expand_insn call, set ops[0].target. If still set after call,
28733 set alt_rtl. Add extra arg to recursive calls.
28734 (extract_bit_field): Add alt_rtl argument. Pass to
28736 * expmed.h (extract_bit_field): Fix prototype.
28737 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28738 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28739 to extract_bit_field_calls.
28740 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28741 Pass alt_rtl to extract_bit_field calls.
28742 * calls.c (store_unaligned_arguments_into_psuedos)
28743 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28744 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28746 * optabs.h (struct expand_operand): Add target bitfield.
28748 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28750 * compare-elim.c (try_eliminate_compare): Canonicalize
28751 operation with embedded compare to
28752 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28753 (set (reg) (operation)].
28755 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28757 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28760 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28761 cost of adding a carry flag for ADC instruction.
28762 [case MINUS]: Ignore the cost of subtracting a carry flag
28763 for SBB instruction.
28765 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28767 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28769 * config/rs6000/bmiintrin.h: New file.
28770 * config/rs6000/bmi2intrin.h: New file.
28771 * config/rs6000/x86intrin.h: New file.
28773 2017-05-12 Jeff Law <law@redhat.com>
28775 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28778 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28780 PR middle-end/80707
28781 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28783 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28785 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28786 early expansion of vector divide builtins.
28787 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28788 builtins identified as having unsigned arguments.
28790 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28792 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28793 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28794 expansion of vector logical operations (and, andc, or, xor,
28797 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28799 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28800 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28802 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28804 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28805 early GIMPLE expansion of vector multiplies.
28807 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28809 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28810 TARGET_HAVE_MOVT conditional.
28811 (movt splitter): Likewise.
28813 2017-05-12 Richard Biener <rguenther@suse.de>
28815 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28816 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28817 Fold all stmts not inplace.
28819 2017-05-12 Richard Biener <rguenther@suse.de>
28821 PR tree-optimization/80713
28822 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28823 inserted_exprs bit for not removed stmts.
28825 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28827 PR middle-end/69921
28828 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28829 parallelized" attribute for parallelized OpenACC kernels.
28830 * omp-offload.c (execute_oacc_device_lower): Use it.
28832 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28833 Set "oacc kernels" attribute.
28834 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28835 parameter. Adjust all users.
28836 (oacc_fn_attrib_kernels_p): Remove function.
28837 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28838 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28839 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28840 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28841 assert "oacc kernels" attribute is set.
28843 2017-05-11 Carl Love <cel@us.ibm.com>
28845 * config/rs6000/rs6000-c: Add support for built-in functions
28846 vector unsigned char vec_popcnt (vector signed char)
28847 vector unsigned char vec_popcnt (vector unsigned char)
28848 vector unsigned short vec_popcnt (vector signed short)
28849 vector unsigned short vec_popcnt (vector unsigned short)
28850 vector unsigned int vec_popcnt (vector signed int)
28851 vector unsigned int vec_popcnt (vector unsigned int)
28852 vector unsigned long long vec_popcnt (vector signed long long)
28853 vector unsigned long long vec_popcnt (vector unsigned long long)
28854 vector signed long long vec_slo (vector signed long long,
28855 vector signed char)
28856 vector signed long long vec_slo (vector signed long long,
28857 vector unsigned char)
28858 vector unsigned long long vec_slo (vector unsigned long long,
28859 vector signed char)
28860 vector unsigned long long vec_slo (vector unsigned long long,
28861 vector unsigned char)
28862 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28863 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28864 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28865 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28866 * doc/extend.texi: Update the built-in documentation file for the
28867 new built-in functions.
28869 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28871 * attribs.h (sorted_attr_string): Move machine independent
28872 functions for target clone support from the i386 port to common
28873 code. Rename ix86_function_versions to common_function_versions.
28874 Rename make_name to make_unique_name.
28875 (common_function_versions): Likewise.
28876 (make_unique_name): Likewise.
28877 (make_dispatcher_decl): Likewise.
28878 (is_function_default_version): Likewise.
28879 * attribs.c (attr_strcmp): Likewise.
28880 (sorted_attr_string): Likewise.
28881 (common_function_versions): Likewise.
28882 (make_unique_name): Likewise.
28883 (make_dispatcher_decl): Likewise.
28884 (is_function_default_version): Likewise.
28885 * config/i386/i386.c (attr_strcmp): Likewise.
28886 (sorted_attr_string): Likewise.
28887 (ix86_function_versions): Likewise.
28888 (make_name): Likewise.
28889 (make_dispatcher_decl): Likewise.
28890 (is_function_default_version): Likewise.
28891 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28893 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28896 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28897 Account for direct move costs for vec_construct of integer
28900 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28903 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28904 (UNSPEC_STX_ATOMIC): Ditto.
28905 (loaddi_via_sse): New insn.
28906 (storedi_via_sse): Ditto.
28907 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28908 Update corresponding peephole2 patterns.
28909 (atomic_storedi_fpu): Ditto.
28911 2017-05-11 Julia Koval <julia.koval@intel.com>
28913 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28914 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28916 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28917 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28918 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28920 2017-05-11 Nathan Sidwell <nathan@acm.org>
28922 * graphite-poly.c: Include dumpfile.h.
28924 * dumpfle.h (dump_function): Declare here ...
28925 * tree-dump.h (dump_function): ... not here.
28926 * dumpfile.c: #include tree-cfg.h.
28927 (dump_function): Move here from ...
28928 * tree-dump.c (dump_function): ... here.
28929 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28930 * graphite-poly.c: Don't include tree-dump.h.
28931 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28932 * print-tree.c: Likewise.
28933 * stor-layout.c: Likewise.
28934 * tree-nested.c: Likewise.
28936 * dumpfile.c (dump_start): Use TDF_FLAGS.
28937 (dump_enable_all): Fix TDF_KIND check thinko.
28939 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28942 array entries to represent two legal parameterizations of the
28943 overloaded __builtin_cmpb function, as represented by the
28944 P6_OV_BUILTIN_CMPB constant.
28945 (altivec_resolve_overloaded_builtin): Add special case handling
28946 for the __builtin_cmpb function, as represented by the
28947 P6_OV_BUILTIN_CMPB constant.
28948 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28949 (BU_P6_64BIT_2): New macro.
28950 (BU_P6_OVERLOAD_2): New macro
28951 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28952 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28953 (CMPB): Add overload support to represent both 32-bit and 64-bit
28954 compare-bytes function.
28955 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28956 support for TARGET_CMPB.
28957 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28958 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28959 documentation of the __builtin_cmpb overloaded built-in function.
28961 2017-05-11 Richard Biener <rguenther@suse.de>
28963 PR tree-optimization/80705
28964 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28965 bases are not vectorizable.
28967 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28969 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28970 when counting register pressure.
28972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28974 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28975 (iv_ca_more_deps): Renamed to ...
28976 (iv_ca_compare_deps): ... this.
28977 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28979 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28981 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28983 (determine_group_iv_costs): ... here.
28984 (find_inv_vars_cb): Record inv var if it's not recorded before.
28986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28988 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28989 (get_shiftadd_cost): Ditto.
28991 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28993 * tree-ssa-address.c: Include header file.
28994 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28996 (add_to_parts): Refactor.
28997 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28998 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29001 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29003 PR tree-optimization/53090
29004 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29006 (extract_cond_operands): Detect condition with IV on both sides
29007 and return COMP_IV_EXPR_2.
29008 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29009 (rewrite_use_compare): Simplify by removing call to function
29010 extract_cond_operands.
29012 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29014 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29015 (extract_cond_operands): Detect condition comparing against non-
29016 invariant bound and return appropriate enum value.
29017 (find_interesting_uses_cond): Update use of extract_cond_operands.
29018 Handle its return value accordingly.
29019 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29021 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29023 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29024 nonlinear iv_use computation in loop invariant sensitive way.
29026 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29028 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29029 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29031 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29033 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29034 (dump_cand): Support iv_cand.inv_exprs.
29035 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29037 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29042 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29044 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29045 as local function. Include necessary header files.
29046 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29050 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29054 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29055 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29056 RSHIFT_EXPR and BIT_NOT_EXPR.
29058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29060 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29061 (adjust_setup_cost): New parameter supporting round up adjustment.
29062 (struct address_cost_data): Delete.
29063 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29064 (split_address_cost, ptr_difference_cost): Delete.
29065 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29066 (struct ainc_cost_data): New struct.
29067 (get_address_cost_ainc): New function.
29068 (get_address_cost, get_computation_cost): Reimplement.
29069 (determine_group_iv_cost_address): Record inv_expr for all uses of
29071 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29072 (iv_ca_has_deps): Reimplemented to ...
29073 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29075 (iv_ca_extend): Call iv_ca_more_deps.
29077 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29079 * tree-ssa-address.c (struct mem_address): Move to header file.
29080 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29081 * tree-ssa-address.h (struct mem_address): Move from C file.
29082 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29084 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29086 * tree-affine.h (aff_combination_type): New interface.
29087 (aff_combination_zero_p): Remove static.
29088 (aff_combination_const_p): New interface.
29089 (aff_combination_singleton_var_p): New interfaces.
29091 2017-05-11 Richard Biener <rguenther@suse.de>
29093 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29094 Skip unreachable blocks and destinations.
29095 (eliminate): Move stmt removal and fixup ...
29096 (fini_eliminate): ... here. Skip inserted exprs.
29097 (pass_pre::execute): Move fini_pre after fini_eliminate.
29098 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29099 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29100 PRE to get rid of dead code that has invalid SSA form and
29101 split critical edges again.
29103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29105 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29107 2017-05-11 Richard Biener <rguenther@suse.de>
29109 * passes.c (execute_function_todo): Verify loops if they are
29110 said to be up-to-date.
29111 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29112 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29114 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29117 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29118 handle calling assemble_external ourself.
29121 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29122 modes with zero size. Enhance comment.
29124 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29126 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29127 built-ins for vec_xl and vec_xst with short and char pointer
29130 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29132 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29133 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29134 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29135 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29136 (_mm_maskz_min_round_ss): New intrinsics.
29137 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29138 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29139 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29140 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29141 (__builtin_ia32_minss_mask_round): New builtins.
29142 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29143 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29144 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29146 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29147 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29149 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29152 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29154 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29155 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29156 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29157 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29158 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29159 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29160 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29161 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29162 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29163 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29164 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29165 (__builtin_ia32_mulss_mask_round): New builtins.
29166 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29167 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29168 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29170 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29171 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29173 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29176 2017-05-10 Julia Koval <julia.koval@intel.com>
29178 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29179 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29180 (_mm256_setr_m128i): New intrinsics.
29182 2017-05-10 Julia Koval <julia.koval@intel.com>
29184 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29185 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29186 (_mm_maskz_rcp14_ss): New intrinsics.
29187 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29188 (__builtin_ia32_rcp14ss_mask): New builtins.
29189 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29191 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29193 PR tree-optimization/51513
29194 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29195 (assert_unreachable_fallthru_edge_p): Use it.
29196 (group_case_labels_stmt): Likewise.
29197 * tree-cfg.h: Prototype it.
29198 * stmt.c: Include cfghooks.h and tree-cfg.h.
29199 (emit_case_dispatch_table) <gap_label>: New local variable.
29200 Use it to fill dispatch table gaps.
29201 Test for default_label before updating probabilities.
29202 (expand_case) <default_label>: Remove unneeded initialization.
29203 Test for unreachable default case statement and remove its edge.
29204 Set default_label accordingly.
29205 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29207 2017-05-10 Carl Love <cel@us.ibm.com>
29209 * config/rs6000/rs6000-c: Add support for built-in functions
29210 vector signed char vec_neg (vector signed char)
29211 vector signed short int vec_neg (vector short int)
29212 vector signed int vec_neg (vector signed int)
29213 vector signed long long vec_neg (vector signed long long)
29214 vector float vec_neg (vector float)
29215 vector double vec_neg (vector double)
29216 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29218 * config/rs6000/altivec.h: Add define for vec_neg
29219 * doc/extend.texi: Update the built-in documentation for the
29220 new built-in functions.
29222 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29224 PR tree-optimization/77644
29225 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29227 2017-05-10 Nathan Sidwell <nathan@acm.org>
29229 * dumpfile.h (TDI_lang_all): New.
29230 (TDF_KIND): New. Renumber others
29231 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29233 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29235 (get_dump_file_name): Adjust suffix generation.
29236 (dump_enable_all): Use TDF_KIND.
29237 * doc/invoke.texi (-fdump-lang-all): Document.
29239 * dumpfile.h: Tabify.
29241 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29244 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29245 Move member access before delete.
29247 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29249 * tree-inline.c (expand_call_inline): Split block at stmt
29252 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29255 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29256 are now unused after splitting mov{sf,sd}_hardfloat.
29257 (f32_lr2): Likewise.
29258 (f32_lm): Likewise.
29259 (f32_lm2): Likewise.
29260 (f32_li): Likewise.
29261 (f32_li2): Likewise.
29262 (f32_lv): Likewise.
29263 (f32_sr): Likewise.
29264 (f32_sr2): Likewise.
29265 (f32_sm): Likewise.
29266 (f32_sm2): Likewise.
29267 (f32_si): Likewise.
29268 (f32_si2): Likewise.
29269 (f32_sv): Likewise.
29270 (f32_dm): Likewise.
29271 (f32_vsx): Likewise.
29272 (f32_av): Likewise.
29273 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29274 For movsf, order stores so the VSX stores occur before the GPR
29275 store which encourages the register allocator to use a traditional
29276 FPR instead of a GPR. For movsd, order the stores so that the GPR
29277 store comes before the VSX stores to allow the power6 to work.
29278 This is due to the power6 not having a 32-bit integer store
29279 instruction from a FPR.
29280 (movsf_hardfloat): Likewise.
29281 (movsd_hardfloat): Likewise.
29283 2017-05-09 Martin Sebor <msebor@redhat.com>
29285 PR translation/80280
29286 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29289 PR translation/80280
29290 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29291 data member added in r247778.
29292 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29294 2017-05-09 Nathan Sidwell <nathan@acm.org>
29296 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29298 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29301 2017-05-09 Marek Polacek <polacek@redhat.com>
29303 * doc/invoke.texi: Fix typo.
29305 2017-05-09 Richard Biener <rguenther@suse.de>
29307 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29308 (vrp_val_is_min): Likewise.
29309 (set_value_range_to_value): Likewise.
29310 (set_value_range_to_nonnegative): Likewise.
29311 (gimple_assign_nonzero_p): Likewise.
29312 (gimple_stmt_nonzero_p): Likewise.
29313 (vrp_int_const_binop): Likewise. Remove unreachable case.
29314 (adjust_range_with_scev): Adjust comments.
29315 (compare_range_with_value): Likewise.
29316 (extract_range_from_phi_node): Likewise.
29317 (test_for_singularity): Likewise.
29319 2017-05-09 Richard Biener <rguenther@suse.de>
29321 * tree-vrp.c (get_single_symbol): Add assert that we don't
29322 get overflowed constants as invariant part.
29323 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29324 checks. Use wi::cmp instead of recursing for integer constants.
29325 (compare_values): Just ignore whether we assumed undefined
29326 overflow instead of failing the compare.
29327 (extract_range_for_var_from_comparison_expr): Add comment before the
29328 TREE_NO_WARNING sets.
29329 (test_for_singularity): Likewise.
29330 (extract_range_from_comparison): Do not disable optimization
29331 when we assumed undefined overflow.
29332 (extract_range_basic): Remove init of unused var.
29334 2017-05-09 Richard Biener <rguenther@suse.de>
29336 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29337 (extract_range_from_multiplicative_op_1): Adjust.
29338 (extract_range_from_binary_expr_1): Use int_const_binop.
29340 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29343 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29344 rs6000_store_data_bypass_p in seven define_bypass directives and
29345 in several comments.
29346 * config/rs6000/rs6000-protos.h: Add prototype for
29347 rs6000_store_data_bypass_p function.
29348 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29349 function implements slightly different (rs6000-specific) semantics
29350 than store_data_bypass_p, returning false rather than aborting
29351 with assertion error when arguments do not satisfy the
29352 requirements of store data bypass.
29353 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29354 rs6000_store_data_bypass_p.
29356 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29358 * config/xtensa/xtensa-protos.h
29359 (xtensa_initial_elimination_offset): New declaration.
29360 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29361 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29362 macro definition, add case for FRAME_POINTER_REGNUM when
29363 FRAME_GROWS_DOWNWARD.
29364 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29365 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29366 xtensa_initial_elimination_offset.
29368 2017-05-08 Nathan Sidwell <nathan@acm.org>
29370 * doc/invoke.texi: Alphabetize -fdump options.
29372 2017-05-08 Martin Sebor <msebor@redhat.com>
29374 PR translation/80280
29375 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29377 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29379 * target.def (compute_frame_layout): New optional target hook.
29380 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29381 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29382 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29384 * reload1.c (verify_initial_elim_offsets): Likewise.
29385 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29386 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29387 (arm_get_frame_offsets): Split up into this ...
29388 (arm_compute_frame_layout): ... and this function.
29390 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29392 * config/aarch64/constraints.md (Usa): New constraint.
29393 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29395 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29397 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29398 with_multilib_list after it has been checked.
29400 2017-05-08 Richard Biener <rguenther@suse.de>
29402 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29403 (bitmap_set_subtract_values): Likewise.
29405 2017-05-08 Richard Biener <rguenther@suse.de>
29407 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29408 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29410 (gimple_stmt_nonzero_warnv_p): Rename to ...
29411 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29413 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29414 (extract_range_basic): Adjust, do not disable propagation on
29415 strict overflow sensitive simplification.
29416 (vrp_visit_cond_stmt): Likewise.
29418 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29420 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29421 body size unconditionally.
29423 2017-05-07 Jeff Law <law@redhat.com>
29426 2017-05-06 Jeff Law <law@redhat.com>
29427 PR tree-optimization/78496
29428 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29431 PR tree-optimization/78496
29432 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29433 (simplify_stmt_using_ranges): Call it.
29434 (vrp_dom_walker::before_dom_children): Extract equivalences
29435 from an ASSERT_EXPR with an equality comparison against a
29438 2017-05-06 Jeff Law <law@redhat.com>
29440 PR tree-optimization/78496
29441 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29444 PR tree-optimization/78496
29445 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29446 (simplify_stmt_using_ranges): Call it.
29447 (vrp_dom_walker::before_dom_children): Extract equivalences
29448 from an ASSERT_EXPR with an equality comparison against a
29451 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29453 * lra-constraints.c (lra_copy_reg_equiv): New function.
29454 (split_reg): Use it to copy equivalence information from the
29455 original register to the spill register.
29457 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29459 PR rtl-optimization/75964
29460 * simplify-rtx.c (simplify_const_relational_operation): Remove
29461 invalid handling of comparisons of integer ABS.
29463 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29465 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29466 initialize to zero.
29467 (init_regs): Remove declaration.
29468 (function_arg_advance_32): Initialize error_p as boolean variable.
29470 2017-05-05 Nathan Sidwell <nathan@acm.org>
29472 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29473 lines. Use for (;;).
29475 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29477 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29478 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29479 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29480 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29481 VF=2 that require versioning.
29483 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29485 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29488 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29490 * diagnostic.h (diagnostic_override_option_index): Convert from
29491 macro to inline function.
29493 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29495 * diagnostic.c (last_module_changed_p): New function.
29496 (set_last_module): New function.
29497 (diagnostic_report_current_module): Convert macro usage to
29498 the above functions.
29499 * diagnostic.h (diagnostic_context::last_module): Strengthen
29500 from const line_map * to const line_map_ordinary *.
29501 (diagnostic_last_module_changed): Delete macro.
29502 (diagnostic_set_last_module): Delete macro.
29504 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29506 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29507 with diagnostic_report_diagnostic.
29508 (diagnostic_n_impl_richloc): Likewise.
29509 * diagnostic.h (report_diagnostic): Delete macro.
29510 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29511 with diagnostic_report_diagnostic.
29512 * substring-locations.c (format_warning_va): Likewise.
29514 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29516 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29517 save/restor of format_spec. Move option-printing code to...
29518 (print_option_information): ...this new function, and
29519 reimplement by simply printing to the pretty_printer,
29520 rather than appending to the format string.
29522 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29524 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29525 handling logic into...
29526 (update_effective_level_from_pragmas): ...this new function.
29528 2017-05-04 Andrew Waterman <andrew@sifive.com>
29530 * config/riscv/riscv.opt (mstrict-align): New option.
29531 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29532 (SLOW_UNALIGNED_ACCESS): Define.
29533 (riscv_slow_unaligned_access): Declare.
29534 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29536 (riscv_slow_unaligned_access): New variable.
29537 (rocket_tune_info): Set slow_unaligned_access to true.
29538 (optimize_size_tune_info): Set slow_unaligned_access to false.
29539 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29540 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29541 (riscv_option_override): Set riscv_slow_unaligned_access.
29542 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29544 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29546 * config/riscv/riscv.md: Unify indentation.
29548 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29553 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29555 (extendsi<mode>2): Add support for doing sign extension via
29556 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29557 don't have ISA 3.0 instructions.
29558 (extendsi<mode>2 splitter): Likewise.
29559 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29560 generate the normal insns since SImode can now go in vector
29561 registers. Disallow the special UNSPECs needed for previous
29562 machines to hide SImode being used. Add new insns
29563 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29564 (fix_trunc<mode>si2_stfiwx): Likewise.
29565 (fix_trunc<mode>si2_internal): Likewise.
29566 (fixuns_trunc<mode>si2): Likewise.
29567 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29568 (fctiw<u>z_<mode>_smallint): Likewise.
29569 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29570 of floating point to 32-bit integer from doing a direct move to
29571 the GPR registers to do a store.
29572 (fctiwz_<mode>): Break long line.
29574 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29576 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29577 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29578 (addr_list, addr_offset_valid_p): New.
29579 (split_address_groups): Check offset validity with above function.
29580 (gt-tree-ssa-loop-ivopts.h): Include header file.
29582 2017-05-05 Nathan Sidwell <nathan@acm.org>
29584 * config.gcc (arm*-*-*): Add missing 'fi'.
29586 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29588 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29589 included in -fopt-info does not matter.
29590 * doc/optinfo.texi (-fopt-info): Fix description of default
29591 behavour. Explicitly say order of options included in -fopt-info
29594 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29596 * config.gcc: Allow combinations of aprofile and rmprofile values for
29597 --with-multilib-list.
29598 * config/arm/t-multilib: New file.
29599 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29600 variables. Remove setting of ISA and floating-point ABI in
29601 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29602 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29603 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29604 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29606 * config/arm/t-rmprofile: Likewise except for the matches changes.
29607 * doc/install.texi (--with-multilib-list): Document the combination of
29608 aprofile and rmprofile values and warn about pitfalls in doing that.
29610 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29612 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29613 (movdi_aarch64): Likewise.
29615 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29617 PR tree-optimization/80632
29618 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29620 (build_arrays): Initialize it for virtual phis.
29621 (fix_phi_nodes): Use it for virtual phis.
29623 PR tree-optimization/80558
29624 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29625 [x, y] op z into [x op, y op z] for op & or | if conditions
29628 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29629 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29632 * config/arm/arm.md (use_literal_pool): Remove.
29633 (64-bit immediate split): No longer takes cost into consideration
29634 if arm_disable_literal_pool is enabled.
29635 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29636 used when arm_disable_literal_pool is enabled.
29637 (arm_max_const_double_inline_cost): Remove use of
29638 arm_disable_literal_pool.
29639 (push_minipool_fix): Add assert.
29640 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29641 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29642 (no_literal_pool_sf_immediate): New.
29644 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29646 PR tree-optimization/80613
29647 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29648 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29650 2017-05-05 Richard Biener <rguenther@suse.de>
29652 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29654 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29656 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29657 of this flag from insn conditions due to removal from r247495.
29659 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29661 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29663 (arm_early_store_addr_dep_ptr): Likewise.
29664 * config/arm/aarch-common-protos.h
29665 (arm_early_load_addr_dep_ptr): Add prototype.
29666 (arm_early_store_addr_dep_ptr): Likewise.
29667 * config/arm/cortex-a53.md: Add new bypasses.
29669 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29671 * tree.c (next_type_uid): Change type to unsigned.
29672 (type_hash_canon): Decrement back next_type_uid if
29673 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29674 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29677 2017-05-04 Martin Sebor <msebor@redhat.com>
29679 * builtins.c: Fix a trivial typo in a comment.
29681 PR middle-end/79234
29682 * builtins.c (check_sizes): Adjust to handle reading past the end.
29683 Avoid printing excessive upper bound of ranges. Use %E to print
29684 tree nodes instead of converting them to %wu.
29685 (expand_builtin_memchr): New function.
29686 (compute_dest_size): Rename...
29687 (compute_objsize): ...to this.
29688 (expand_builtin_memcpy): Adjust.
29689 (expand_builtin_mempcpy): Adjust.
29690 (expand_builtin_strcat): Adjust.
29691 (expand_builtin_strcpy): Adjust.
29692 (check_strncat_sizes): Adjust.
29693 (expand_builtin_strncat): Adjust.
29694 (expand_builtin_strncpy): Adjust and simplify.
29695 (expand_builtin_memset): Adjust.
29696 (expand_builtin_bzero): Adjust.
29697 (expand_builtin_memcmp): Adjust.
29698 (expand_builtin): Handle memcmp.
29699 (maybe_emit_chk_warning): Check strncat just once.
29701 2017-05-04 Martin Sebor <msebor@redhat.com>
29703 PR preprocessor/79214
29704 PR middle-end/79222
29705 PR middle-end/79223
29706 * builtins.c (check_sizes): Add inlining context and issue
29707 warnings even when -Wno-system-headers is set.
29708 (check_strncat_sizes): Same.
29709 (expand_builtin_strncat): Same.
29710 (expand_builtin_memmove): New function.
29711 (expand_builtin_stpncpy): Same.
29712 (expand_builtin): Handle memmove and stpncpy.
29714 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29716 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29717 which is not used any more.
29719 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29721 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29723 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29725 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29726 (cortexa53_tunings): Likewise.
29727 (cortexa57_tunings): Likewise.
29728 (cortexa72_tunings): Likewise.
29729 (cortexa73_tunings): Likewise.
29731 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29733 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29734 Set loop alignment to 8.
29736 2017-05-04 Martin Sebor <msebor@redhat.com>
29738 PR translation/80280
29739 * builtins.c (expand_builtin_object_size): Add missing quoting to
29740 %D and like directives.
29741 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29742 (hsa_type_for_tree_type): Same.
29743 (verify_function_arguments): Same.
29744 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29745 * varasm.c (get_section): Same.
29748 2017-05-04 Martin Sebor <msebor@redhat.com>
29750 PR translation/80280
29751 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29753 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29755 * config/aarch64/aarch64.c (generic_addrcost_table):
29756 Change HI/TI mode setting.
29758 2017-05-04 Martin Jambor <mjambor@suse.cz>
29760 PR tree-optimization/80622
29761 * tree-sra.c (comes_initialized_p): New function.
29762 (build_accesses_from_assign): Only set write lazily when
29763 comes_initialized_p is false.
29764 (analyze_access_subtree): Use comes_initialized_p.
29765 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29766 instead of testing for PARM_DECL.
29768 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29770 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29771 constraint on operand 0 to allow more general addressing modes.
29772 Adjust output template.
29773 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29775 * config/aarch64/aarch64-protos.h
29776 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29777 * config/aarch64/constraints.md (Dp): New address constraint.
29778 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29781 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29783 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29784 update use of estimate_ipcp_clone_size_and_time.
29785 (estimate_local_effects): Update use of
29786 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29787 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29788 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29789 Return nonspecialized time.
29791 2017-05-04 Richard Biener <rguenther@suse.de>
29793 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29794 for the last VUSE which def dominates the PHI. Directly call
29796 (get_continuation_for_phi_1): Remove.
29798 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29800 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29801 to explain the use of truncating division. Cap the number of
29802 iterations to the maximum given by nb_iterations_upper_bound,
29805 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29807 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29808 * configure: Regenerate.
29809 * config.in: Regenerate.
29810 * config/i386/driver-mingw32.c: new file.
29811 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29812 * config.host: Link driver-mingw32.o on MinGW host.
29813 * doc/install.texi: Document new --enable-mingw-wildcard configure
29816 2017-05-04 Marek Polacek <polacek@redhat.com>
29818 PR tree-optimization/80612
29819 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29821 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29822 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29824 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29825 (movt splitter): Likewise.
29826 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29827 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29828 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29829 block for Thumb-1 with MOVT.
29830 (thumb2_legitimate_address_p): Move code block ...
29831 (can_avoid_literal_pool_for_label_p): ... into this new function.
29832 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29834 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29835 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29836 "M-profile targets with the MOVT instruction".
29838 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29840 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29841 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29842 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29844 2017-05-04 Martin Liska <mliska@suse.cz>
29846 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29847 variable cond_code.
29849 2017-05-04 Richard Biener <rguenther@suse.de>
29851 * tree.c (array_at_struct_end_p): Handle arrays at struct
29852 end with flexarrays more conservatively. Refactor and treat
29853 arrays of arrays or aggregates more strict. Fix
29854 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29855 * tree.c (array_at_struct_end_p): Adjust prototype.
29856 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29857 * gimple-fold.c (get_range_strlen): Likewise.
29858 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29860 2017-05-04 Richard Biener <rguenther@suse.de>
29862 PR tree-optimization/31130
29863 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29865 (supports_overflow_infinity): Likewise.
29866 (is_negative_overflow_infinity): Likewise.
29867 (is_positive_overflow_infinity): Likewise.
29868 (is_overflow_infinity): Likewise.
29869 (stmt_overflow_infinity): Likewise.
29870 (overflow_infinity_range_p): Likewise.
29871 (usable_range_p): Remove as always returning true.
29872 (make_overflow_infinity): Remove.
29873 (negative_overflow_infinity): Likewise.
29874 (positive_overflow_infinity): Likewise.
29875 (avoid_overflow_infinity): Likewise.
29876 (set_value_range): Adjust accordingly.
29877 (set_value_range_to_nonnegative): Likewise, remove now unused
29878 overflow_infinity arg.
29879 (vrp_operand_equal_p): Adjust.
29880 (update_value_range): Likewise.
29881 (range_int_cst_singleton_p): Likewise.
29882 (operand_less_p): Likewise.
29883 (compare_values_warnv): Likewise.
29884 (extract_range_for_var_from_comparison_expr): Likewise.
29885 (vrp_int_const_binop): Likewise.
29886 (zero_nonzero_bits_from_vr): Likewise.
29887 (extract_range_from_multiplicative_op_1): Likewise.
29888 (extract_range_from_binary_expr_1): Likewise.
29889 (extract_range_from_unary_expr): Likewise.
29890 (extract_range_from_comparison): Likewise.
29891 (extract_range_basic): Likewise.
29892 (adjust_range_with_scev): Likewise.
29893 (compare_ranges): Likewise.
29894 (compare_range_with_value): Likewise.
29895 (dump_value_range): Likewise.
29896 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29898 (simplify_cond_using_ranges): Adjust.
29900 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29902 * brig-builtins.def: Added a builtin for class_f64.
29903 * builtin-types.def: Added a builtin type needed by class_f64.
29905 2017-05-03 Jason Merrill <jason@redhat.com>
29907 * timevar.def: Add TV_CONSTEXPR.
29909 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29911 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29913 2017-05-03 Martin Jambor <mjambor@suse.cz>
29915 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29916 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29917 * ipa-cp.c (ipcp_propagate_stage): Do not call
29918 ipa_update_after_lto_read.
29919 * ipa-inline.c (ipa_inline): Likewise.
29921 2017-05-03 Martin Jambor <mjambor@suse.cz>
29923 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29924 tag. Added a default constructor and a destructor.
29925 (ipa_edge_args_sum_t): New class;
29926 (ipa_edge_args_sum): Declare.
29927 (ipa_edge_args_vector): Remove declaration.
29928 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29929 (ipa_free_edge_args_substructures): Remove declaration.
29930 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29931 (ipa_edge_args_info_available_for_edge_p): Likewise.
29932 * ipa-prop.c (ipa_edge_args_vector): Removed.
29933 (edge_removal_hook_holder): Likewise.
29934 (edge_duplication_hook_holder): Likewise.
29935 (ipa_edge_args_sum): New variable.
29936 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29937 ipa_edge_args_vector.
29938 (ipa_free_edge_args_substructures): Likewise.
29939 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29940 ipa_edge_args_vector.
29941 (ipa_edge_removal_hook): Turned into method
29942 ipa_edge_args_sum_t::remove.
29943 (ipa_edge_duplication_hook): Turned into method
29944 ipa_edge_args_sum_t::duplicate.
29945 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29946 registering edge hooks.
29947 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29948 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29949 ipa_edge_args_sum instead of ipa_edge_args_vector.
29950 * ipa-profile.c (ipa_profile): Likewise.
29952 2017-05-03 Martin Jambor <mjambor@suse.cz>
29954 * symbol-summary.h (function_summary): New method exists.
29955 (function_summary::symtab_removal): Deallocate through release.
29956 (call_summary): New class.
29957 (gt_ggc_mx): New overload.
29958 (gt_pch_nx): Likewise.
29959 (gt_pch_nx): Likewise.
29961 2017-05-03 Jeff Law <law@redhat.com>
29963 PR tree-optimization/78496
29964 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29965 from simplify_cond_using_ranges. Split off code to walk
29966 backwards through casts into ...
29967 (simplify_cond_using_ranges_2): New function.
29968 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29969 (execute_vrp): After identifying jump threads, call
29970 simplify_cond_using_ranges_2.
29972 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29975 * ipa-inline.h (inline_summary): Add ctor.
29976 (create_ggc): Do not use ggc_cleared_alloc.
29978 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29979 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29981 * gcc.c (handle_braces): Support escaping in switch matching
29983 * doc/invoke.texi (Spec Files): Document it.
29984 Remove superfluous @code markup in items.
29986 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29988 * diagnostic-show-locus.c (struct column_range): New struct.
29989 (get_affected_columns): New function.
29990 (get_printed_columns): New function.
29991 (struct correction): New struct.
29992 (correction::ensure_capacity): New function.
29993 (correction::ensure_terminated): New function.
29994 (struct line_corrections): New struct.
29995 (line_corrections::~line_corrections): New dtor.
29996 (line_corrections::add_hint): New function.
29997 (layout::print_trailing_fixits): Reimplement in terms of the new
29999 (selftest::test_overlapped_fixit_printing): New function.
30000 (selftest::diagnostic_show_locus_c_tests): Call it.
30002 2017-05-03 Nathan Sidwell <nathan@acm.org>
30004 Canonicalize canonical type hashing
30005 * tree.h (type_hash_canon_hash): Declare.
30006 * tree.c (type_hash_list, attribute_hash_list): Move into
30007 type_hash_canon_hash.
30008 (build_type_attribute_qual_variant): Break out hash code calc into
30009 type_hash_canon_hash.
30010 (type_hash_canon_hash): New. Generic type hash computation.
30011 (build_range_type_1, build_array_type_1, build_function_type,
30012 build_method_type_directly, build_offset_type, build_complex_type,
30013 make_vector_type): Call it.
30015 2017-05-03 Richard Biener <rguenther@suse.de>
30017 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30018 When all DRs have unknown misaligned do not always peel
30019 when there is a store but apply the same costing model as if
30020 there were only loads.
30022 2017-05-03 Richard Biener <rguenther@suse.de>
30025 PR tree-optimization/80492
30026 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30027 compare_base_decls returning dont-know properly.
30029 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30031 * config/arm/iterators.md (CCSI): New mode iterator.
30032 (arch): New mode attribute.
30033 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30034 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30035 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30036 code iterator for success result mode.
30037 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30038 the corresponding new insn generators.
30040 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30043 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30044 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30046 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30048 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30049 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30050 (DDR_A): Wrap DDR argument in brackets.
30051 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30052 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30053 (DDR_REVERSED_P): Likewise.
30055 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30057 PR tree-optimization/79472
30058 * tree-switch-conversion.c (struct switch_conv_info): Add
30059 contiguous_range and default_case_nonstandard fields.
30060 (collect_switch_conv_info): Compute contiguous_range and
30061 default_case_nonstandard fields, don't clear final_bb if
30062 contiguous_range and only the default case doesn't have the required
30064 (check_all_empty_except_final): Set default_case_nonstandard instead
30065 of failing if contiguous_range and the default case doesn't have empty
30067 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30068 and only the default case doesn't have the required constants. Skip
30070 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30071 if default_case_nonstandard.
30072 (build_constructors): Build constant 1 just once. Assert that default
30073 values aren't inserted in between cases if contiguous_range. Skip
30075 (build_arrays): Skip virtual phis.
30076 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30077 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30078 Handle virtual phis.
30079 (gen_inbound_check): Handle default_case_nonstandard case.
30080 (process_switch): Adjust check_final_bb caller. Call
30081 gather_default_values with the first non-default case instead of
30082 default case if default_case_nonstandard.
30084 2017-05-02 Nathan Sidwell <nathan@acm.org>
30086 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30087 check. Fix formatting.
30089 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30091 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30092 errors when comparing specialized and unspecialized times.
30094 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30096 * diagnostic-show-locus.c
30097 (layout::should_print_annotation_line_p): Make private.
30098 (layout::print_annotation_line): Make private.
30099 (layout::annotation_line_showed_range_p): Make private.
30100 (layout::show_ruler): Make private.
30101 (layout::print_source_line): Make private. Pass in line and
30102 line_width, rather than calling location_get_source_line. Drop
30104 (layout::print_leading_fixits): New method.
30105 (layout::print_any_fixits): Rename to...
30106 (layout::print_trailing_fixits): ...this, and make private.
30107 Don't print newline fixits.
30108 (diagnostic_show_locus): Move logic for printing one row into...
30109 (layout::print_line): ...this new function. Move the
30110 location_get_source_line call and error-handling from
30111 print_source_line to here. Call print_leading_fixits, and rename
30112 print_any_fixits to print_trailing_fixits.
30113 (selftest::test_fixit_insert_containing_newline): Update now that
30114 newlines are partially supported.
30115 (selftest::test_fixit_insert_containing_newline_2): New test.
30116 (selftest::test_fixit_replace_containing_newline): Update comments.
30117 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30118 * edit-context.c (class added_line): New class.
30119 (class edited_line): Describe newline handling in comment.
30120 (edited_line::actually_edited_p): New method.
30121 (edited_line::print_content): Delete redundant decl.
30122 (edited_line::m_predecessors): New field.
30123 (edited_file::print_content): Call edited_line::print_content.
30124 (edited_file::print_diff): Update to support newlines.
30125 (edited_file::print_diff_hunk): Likewise.
30126 (edited_file::print_run_of_changed_lines): New function.
30127 (edited_file::print_diff_line): Convert to...
30128 (print_diff_line): ...this.
30129 (edited_file::get_effective_line_count): New function.
30130 (edited_line::edited_line): Initialize new field m_predecessors.
30131 (edited_line::~edited_line): Clean up m_predecessors.
30132 (edited_line::apply_fixit): Handle newlines.
30133 (edited_line::get_effective_line_count): New function.
30134 (edited_line::print_content): New function.
30135 (edited_line::print_diff_lines): New function.
30136 (selftest::test_applying_fixits_insert_containing_newline): New
30138 (selftest::test_applying_fixits_replace_containing_newline): New
30140 (selftest::insert_line): New function.
30141 (selftest::test_applying_fixits_multiple_lines): Add example of
30143 (selftest::edit_context_c_tests): Call the new tests.
30145 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30147 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30148 parameter cand. Update dump information.
30149 (get_computation_cost): Update uses.
30151 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30153 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30154 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30155 (get_computation_at, rewrite_use_address): Update use of
30156 get_computation_aff.
30158 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30160 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30161 (get_computation): Delete.
30162 (get_computation_cost): Implement like get_computation_cost_at.
30163 Use get_computation_at.
30164 (get_computation_cost_at): Delete.
30165 (rewrite_use_nonlinear_expr): Use get_computation_at.
30166 (rewrite_use_compare, remove_unused_ivs): Ditto.
30168 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30170 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30174 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30175 (ivopts_global_cost_for_size): Rename parameter and update uses.
30176 (iv_ca_recount_cost): Update uses.
30177 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30178 candidates seperately in n_invs and n_cands.
30179 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30181 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30183 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30184 (find_inv_vars_cb): New.
30185 (find_depends): Renamed to ...
30186 (find_inv_vars): ... this.
30187 (add_candidate_1, force_var_cost): Call find_inv_vars.
30188 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30190 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30192 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30193 inv_vars. Add inv_exprs.
30194 (struct iv_cand): Rename depends_on to inv_vars.
30195 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30196 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30197 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30198 (dump_cand): Dump inv_vars.
30199 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30200 (record_invariant, find_depends, add_candidate_1): Ditto.
30201 (set_group_iv_cost, force_var_cost): Ditto.
30202 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30203 (get_computation_cost_at, get_computation_cost): Ditto.
30204 (determine_group_iv_cost_generic): Ditto.
30205 (determine_group_iv_cost_address): Ditto.
30206 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30207 (determine_group_iv_costs): Ditto.
30208 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30209 (iv_ca_set_remove_invariants): Renamed to ...
30210 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30211 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30212 (iv_ca_set_add_invariants): Renamed to ...
30213 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30214 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30215 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30216 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30217 (create_new_ivs): Remove useless dump.
30219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30221 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30223 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30224 (iv_ca_set_no_cp, create_new_iv): Ditto.
30226 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30228 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30230 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30232 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30233 function tree_check2.
30235 2017-05-02 Martin Liska <mliska@suse.cz>
30237 * doc/gcov.texi: Add missing preposition.
30238 * gcov.c (function_info::function_info): Properly fill up
30239 all member variables.
30241 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30243 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30245 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30247 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30249 2017-05-02 Martin Liska <mliska@suse.cz>
30252 * lto-streamer-in.c (lto_read_tree_1): Remove
30253 LTO_STREAMER_DEBUG.
30254 * lto-streamer.c (struct tree_hash_entry): Likewise.
30255 (struct tree_entry_hasher): Likewise.
30256 (tree_entry_hasher::hash): Likewise.
30257 (tree_entry_hasher::equal): Likewise.
30258 (lto_streamer_init): Likewise.
30259 (lto_orig_address_map): Likewise.
30260 (lto_orig_address_get): Likewise.
30261 (lto_orig_address_remove): Likewise.
30262 * lto-streamer.h: Likewise.
30263 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30264 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30266 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30268 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30269 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30270 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30271 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30272 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30273 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30274 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30275 (mm_maskz_sub_ss): New intrinsics.
30276 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30277 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30278 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30279 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30280 (__builtin_ia32_subss_mask_round): New builtins.
30281 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30282 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30283 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30285 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30286 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30288 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30291 2017-05-02 Martin Jambor <mjambor@suse.cz>
30293 PR tree-optimization/78687
30294 * tree-sra.c (access): New field parent.
30295 (process_subtree_disqualification): New function.
30296 (disqualify_candidate): Call it.
30297 (build_accesses_from_assign): Reset write flag if creating an
30299 (build_access_subtree): Fill in parent field and also prpagate
30300 down grp_write flag.
30301 (create_artificial_child_access): New parameter set_grp_write, set
30302 grp_write to its value.
30303 (propagate_subaccesses_across_link): Also propagate grp_write flag
30305 (propagate_all_subaccesses): Push the closest parent back to work
30306 queue if add_access_to_work_queue returned true.
30308 2017-05-02 Richard Biener <rguenther@suse.de>
30310 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30311 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30312 -fstrict-overflow documentation.
30313 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30314 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30315 flag_strict_overflow.
30316 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30317 * lto-opts.c (lto_write_options): Do not stream it.
30318 * lto-wrapper.c (merge_and_complain): Do not handle it.
30319 * opts.c (default_options_table): Do not set -fstrict-overflow.
30320 (finish_options): Likewise do not clear it when sanitizing.
30321 * simplify-rtx.c (simplify_const_relational_operation): Do not
30322 test flag_strict_overflow.
30324 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30326 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30327 using enabled attribute.
30328 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30329 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30330 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30331 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30332 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30333 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30334 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30335 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30336 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30337 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30339 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30341 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30343 2017-05-02 Richard Biener <rguenther@suse.de>
30345 PR tree-optimization/80591
30347 2017-04-10 Richard Biener <rguenther@suse.de>
30349 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30352 2017-05-02 Richard Biener <rguenther@suse.de>
30354 PR tree-optimization/80549
30355 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30356 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30357 headers if they do not have a preheader.
30359 2017-05-02 Martin Liska <mliska@suse.cz>
30362 * common.opt: Fix typo.
30363 * doc/invoke.texi: Likewise.
30365 2017-05-01 Jan Beulich <jbeulich@suse.com>
30367 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30368 swapping, add (x,x,m,x,n) alternative.
30370 2017-05-01 Nathan Sidwell <nathan@acm.org>
30372 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30373 unnecessary unadjusted_alignment check.
30375 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30378 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30380 * gimplify.c (gimplify_cilk_detach): New function.
30381 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30382 * tree-core.h: Document EXPR_CILK_SPAWN.
30383 * tree.h (EXPR_CILK_SPAWN): Define.
30385 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30387 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30388 to use new fixit_hint representation, using the "replace" logic.
30389 (get_line_span_for_fixit_hint): Likewise.
30390 (layout::print_any_fixits): Likewise.
30391 (selftest::test_one_liner_many_fixits): Rename to...
30392 (selftest::test_one_liner_many_fixits_1): ...this, and update
30393 comment and expected output to reflect that the multiple fix-it
30394 hints are now consolidated into one insertion.
30395 (selftest::test_one_liner_many_fixits_2): New test.
30396 (selftest::test_diagnostic_show_locus_one_liner): Update for
30398 (selftest::test_fixit_consolidation): Update for fix-it API
30400 * diagnostic.c (print_parseable_fixits): Likewise.
30401 * edit-context.c (edited_line::m_line_events): Convert from
30402 auto_vec <line_event *> to auto_vec <line_event>.
30403 (class line_event): Convert from abstract base class to a concrete
30404 class, taking over the role of replace_event.
30405 (class insert_event): Delete.
30406 (class replace_event): Rename to class line_event. Convert to
30408 (edit_context::add_fixits): Reimplement.
30409 (edit_context::apply_insert): Delete.
30410 (edit_context::apply_replace): Rename to...
30411 (edit_context::apply_fixit): ...this. Convert to half-open range.
30412 (edited_file::apply_insert): Delete.
30413 (edited_file::apply_replace): Rename to...
30414 (edited_file::apply_fixit): ...this.
30415 (edited_line::~edited_line): Drop deletion of events.
30416 (edited_line::apply_insert): Delete.
30417 (edited_line::apply_replace): Rename to...
30418 (edited_line::apply_fixit): ...this. Convert to half-open range.
30419 Update for change to type of m_line_events.
30420 * edit-context.h (edit_context::apply_insert): Delete.
30421 (edit_context::apply_replace): Rename to...
30422 (edit_context::apply_fixit): ...this.
30424 2017-05-01 Martin Sebor <msebor@redhat.com>
30426 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30429 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30432 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30433 __get_cpuid_max returns 0.
30434 (__get_cpuid_count): Ditto.
30436 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30438 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30439 replacement expression is another instance of one of its arguments.
30441 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30444 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30445 check for stack push/pop autoinc.
30446 * config/i386/i386.c (ix86_agi_dependent): Return false
30447 if the only reason why modified_in_p returned true is that
30448 addr is SP based and set_insn is a push or pop.
30450 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30452 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30455 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30458 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30459 (account_size_time): Use two predicates - exec_pred and
30461 (evaluate_conditions_for_known_args): Compute both clause and
30463 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30464 (inline_summary_t::duplicate): Update.
30465 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30467 (compute_inline_parameters): Likewise.
30468 (estimate_edge_size_and_time): Update caluclation of time.
30469 (estimate_node_size_and_time): Compute both time and nonspecialized
30471 (estimate_ipcp_clone_size_and_time): Update.
30472 (inline_merge_summary): Update.
30473 (do_estimate_edge_time): Update.
30474 (do_estimate_edge_size): Update.
30475 (do_estimate_edge_hints): Update.
30476 (inline_read_section, inline_write_summary): Stream both new predicates.
30477 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30479 (compute_inlined_call_time): Cleanup.
30480 (big_speedup_p): Update.
30481 (edge_badness): Update.
30482 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30483 (size_time_entry): Replace predicate by exec_predicate and
30484 nonconst_predicate.
30485 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30486 (estimate_edge_time): Return also nonspec_time.
30487 (reset_edge_growth_cache): Update.
30489 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30491 PR rtl-optimization/80491
30492 * ifcvt.c (noce_process_if_block): When looking for x setter
30493 with missing else_bb, don't check only the insn right before
30494 cond_earliest, but look for the last insn that x is modified in
30495 within the same bb.
30497 PR rtl-optimization/80491
30498 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30500 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30502 PR tree-optimization/80487
30503 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30505 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30507 PR tree-optimization/79697
30508 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30509 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30510 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30512 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30513 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30515 2017-04-28 Martin Sebor <msebor@redhat.com>
30517 PR tree-optimization/80523
30518 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30519 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30521 (maybe_warn, format_directive, parse_directive): Use new functions.
30522 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30524 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30526 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30528 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30530 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30531 target_header_dir): Set correctly.
30532 * configure: Regenerated.
30533 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30534 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30535 instead of SYSTEM_HEADER_DIR.
30537 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30539 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30540 (estimate_local_effects): Likewise.
30541 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30542 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30543 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30544 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30545 do_estimate_edge_time, estimate_edge_time): Likewise.
30546 * ipa-inline-analysis.c (estimate_node_size_and_time,
30547 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30548 (estimate_time_after_inlining): Remove.
30550 2017-04-28 Martin Liska <mliska@suse.cz>
30552 * doc/gcov.texi: Enhance documentation of gcov.
30554 2017-04-28 Martin Liska <mliska@suse.cz>
30556 * doc/gcov.texi: Sort options in alphabetic order.
30557 * doc/gcov-dump.texi: Likewise.
30558 * doc/gcov-tool.texi: Likewise.
30559 * gcov.c (print_usage): Likewise.
30560 * gcov-dump.c (print_usage): Likewise.
30561 * gcov-tool.c (print_merge_usage_message): Likewise.
30562 (print_rewrite_usage_message): Likewise.
30563 (print_overlap_usage_message): Likewise.
30565 2017-04-28 Martin Liska <mliska@suse.cz>
30567 PR gcov-profile/53915
30568 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30570 2017-04-28 Martin Liska <mliska@suse.cz>
30572 PR gcov-profile/79891
30573 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30574 is marked by compiler as living on a line.
30575 (get_cycles_count): Remove usage of the union.
30576 (output_intermediate_file): Likewise.
30577 (find_source): Fix GNU coding style.
30578 (accumulate_line_counts): Remove old non-all block mode.
30579 (output_lines): Remove usage of the union.
30580 * profile.c (output_location): Include all BBs, even if
30581 belonging to a same line (and file) as a previous BB.
30583 2017-04-28 Martin Liska <mliska@suse.cz>
30585 * gcov.c (process_args): Handle new argument 'w'.
30586 (read_graph_file): Assign ID to BBs.
30587 (output_branch_count): Display BB # if verbose flag is set.
30588 (output_lines): Likewise for arcs.
30589 (print_usage): Add '--verbose' option help.
30590 * doc/gcov.texi: Document --verbose (-w) option.
30592 2017-04-28 Martin Liska <mliska@suse.cz>
30594 * gcov.c (struct block_location_info): New struct.
30595 (process_file): Fill up the new structure.
30596 (read_graph_file): Replace usage of encoding by the newly added
30598 (add_line_counts): Likewise.
30599 (accumulate_line_counts): Remove usage of the union.
30600 (function_info::function_info): New function.
30601 (function_info::~function_info): Likewise.
30602 (process_file): Call delete instead of release_function.
30603 (release_function): Release the function.
30604 (release_structures): Call delete instead of release_function.
30605 (solve_flow_graph): Replace usage of num_blocks.
30606 (find_exception_blocks): Likewise.
30607 (output_lines): Fix GNU coding style.
30609 2017-04-28 Martin Liska <mliska@suse.cz>
30612 * coverage.c (coverage_remove_note_file): New function.
30613 * coverage.h: Declare the function.
30614 * toplev.c (finalize): Clean if an error has been seen.
30616 2017-04-28 Martin Liska <mliska@suse.cz>
30618 PR gcov-profile/80031
30619 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30620 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30621 * gcov.c (read_graph_file): Read just number of blocks.
30622 * profile.c (branch_prob): Do not stream 0 flags per a basic
30625 2017-04-28 Martin Liska <mliska@suse.cz>
30627 * gcov-dump.c (tag_*): Add new argument to declarations.
30628 (dump_gcov_file): Likewise.
30629 (tag_blocks): Add and use new argument depth.
30630 (tag_arcs): Likewise.
30631 (tag_lines): Likewise.
30632 (tag_counters): Likewise.
30633 (tag_summary): Likewise.
30634 (dump_working_sets): Use depth to do a proper indentation.
30636 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30639 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30640 * symtab.c (symtab_node::debug_symtab): Move definition here.
30642 2017-04-28 Richard Biener <rguenther@suse.de>
30644 * lto-streamer.h (LTO_major_version): Bump to 7.
30646 2017-04-28 Richard Biener <rguenther@suse.de>
30648 * tree-vrp.c (assert_info): New struct.
30649 (add_assert_info): New helper.
30650 (register_edge_assert_for_2): Refactor to add asserts to a vector
30652 (register_edge_assert_for_1): Likewise.
30653 (register_edge_assert_for): Likewise.
30654 (finish_register_edge_assert_for): New helper actually registering
30655 asserts where live on edge.
30656 (find_conditional_asserts): Adjust.
30657 (find_switch_asserts): Likewise.
30658 (evrp_dom_walker::try_find_new_range): Generalize.
30659 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30661 2017-04-27 Marek Polacek <polacek@redhat.com>
30664 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30665 arg10 and arg11 to itype.
30667 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30669 * doc/extend.texi (Object Size Checking): Improve grammar.
30671 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30674 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30675 that the logic for permitting reciprocal estimates matches that
30678 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30681 * tree.c (type_cache_hasher::equal): Only compare
30682 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30683 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30684 non-aggregate element types.
30685 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30686 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30688 2017-04-27 Richard Biener <rguenther@suse.de>
30690 PR middle-end/80533
30691 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30692 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30693 keeping a reference to a trailing array.
30695 2017-04-27 Richard Biener <rguenther@suse.de>
30697 PR middle-end/80539
30698 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30699 being in loop-closed SSA form conservatively.
30700 (chrec_fold_multiply_poly_poly): Likewise.
30702 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30704 PR middle-end/79665
30705 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30706 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30708 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30711 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30712 (aarch64_function_arg_alignment): Return unsigned int again, but still
30713 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30714 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30715 Don't emit -Wpsabi note.
30716 (aarch64_function_arg_boundary): Likewise.
30717 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30720 2017-04-26 Nathan Sidwell <nathan@acm.org>
30722 * tree.h (crc32_unsigned_n): Declare.
30723 (crc32_unsigned, crc32_unsigned): Make inline.
30724 * tree.c (crc32_unsigned_bits): Replace with ...
30725 (crc32_unsigned_n): ... this.
30726 (crc32_unsigned, crc32_byte): Remove.
30727 (crc32_string): Remove unnecessary braces.
30729 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30731 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30732 * ipa-inline-analysis.c (MAX_TIME): Remove.
30733 (account_size_time): Use sreal for time.
30734 (dump_inline_summary): Update.
30735 (estimate_function_body_sizes): Update.
30736 (estimate_edge_size_and_time): Update.
30737 (estimate_calls_size_and_time): Update.
30738 (estimate_node_size_and_time): Update.
30739 (inline_merge_summary): Update.
30740 (inline_update_overall_summary): Update.
30741 (estimate_time_after_inlining): Update.
30742 (inline_read_section): Update.
30743 (inline_write_summary): Update.
30744 * ipa-inline.c (compute_uninlined_call_time): Update.
30745 (compute_inlined_call_time): Update.
30746 (recursive_inlining): Update.
30747 (inline_small_functions): Update.
30748 (dump_overall_stats): Update.
30749 * ipa-inline.h: Include sreal.h.
30750 (size_time_entry): Turn time to sreal.
30751 (inline_summary): Turn self_time nad time to sreal.
30753 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30755 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30757 (sreal::stream_out, sreal::stream_in): New.
30758 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30760 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30762 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30765 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30768 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30770 (dimode_scalar_chain::compute_convert_gain): Ditto.
30771 (dimode_scalar_chain::make_vector_copies): Ditto.
30772 (dimode_scalar_chain::convert_reg): Ditto.
30773 (dimode_scalar_chain::convert_insn): Ditto.
30774 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30775 (VI248_AVX512BW_1): New mode iterator.
30776 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30777 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30780 2017-04-25 Martin Sebor <msebor@redhat.com>
30782 PR tree-optimization/80497
30783 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30784 constants are representable in HOST_WIDE_INT.
30785 (parse_directive): Ditto.
30787 2017-04-25 Martin Sebor <msebor@redhat.com>
30790 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30791 (new_zero_array): Adjust signature.
30792 (dom_info::dom_init): Used unsigned rather that size_t.
30793 (dom_info::dom_info): Same.
30795 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30796 Jakub Jelinek <jakub@redhat.com>
30799 * config/arm/arm.c: Include gimple.h.
30800 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30801 returns negative, increment ncrn only if it returned positive.
30802 (arm_needs_doubleword_align): Return int instead of bool,
30803 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30804 members, but if there is any such non-FIELD_DECL
30805 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30806 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30807 returns negative, increment nregs only if it returned positive.
30808 (arm_setup_incoming_varargs): Likewise.
30809 (arm_function_arg_boundary): Emit -Wpsabi note if
30810 arm_needs_doubleword_align returns negative, return
30811 DOUBLEWORD_ALIGNMENT only if it returned positive.
30813 2017-04-25 Marek Polacek <polacek@redhat.com>
30816 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30817 first argument to type.
30819 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30822 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30823 type checks to test for compatibility instead of equality.
30825 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30826 Jakub Jelinek <jakub@redhat.com>
30829 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30831 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30832 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30833 the alignment computation, but return their maximum in warn_alignment.
30834 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30835 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30837 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30838 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30841 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30843 * config/arc/simdext.md (dmpyh): Fix typo.
30845 2017-04-25 Richard Biener <rguenther@suse.de>
30847 PR tree-optimization/80492
30848 * alias.c (compare_base_decls): Handle registers with asm
30849 specification conservatively.
30850 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30851 compare_base_decls returning dont-know properly.
30853 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30855 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30856 (legitimate_offset_address_p): New function.
30857 (arc_legitimate_address_p): Use above function.
30859 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30861 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30863 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30865 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30866 ACCH registers whenever they are available.
30868 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30870 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30871 double regs fix when not used.
30873 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30875 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30877 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30878 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30880 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30882 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30883 when emitting short ADD instructions.
30885 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30887 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30889 (cmpsi_cc_c_insn): Likewise.
30890 (cbranchsi4_scratch): Compute proper instruction length using
30891 compact_hreg_operand.
30892 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30894 2017-04-25 Richard Biener <rguenther@suse.de>
30896 PR middle-end/80509
30897 * passes.c (pass_manager::pass_manager): Initialize
30898 m_name_to_pass_map.
30900 2017-04-25 Richard Biener <rguenther@suse.de>
30902 PR tree-optimization/79201
30903 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30905 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30908 * config/s390/vector.md: Split MEM->GPR vector moves for
30909 non-s_operand addresses.
30911 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30914 * config/s390/predicates.md (reload_const_wide_int_operand): New
30916 * config/s390/s390.md ("movti"): Remove d/P alternative.
30917 ("movti_bigconst"): New pattern definition.
30919 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30922 * s390-protos.h (s390_expand_cs_hqi): Removed.
30923 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30924 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30925 modes as well as CCZ1mode and CCZmode.
30926 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30927 signature of s390_emit_compare_and_swap.
30928 (s390_expand_cs_hqi): Likewise, make static.
30929 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30930 compare-and-swap, in some cases.
30931 (s390_expand_cs): Wrapper function.
30932 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30934 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30935 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30936 patterns for small and large integers. Forbid symref memory operands.
30937 Move expander to s390.c. Require cc register.
30938 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30939 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30940 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30941 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30942 symref memory operands. Remove CC mode and call s390_match_ccmode
30944 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30946 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30948 * config/s390/s390.md (define_peephole2): New peephole to help
30949 combining the load-and-test pattern with volatile memory.
30951 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30953 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30954 with CCZmode for TARGET_Z196.
30956 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30958 PR rtl-optimization/80501
30959 * combine.c (make_compound_operation_int): Set subreg_code to SET
30960 even for AND with mask of the sign bit of mode.
30962 PR rtl-optimization/80500
30963 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30964 sum's initial value.
30966 2017-04-25 Julian Brown <julian@codesourcery.com>
30967 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30969 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30971 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30973 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30975 2017-04-25 Julian Brown <julian@codesourcery.com>
30976 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30978 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30979 (thunderx2t99_sha): New Reservation.
30981 2017-04-25 Julian Brown <julian@codesourcery.com>
30982 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30984 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30985 type for 1-element load.
30987 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30989 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30991 2017-04-24 Martin Jambor <mjambor@suse.cz>
30993 PR tree-optimization/80293
30994 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30995 char arrays not totally scalarizable if it is false.
30996 (analyze_all_variable_accesses): Pass correct value in the new
30997 parameter. Add a statistics counter.
30999 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31001 PR middle-end/79931
31002 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31004 2017-04-24 Richard Biener <rguenther@suse.de>
31006 PR tree-optimization/80494
31007 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31008 out for complex types.
31010 2017-04-24 Richard Biener <rguenther@suse.de>
31012 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31013 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31014 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31015 (DFS): Adjust and never fail.
31016 (sccvn_dom_walker::fail): Remove.
31017 (sccvn_dom_walker::before_dom_children): Adjust.
31018 (run_scc_vn): Likewise and never fail.
31019 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31020 (pass_fre::execute): Likewise.
31022 2017-04-24 Richard Biener <rguenther@suse.de>
31024 PR tree-optimization/79725
31025 * tree-ssa-sink.c (statement_sink_location): Return whether
31026 failure reason was zero uses. Move that check later.
31027 (sink_code_in_bb): Deal with zero uses by removing the stmt
31030 2017-04-24 Richard Biener <rguenther@suse.de>
31033 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31034 pointer-based references.
31036 2017-04-24 Richard Biener <rguenther@suse.de>
31039 * pass_manager.h (pass_manager::operator new): Remove.
31040 (pass_manager::operator delete): Likewise.
31041 * passes.c (pass_manager::operator new): Remove.
31042 (pass_manager::operator delete): Likewise.
31043 (pass_manager::pass_manager): Zero individual pass members.
31045 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31048 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31049 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31050 Check "XEXP (src, 1)" operand here.
31051 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31052 Check "XEXP (src, 1)" operand here.
31053 (dimode_scalar_chain::make_vector_copies): Detect count register
31054 of a shift instruction. Zero extend count register from QImode
31055 to DImode to satisfy vector shift pattern count operand predicate.
31056 Substitute vector shift count operand with a DImode copy.
31057 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31060 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31062 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31063 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31064 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31065 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31066 (UNSPEC_NOREX_MEM): Remove definition.
31068 2017-04-21 Richard Biener <rguenther@suse.de>
31070 PR tree-optimization/79547
31071 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31072 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31073 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31074 without any constraints.
31076 2017-04-21 Richard Biener <rguenther@suse.de>
31078 PR tree-optimization/78847
31079 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31081 2017-04-21 Richard Biener <rguenther@suse.de>
31083 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31084 (build_distinct_type_copy): Likewise.
31085 (build_variant_type_copy): Likewise.
31086 * tree.c (build_qualified_type): Pass down mem-stat info.
31087 (build_distinct_type_copy): Likewise.
31088 (build_variant_type_copy): Likewise.
31090 2017-04-21 Richard Biener <rguenther@suse.de>
31092 PR tree-optimization/80237
31093 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31095 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31096 for a simplified result.
31098 2016-04-21 Richard Biener <rguenther@suse.de>
31100 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31101 sth as strict as a simple_iv but a chrec without symbols and an
31102 operand defined in the loop we are peeling (and not some subloop).
31103 (propagate_constants_for_unrolling): Propagate all constants.
31105 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31108 * config/i386/i386.c (print_reg): Remove assert for disalowed
31109 regno values, call output_operand_lossage instead.
31111 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31114 * config/i386/constraints.md (Yc): New register constraint.
31115 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31116 Use Yc constraint for alternative 2 of operand 0. Remove
31117 preferred_for_speed attribute.
31119 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31121 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31122 lastprivate clauses in SIMT case.
31124 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31126 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31128 2017-04-20 Richard Biener <rguenther@suse.de>
31130 PR tree-optimization/57796
31131 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31132 as N scalar stores.
31133 (vect_model_load_cost): Cost gathers as N scalar loads.
31135 2017-04-20 Richard Biener <rguenther@suse.de>
31137 * ggc-page.c (ggc_allocated_p): Rename to ...
31138 (safe_lookup_page_table_entry): ... this and return the lookup
31140 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31142 2017-04-20 Richard Biener <rguenther@suse.de>
31144 PR tree-optimization/80453
31145 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31146 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31147 from the conditions.
31148 (vn_phi_eq): Pass them down.
31149 (vn_phi_lookup): Record them.
31150 (vn_phi_insert): Likewise.
31152 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31154 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31155 uninitialized variable warning to avoid buffer overrun.
31157 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31160 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31161 is suppressed for '{ 0 }' in C.
31163 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31165 * BASE-VER: Set to 8.0.0.
31167 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31169 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31170 priority .init_array and .fini_array section with SECTION_NOTYPE
31173 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31175 PR middle-end/80423
31176 * tree.h (build_array_type): Add typeless_storage default argument.
31177 * tree.c (type_cache_hasher::equal): Also compare
31178 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31179 (build_array_type): Add typeless_storage argument, set
31180 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31182 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31183 (build_array_type): Likewise. Add typeless_storage argument.
31185 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31186 Jakub Jelinek <jakub@redhat.com>
31188 PR tree-optimization/80426
31189 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31190 operation on symbolic operands, also compute the overflow for the
31191 invariant part when the operation degenerates into a negation.
31193 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31196 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31197 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31200 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31202 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31205 * config/avr/avr.c (tree.h): Include it.
31206 (cgraph.h): Include it.
31207 (avr_encode_section_info): Don't warn for uninitialized progmem
31208 variable if it's just an alias.
31210 2017-04-19 Richard Biener <rguenther@suse.de>
31213 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31214 when needed by AutoPGO.
31216 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31219 * doc/lto.texi: Remove an extra 'that'.
31221 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31223 PR rtl-optimization/80429
31224 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31225 are only used in debug insns.
31227 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31228 Vladimir Makarov <vmakarov@redhat.com>
31230 * config/sparc/predicates.md (input_operand): Add comment. Return
31231 true for any memory operand when LRA is in progress.
31232 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31234 2017-04-18 Jeff Law <law@redhat.com>
31237 * mips.md ({return,simple_return}_internal): Do not overwrite
31240 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31242 PR tree-optimization/80443
31243 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31244 instead of adding 1, subtract -1 and similarly instead of subtracting
31247 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31249 PR rtl-optimization/80357
31250 * haifa-sched.c (tmp_bitmap): New variable.
31251 (model_recompute): Handle duplicate use records.
31252 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31253 (free_global_sched_pressure_data): Free it.
31255 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31258 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31259 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31260 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31261 instead of SYSTEM_HEADER_DIR.
31263 2017-04-18 Jeff Law <law@redhat.com>
31265 PR middle-end/80422
31266 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31267 predecessors after walking up the insn chain.
31269 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31272 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31273 sizetype type into debug info.
31275 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31278 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31279 unneeded test for TARGET_UPPER_REGS_SF.
31280 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31282 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31285 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31286 instead of gsi_after_labels.
31288 2017-04-18 Jeff Law <law@redhat.com>
31290 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31294 2017-04-13 Jeff Law <law@redhat.com>
31295 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31296 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31298 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31301 * config/avr/avr.c (intl.h): Include it.
31302 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31304 2017-04-18 Martin Liska <mliska@suse.cz>
31306 PR gcov-profile/78783
31307 * gcov-tool.c (gcov_output_files): Validate that destination
31308 file is either removed by the tool or by a user.
31310 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31311 Guy Benyei <guybe@mellanox.com>
31313 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31314 block, and do not negate it, the stored id is already negative.
31316 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31318 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31320 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31323 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31324 masks of options that should be turned off if the VSX vector
31325 options are turned off.
31326 (OTHER_P8_VECTOR_MASKS): Likewise.
31327 (OTHER_VSX_VECTOR_MASKS): Likewise.
31328 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31329 rs6000_disable_incompatible_switches to validate no type switches
31331 (rs6000_incompatible_switch): New function to disallow turning on
31332 other vector options if -mno-vsx, -mno-power8-vector, or
31333 -mno-power9-vector are specified.
31335 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31337 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31339 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31341 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31342 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31343 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31344 (ARG_POINTER_CFA_OFFSET): Likewise.
31346 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31348 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31349 conditions to take advantage of various optimizations.
31351 2017-04-13 Jeff Law <law@redhat.com>
31353 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31354 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31355 (zero_extendsidi2_dext): Likewise.
31357 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31360 * fold-const.c (fold_ternary_loc): Revert
31361 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31364 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31366 PR rtl-optimization/80343
31367 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31368 new scratch pseudo.
31370 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31373 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31374 to ubsan_encode_value.
31376 2017-04-13 Jeff Law <law@redhat.com>
31378 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31379 appearing in DEBUG_INSNs.
31381 2017-04-13 Martin Liska <mliska@suse.cz>
31383 PR gcov-profile/80413
31384 * gcov-io.c (gcov_write_string): Copy to buffer just when
31385 allocated size is greater than zero.
31387 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31390 * dwarf2out.c (decls_for_scope): Ignore declarations of
31391 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31393 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31396 * ipa-visibility.c (non_local_p): Fix typos.
31397 (localize_node): When localizing symbol in same comdat group,
31398 dissolve the group only when we know external symbols are going
31400 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31402 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31404 PR tree-optimization/79390
31405 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31406 order does not result in usable sequence, retry with reversed operand
31412 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31413 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31414 op0 instead of fold_convert_loc (loc, type, arg0).
31416 2017-04-12 Jeff Law <law@redhat.com>
31418 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31419 has a delay slot in the generated code.
31421 * config/cris/cris.md (cris_preferred_reload_class): Return
31422 GENNONACR_REGS rather than GENERAL_REGS.
31424 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31427 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31428 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31429 signedness of the result type.
31431 2017-04-12 Richard Biener <rguenther@suse.de>
31432 Jeff Law <law@redhat.com>
31434 PR tree-optimization/80359
31435 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31436 trim stores to TARGET_MEM_REFs.
31438 2017-04-12 Richard Biener <rguenther@suse.de>
31440 PR tree-optimization/79390
31441 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31442 threading case even more.
31444 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31447 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31448 for quad_address_p for TImode, instead of just not indexed_address.
31450 2017-04-12 Richard Biener <rguenther@suse.de>
31451 Bernd Edlinger <bernd.edlinger@hotmail.de>
31453 PR middle-end/79671
31454 * alias.c (component_uses_parent_alias_set_from): Handle
31455 TYPE_TYPELESS_STORAGE.
31456 (get_alias_set): Likewise.
31457 * tree-core.h (tree_type_common): Add typeless_storage flag.
31458 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31459 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31460 for types containing members with TYPE_TYPELESS_STORAGE.
31461 (place_field): Likewise.
31462 (layout_type): Likewise for ARRAY_TYPE.
31463 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31464 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31465 TYPE_TYPELESS_STORAGE.
31466 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31468 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31471 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31472 first argument to type.
31474 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31478 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31479 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31480 (rs6000_expand_binop_builtin): Likewise.
31481 (rs6000_expand_ternop_builtin): Likewise; also add missing
31482 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31483 vshasigma built-ins.
31484 * doc/extend.texi: Document that vec_xxpermdi's third argument
31485 must be a constant.
31487 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31489 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31490 Use shift_const cost parameter when calculating gain of STV shifts.
31492 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31494 PR rtl-optimization/70478
31495 * lra-constraints.c (process_alt_operands): Check memory for
31496 disfavoring memory insn operand.
31498 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31500 PR middle-end/80100
31501 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31502 left shift in unsigned HOST_WIDE_INT type.
31504 PR rtl-optimization/80385
31505 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31506 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31509 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31510 if they have any depend clauses.
31512 2017-04-11 Martin Liska <mliska@suse.cz>
31515 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31516 * ipa-split.c (split_function): Create a local comdat symbol
31517 if caller is in a comdat group.
31519 2017-04-11 Martin Liska <mliska@suse.cz>
31522 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31525 2017-04-11 Martin Sebor <msebor@redhat.com>
31527 PR middle-end/80364
31528 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31529 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31531 (directive::set_width, directive::set_precision, format_character):
31533 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31536 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31539 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31540 conflict, set target->arch_name instead of target->cpu_name.
31542 2017-04-11 Richard Biener <rguenther@suse.de>
31544 PR tree-optimization/80374
31545 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31546 build_zero_cst, remove fold_convertible_p check again.
31548 2017-04-11 Martin Liska <mliska@suse.cz>
31551 * ubsan.c (instrument_object_size): Do not instrument register
31554 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31557 * config/i386/i386-builtin-types.def
31558 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31559 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31560 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31561 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31562 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31563 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31564 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31565 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31566 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31567 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31568 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31569 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31570 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31571 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31572 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31573 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31574 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31575 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31576 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31577 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31578 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31579 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31580 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31581 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31582 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31583 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31584 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31585 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31586 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31587 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31588 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31589 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31590 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31591 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31592 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31593 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31594 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31595 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31596 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31597 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31598 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31599 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31600 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31601 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31602 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31603 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31605 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31606 flag to second_arg_count, handle 4 argument function type _COUNT
31607 aliases, handle second_arg_count on second argument rather than last.
31609 2017-04-10 Jeff Law <law@redhat.com>
31611 PR tree-optimization/80374
31612 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31613 record anything if we can not convert integer_zero_node to the
31616 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31619 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31620 Enhance special handling given to the TARGET_P9_MINMAX option in
31621 relation to certain other options.
31623 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31625 PR tree-optimization/80153
31626 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31627 remove POINTER_PLUS_EXPR's base part directly, rather than through
31630 2017-04-10 Richard Biener <rguenther@suse.de>
31631 Bin Cheng <bin.cheng@arm.com>
31633 PR tree-optimization/80153
31634 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31635 the first element of pointer type aff_tree. Build result expr in
31637 (add_elt_to_tree): Convert to type unconditionally. Remove other
31638 fold_convert calls.
31639 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31640 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31642 2017-04-10 Richard Biener <rguenther@suse.de>
31644 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31647 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31649 PR rtl-optimization/70478
31650 * lra-constraints.c (curr_small_class_check): New.
31651 (update_and_check_small_class_inputs): New.
31652 (process_alt_operands): Update curr_small_class_check. Disfavor
31653 alternative insn memory operands. Check available regs for small
31656 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31659 * config/mips/mips.opt (-mvirt): Update description.
31660 * doc/invoke.texi (-mvirt): Likewise.
31662 2017-04-10 Richard Biener <rguenther@suse.de>
31664 PR middle-end/80362
31665 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31666 looking for NEGATE_EXPR in -A / -B to A / B folding.
31668 2017-04-10 Martin Liska <mliska@suse.cz>
31670 PR gcov-profile/80224
31671 * gcov.c (print_usage): Fix usage string.
31672 (get_gcov_intermediate_filename): Remove.
31673 (output_gcov_file): Use both for normal and intermediate format.
31674 (generate_results): Do not initialize special file for
31675 intermediate format.
31677 2017-04-10 Richard Biener <rguenther@suse.de>
31679 PR tree-optimization/80304
31680 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31683 2017-04-10 Nathan Sidwell <nathan@acm.org>
31686 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31687 (rs6000_init_builtins): Use it.
31689 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31691 * config/arm/arm.md (<mrc>): Add mode to SET source.
31692 (<mrrc>): Likewise.
31694 2017-04-10 Richard Biener <rguenther@suse.de>
31696 PR middle-end/80344
31697 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31699 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31702 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31703 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31704 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31705 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31706 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31707 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31708 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31709 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31710 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31711 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31712 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31713 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31714 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31715 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31716 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31717 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31718 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31719 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31720 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31721 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31722 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31723 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31724 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31726 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31728 PR rtl-optimization/70478
31729 * lra-constraints.c: Reverse the last patch.
31731 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31733 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31734 Add comment for WCHAR_T.
31736 2017-04-08 Martin Liska <mliska@suse.cz>
31739 2017-04-07 Martin Liska <mliska@suse.cz>
31742 * ipa-split.c (split_function): Add function part to a same comdat
31745 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31748 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31750 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31752 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31753 (vec_pairing): ...to this.
31754 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31755 (rs6000_sched_init): Adjust for name change.
31756 (struct rs6000_sched_context): Likewise.
31757 (rs6000_init_sched_context): Likewise.
31758 (rs6000_set_sched_context): Likewise.
31760 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31766 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31768 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31769 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31770 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31772 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31774 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31776 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31778 PR rtl-optimization/70703
31779 * ira-color.c (update_conflict_hard_regno_costs): Use
31780 int64_t instead of HOST_WIDE_INT.
31782 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31784 PR rtl-optimization/70478
31785 * lra-constraints.c (process_alt_operands): Disfavor alternative
31786 insn memory operands.
31788 2017-04-07 Jeff Law <law@redhat.com>
31790 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31791 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31793 2017-04-07 Martin Liska <mliska@suse.cz>
31796 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31797 Show error message instead of an ICE.
31799 2017-04-07 Martin Liska <mliska@suse.cz>
31802 * ipa-split.c (split_function): Add function part to a same comdat
31805 2017-04-07 Richard Biener <rguenther@suse.de>
31807 PR middle-end/80341
31808 * tree.c (get_unwidened): Also handle ! for_type case for
31810 * convert.c (do_narrow): Split out from ...
31811 (convert_to_integer_1): ... here. Do not pass final truncation
31812 type to get_unwidened for TRUNC_DIV_EXPR.
31814 2017-04-07 Richard Biener <rguenther@suse.de>
31816 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31818 (aff_combination_const): Adjust.
31819 (aff_combination_scale): Likewise.
31820 (aff_combination_add_elt): Likewise.
31821 (aff_combination_add_cst): Likewise.
31822 (aff_combination_convert): Likewise.
31823 (add_elt_to_tree): Likewise. Remove unused argument.
31824 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31826 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31828 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31830 * config/arm/arm.c (arm_default_short_enums): Use
31831 ARM_DEFAULT_SHORT_ENUMS.
31832 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31834 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31837 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31838 members with redundant out-of-class redeclaration.
31840 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31843 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31844 * config/i386/i386.md (*zero_extendsidi2):
31845 Add (?*x,*x) and (?*v,*v) alternatives.
31847 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31850 * config/i386/i386.c (ix86_expand_builtin)
31851 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31852 mode from insn data. Convert operands to insn operand mode.
31853 Copy operands that don't satisfy insn predicate to a register.
31855 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31857 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31860 2017-04-06 Richard Biener <rguenther@suse.de>
31862 PR tree-optimization/80334
31863 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31864 preserve alignment of accesses.
31866 2017-04-06 Richard Biener <rguenther@suse.de>
31868 PR tree-optimization/80262
31869 * tree-sra.c (build_ref_for_offset): Preserve address-space
31871 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31872 Drop useless address-space information on MEM_REF offsets.
31874 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31876 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31878 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31880 PR rtl-optimization/70703
31881 * ira-color.c (update_conflict_hard_regno_costs): Use
31882 HOST_WIDE_INT instead of long.
31884 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31887 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31888 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31890 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31891 for x86_64 target. Handle -m3dnowa option.
31893 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31895 PR rtl-optimization/70703
31896 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31897 (update_conflict_hard_regno_costs): Use long instead of unsigned
31898 arithmetic for cost calculation.
31900 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31901 Bernd Edlinger <bernd.edlinger@hotmail.de>
31904 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31907 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31910 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31911 ptr_mode with Pmode throughout.
31912 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31913 into probe_stack_range and use DImode.
31915 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31918 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31919 call_eh_return is true.
31921 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31923 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31924 Initialize last_match_fntype_index.
31926 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31929 * tree-nvr.c: Include internal-fn.h.
31930 (pass_return_slot::execute): Ignore internal calls without
31933 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31934 Richard Biener <rguenther@suse.de>
31937 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31938 captures used multiple times, except for the last use.
31939 * generic-match-head.c: Include gimplify.h.
31941 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31943 PR tree-optimization/79390
31944 * target.h (struct noce_if_info): Declare.
31945 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31946 * target.def (noce_conversion_profitable_p): New target hook.
31947 * ifcvt.h (struct noce_if_info): New type, moved from ...
31948 * ifcvt.c (struct noce_if_info): ... here.
31949 (noce_conversion_profitable_p): Renamed to ...
31950 (default_noce_conversion_profitable_p): ... this. No longer
31952 (noce_try_store_flag_constants, noce_try_addcc,
31953 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31954 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31955 instead of noce_conversion_profitable_p.
31956 * config/i386/i386.c: Include ifcvt.h.
31957 (ix86_option_override_internal): Don't override
31958 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31959 (ix86_noce_conversion_profitable_p): New function.
31960 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31961 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31962 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31963 * doc/tm.texi: Regenerated.
31965 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31967 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31970 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31973 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31974 instructions for small multiply cores.
31976 2017-04-04 Jeff Law <law@redhat.com>
31978 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31980 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31981 that are not set by the loop over the elements.
31983 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31986 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31987 int mode, convert_modes it to mode as unsigned, otherwise use
31988 lowpart_subreg to mode rather than SImode.
31989 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31990 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31991 Use DImode instead of SImode for the shift count operand.
31992 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31995 2017-04-04 Richard Biener <rguenther@suse.de>
31997 PR middle-end/80281
31998 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31999 arithmetic done for the negate or the plus. Simplify.
32000 (A - (-B) -> A + B): Likewise.
32001 * fold-const.c (split_tree): Make sure to not negate pointers.
32003 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32005 PR rtl-optimization/60818
32006 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32007 a compare of comparisons with the thing compared if this results
32008 in a different machine mode.
32010 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32012 * alias.c (base_alias_check): Fix typo in comment.
32013 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32014 * cgraphunit.c (symbol_table::compile): Likewise.
32015 * collect2.c (maybe_run_lto_and_relink): Likewise.
32016 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32017 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32018 * config/avr/avr.c (avr_map_op_t): Likewise.
32019 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32020 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32021 * config/epiphany/epiphany.md (movcc): Likewise.
32022 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32023 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32025 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32026 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32027 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32028 * config/sh/sh.c (sh_rtx_costs): Likewise.
32029 * fold-const.c (fold_truth_andor): Likewise.
32030 * genautomata.c (collapse_flag): Likewise.
32031 * gengtype.h (struct type::u::s): Likewise.
32032 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32033 * input.c (FORMAT_AMOUNT): Likewise.
32034 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32035 (known_aggs_to_agg_replacement_list): Likewise.
32036 * ipa-inline-analysis.c: Likewise.
32037 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32038 * ipa-polymorphic-call.c
32039 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32040 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32041 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32043 * modulo-sched.c (apply_reg_moves): Likewise.
32044 * omp-expand.c (build_omp_regions_1): Likewise.
32045 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32046 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32047 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32048 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32049 * value-prof.c: Likewise.
32050 * var-tracking.c (val_reset): Likewise.
32052 2017-04-03 Richard Biener <rguenther@suse.de>
32054 PR tree-optimization/80275
32055 * fold-const.c (split_address_to_core_and_offset): Handle
32058 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32060 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32061 descriptors is at least equal to that of functions.
32063 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32065 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32067 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32070 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32071 (mov<IMOD4:mode>): New expander.
32072 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32074 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32076 PR rtl-optimization/79405
32077 * fwprop.c (propagations_left): New variable.
32078 (forward_propagate_into): Decrement it.
32079 (fwprop_init): Initialize it.
32080 (fw_prop): If the variable has reached zero, stop propagating.
32081 (fwprop_addr): Ditto.
32083 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32086 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32087 a FUNCTION_DECL, pass it as decl instead of origin to
32090 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32092 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32095 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32098 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32099 TARGET_VSX_SMALL_INTEGER.
32101 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32103 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32104 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32106 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32108 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32109 extraction from odd-numbered MSA register.
32111 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32113 PR middle-end/80173
32114 * expmed.c (store_bit_field_1): Don't attempt to create
32115 a word subreg out of hard registers wider than word if they
32116 have HARD_REGNO_NREGS of 1 for their mode.
32118 PR middle-end/80163
32119 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32120 conversions to integer types wider than word and pointer.
32123 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32124 (rtx_equal_for_cselib_p): Pass 0 to it.
32125 * cselib.c (cselib_hasher::equal): Likewise.
32126 (rtx_equal_for_cselib_1): Add depth argument. If depth
32127 is 128, don't look up VALUE locs and punt. Increment
32128 depth in recursive calls when walking VALUE locs.
32130 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32132 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32133 (make_gcov_file_name): Use the canonical path name for generating
32135 (read_line): Fix handling of files with ascii null bytes.
32137 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32139 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32140 to initialise a vector register instead
32141 of using a const_int.
32143 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32145 PR translation/80189
32146 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32147 diagnostic messages.
32149 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32152 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32153 (dfp_diex_<mode>): Update mode of operand 1.
32154 * doc/extend.texi (dxex, dxexq): Document change to return type.
32155 (diex, diexq): Document change to argument type.
32157 2017-03-30 Martin Jambor <mjambor@suse.cz>
32160 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32161 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32162 it reflects the signature changes performed at the callee side.
32163 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32164 to cgraph_build_function_type_skip_args.
32165 (build_function_decl_skip_args): Adjust call to the above function.
32167 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32170 * config/i386/sse.md
32171 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32172 register as dest whenever it is a MEM not rtx_equal_p to the
32173 corresponding dup operand, and when forcing into reg move the
32174 reg into the memory afterwards.
32175 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32176 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32177 for the force_reg mode.
32178 (avx512vl_vextractf128<mode>): Use register as dest either
32179 always when a MEM, or when it is a MEM not rtx_equal_p to the
32180 corresponding dup operand, or even not when it is a CONST_VECTOR
32181 depending on the mode and lo vs. hi.
32182 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32184 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32185 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32186 Likewise. Require that operands[2] is even.
32187 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32188 Remove extraneous parens. Require that operands[2] is a multiple
32190 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32191 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32192 disallow memory then.
32194 2017-03-30 Richard Biener <rguenther@suse.de>
32196 PR tree-optimization/77498
32197 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32198 to non-constants over backedges.
32200 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32202 PR rtl-optimization/80233
32203 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32204 as last_combined_insn. Do not test for BARRIER_P separately.
32206 2017-03-29 Andreas Schwab <schwab@suse.de>
32209 * calls.c (prepare_call_address): Convert funexp to Pmode before
32210 copying to temp reg.
32212 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32214 PR tree-optimization/80158
32215 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32216 Handle possible future case of more than one alternate
32218 (replace_rhs_if_not_dup): Likewise.
32219 (replace_one_candidate): Likewise.
32221 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32223 PR rtl-optimization/80193
32224 * ira.c (ira): Do not check allocation for LRA.
32226 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32228 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32229 (nvptx_output_simt_exit): Declare.
32230 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32231 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32232 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32233 (nvptx_declare_function_name): ...here. Emit declaration of local
32234 memory space buffer for omp_simt_enter insn.
32235 (nvptx_output_unisimt_switch): New.
32236 (nvptx_output_softstack_switch): New.
32237 (nvptx_output_simt_enter): New.
32238 (nvptx_output_simt_exit): New.
32239 * config/nvptx/nvptx.h (struct machine_function): New fields
32240 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32241 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32242 (UNSPECV_SIMT_EXIT): Ditto.
32243 (omp_simt_enter_insn): New insn.
32244 (omp_simt_enter): New expansion.
32245 (omp_simt_exit): New insn.
32246 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32248 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32249 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32250 (expand_GOMP_SIMT_EXIT): New.
32251 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32252 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32253 (GOMP_SIMT_EXIT): Ditto.
32254 * target-insns.def (omp_simt_enter): New insn.
32255 (omp_simt_exit): Ditto.
32256 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32258 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32259 (lower_rec_input_clauses): Likewise.
32260 (lower_lastprivate_clauses): Handle SIMT privatization.
32262 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32263 (ompdevlow_adjust_simt_enter): New.
32264 (find_simtpriv_var_op): New.
32265 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32266 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32268 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32269 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32270 (copy_decl_for_dup_finish): Ditto.
32272 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32274 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32277 * config/i386/i386.c (ix86_option_override_internal): Always
32278 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32280 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32282 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32284 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32286 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32287 mark new edge's irreducible flag accordign to it.
32288 (vect_do_peeling): Check loop preheader edge's irreducible flag
32289 and pass it to function slpeel_add_loop_guard.
32291 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32293 PR tree-optimization/80218
32294 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32295 Update block frequencies and counts.
32297 2017-03-28 Richard Biener <rguenther@suse.de>
32299 PR tree-optimization/78644
32300 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32301 of a simplification result we may not use it at all.
32303 2017-03-28 Richard Biener <rguenther@suse.de>
32306 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32307 without arguments, generate default definition of a SSA name.
32309 2017-03-28 Richard Biener <rguenther@suse.de>
32311 PR middle-end/80222
32312 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32313 TYPE_REF_CAN_ALIAS_ALL references.
32314 * fold-const.c (fold_indirect_ref_1): Likewise.
32316 2017-03-28 Martin Liska <mliska@suse.cz>
32319 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32320 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32322 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32323 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32325 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32326 (EXTRA_SPECS): Define.
32327 (SUBTARGET_EXTRA_SPECS): Likewise.
32328 (SUBTARGET_CPP_SPEC): Likewise.
32329 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32330 SUBTARGET_EXTRA_SPECS.
32331 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32333 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32335 * config/arc/simdext.md (vst64_insn): Update pattern.
32336 (vld32wh_insn): Likewise.
32337 (vld32wl_insn): Likewise.
32338 (vld64_insn): Likewise.
32339 (vld32_insn): Likewise.
32341 2017-03-28 Marek Polacek <polacek@redhat.com>
32344 * fold-const.c (fold_comparison): Use protected_set_expr_location
32345 instead of SET_EXPR_LOCATION.
32347 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32349 * tree.c (add_expr): Avoid name lookup warning.
32351 2017-03-27 Jeff Law <law@redhat.com>
32353 PR tree-optimization/80216
32354 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32355 function name. Limit recursion depth.
32356 (record_temporary_equivalences): Corresponding changes.
32358 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32360 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32363 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32366 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32368 * cfgcleanup.c (reg_note_cfa_p): New array.
32369 (insns_have_identical_cfa_notes): New function.
32370 (old_insns_match_p): Don't cross-jump in between /f
32371 and non-/f instructions. If both i1 and i2 are frame related,
32372 verify all CFA notes, their order and content.
32374 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32377 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32378 HImode and SImode with zero extend to DImode to one insn.
32379 (bswap<mode>2_extenddi): Likewise.
32380 (bswapsi2_extenddi): Likewise.
32381 (bswaphi2_extendsi): Likewise.
32382 (bswaphi2): Combine bswap HImode and SImode into one insn.
32383 Separate memory insns from swapping register.
32384 (bswapsi2): Likewise.
32385 (bswap<mode>2): Likewise.
32386 (bswaphi2_internal): Delete, no longer used.
32387 (bswapsi2_internal): Likewise.
32388 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32389 store, and gpr<-gpr swap insns.
32390 (bswap<mode>2_store): Likewise.
32391 (bswaphi2_reg): Register only splitter, combine with the splitter.
32392 (bswaphi2 splitter): Likewise.
32393 (bswapsi2_reg): Likewise.
32394 (bswapsi2 splitter): Likewise.
32395 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32396 the insns into load, store, and register/register insns.
32397 (bswapdi2_ldbrx): Likewise.
32398 (bswapdi2_load): Likewise.
32399 (bswapdi2_store): Likewise.
32400 (bswapdi2_reg): Likewise.
32402 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32404 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32405 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32407 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32410 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32412 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32413 special handling for target option conflicts between dform
32414 options (-mpower9-dform, -mpower9-dform-vector,
32415 -mpower9-dform-scalar) and -mno-direct-move.
32417 2017-03-27 Richard Biener <rguenther@suse.de>
32419 PR tree-optimization/80181
32420 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32422 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32424 * config/arc/predicates.md (move_double_src_operand): Replace the
32425 call to move_double_src_operand with a call to address_operand.
32427 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32429 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32430 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32431 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32433 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32435 * config/arc/predicates.md (long_immediate_loadstore_operand):
32436 Consider scaled addresses cases.
32438 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32440 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32441 restored when in interrupt.
32442 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32443 doesn't have delay slot.
32445 2017-03-27 Richard Biener <rguenther@suse.de>
32448 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32449 inlined thunk clones.
32451 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32454 * asan.c (instrument_derefs): Copy over last operand from
32455 original COMPONENT_REF to the new COMPONENT_REF with
32456 DECL_BIT_FIELD_REPRESENTATIVE.
32457 * ubsan.c (instrument_object_size): Likewise.
32459 2017-03-27 Richard Biener <rguenther@suse.de>
32461 PR tree-optimization/80170
32462 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32463 sure DR/SCEV didnt fold in constants we do not see when looking
32464 at the reference base alignment.
32466 2017-03-27 Richard Biener <rguenther@suse.de>
32468 PR middle-end/80171
32469 * gimple-fold.c (fold_ctor_reference): Properly guard against
32470 NULL return value from canonicalize_constructor_val.
32472 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32475 * config/i386/i386.c (ix86_expand_builtin)
32476 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32477 flags reg setting and flags reg using instructions.
32478 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32479 clobbering instructions to zero extend op2.
32481 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32483 * doc/install.texi (Configuration) <--with-aix-soname>:
32484 Update link to AIX ld.
32486 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32488 PR rtl-optimization/80160
32489 PR rtl-optimization/80159
32490 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32491 reg_alternate_class into account.
32493 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32496 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32497 to consider in curr_insn_transform.
32499 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32501 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32502 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32503 and emit_mode_inner.
32505 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32507 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32508 argument to the overloaded builtin variants. Use the new flag to
32509 deprecate certain builtin variants.
32510 * config/s390/s390-builtin-types.def: Add new builtin types.
32511 * config/s390/s390-builtins.h: Support new flags field for
32512 overloaded builtins.
32513 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32514 (s390_macro_to_expand): Enable vector float data type.
32515 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32516 builtins by incrementing the __VEC__ version number.
32517 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32519 (s390_resolve_overloaded_builtin): Emit error messages depending
32520 on the builtin flags.
32521 * config/s390/s390.c (s390_expand_builtin): Support additional
32522 flags argument. Change error message to match the messages
32523 emitted in s390-c.c.
32524 * config/s390/s390.md: New UNSPEC_* constants.
32525 (op_type): Add new instruction types.
32526 * config/s390/vecintrin.h: Add new builtins and test data class
32528 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32529 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32530 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32531 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32532 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32533 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32535 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32536 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32537 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32538 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32540 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32541 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32542 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32543 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32545 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32546 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32547 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32548 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32549 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32550 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32551 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32553 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32554 ("vec_scatter_element<V_HW_4:mode>_DI")
32555 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32556 ("vec_fpint<mode>", "vflls")
32557 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32558 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32559 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32560 ("*vec_cmphe<mode>_cc"): ... these.
32562 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32563 mode constant instead of magic value.
32565 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32567 * config/s390/s390.c (s390_expand_vec_compare): Support other
32568 vector floating point modes than just V2DF.
32569 (s390_expand_vcond): Likewise.
32570 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32571 (s390_cannot_change_mode_class): Prevent mode changes between TF
32572 and V1TF in vector registers.
32573 * config/s390/s390.md (DF, SF): New mode attributes.
32574 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32575 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32576 SFmode support for VRs.
32577 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32579 (VFT, VF_HW): New mode iterators.
32580 (vw, sdx): New mode attributes.
32581 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32582 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32583 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32584 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32585 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32586 also the new vector floating point modes. Renaming to ...
32588 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32589 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32590 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32591 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32592 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32593 ("vec_unordered<mode>"): ... these.
32595 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32596 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32597 ("*vec_extendv2df"): New insn definitions.
32599 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32601 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32602 ("mulditi3_2", "*muldi3_sign"): New patterns.
32603 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32604 rename the pattern definition.
32606 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32608 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32610 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32612 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32614 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32615 instruction if possible.
32616 * config/s390/vector.md (vec_halfnumelts): New mode
32618 ("*vec_vllezlf<mode>"): New pattern.
32620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32622 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32623 ("popcountv4si2", "popcountv2di2"): Rename to ...
32624 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32625 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32627 ("popcount<mode>2_vxe"): New pattern.
32629 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32631 * common/config/s390/s390-common.c (processor_flags_table): Add
32633 * config.gcc: Add arch12.
32634 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32635 Default to arch12 for unknown CPU model numbers.
32636 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32637 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32638 PROCESSOR_max sanity check.
32639 * config/s390/s390-opts.h (enum processor_type): Add
32641 * config/s390/s390.c (processor_table): Add arch12.
32642 (s390_expand_builtin): Add check for B_VXE flag.
32643 (s390_issue_rate): Add PROCESSOR_ARCH12.
32644 (s390_get_sched_attrmask): Likewise.
32645 (s390_get_unit_mask): Likewise.
32646 (s390_sched_score): Enable z13 scheduling for arch12.
32647 (s390_sched_reorder): Likewise.
32648 (s390_sched_variable_issue): Likewise.
32649 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32651 (s390_tune_attr): Use z13 scheduling also for arch12.
32652 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32653 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32654 (TARGET_VXE_P): New macros.
32655 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32656 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32657 * config/s390/s390.opt: Add arch12 as processor_type.
32659 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32661 * config/s390/s390.md
32662 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32663 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32664 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32666 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32667 Rename expanders to ...
32669 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32670 ("fixuns_truncdddi2_emu"): ... these.
32672 ("fixuns_trunc<mode>si2_emu"): New expander.
32674 ("*fixuns_truncdfdi2_z13"): Rename to ...
32675 ("*fixuns_truncdfdi2_vx"): ... this.
32677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32679 * config/s390/2964.md: Remove the single element vector compare
32680 instructions which are no longer used.
32681 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32683 (s390_canonicalize_comparison): Remove handling of DFmode
32685 (s390_expand_vec_compare_scalar): Remove function.
32686 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32687 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32689 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32691 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32693 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32695 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32696 will anyway by matched by mov<mode>_64dfp.
32698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32700 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32701 vlef/vstef. Add missing operand to vleif.
32703 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32705 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32706 pair for all vector types with 64 bit elements.
32707 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32708 * config/s390/vector.md (V_HW_64): ... here.
32709 (V_128_NOSINGLE): New mode iterator.
32710 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32711 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32712 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32713 ("*vec_load_pairv2di"): Change to ...
32714 ("*vec_load_pair<mode>"): ... this one.
32716 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32718 * config/s390/constraints.md: Add comments.
32719 (jKK): Reject element sizes > 8 bytes.
32720 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32722 * config/s390/s390.md: Add the s_operand checks formerly in
32723 s390_split_ok_p to various splitters where they are still
32725 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32726 for 128 bit vectors. Plus two splitters.
32728 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32730 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32733 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32736 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32737 error if the boundary argument is not constant.
32739 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32741 PR rtl-optimization/80112
32742 * loop-doloop.c (doloop_condition_get): Don't check condition
32743 if cmp isn't SET with IF_THEN_ELSE src.
32745 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32747 PR tree-optimization/80158
32748 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32749 replacing a candidate statement, also replace it for the
32750 candidate's alternate interpretation.
32751 (replace_rhs_if_not_dup): Likewise.
32752 (replace_one_candidate): Likewise.
32754 2017-03-24 Richard Biener <rguenther@suse.de>
32756 PR tree-optimization/80167
32757 * graphite-isl-ast-to-gimple.c
32758 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32760 (translate_isl_ast_to_gimple::get_rename): Likewise.
32762 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32764 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32765 handling of certain combinations of target options, including the
32766 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32767 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32769 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32772 * config/arm/arm.md (*load_multiple): Add reload_completed to
32773 matching condition.
32775 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32776 Richard Biener <rguenth@suse.de>
32778 PR tree-optimization/79908
32779 PR tree-optimization/80136
32780 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32781 been cast away, gimplify_and_add suffices.
32783 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32785 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32787 2017-03-23 Richard Biener <rguenther@suse.de>
32789 PR tree-optimization/80032
32790 * gimplify.c (gimple_push_cleanup): Forced unconditional
32791 cleanups still have to go to the conditional_cleanups
32794 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32796 PR tree-optimization/80072
32797 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32799 (next_operand_entry_id): Change type to unsigned int.
32800 (sort_by_operand_rank): Make sure to return the right return value
32801 even if unsigned fields are bigger than INT_MAX.
32802 (struct oecount): Change cnt and id type to unsigned int.
32803 (oecount_hasher::equal): Formatting fix.
32804 (oecount_cmp): Make sure to return the right return value
32805 even if unsigned fields are bigger than INT_MAX.
32806 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32809 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32810 TREE_READONLY on result if writing it more than once.
32813 * doc/invoke.texi (-fsanitize=thread): Document that with
32814 -fnon-call-exceptions atomics are not able to throw
32818 * tsan.c: Include tree-eh.h.
32819 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32820 maybe_clean_or_replace_eh_stmt where needed.
32821 (instrument_memory_accesses): Add cfg_changed argument.
32822 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32823 if it returned true.
32824 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32826 PR rtl-optimization/63191
32827 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32828 wrapper function, moved the whole old content into ...
32829 (ix86_delegitimize_address_1): ... this. New inline function.
32830 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32831 true as last argument instead of ix86_delegitimize_address.
32833 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32835 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32836 cortexa57_branch_cost.
32838 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32840 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32842 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32845 * doc/md.texi (Constraints): Document wA constraint.
32846 * config/rs6000/constraints.md (wA): New.
32847 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32848 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32849 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32850 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32852 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32855 * gimplify.c (is_oacc_declared): New function.
32856 (oacc_default_clause): Use it to set default flags for acc declared
32857 variables inside parallel regions.
32858 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32859 declared variables.
32860 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32861 declare attribute to any decl as necessary.
32863 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32866 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32867 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32868 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32869 (arm_arch_lpae): This.
32870 * config/arm/arm.c (arm_arch7ve): Rename into ...
32871 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32872 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32875 2017-03-22 Martin Liska <mliska@suse.cz>
32878 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32879 error message instead of an ICE.
32881 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32883 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32885 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32887 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32889 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32892 2017-03-21 Martin Sebor <msebor@redhat.com>
32894 * doc/extend.texi: Use "cannot" instead of "can't."
32895 * doc/hostconfig.texi: Same.
32896 * doc/install.texi: Same.
32897 * doc/invoke.texi: Same.
32898 * doc/loop.texi: Same.
32899 * doc/md.texi: Same.
32900 * doc/objc.texi: Same.
32901 * doc/rtl.texi: Same.
32902 * doc/tm.texi: Same.
32903 * doc/tm.texi.in: Same.
32904 * doc/trouble.texi: Same.
32906 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32909 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32910 (collect_checksum_attributes): Set it.
32911 (die_checksum_ordered): Use it.
32913 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32915 PR tree-optimization/79908
32916 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32917 change: For a VA_ARG whose LHS has been cast away, use
32918 force_gimple_operand to construct the side effects.
32920 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32922 PR translation/80001
32923 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32924 more amenable to translation.
32925 (oacc_loop_auto_partitions): Likewise.
32927 2017-03-21 Marek Polacek <polacek@redhat.com>
32928 Martin Sebor <msebor@redhat.com>
32930 PR tree-optimization/80109
32931 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32932 on INTEGRAL_TYPE_P.
32934 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32935 Segher Boessenkool <segher@kernel.crashing.org>
32938 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32939 check reg_used_between_p between insn and one of succ or succ2
32940 depending on if succ is artificial insn not inserted into insn
32943 2017-03-21 Martin Liska <mliska@suse.cz>
32945 PR gcov-profile/80081
32946 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32947 * doc/gcc.texi: Include gcov-dump stuff.
32948 * doc/gcov-dump.texi: New file.
32950 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32952 PR rtl-optimization/79150
32953 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32954 conditional jump, if the jump is the last insn of the loop.
32956 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32957 Richard Biener <rguenth@suse.de>
32959 PR tree-optimization/79908
32960 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32961 been cast away, use force_gimple_operand to construct the side
32964 2017-03-21 Martin Liska <mliska@suse.cz>
32966 PR libfortran/79956
32967 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32970 2017-03-21 Brad Spengler <spender@grsecurity.net>
32973 * plugin.c (htab_hash_plugin): New function.
32974 (add_new_plugin): Use it and adjust.
32975 (parse_plugin_arg_opt): Adjust.
32976 (init_one_plugin): Likewise.
32978 2017-03-21 Richard Biener <rguenther@suse.de>
32980 PR tree-optimization/80032
32981 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32982 if set force the cleanup to happen unconditionally.
32983 (gimplify_target_expr): Push inserted clobbers with force_uncond
32984 to avoid them being removed by control-dependent DCE.
32986 2017-03-21 Richard Biener <rguenther@suse.de>
32988 PR tree-optimization/80122
32989 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32990 va_arg_pack_len when the inlined call stmt requires pack
32992 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32994 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32997 * tsan.c (instrument_builtin_call): If the memory model argument
32998 is not a constant, assume it is valid.
33001 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33004 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33006 PR rtl-optimization/79910
33007 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33008 if its dest is used by an insn before I2 (other than the combined
33009 insns themselves, which are properly handled already).
33011 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33014 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33016 * combine.c (record_used_regs): New static function.
33017 (try_combine): Handle situations where there is an additional
33018 instruction between I2 and I3 which needs to have a LOG_LINK
33022 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33024 * combine.c (try_combine): Delete redundant i1 test. Call
33025 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33027 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33030 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33031 alternatives 13/14.
33033 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33035 PR tree-optimization/80054
33036 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33037 the optimization if a PHI or any of its arguments is not dominated
33038 by the candidate's basis. Use gphi* rather than gimple* as
33040 (replace_profitable_candidates): Clean up a gimple* variable that
33041 should be a gphi* variable.
33043 2017-03-20 Martin Sebor <msebor@redhat.com>
33046 * doc/extend.texi (attribute constructor): Document present limitation.
33048 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33051 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33052 __POWER9_VECTOR__ #ifdef control, change template definition to
33053 use Power9-specific built-in function.
33054 (vec_any_eq): Likewise.
33055 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33056 to control outcomes from this test.
33057 (vector_ae_<mode>p): For VEC_F modes, likewise.
33059 2017-03-20 Ian Lance Taylor <iant@google.com>
33061 * config/i386/i386.c (ix86_function_regparm): Save an extra
33062 register for -fsplit-stack with DECL_STATIC_CHAIN.
33064 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33067 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33068 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33070 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33072 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33074 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33077 2017-03-20 Marek Polacek <polacek@redhat.com>
33080 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33082 2017-03-20 Richard Biener <rguenther@suse.de>
33084 PR tree-optimization/80113
33085 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33086 allocate extra SSA name for PHI def.
33087 (add_close_phis_to_outer_loops): Likewise.
33088 (add_close_phis_to_merge_points): Likewise.
33089 (copy_loop_close_phi_args): Likewise.
33090 (copy_cond_phi_nodes): Likewise.
33092 2017-03-20 Martin Liska <mliska@suse.cz>
33094 PR middle-end/79753
33095 * tree-chkp.c (chkp_build_returned_bound): Do not build
33096 returned bounds for a LHS that's not a BOUNDED_P type.
33098 2017-03-20 Martin Liska <mliska@suse.cz>
33102 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33103 COMPLEX_CST and VECTOR_CST.
33105 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33108 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33109 target operand. A new splitter adds the clobber statement in case
33110 the target operand is dead anyway.
33112 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33114 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33115 to age-old versions of binutils and glibc.
33117 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33119 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33121 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33123 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33125 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33127 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33128 requirement for binutils 2.13.
33130 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33132 * combine.c (try_combine): Delete redundant i1 test. Call
33133 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33135 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33137 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33138 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33140 <riscv64-*-elf>: Re-arrange section
33141 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33142 <riscv32-*-linux>: Likewise.
33143 <riscv64-*-elf>: Likewise
33144 <riscv64-*-linux>: Likewise.
33146 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33149 * aarch64.opt(verbose-cost-dump): Fix typo.
33151 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33154 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33155 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33157 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33159 * reload.c (find_reloads): When reloading a nonoffsettable address,
33160 use RELOAD_OTHER for it and its address reloads.
33162 PR rtl-optimization/79910
33163 * combine.c (record_used_regs): New static function.
33164 (try_combine): Handle situations where there is an additional
33165 instruction between I2 and I3 which needs to have a LOG_LINK
33168 2017-03-17 Jeff Law <law@redhat.com>
33170 PR tree-optimization/71437
33171 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33172 conditional in the hash table first.
33173 (vrp_dom_walker::before_dom_children): Extract condition from
33174 ASSERT_EXPR. Record condition, its inverion and any implied
33175 conditions as well.
33177 2017-03-17 Marek Polacek <polacek@redhat.com>
33178 Markus Trippelsdorf <markus@trippelsdorf.de>
33180 PR tree-optimization/80079
33181 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33184 2017-03-17 Richard Biener <rguenther@suse.de>
33186 PR middle-end/80075
33187 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33188 Properly verify the LHS before the RHS possibly claims to be
33190 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33193 2017-03-17 Martin Jambor <mjambor@suse.cz>
33195 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33196 (List of -O2 options): Likewise.
33197 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33200 2017-03-17 Tom de Vries <tom@codesourcery.com>
33202 * gcov-dump.c (print_usage): Print bug_report_url.
33204 2017-03-17 Richard Biener <rguenther@suse.de>
33206 PR middle-end/80050
33207 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33208 (parser::peek): Likewise.
33210 2017-03-17 Richard Biener <rguenther@suse.de>
33212 PR tree-optimization/80048
33213 * sese.c (free_sese_info): Properly release rename_map and
33214 copied_bb_map elements.
33216 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33218 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33219 Add linked-list forward and backlinks. Insert on
33220 construction, remove on destruction.
33221 (class pass_store_merging): Add m_stores_head field.
33222 (pass_store_merging::terminate_and_process_all_chains):
33223 Iterate over m_stores_head list.
33224 (pass_store_merging::terminate_all_aliasing_chains):
33226 (pass_store_merging::execute): Check for debug stmts first.
33227 Push new chains onto the m_stores_head stack.
33229 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33232 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33233 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33234 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33236 2017-03-16 Jeff Law <law@redhat.com>
33238 PR tree-optimization/71437
33239 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33240 member function. Implementation moved into after_dom_children
33241 member function and into the threader's thread_outgoing_edges
33243 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33244 some code into new thread_outgoing_edges.
33245 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33246 definition. Simplify marker handling (do it here). Assume we always
33247 have the available expression and the const/copies tables.
33248 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33250 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33251 * tree-vrp.c (equiv_stack): No longer file scoped.
33252 (vrp_dom_walker): New class.
33253 (vrp_dom_walker::before_dom_children): New member function.
33254 (vrp_dom_walker::after_dom_children): Likewise.
33255 (identify_jump_threads): Setup domwalker. Use it rather than
33256 walking edges in a random order by hand. Simplify setup/finalization.
33257 (finalize_jump_threads): Remove.
33258 (vrp_finalize): Do not call identify_jump_threads here.
33259 (execute_vrp): Do it here instead and call thread_through_all_blocks
33262 PR tree-optimization/71437
33263 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33265 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33267 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33268 (dom_opt_dom_walker::thread_across_edge): Remove
33269 handle_dominating_asserts argument. All callers changed.
33270 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33271 changes. Remove calls to lhs_of_dominating_assert. Other
33272 uses of handle_dominating_asserts turn into unconditional code
33273 (simplify_control_stmt_condition_1): Likewise.
33274 (simplify_control_stmt_condition): Likewise.
33275 (thread_through_normal_block, thread_across_edge): Likewise.
33276 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33277 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33278 object if it is not an SSA_NAME.
33279 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33280 before calling into the VRP specific simplifiers.
33281 (identify_jump_threads): Remove handle_dominating_asserts
33284 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33287 * tree-diagnostic.c (default_tree_printer): No longer static.
33288 * tree-diagnostic.h (default_tree_printer): New prototype.
33290 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33292 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33293 Change ins into fmov.
33295 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33297 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33298 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33299 Use h_con constraint for operand 1.
33300 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33301 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33303 2017-03-15 Jeff Law <law@redhat.com>
33305 PR tree-optimization/71437
33306 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33307 (record_temporary_equivalences): Use it.
33309 PR tree-optimization/71437
33310 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33311 tree-ssa-scopedtables.
33312 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33313 (record_conditions, record_cond, vuse_eq): Likewise.
33314 (record_edge_info): Adjust to API tweak of record_conditions.
33315 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33316 (record_temporary_equivalences, optimize_stmt): Likewise.
33317 (eliminate_redundant_computations): Likewise.
33318 (record_equivalences_from_stmt): Likewise.
33319 * tree-ssa-scopedtables.c: Include options.h and params.h.
33320 (vuse_eq): New function, moved from tree-ssa-dom.c
33321 (build_and_record_new_cond): Likewise.
33322 (record_conditions): Likewise. Accept vector of conditions rather
33323 than edge_equivalence structure for first argument.
33324 for the first argument.
33325 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33326 from tree-ssa-dom.c.
33327 (avail_exprs_stack::record_cond): Likewise.
33328 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33329 from tree-ssa-dom.c.
33330 (avail_exprs_stack): Add new member functions lookup_avail_expr
33332 (record_conditions): Declare.
33334 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33337 * lra-constraints.c (process_alt_operands): Increase reject for
33338 reloading an input/output operand.
33340 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33343 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33344 insns to convert from signed/unsigned char/short to IEEE 128-bit
33346 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33348 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33351 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33352 subreg of inner mode for values already in registers.
33354 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33356 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33357 iteration reg is used after the loop.
33359 2017-03-14 Martin Sebor <msebor@redhat.com>
33361 PR tree-optimization/79800
33362 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33363 precision in negative-positive range.
33364 (format_floating): Call non-const overload with adjusted precision.
33366 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33369 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33372 2017-03-14 Martin Sebor <msebor@redhat.com>
33374 PR middle-end/80020
33375 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33376 * builtins.def (aligned_alloc): Use it.
33379 * Makefile.in (GTFILES): Add calls.c.
33380 * calls.c: Include "gt-calls.h".
33382 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33384 PR rtl-optimization/79728
33385 * regs.h (struct target_regs): New field
33386 x_contains_allocatable_regs_of_mode.
33387 (contains_allocatable_regs_of_mode): New macro.
33388 * reginfo.c (init_reg_sets_1): Initialize it, and change
33389 contains_reg_of_mode so it includes global regs as well.
33390 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33391 rather than contains_regs_of_mode.
33393 2017-03-14 Martin Liska <mliska@suse.cz>
33395 * doc/invoke.texi: Document options that can't be combined with
33396 -fcheck-pointer-bounds.
33398 2017-03-14 Martin Liska <mliska@suse.cz>
33400 PR middle-end/79831
33401 * doc/invoke.texi (-Wchkp): Document the option.
33403 2017-03-14 Martin Liska <mliska@suse.cz>
33405 * Makefile.in: Install gcov-dump.
33407 2017-03-14 Martin Liska <mliska@suse.cz>
33409 * multiple_target.c (expand_target_clones): Bail out for
33410 an invalid attribute.
33412 2017-03-14 Richard Biener <rguenther@suse.de>
33414 * alias.c (struct alias_set_entry): Pack properly.
33415 * cfgloop.h (struct loop): Likewise.
33416 * cse.c (struct set): Likewise.
33417 * ipa-utils.c (struct searchc_env): Likewise.
33418 * loop-invariant.c (struct invariant): Likewise.
33419 * lra-remat.c (struct cand): Likewise.
33420 * recog.c (struct change_t): Likewise.
33421 * rtl.h (struct address_info): Likewise.
33422 * symbol-summary.h (function_summary): Likewise.
33423 * tree-loop-distribution.c (struct partition): Likewise.
33424 * tree-object-size.c (struct object_size_info): Likewise.
33425 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33426 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33427 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33428 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33429 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33430 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33431 (struct _stmt_vec_info): Likewise.
33433 2017-03-14 Martin Liska <mliska@suse.cz>
33436 * multiple_target.c (create_dispatcher_calls): Check that
33437 a target can create a function dispatcher.
33439 2017-03-14 Martin Liska <mliska@suse.cz>
33442 * multiple_target.c (expand_target_clones): Drop local.local
33443 flag for default implementation.
33445 2017-03-14 Richard Biener <rguenther@suse.de>
33447 PR tree-optimization/80030
33448 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33450 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33452 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33453 gcc_fallthrough() instead of __attribute__((fallthrough));
33455 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33457 * doc/gcc.texi: Remove "up" link to (DIR).
33458 * doc/gccint.texi: Ditto.
33460 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33462 * doc/install.texi (Specific) <avr>: Remove reference to
33465 2017-03-13 Jeff Law <law@redhat.com>
33467 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33468 attribute rather than comments.
33470 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33471 match_scratch operand is highest.
33473 2017-03-13 Martin Liska <mliska@suse.cz>
33475 PR middle-end/78339
33476 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33477 is a CHKP clone, use original declaration.
33479 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33481 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33482 (arc_conditional_register_usage): Use a different allocation order
33483 when optimizing for size.
33484 * common/config/arc/arc-common.c (arc_option_optimization_table):
33485 Section anchors default on when optimizing for size.
33487 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33489 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33491 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33493 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33494 * config/arc/arc.md (cpu_facility): Add cd variant.
33495 (*movqi_insn): Add code density variant.
33496 (*movhi_insn): Likewise.
33497 (*movqi_insn): Likewise.
33498 (*addsi3_mixed): Likewise.
33499 (subsi3_insn): Likewise.
33501 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33503 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33505 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33507 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33508 expressions with MINUS and UNARY ops.
33510 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33513 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33515 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33516 between vec_select and vector argument.
33517 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33518 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33519 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33520 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33521 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33522 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33524 2017-03-13 Richard Biener <rguenther@suse.de>
33527 * params.def (vect-max-peeling-for-alignment): Fix typo.
33529 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33531 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33532 issue that only occurred with binutils below 2.18.
33534 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33536 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33537 refer to binutils 2.11/2.12 minimum.
33539 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33541 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33542 ftp.kernel.org and simplify binutils requirement.
33544 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33546 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33548 (Optimize Options): Ditto. Also remove redundancy.
33550 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33552 PR translation/79848
33553 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33555 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33556 to G_ to avoid double translation.
33558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33560 PR translation/79923
33561 * auto-profile.c (get_combined_location): Convert leading
33562 character of diagnostics to lower case and remove trailing period.
33563 (read_profile): Likewise for various diagnostics.
33564 * config/arm/arm.c (arm_option_override): Remove trailing period
33565 from various diagnostics.
33566 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33567 (msp430_expand_delay_cycles): Likewise.
33569 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33572 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33573 full command-line argument, rather than just "str".
33574 (aarch64_validate_march): Likewise.
33575 (aarch64_validate_mtune): Likewise.
33577 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33579 PR rtl-optimization/78911
33580 * lra-assigns.c (must_not_spill_p): New function.
33581 (spill_for): Use it.
33583 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33585 PR tree-optimization/79981
33586 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33587 ATOMIC_COMPARE_EXCHANGE ifn result.
33588 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33589 IFN_ATOMIC_COMPARE_EXCHANGE.
33591 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33594 * opts.c (parse_sanitizer_options): Add missing question mark to
33595 "did you mean" message.
33597 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33599 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33601 (VMULEUH_UNS): Likewise.
33602 (VMULOUB_UNS): Likewise.
33603 (VMULOUH_UNS): Likewise.
33604 * config/rs6000/rs6000.c (builtin_function_type): Remove
33605 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33607 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33610 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33611 x with result of extra_parsing_for_operand_code_0.
33612 (function_reader::extra_parsing_for_operand_code_0): Convert
33613 return type from void to rtx, returning x. When reading
33614 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33615 larger size containing struct block_symbol.
33617 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33620 -mfloat128-hardware without -m64.
33622 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33625 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33626 entries to the case statement that marks unsigned arguments to
33627 overloaded functions.
33629 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33632 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33634 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33637 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33638 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33640 2017-03-10 Martin Liska <mliska@suse.cz>
33644 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33645 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33648 2017-03-10 Olivier Hainque <hainque@adacore.com>
33650 * tree-switch-conversion (array_value_type): Start by resetting
33651 candidate type to it's main variant.
33653 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33655 PR rtl-optimization/79909
33656 * combine.c (try_combine): Use simplify_replace_rtx on individual
33657 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33658 of the whole CALL_INSN_FUNCTION_USAGE.
33660 PR tree-optimization/79972
33661 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33662 get_range_info on SSA_NAMEs. Formatting fixes.
33664 2017-03-10 Richard Biener <rguenther@suse.de>
33665 Jakub Jelinek <jakub@redhat.com>
33667 PR tree-optimization/77975
33668 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33669 edge to be constant.
33670 (get_val_for): For constant x return it. Formatting fix.
33671 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33672 would use the same bases as the current one.
33674 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33676 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33677 instead of vec_select for V1TImode.
33678 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33680 (VSX_LE_128): Add V1TI to this mode iterator.
33681 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33682 (*vsx_le_perm_store_<mode>): Likewise.
33683 (pre-reload splitter for VSX stores): Likewise.
33684 (post-reload splitter for VSX stores): Likewise.
33685 (*vsx_xxpermdi2_le_<mode>): Likewise.
33686 (*vsx_lxvd2x2_le_<mode>): Likewise.
33687 (*vsx_stxvd2x2_le_<mode>): Likewise.
33689 2017-03-09 Michael Eager <eager@eagercon.com>
33691 Correct failures with --enable-checking=yes,rtl.
33693 * config/microblaze/microblaze.c (microblaze_expand_shift):
33694 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33695 test for const0_rtx.
33696 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33697 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33699 2017-03-09 Richard Biener <rguenther@suse.de>
33701 PR tree-optimization/79977
33702 * graphite-scop-detection.c (scop_detection::merge_sese):
33703 Handle the case of extra exits to blocks dominating the entry.
33705 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33707 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33710 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33712 PR rtl-optimization/79949
33713 * lra-constraints.c (process_alt_operands): Check memory when
33714 trying to predict a cycle. Print about the overall increase.
33716 2017-03-09 Richard Biener <rguenther@suse.de>
33718 PR middle-end/79971
33719 * gimple-expr.c (useless_type_conversion_p): Preserve
33720 TYPE_SATURATING for fixed-point types.
33722 2017-03-09 Richard Biener <rguenther@suse.de>
33725 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33726 alignment of BLKmode params.
33728 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33731 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33732 (VALL_NO_V2Q): Likewise.
33734 * config/aarch64/aarch64-simd.md
33735 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33737 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33738 VALL_NO_V2Q mode iterator.
33739 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33741 2017-03-09 Martin Liska <mliska@suse.cz>
33743 PR tree-optimization/79631
33744 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33745 tree_int_cst_sign_bit just for INTEGER constants.
33747 2017-03-09 Martin Liska <mliska@suse.cz>
33751 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33754 2017-03-09 Marek Polacek <polacek@redhat.com>
33757 * tree.c (inchash::add_expr): Handle TREE_VEC.
33759 2017-03-09 Martin Liska <mliska@suse.cz>
33762 (chkp_narrow_size_and_offset): New function.
33763 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33764 (void chkp_parse_bit_field_ref): New function.
33765 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33766 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33768 2017-03-09 Martin Liska <mliska@suse.cz>
33771 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33772 (chkp_find_bounds_1): Remove gcc_unreachable.
33774 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33777 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33778 BUILT_IN_SYNC*, determine the access type from the size suffix and
33779 always build a MEM_REF with that type. Handle forgotten
33780 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33783 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33784 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33785 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33786 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33787 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33788 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33789 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33790 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33791 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33792 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33793 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33794 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33795 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33796 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33797 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33798 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33799 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33800 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33801 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33802 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33803 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33804 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33805 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33806 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33807 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33808 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33809 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33810 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33811 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33812 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33813 definitions outside of __OPTIMIZE__ guarded section.
33816 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33817 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33818 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33819 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33822 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33824 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33825 ("vfenez<mode>"): Add missing constraints.
33827 2017-03-08 Martin Sebor <msebor@redhat.com>
33830 * config/nds32/nds32.c (nds32_option_override):
33831 Fix misspelled diagnostic.
33833 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33836 * gimplify.c (gimplify_omp_for): Replace index var in outer
33837 taskloop statement with an artificial variable and add
33838 OMP_CLAUSE_PRIVATE clause for it.
33840 2017-03-08 Richard Biener <rguenther@suse.de>
33842 PR tree-optimization/79955
33843 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33844 for accesses that are completely outside of the variable.
33846 2017-03-08 Andrew Haley <aph@redhat.com>
33848 PR tree-optimization/79943
33849 * tree-ssa-loop-split.c (compute_new_first_bound): When
33850 calculating the new upper bound, (END-BEG) should be added, not
33853 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33855 * config/avr/avr.md (setmemhi): Make sure match_dup
33856 operand number comes before match_scratch.
33858 2017-03-08 Richard Biener <rguenther@suse.de>
33860 PR tree-optimization/79920
33861 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33862 with ncopies == 1 to ...
33863 (vect_transform_slp_perm_load): ... here. Properly compute
33864 all element loads by iterating VF times over the group. Do
33865 not handle ncopies (computed in a broken way) in
33866 vect_create_mask_and_perm.
33868 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33871 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33872 is a uniform vector, use uniform_vector_p return value instead of
33873 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33875 2017-03-07 Marek Polacek <polacek@redhat.com>
33877 PR middle-end/79809
33878 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33879 (alloca_call_type): Likewise.
33881 2017-03-07 Martin Liska <mliska@suse.cz>
33883 * gcov.c (process_args): Put comment to correct location.
33885 2017-03-07 Martin Liska <mliska@suse.cz>
33887 PR middle-end/68270
33888 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33889 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33890 (chkp_narrow_bounds_for_field): Likewise.
33891 (chkp_parse_array_and_component_ref): Pass one more argument to
33894 2017-03-07 Richard Biener <rguenther@suse.de>
33896 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33899 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33901 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33902 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33904 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33907 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33908 to end of description.
33909 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33911 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33913 PR rtl-optimization/79901
33914 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33916 (*avx512f_<code><mode>3<mask_name>): ... this.
33917 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33918 iterator instead of VI8_AVX2_AVX512BW.
33920 PR rtl-optimization/79901
33921 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33922 min/max expander, expand it using expand_vec_cond_expr.
33925 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33928 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33931 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33932 to void * for PCH reasons.
33933 * dwarf2out.c (output_loc_operands, output_die): Cast
33934 v.val_vec.array to unsigned char *.
33936 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33939 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33942 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33944 PR rtl-optimization/79571
33945 * lra-constraints.c (process_alt_operands): Calculate static
33946 reject and subtract it from overall when only addresses will be
33949 2017-03-06 Julia Koval <julia.koval@intel.com>
33952 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33953 incoming stack boundary to 128 for 64-bit targets.
33955 2017-03-06 Richard Biener <rguenther@suse.de>
33957 PR tree-optimization/79894
33958 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33959 to NULL after folding it.
33961 2017-03-06 Richard Biener <rguenther@suse.de>
33963 PR tree-optimization/79824
33964 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33965 check disabling peeling for gaps.
33967 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33969 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33970 attributes): Document gettimeofday.
33972 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33974 * config/s390/s390.c (s390_option_override_internal): Set
33975 PARAM_MIN_VECT_LOOP_BOUND
33977 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33979 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33980 * config/s390/s390.md: Likewise.
33982 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33985 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33986 (<avx2_avx512>_perm<mode>): Rename to ...
33987 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33989 (<avx512>_perm<mode>_mask): Rename to ...
33990 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33992 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33993 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33994 instead of VI8F_256_512.
33995 (avx512f_perm<mode>): New define_expand.
33996 (avx512f_perm<mode>_mask): Likewise.
33997 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33998 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34000 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34002 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34003 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34005 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34007 2017-03-06 Martin Liska <mliska@suse.cz>
34010 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34011 when having a SSA NAME w/o VAR_DECL assigned to it.
34013 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34015 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34016 msa_dpsub_<su>_d): Fix MODE for vec_select.
34018 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34020 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34022 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34024 2017-03-06 Richard Biener <rguenther@suse.de>
34026 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34027 * plugin.c (register_plugin_info): Likewise.
34028 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34030 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34032 * config/i386/sse.md (sse_storehps, sse_storelps,
34033 avx_<castmode><avxsizesuffix>_<castmode>,
34034 avx512f_<castmode><avxsizesuffix>_<castmode>,
34035 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34036 in condition that at least one operand is not a MEM.
34038 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34040 PR middle-end/79805
34041 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34042 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34044 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34045 gimple_call_nothrow_p flag based on whether original builtin can throw.
34046 If it can, emit following stmts on the fallthrough edge.
34047 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34048 don't create new bb if inserting just debug stmts on the edge, try to
34049 insert them on the fallthru bb or just reset debug stmts.
34051 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34054 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34055 restore recog_data (including the operand rtxes inside it) around
34056 the call to get_insn_template.
34058 2017-03-03 Martin Sebor <msebor@redhat.com>
34060 PR tree-optimization/79699
34061 * context.c (context::~context): Free MPFR caches to avoid
34062 a memory leak on program exit.
34064 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34066 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34067 Use wide_int::ulow () instead of .elt (0).
34069 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34071 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34072 (*pushxf): Limit oF constraint to 32bit targets and add oC
34073 constraint for 64bit targets.
34074 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34075 (*pushdf): Change rmF constraint to rmC.
34077 2017-03-03 Martin Liska <mliska@suse.cz>
34079 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34080 Remove unused variable.
34082 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34085 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34086 is a memory operand, increase num_memory.
34087 (ix86_expand_args_builtin): Likewise.
34089 2017-03-03 Jan Hubicka <jh@suse.cz>
34092 * ipa-devirt.c (maybe_record_node): Properly handle
34093 __cxa_pure_virtual visibility.
34095 2017-03-03 Martin Liska <mliska@suse.cz>
34097 PR tree-optimization/79803
34098 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34100 (pass_loop_prefetch::execute): Disabled optimization if an
34101 assumption about L1 cache size is not met.
34103 2017-03-03 Martin Liska <mliska@suse.cz>
34105 PR rtl-optimization/79574
34106 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34107 (hash_scan_set): Likewise.
34108 (dump_hash_table): Likewise.
34109 (hoist_code): Likewise.
34111 2017-03-03 Richard Biener <rguenther@suse.de>
34113 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34114 in place of uhigh ().
34115 (fixed_convert_from_real): Likewise.
34117 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34120 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34122 2017-03-03 Richard Biener <rguenther@suse.de>
34124 PR middle-end/79818
34125 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34126 TYPE_OVERFLOW_UNDEFINED check.
34128 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34130 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34132 (vector_ae_<mode>_p): Likewise.
34133 (vector_nez_<mode>_p): Likewise.
34134 (vector_ne_v2di_p): Likewise.
34135 (vector_ae_v2di_p): Likewise.
34136 (vector_ne_<mode>_p): Likewise.
34137 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34139 (vsx_tsqrt<mode>2_fe): Likewise.
34141 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34144 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34146 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34148 PR rtl-optimization/79780
34149 * cprop.c (one_cprop_pass): When second and further conditional trap
34150 in a single basic block is turned into an unconditional trap, turn it
34151 into a deleted note to avoid RTL verification failures.
34153 2017-03-02 Richard Biener <rguenther@suse.de>
34155 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34157 2017-03-02 Richard Biener <rguenther@suse.de>
34159 PR tree-optimization/79345
34161 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34162 param and abort the walk, returning -1 if it is hit.
34163 (walk_aliased_vdefs): Take a limit param and pass it on.
34164 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34165 defaulting to 0 and return a signed int.
34166 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34167 (check_defs): New helper.
34168 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34169 about uninitialized memory.
34170 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34171 bogus uninitialized warning.
34172 (fixed_convert_from_real): Likewise.
34174 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34176 PR tree-optimization/66768
34177 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34178 iv_use if base object can't be determined.
34180 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34182 PR tree-optimization/79345
34183 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34184 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34185 (get_pattern_stats): Initialize it.
34186 * genemit.c (gen_expand): Verify match_scratch numbers come after
34187 match_operand/match_dup numbers.
34188 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34189 match_scratch numbers.
34190 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34192 * config/s390/s390.md (trunctdsd2): Likewise.
34194 2017-03-02 Richard Biener <rguenther@suse.de>
34196 * wide-int.h (wide_int_storage::operator=): Implement in terms
34199 2017-03-02 Richard Biener <rguenther@suse.de>
34201 PR tree-optimization/79777
34202 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34203 the to insert expression to sth existing.
34205 2017-03-01 Martin Sebor <msebor@redhat.com>
34207 PR middle-end/79692
34208 * gimple-ssa-sprintf.c
34209 (directive::known_width_and_precision): New function.
34210 (format_integer): Use it.
34211 (get_mpfr_format_length): Consider the full range of precision
34212 when computing %g output with the # flag. Set the likely byte
34213 count to 3 rather than 1 when precision is indeterminate.
34214 (format_floating): Correct the lower bound of precision.
34216 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34218 * doc/invoke.texi: Document default code model for 64-bit Linux.
34220 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34223 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34224 udiv rather than div since input pattern is unsigned.
34226 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34228 * config/i386/i386.c (print_reg): Warn for values of
34229 unsupported size in integer register.
34231 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34234 * config/rs6000/predicates.md (current_file_function_operand): Do
34235 not allow self calls to be local if the function is replaceable.
34237 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34240 * config/rs6000/altivec.h (vec_ctz and others): Change the
34241 preprocessor macro that controls conditional compilation from
34242 _ARCH_PWR9 to __POWER9_VECTOR__.
34243 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34244 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34245 control (instead of _ARCH_PWR9 control) so that template
34246 definition uses power9-specific function.
34247 (vec_any_eq): Likewise.
34248 (vec_all_ne): Change macro definition to use a power9-specific
34249 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34250 _ARCH_PWR9 control).
34251 (vec_any_eq) Likewise.
34252 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34253 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34254 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34255 support for xvcmpnedp instruction.
34256 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34257 macro expansion so that Power9 implementation of vec_all_ne does
34258 not use the AltiVec predicate framework.
34259 (VCMPNEH_P): Likewise.
34260 (VCMPNEW_P): Likewise.
34261 (VCMPNED_P): Likewise.
34262 (VCMPNEFP_P): Likewise.
34263 (VCMPNEDP_P): Likewise.
34264 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34265 implementation of vec_any_eq to not use AltiVec predicate
34267 (VCMPAEH_P): Likewise.
34268 (VCMPAEW_P): Likewise.
34269 (VCMPAED_P): Likewise.
34270 (VCMPAEFP_P): Likewise.
34271 (VCMPAEDP_P): Likewise.
34272 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34273 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34274 not use the AltiVec predicate framework.
34275 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34276 of vec_any_eq to not use AltiVec predicate framework.
34277 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34278 support for predefined __POWER9_VECTOR__ macro to indicate that
34279 Power9 instruction selection is enabled.
34280 (altivec_overloaded_builtins): Remove extraneous
34281 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34282 function argument types RS6000_BTI_bool_V16QI and
34283 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34284 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34285 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34286 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34287 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34288 Power9 for implementations of vec_cmpne. Change the signature for
34289 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34290 (representing vec_all_ne) to remove the previously described first
34291 argument of type RS6000_BTI_INTSI, as this was an artifact of
34292 reliance on the AltiVec predicate framework, which is no longer
34293 used in the implementation of these functions. Add
34294 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34295 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34296 since, unlike the AltiVec predicate framework implementation, we
34297 do not share function descriptors between vec_alle and vec_anyeq.
34298 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34299 set of modes that receive special treatment even when
34300 TARGET_P9_VECTOR is true. The special treatment emits code that
34301 does not depend on Power9 instructions.
34302 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34303 define_expand to not rely on AltiVec predicate framework.
34304 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34306 (vector_ne_v2di_p): Change this define_expand to not rely on
34307 AltiVec predicate framework.
34308 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34310 (vector_ne_<mode>_p): Change this define_expand to not rely on
34311 AltiVec predicate framework.
34312 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34314 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34315 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34316 define_insn pattern.
34317 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34318 define_insn pattern because the xvcmpne<VSs>. instruction is not
34320 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34321 instruction is not supported.
34323 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34325 * config/nvptx/nvptx.c: Include intl.h.
34327 2017-03-01 Martin Jambor <mjambor@suse.cz>
34330 * ipa-prop.h (ipa_bits): Removed field known.
34331 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34332 to pointers. Adjusted their comments to warn about their sharing.
34333 (ipcp_transformation_summary): Change bits to a vector of pointers.
34334 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34335 (ipa_get_ipa_bits_for_value): Declare.
34336 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34337 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34338 (ipa_bits_hash_table): Likewise.
34339 (ipa_vr_ggc_hash_traits): Likewise.
34340 (ipa_vr_hash_table): Likewise.
34341 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34342 being pointers and vr_known being removed.
34343 (ipa_set_jf_unknown): Likewise.
34344 (ipa_get_ipa_bits_for_value): New function.
34345 (ipa_set_jfunc_bits): Likewise.
34346 (ipa_get_value_range): New overloaded functions.
34347 (ipa_set_jfunc_vr): Likewise.
34348 (ipa_compute_jump_functions_for_edge): Use the above functions to
34349 construct bits and vr parts of jump functions.
34350 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34351 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34353 (ipcp_grow_transformations_if_necessary): Also allocate
34354 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34356 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34357 them. Fix too long lines.
34358 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34359 vr_known being removed.
34360 (ipa_read_jump_function): Use new setter functions to construct bits
34361 and vr parts of jump functions or set them to NULL.
34362 (write_ipcp_transformation_info): Adjust for bits being pointers.
34363 (read_ipcp_transformation_info): Likewise.
34364 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34366 Include gt-ipa-prop.h.
34367 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34369 (ipcp_store_bits_results): Likewise.
34370 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34371 Do not write to existing jump functions but use a temporary instead.
34373 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34376 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34377 attempt to use its first operand as BIT_FIELD_REF base.
34379 2017-03-01 Richard Biener <rguenther@suse.de>
34381 PR middle-end/79721
34382 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34383 interpolating formula in wrapping arithmetic.
34384 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34386 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34388 PR tree-optimization/79734
34389 * tree-vect-generic.c (expand_vector_condition): Optimize
34390 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34391 Handle VEC_COND_EXPR where comparison has different inner width from
34392 type's inner width.
34394 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34396 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34397 markup, and similar issues. Remove @opindex entries for things
34398 that aren't options. Add missing -mmpy-option entries.
34400 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34402 PR tree-optimization/79737
34403 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34404 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34405 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34406 instead of byte_size. Formatting fix.
34407 (shift_bytes_in_array_right): Formatting fix.
34409 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34412 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34413 condition on optimize for the leaf function test.
34415 2017-02-28 Martin Liska <mliska@suse.cz>
34418 * read-rtl-function.c (function_reader::handle_unknown_directive):
34419 Bail out when one uses -flto.
34421 2017-02-28 Martin Liska <mliska@suse.cz>
34423 * common.opt: Replace space with tabular for options of <number>
34425 * config/i386/i386.opt: Show <number> value for
34426 -mlarge-data-threshold.
34427 * opts.c (print_filtered_help): Do not display number in hexadecimal
34430 2017-02-28 Martin Liska <mliska@suse.cz>
34432 * common.opt: Fix --help=option -Q for options which are of
34435 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34437 * config/i386/i386.c (print_reg): Error out for values
34438 of 8-bit size in invalid integer register.
34440 2017-02-28 Martin Sebor <msebor@redhat.com>
34442 PR tree-optimization/79691
34443 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34445 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34448 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34449 gcc_unreachable with output_operand_lossage.
34451 2017-02-28 Richard Biener <rguenther@suse.de>
34453 PR tree-optimization/79740
34454 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34456 (visit_nary_op): Insert the nary into the hashtable if we
34457 pattern-matched sth.
34458 * tree-ssa-pre.c (eliminate_insert): Robustify.
34460 2017-02-28 Richard Biener <rguenther@suse.de>
34462 PR middle-end/79731
34463 * fold-const.c (decode_field_reference): Reject out-of-bound
34466 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34468 * config/i386/i386.c: Include intl.h.
34469 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34470 instead of just cond ? "..." : "...".
34471 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34472 * coverage.c (read_counts_file): Likewise.
34473 * omp-offload.c: Include intl.h.
34474 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34475 of just cond ? "..." : "...".
34476 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34477 of just cond ? "..." : "...".
34479 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34482 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34484 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34485 'tune for' CPU name.
34486 * config/arm/arm-cpu-data.h: Regenerated.
34488 2017-02-28 Richard Biener <rguenther@suse.de>
34490 PR tree-optimization/79732
34491 * tree-inline.c (expand_call_inline): Do not shadow var.
34493 2017-02-28 Richard Biener <rguenther@suse.de>
34495 PR tree-optimization/79723
34496 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34497 address-space properly.
34499 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34501 * doc/optinfo.texi (Optimization groups): Fix option used for
34503 * doc/invoke.texi (-fopt-info): Document "omp".
34504 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34505 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34506 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34507 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34508 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34510 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34512 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34513 OPTGROUP_OMP with "omp".
34515 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34518 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34519 for arithmetic shift of unsigned V2DI.
34521 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34523 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34524 arc/linux.h headers.
34525 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34526 (LINK_SPEC): Likewise.
34527 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34528 (EXTRA_SPECS): Likewise.
34529 (STARTFILE_SPEC): Likewise.
34530 (ENDFILE_SPEC): Likewise.
34531 (LIB_SPEC): Likewise.
34532 (TARGET_SDATA_DEFAULT): Likewise.
34533 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34534 (MULTILIB_DEFAULTS): Likewise.
34535 (DWARF2_UNWIND_INFO): Likewise.
34536 * config/arc/big.h: New file.
34537 * config/arc/elf.h: Likewise.
34538 * config/arc/linux.h: Likewise.
34539 * config/arc/t-uClibc: Remove.
34541 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34543 PR tree-optimization/77536
34544 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34545 (tree_transform_and_unroll_loop): Use above function to compute the
34546 estimated niter of unrolled loop and use it when scaling profile.
34547 Also use count info rather than frequency if it's non-zero.
34548 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34549 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34550 (vect_transform_loop): Call above function.
34552 2017-02-27 Richard Biener <rguenther@suse.de>
34554 PR tree-optimization/45397
34555 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34556 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34557 (visit_nary_op): Add pattern matching for CSEing sign-changed
34558 or truncated operations with wider ones.
34560 2017-02-27 Richard Biener <rguenther@suse.de>
34562 PR tree-optimization/79690
34563 * tree-vect-stmts.c (vectorizable_store): Use vector type
34564 built from the DR with address-space.
34566 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34568 * doc/invoke.texi (Optimize Options): Refine the description
34569 of asan-use-after-return.
34571 2017-02-25 Alan Modra <amodra@gmail.com>
34573 PR rtl-optimization/79584
34574 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34575 base, not ad->base_term, the reg within base. Remove assertion
34576 that ad->base == ad->base_term. Replace gen_int_mode using
34577 bogus mode with const0_rtx.
34579 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34581 PR middle-end/79396
34582 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34583 FMA_EXPR like tcc_binary or tcc_unary.
34585 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34588 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34590 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34591 (output_loc_operands): Handle DW_OP_call_ref and
34592 DW_OP_GNU_variable_value.
34593 (struct variable_value_struct): New type.
34594 (struct variable_value_hasher): Likewise.
34595 (variable_value_hash): New variable.
34596 (string_types): Remove.
34597 (copy_loc_descr): New function.
34598 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34599 (prepend_loc_descr_to_each): New function.
34600 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34601 instead of add_loc_descr_to_each if the first argument is single
34602 location list and the second has multiple.
34603 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34604 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34605 when looking for variable value which doesn't have other location info.
34606 (loc_list_from_tree): Formatting fix.
34607 (gen_array_type_die): Simplify DW_AT_string_length handling.
34608 (adjust_string_types): Remove.
34609 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34610 string_types. Call resolve_variable_values.
34611 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34612 (resolve_addr_in_expr): Likewise. Add A argument.
34613 (copy_deref_exprloc): Remove deref argument. Adjust for the
34614 original expression being DW_OP_GNU_variable_value with optionally
34615 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34616 optionally after it.
34617 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34618 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34619 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34620 (variable_value_hasher::hash, variable_value_hasher::equal): New
34622 (resolve_variable_value_in_expr, resolve_variable_value,
34623 resolve_variable_values, note_variable_value_in_expr,
34624 note_variable_value): New functions.
34625 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34628 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34631 * opts.h (handle_generated_option): Add GENERATED_P argument.
34632 * opts-common.c (handle_option): Adjust function comment.
34633 (handle_generated_option): Add GENERATED_P argument, pass it to
34635 (control_warning_option): Pass false to handle_generated_option
34637 * opts.c (maybe_default_option): Pass true to handle_generated_option
34639 * optc-gen.awk: Likewise.
34641 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34643 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34644 a REG, look at the REG it is a SUBREG of.
34645 (splitter for cmpeqsi_t): Ditto.
34647 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34649 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34650 the special USEs with the pattern of the insn, not the insn itself.
34652 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34655 * doc/invoke.texi: Document -mload-store-pairs.
34657 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34658 Sandra Loosemore <sandra@codesourcery.com>
34660 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34661 argument isn't a CONST_INT.
34662 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34663 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34664 (nios2_validate_compare): Bypass alternate compare logic if *op2
34665 is not a CONST_INT.
34666 (ldstwm_operation_p): Return false if first_base is not a REG or
34667 if first_offset is not a CONST_INT.
34669 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34671 * config/cris/cris.md: Use correct operand in a define_peephole2.
34673 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34675 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34677 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34679 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34680 this_insn if it is an INSN or JUMP_INSN.
34681 (force_offsettable): Look at base, not at addr.
34682 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34683 on things that aren't necessarily CONST_INTs.
34685 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34687 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34688 -mfpmath=sse is the default also for x86-32 targets with SSE2
34689 instruction set when @option{-ffast-math} is enabled
34691 2017-02-24 Jeff Law <law@redhat.com>
34693 PR rtl-optimizatoin/79286
34694 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34697 2017-02-24 Richard Biener <rguenther@suse.de>
34699 PR tree-optimization/79389
34700 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34703 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34705 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34706 function comment to reflect reality.
34707 (loop_exits_before_overflow): Fix typo in function description.
34709 2017-02-24 Richard Biener <rguenther@suse.de>
34711 PR tree-optimization/79389
34712 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34713 properly that a threading opportunity exists. Detect conditional
34714 copy/constant propagation opportunities.
34716 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34718 * config/visium/visium.md (type): Add trap.
34719 (b): New mode attribute.
34720 (*btst): Rename into...
34721 (*btst<mode>): ...this and adjust.
34722 (*cbranchsi4_btst_insn): Rename into...
34723 (*cbranch<mode>4_btst_insn): ...this and adjust.
34724 (trap): New define_insn.
34726 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34728 PR tree-optimization/79389
34729 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34730 (noce_reversed_cond_code): New function.
34731 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34732 reversed_comparison_code. Formatting fix.
34733 (noce_try_store_flag): Test rev_cond != NULL in addition to
34734 reversed_comparison_code.
34735 (noce_try_store_flag_constants): Likewise.
34736 (noce_try_store_flag_mask): Likewise.
34737 (noce_try_addcc): Use rev_cond if non-NULL instead of
34738 reversed_comparison_code.
34739 (noce_try_cmove_arith): Likewise. Formatting fixes.
34740 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34741 (noce_find_if_block): Initialize rev_cond.
34742 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34743 instead of false as last argument never attempt to reverse it
34746 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34748 PR tree-optimization/79663
34749 * tree-predcom.c (combine_chains): Process refs in reverse order
34750 only for ZERO length chains, and add explaining comment.
34752 2017-02-23 Jeff Law <law@redhat.com>
34754 PR tree-optimization/79578
34755 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34756 in call to operand_equal_p.
34758 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34761 * config/i386/cpuid.h: Fix another undefined behavior.
34763 2017-02-23 Richard Biener <rguenther@suse.de>
34765 PR tree-optimization/79683
34766 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34767 vector types for data-refs.
34769 2017-02-23 Martin Liska <mliska@suse.cz>
34771 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34773 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34775 PR middle-end/79665
34776 * internal-fn.c (get_range_pos_neg): Moved to ...
34777 * tree.c (get_range_pos_neg): ... here. No longer static.
34778 * tree.h (get_range_pos_neg): New prototype.
34779 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34780 are known to be in between 0 and signed maximum inclusive, try to
34781 expand both unsigned and signed divmod and use the cheaper one from
34784 2017-02-22 Jeff Law <law@redhat.com>
34786 PR tree-optimization/79578
34787 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34788 to compare base operands.
34790 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34793 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34794 gpc_reg_operand instead of fpr_reg_operand.
34796 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34798 * config/mips/mips.c (mips_return_in_memory): Force FP
34799 vector types to be returned in memory for o32 ABI.
34801 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34803 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34804 instead of DW_TAG_member for static data member declarations and don't
34805 set no_linkage_name for static inline data members.
34806 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34809 2017-02-22 Martin Liska <mliska@suse.cz>
34811 * doc/invoke.texi: Replace inequality signs with square brackets
34814 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34817 * lra-constraints.c (simplify_operand_subreg): Handle
34818 WORD_REGISTER_OPERATIONS targets.
34820 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34823 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34824 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34825 elimination by swapping fld*.
34827 2017-02-22 Richard Biener <rguenther@suse.de>
34829 PR tree-optimization/79673
34830 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34831 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34832 irrelevant address-space qualifiers and avoiding a
34833 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34835 2017-02-22 Richard Biener <rguenther@suse.de>
34837 PR tree-optimization/79666
34838 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34839 to not symbolically negate if that may introduce undefined
34842 2017-02-22 Martin Liska <mliska@suse.cz>
34845 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34846 * data-streamer-out.c (streamer_write_gcov_count_stream):
34848 * value-prof.c (stream_out_histogram_value): Make assert more
34849 precise based on type of counter.
34851 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34854 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34855 Use nonimmediate_operand instead of memory_operand for operand 1.
34856 (float-extend standard_x87sse_constant_load splitter): Ditto.
34858 2017-02-21 Jeff Law <law@redhat.com>
34860 PR tree-optimization/79621
34861 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34862 blocks with edges to themselves.
34864 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34867 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34868 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34869 Use gimple_call_builtin_p.
34872 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34873 on temporarily removed DEBUG_INSNs.
34875 PR tree-optimization/79649
34876 * tree-loop-distribution.c (classify_partition): Give up on
34877 non-generic address space loads/stores.
34879 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34881 * doc/loop.texi (Loop manipulation): Remove nonexistent
34882 tree_ssa_loop_version from the documentation.
34883 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34885 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34888 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34889 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34890 * config/rs6000/rs6000.c: Include except.h.
34891 (rs6000_expand_split_stack_prologue): Call
34892 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34894 2017-02-21 Martin Jambor <mjambor@suse.cz>
34897 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34898 have been analyzed.
34900 2017-02-21 Martin Jambor <mjambor@suse.cz>
34902 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34903 for backward compatibility only.
34904 * doc/invoke.texi (Option Summary): Remove all references to
34905 -fipa-cp-alignment.
34907 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34911 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34913 * lra-constraints.c (curr_insn_transform): Handle
34914 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34916 2017-02-21 Martin Liska <mliska@suse.cz>
34918 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34920 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34922 PR translation/79638
34923 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34925 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34928 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34929 (arm_function_ok_for_sibcall): Return false for an indirect call by
34930 descriptor if all the argument registers are used.
34931 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34932 alignment of the function.
34934 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34936 PR tree-optimization/61441
34937 * simplify-rtx.c (simplify_const_unary_operation): For
34938 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34939 the sNaN unmodified.
34941 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34943 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34944 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34945 instead of SYSTEM_HEADER_DIR.
34947 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34948 Martin Liška <mliska@suse.cz>
34950 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34951 Fix typos and grammar, use active voice, and clarify.
34953 2017-02-20 Marek Polacek <polacek@redhat.com>
34955 PR middle-end/79537
34956 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34959 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34961 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34964 * config/i386/i386.c (ix86_expand_builtin): Handle
34965 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34966 ix86_builtins_isa[fcode].isa as a requirement of those
34967 flags and any other flag in the bitmask.
34968 (ix86_init_mmx_sse_builtins): Use 0 instead of
34969 ~OPTION_MASK_ISA_64BIT as mask.
34970 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34971 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34972 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34973 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34975 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34978 * lra-constraints.c (split_reg): Check requested split mode
34979 is supported by the register.
34981 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34983 * lra-constraints.c (simplify_operand_subreg): Remove early
34986 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34989 * lra-constraints.c (curr_insn_transform): Tighten condition
34990 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34992 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34995 * lra-constraints.c (curr_insn_transform): Handle
34996 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34998 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35001 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35003 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35005 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35008 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35011 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35013 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35016 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35018 * final.c (last_columnnum, override_columnnum): New variables.
35019 (final_start_function): Set last_columnnum, pass it to begin_prologue
35020 hook and pass 0 to dwarf2out_begin_prologue.
35021 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35022 to source_line debug hook.
35023 (notice_source_line): Compute last_columnnum and for debug_column_info
35024 return true on column changes.
35025 * debug.h (struct gcc_debug_hooks): Add column argument to
35026 source_line and begin_prologue hooks.
35027 (debug_nothing_int_charstar_int_bool): Remove prototype.
35028 (debug_nothing_int_int_charstar,
35029 debug_nothing_int_int_charstar_int_bool): New prototypes.
35030 (dwarf2out_begin_prologue): Add column argument.
35031 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35032 begin_prologue hooks.
35033 (debug_nothing_int_charstar_int_bool): Remove.
35034 (debug_nothing_int_int_charstar,
35035 debug_nothing_int_int_charstar_int_bool): New functions.
35036 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35037 through to dwarf2out_source_line.
35038 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35039 (dwarf2out_source_line): Add column argument, emit it if requested.
35040 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35042 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35043 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35044 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35045 through to dwarf2out_begin_prologue.
35046 (vmsdbgout_source_line): Add column argument, pass it through to
35047 dwarf2out_source_line.
35048 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35049 dbxout_source_line caller.
35050 (dbxout_source_line): Add column argument.
35052 * common.opt (gno-column-info, gcolumn-info): New options.
35053 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35054 (check_die): Also test for multiple DW_AT_decl_column attributes.
35055 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35056 DW_AT_decl_column if requested.
35057 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35059 (gen_variable_die): Likewise.
35060 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35061 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35064 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35065 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35066 (ix86_handle_option): Handle OPT_m3dnowa.
35067 * doc/invoke.texi (-m3dnowa): Document.
35068 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35069 -m3dnowa instead of -m3dnow -march=athlon.
35072 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35073 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35075 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35078 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35079 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35080 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35081 generator for vsx_xxpermdi_<mode>_be.
35082 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35083 force big-endian semantics.
35084 (vsx_xxpermdi_<mode>_be): New define_expand with same
35085 implementation as previous version of vsx_xxpermdi_<mode>.
35087 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35089 PR tree-optimization/79327
35090 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35091 variable, its initialization and use.
35093 2017-02-17 Julia Koval <julia.koval@intel.com>
35095 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35096 (OPTION_MASK_ISA_PKU_UNSET): New.
35097 (ix86_handle_option): Handle -mrdpid.
35098 * config/i386/cpuid.h (bit_RDPID): New.
35099 * config/i386/driver-i386.c (host_detect_local_cpu):
35100 Detect RDPID feature.
35101 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35102 * config/i386/i386-c.c (ix86_target_macros_internal):
35104 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35105 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35106 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35107 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35108 * config/i386/i386.md (define_insn "rdpid"): New.
35109 * config/i386/i386.opt Add -mrdpid.
35110 * config/i386/immintrin.h (_rdpid_u32): New.
35112 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35114 PR rtl-optimization/79541
35115 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35116 instead of transforming it into USE.
35118 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35120 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35121 If HONOR_SNANS (SFmode) force the input to a register.
35122 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35123 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35124 an frsp or similar insn.
35126 2017-02-17 Martin Liska <mliska@suse.cz>
35128 PR rtl-optimization/79577
35129 * params.def (selsched-max-sched-times): Increase minimum to 1.
35131 2017-02-17 Martin Liska <mliska@suse.cz>
35133 PR rtl-optimization/79574
35134 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35136 2017-02-17 Martin Liska <mliska@suse.cz>
35138 PR tree-optimization/79529
35139 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35140 ssa_defined_default_def_p to handle cases which are implicitly
35142 * tree-ssa.c (ssa_defined_default_def_p): New function.
35143 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35144 which are implicitly defined.
35145 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35147 2017-02-17 Richard Biener <rguenther@suse.de>
35149 PR middle-end/79576
35150 * params.def (max-ssa-name-query-depth): Limit to 10.
35152 2017-02-17 Richard Biener <rguenther@suse.de>
35154 PR tree-optimization/79552
35155 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35158 2017-02-17 Richard Biener <rguenther@suse.de>
35161 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35163 2017-02-17 Marek Polacek <polacek@redhat.com>
35165 PR middle-end/79536
35166 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35167 (fold_negate_expr): New wrapper.
35169 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35171 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35172 Correct terminology and de-emphasize pre-standard behavior.
35174 2017-02-16 Alan Modra <amodra@gmail.com>
35176 PR rtl-optimization/79286
35177 * ira.c (def_dominates_uses): New function.
35178 (update_equiv_regs): Don't create an equivalence for insns that
35179 may trap where the register def does not dominate the use.
35181 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35183 PR rtl-optimization/78127
35184 * lra.c (lra): Call lra_eliminate before finish the loop after
35187 2017-02-16 Richard Biener <rguenther@suse.de>
35189 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35191 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35192 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35193 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35194 (isl_val_int_from_wi): New function.
35195 (extract_affine_gmp): Rename to ...
35196 (extract_affine_wi): ... this, take a widest_int.
35197 (extract_affine_int): Just wrap extract_affine_wi.
35198 (add_param_constraints): Use isl_val_int_from_wi.
35199 (add_loop_constraints): Likewise, and extract_affine_wi.
35201 2017-02-15 Jeff Law <law@redhat.com>
35203 PR middle-end/79521
35204 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35205 ira_init_register_move_cost_if_necessary.
35207 2017-02-15 Martin Sebor <msebor@redhat.com>
35209 PR middle-end/32003
35210 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35211 removed in a prior commit.
35213 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35215 PR tree-optimization/79347
35216 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35217 counters during peeling.
35219 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35221 * Makefile.in (site.exp): Remove "set ISLVER".
35223 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35226 * real.c (real_from_integer): Call real_convert even for decimal.
35228 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35231 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35233 2017-02-14 Andrew Pinski <apinski@cavium.com>
35235 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35236 cores and change the partno/implementer to be correct.
35237 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35238 the 'B" as the implementer.
35239 * config/aarch64/aarch64-tune.md: Regenerate.
35241 2017-02-14 Carl Love <cel@us.ibm.com>
35243 * config/rs6000/rs6000.c: Add case statement entry to make the
35244 xvcvuxdsp built-in argument unsigned.
35245 * config/rs6000/vsx.md: Fix the source and return operand types so they
35246 match the instruction definitions from the ISA document. Fix typo
35247 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35250 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35253 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35254 member early_clobber_alts.
35255 * lra-lives.c (reg_early_clobber_p): New.
35256 (process_bb_lives): Use it.
35257 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35258 (debug_operand_data): Initialize early_clobber_alts.
35259 (setup_operand_alternative): Set up early_clobber_alts.
35260 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35261 alternatives to new_insn_reg.
35262 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35264 (lra_update_insn_regno_info): Pass the new arg.
35266 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35268 PR middle-end/79505
35269 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35270 (new_oacc_loop_raw): Don't clear already cleared fields.
35273 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35274 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35275 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35277 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35280 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35282 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35285 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35286 the extra instruction to the right place to store 128-bit constant
35289 2017-02-14 Martin Sebor <msebor@redhat.com>
35291 PR middle-end/79448
35292 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35293 warning for strings of unknown length.
35295 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35297 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35299 2017-02-14 Jeff Law <law@redhat.com>
35302 * ira-costs.c (scan_one_insn): Initialize register move costs
35303 for pseudos seen in USE/CLOBBER insns.
35305 PR tree-optimization/79095
35306 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35307 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35308 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35309 if the operands are known to be not equal, then the resulting range
35311 (intersect_ranges): If the new range is ~[0,0] and the old range is
35312 wide, then prefer ~[0,0].
35313 * tree-vrp.c (overflow_comparison_p_1): New function.
35314 (overflow_comparison_p): New function.
35315 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35316 if NAME is used in an overflow test.
35317 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35318 overflow check that can be expressed as an equality test, then adjust
35319 ops to be that equality test.
35321 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35323 * config/s390/s390-builtin-types.def: Remove flags argument.
35324 * config/s390/s390.c (s390_init_builtins): Likewise.
35326 2017-02-14 Martin Liska <mliska@suse.cz>
35328 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35329 vector. Fix trailing white spaces.
35331 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35333 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35336 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35338 PR rtl-optimization/68664
35339 * config/arm/arm.c (arm_sched_can_speculate_insn):
35340 New function. Declare prototype.
35341 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35343 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35345 PR rtl-optimization/68664
35346 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35348 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35350 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35352 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35353 max skip bytes for function, loop and jump.
35355 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35357 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35358 ABS_EXPR for gimple dump.
35360 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35363 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35365 PR tree-optimization/79408
35366 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35367 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35368 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35369 also if rhs1 is INTEGER_CST.
35371 2017-02-14 Richard Biener <rguenther@suse.de>
35373 PR middle-end/79432
35374 * tree-into-ssa.c (insert_phi_nodes): When the function can
35375 have abnormal edges rewrite SSA names with broken use-def
35376 dominance out of SSA and register them for PHI insertion.
35378 2017-02-13 Martin Sebor <msebor@redhat.com>
35380 PR middle-end/79496
35381 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35382 clearing info.nowrite flag when snprintf size argument is a range.
35384 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35386 * cprop.c (cprop_jump): Add missing space in string literal.
35387 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35388 (get_constraint_for_component_ref): Likewise.
35389 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35390 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35391 * lra-constraints.c (process_alt_operands): Likewise.
35392 * ipa-inline.c (inline_small_functions): Likewise.
35393 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35394 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35395 * trans-mem.c (diagnose_tm_1_op): Likewise.
35396 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35397 (grid_parallel_clauses_gridifiable): Likewise.
35399 * config/nvptx/mkoffload.c (process): Add space in between
35402 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35403 "MOD4_SSE_REGS" and "ALL_REGS".
35405 * spellcheck.c (test_data): Add , in between "foo" and "food".
35407 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35410 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35411 boundary crossing check and subsequent code generation agree.
35413 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35415 * config/aarch64/aarch64.c (has_memory_op): Delete.
35416 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35419 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35421 PR rtl-optimization/79388
35422 PR rtl-optimization/79450
35423 * combine.c (distribute_notes): When removing TEM_INSN for which
35424 corresponding dest has last value recorded, invalidate that last
35427 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35429 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35430 of explicit '@'. Add missing assembly comment marker on branch costs
35433 2017-02-13 Nathan Sidwell <nathan@acm.org>
35435 * gengtype-lex.l (<in_struct>): Add '/'.
35437 2017-02-13 Martin Liska <mliska@suse.cz>
35440 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35442 2017-02-13 Richard Biener <rguenther@suse.de>
35444 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35446 * configure: Re-generate.
35447 * config.in: Likewise.
35448 * graphite-dependences.c: Simplify as if
35449 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35450 * graphite-isl-ast-to-gimple.c: Likewise.
35451 * graphite-optimize-isl.c: Likewise.
35452 * graphite-poly.c: Likewise.
35453 * graphite-sese-to-poly.c: Likewise.
35454 * graphite.h: Likewise.
35455 * toplev.c: Include isl/version.h and use isl_version () for
35456 printing the ISL version.
35457 * doc/install.texi: Update ISL requirement.
35459 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35461 * doc/standards.texi (Standards): Update reference to
35464 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35466 * doc/extend.texi (Named Address Spaces): sourceware.org now
35468 * doc/install.texi (Binaries): Ditto.
35471 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35473 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35474 standard terminology "stringize"/"stringizing" throughout.
35475 * doc/cppinternals.texi: Likewise.
35477 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35479 * doc/extend.texi: Fix some spelling mistakes and typos.
35480 * doc/invoke.texi: Likewise.
35482 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35485 * params.def (inline-min-speedup) Change from 10 to 8.
35487 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35489 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35492 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35495 * ipa-inline-analysis.c (get_minimal_bb): New function.
35496 (record_modified): Use it.
35497 (remap_edge_change_prob): Handle also ancestor functions.
35499 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35501 * doc/contrib.texi (Contributors): Remove broken link into
35502 the Mauve CVS repository.
35504 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35506 PR middle-end/79454
35507 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35508 result computation whenever lhs doesn't have vector mode, not
35509 just when it has BLKmode.
35511 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35513 * doc/makefile.texi (profiledbootstrap): Refer to the
35514 installation instructions only in textual form.
35516 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35519 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35521 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35523 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35524 (Specific): Update mingw-w64 reference.
35526 (Specific): Remove broken link to Renesas RX processor.
35528 2017-02-10 Richard Biener <rguenther@suse.de>
35530 * toplev.c (process_options): Do not mention obsolete graphite
35531 options when printing sorry message about missing graphite support.
35532 Mention -floop-nest-optimize.
35534 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35536 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35537 (vtst_p16): Likewise.
35538 (vtstq_p8): Likewise.
35539 (vtstq_p16): Likewise.
35541 (vtstq_p64): Likewise.
35542 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35543 (vset_lane_p64): New.
35544 (vsetq_lane_p64): New.
35546 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35548 PR tree-optimization/79411
35549 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35550 stmt operands are SSA_NAMEs used in abnormal phis.
35551 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35554 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35557 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35560 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35562 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35564 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35566 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35570 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35571 not arbitrary TREE_CONSTANT.
35574 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35575 "omp declare target link" attribute unless is_global_var.
35576 * omp-offload.c (find_link_var_op): Likewise.
35578 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35579 Chung-Lin Tang <cltang@codesourcery.com>
35581 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35583 (gimplify_adjust_omp_clauses): Don't delete TILE.
35584 (gimplify_omp_for): Deal with TILE.
35585 * internal-fn.c (expand_GOACC_TILE): New function.
35586 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35588 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35589 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35591 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35592 avoid DIV for outermost collapse var.
35593 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35594 Remove out of date comments, fix whitespace.
35595 * omp-general.c (omp_extract_for_data): Deal with tiling.
35596 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35597 adjust OLF_DIM_BASE value.
35598 (struct omp_for_data): Add tiling field.
35599 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35600 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35601 for auto loops. Remove default auto determining, moved to
35602 oacc_loop_fixed_partitions.
35603 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35604 stmts, add e_mask field.
35605 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35606 (oacc_thread_numbers): Use oacc_dim_call.
35607 (oacc_xform_tile): New.
35608 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35609 (finish_oacc_loop): Adjust for ifns vector.
35610 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35611 add case for GOACC_TILE fns.
35612 (oacc_loop_xform_loop): Delete.
35613 (oacc_loop_process): Iterate over call list directly, and add
35614 handling for GOACC_TILE fns.
35615 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35617 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35618 vector partitioning to outer loops. Assign 2 partitions to loops
35619 when available. Add TILE handling.
35620 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35621 (execite_oacc_device_lower): Process GOACC_TILE fns,
35622 ignore unknown specs.
35623 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35624 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35625 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35627 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35629 * configure.ac (ACX_BUGURL): Update.
35630 * configure: Regenerate.
35632 2017-02-09 Richard Biener <rguenther@suse.de>
35634 PR tree-optimization/69823
35635 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35636 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35638 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35640 * config/arc/arc-c.def: Add __NPS400__ definition.
35641 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35642 (TARGET_NPS400): Define.
35644 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35646 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35648 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35649 pointer, arch_info.
35650 (arc_cpu_types): Fill the arch_info field with a pointer into the
35651 arc_arch_types table.
35652 (arc_selected_cpu): Declare.
35653 * config/arc/arc.c (arc_selected_cpu): Make global.
35654 (arc_selected_arch): Delete.
35655 (arc_base_cpu): Delete.
35656 (arc_override_options): Remove references to deleted variables,
35657 update access to arch information.
35658 (ARC_OPT): Update access to arch information.
35659 (ARC_OPTX): Likewise.
35660 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35661 (TARGET_ARC600): Update access to arch information.
35662 (TARGET_ARC601): Likewise.
35663 (TARGET_ARC700): Likewise.
35664 (TARGET_EM): Likewise.
35665 (TARGET_HS): Likewise.
35666 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35669 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35672 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35673 condition/operands for integer GE/LE/GEU/LEU operations.
35675 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35677 PR translation/79397
35678 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35681 2017-02-08 Martin Jambor <mjambor@suse.cz>
35684 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35685 whether allocation happened.
35686 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35687 nothing was allocated.
35689 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35691 PR tree-optimization/79408
35692 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35693 constant, but SSA_NAME with a known integer range, use the minimum
35694 of that range instead of op1 to determine if modulo can be replaced
35695 with its first operand.
35697 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35699 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35701 2017-02-08 Richard Biener <rguenther@suse.de>
35703 PR tree-optimization/71824
35704 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35705 Check all loops contained in the merged region.
35707 2017-02-07 Andrew Pinski <apinski@cavium.com>
35709 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35711 2017-02-07 Andrew Pinski <apinski@cavium.com>
35713 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35714 (thunderxt88): Likewise.
35715 (thunderxt81): Disable LSE and change v8.1 to v8.
35716 (thunderxt83): Likewise.
35718 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35719 Richard Biener <rguenther@suse.de>
35721 PR middle-end/79399
35722 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35723 type from int to size_t.
35724 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35726 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35728 PR rtl-optimization/79386
35729 * cprop.c (bypass_conditional_jumps): Initialize
35730 bypass_last_basic_block already before splitting bbs after
35731 unconditional traps...
35732 (bypass_conditional_jumps): ... rather than here.
35735 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35736 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35737 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35738 fix -masm=intel patterns.
35740 2017-02-07 Richard Biener <rguenther@suse.de>
35742 PR tree-optimization/79256
35743 PR middle-end/79278
35744 * builtins.c (get_object_alignment_2): Use min_align_of_type
35745 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35746 and ADJUST_FIELD_ALIGN.
35748 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35750 * doc/tm.texi: Regenerate.
35751 * stor-layout.c (layout_decl): Adjust.
35752 (update_alignment_for_field): Likewise.
35753 (place_field): Likewise.
35754 (min_align_of_type): Likewise.
35755 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35756 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35757 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35758 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35759 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35760 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35761 * config/i386/i386.c (x86_field_alignment): Likewise.
35762 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35763 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35764 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35765 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35766 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35767 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35771 2017-01-30 Richard Biener <rguenther@suse.de>
35773 PR tree-optimization/79256
35774 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35775 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35778 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35780 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35781 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35782 builtins to SImode and emit a zero-extend, if necessary.
35784 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35786 * docs/invoke.texi (RISC-V Options): Alphabetize.
35788 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35790 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35793 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35795 * config/riscv/riscv.c: New file.
35796 * common/config/riscv/riscv-common.c: Likewise.
35797 * config.gcc: Likewise.
35798 * config/riscv/constraints.md: Likewise.
35799 * config/riscv/elf.h: Likewise.
35800 * config/riscv/generic.md: Likewise.
35801 * config/riscv/linux.h: Likewise.
35802 * config/riscv/multilib-generator: Likewise.
35803 * config/riscv/peephole.md: Likewise.
35804 * config/riscv/pic.md: Likewise.
35805 * config/riscv/predicates.md: Likewise.
35806 * config/riscv/riscv-builtins.c: Likewise.
35807 * config/riscv/riscv-c.c: Likewise.
35808 * config/riscv/riscv-ftypes.def: Likewise.
35809 * config/riscv/riscv-modes.def: Likewise.
35810 * config/riscv/riscv-opts.h: Likewise.
35811 * config/riscv/riscv-protos.h: Likewise.
35812 * config/riscv/riscv.h: Likewise.
35813 * config/riscv/riscv.md: Likewise.
35814 * config/riscv/riscv.opt: Likewise.
35815 * config/riscv/sync.md: Likewise.
35816 * config/riscv/t-elf-multilib: Likewise.
35817 * config/riscv/t-linux: Likewise.
35818 * config/riscv/t-linux-multilib: Likewise.
35819 * config/riscv/t-riscv: Likewise.
35820 * configure.ac: Likewise.
35821 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35822 Waterman as RISC-V maintainers.
35823 * doc/install.texi: Add RISC-V entries.
35824 * doc/invoke.texi: Add RISC-V options section.
35825 * doc/md.texi: Add RISC-V constraints section.
35826 * configure: Regenerated.
35828 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35831 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35832 false values to be constant vectors with all 0 or all 1 bits set.
35833 (vcondu<mode><mode>): Likewise.
35834 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35836 (fpmask_comparison_operator): Update comment.
35837 (vecint_comparison_operator): New predicate.
35838 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35839 vector conditionals when the true and false values are constant
35840 vectors with all 0 bits or all 1 bits set.
35842 2017-02-06 Martin Sebor <msebor@redhat.com>
35844 PR tree-optimization/79376
35845 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35847 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35849 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35850 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35851 to simplify split condition.
35853 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35855 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35856 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35859 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35861 PR rtl-optimization/68664
35862 * target.def (can_speculate_insn): New hook.
35863 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35864 * doc/tm.texi: Regenerate.
35865 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35866 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35867 (rs6000_sched_can_speculate_insn): New function.
35869 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35871 PR tree-optimization/79284
35872 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35873 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35874 vectorizable_mask_load_store, vectorizable_operation,
35875 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35876 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35877 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35878 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35879 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35880 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35881 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35882 is_gimple_assign (stmt). Replace another such test with
35883 is_gimple_assign (stmt).
35885 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35888 * config/avr/avr.c (rtl-iter.h): Include it.
35889 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35890 (avr_legitimate_combined_insn): ...and implementation.
35892 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35894 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35895 * config/s390/s390.c (s390_const_operand_ok)
35896 (s390_canonicalize_comparison, s390_extract_part)
35897 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35898 (s390_contiguous_bitmask_p, s390_rtx_costs)
35899 (legitimize_pic_address): Likewise.
35900 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35901 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35902 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35903 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35904 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35906 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35908 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35909 REGNO($0) == REGNO($1).
35911 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35913 * config/s390/linux.h(SIZE_TYPE): Add comment.
35915 2017-02-06 Julian Brown <julian@codesourcery.com>
35916 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35917 Virendra Pathak <virendra.pathak@broadcom.com>
35919 * config/aarch64/aarch64-cores.def: Change the scheduler
35921 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35922 * config/aarch64/thunderx2t99.md: New file.
35924 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35926 * doc/standards.texi (Go Language): Update link to language
35929 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35931 * tree-eh.c (lower_resx): Sanitize profile.
35932 (cleanup_empty_eh_move_lp): Likewise.
35934 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35937 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35939 * cfgloopmanip.h (loop_version): Update prototype.
35940 * modulo-sched.c (sms_schedule): Update call of loop_version.
35941 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35942 * tree-parloops.c (gen_parallel_loop): Likewise.
35943 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35944 * tree-ssa-loop-split.c (split_loop): Likewise.
35945 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35946 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35948 2017-02-05 Martin Liska <mliska@suse.cz>
35951 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35953 (print_operand_address): Initialize a struct to zero.
35955 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35957 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35958 garbage collector only in textual form.
35960 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35962 * doc/extend.texi (x86 specific memory model extensions for
35963 transactional memory): Simplify a phrase.
35965 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35968 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35969 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35970 (atomic_storedi_1): Likewise.
35972 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35974 PR tree-optimization/79338
35975 * tree-parloops.c (gather_scalar_reductions): Don't call
35976 vect_analyze_loop_form for loop->inner before destroying loop's
35979 2017-02-03 Martin Sebor <msebor@redhat.com>
35981 PR tree-optimization/79327
35982 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35983 when precision has resulted in leading zeros.
35984 (format_integer): Adjust the likely counter to assume an unknown
35985 argument that may be zero is non-zero.
35987 2017-02-03 Jason Merrill <jason@redhat.com>
35990 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35991 avoid copying non-taken branch.
35993 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35995 PR tree-optimization/79340
35996 * tree-vect-loop.c (vectorizable_reduction): Release
35997 vec_defs elements after safe_splicing them into other vectors.
36000 PR tree-optimization/79327
36001 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36002 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36004 (format_integer): Use wide_int_to_tree instead of build_int_cst
36005 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36006 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36007 of shortest and longest sequence.
36009 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36011 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36012 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36014 2017-02-03 Walter Lee <walt@tilera.com>
36017 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36018 after initial stackframe link reg save.
36019 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36021 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36024 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36025 wu for stxssp alternative.
36027 2017-02-03 Martin Sebor <msebor@redhat.com>
36029 PR tree-optimization/79352
36030 * gimple-fold.c (get_range_strlen): Add argument.
36031 (get_range_strlen): Change return type to bool.
36032 (get_maxval_strlen): Pass in a dummy argument.
36033 * gimple-fold.h (get_range_strlen): Change return type to bool.
36034 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36035 * tree.h (array_at_struct_end_p): Add argument.
36036 * tree.c (array_at_struct_end_p): Handle it.
36038 2017-02-03 Martin Liska <mliska@suse.cz>
36041 * multiple_target.c (create_dispatcher_calls): Redirect edge
36042 from a caller of a dispatcher.
36043 (expand_target_clones): Make the clones local.
36044 (ipa_target_clone): Do both target clones and resolvers.
36045 (ipa_dispatcher_calls): Remove the pass.
36046 (pass_dispatcher_calls::gate): Likewise.
36047 (make_pass_dispatcher_calls): Likewise.
36048 * passes.def (pass_target_clone): Put as very first IPA early
36051 2017-02-03 Martin Liska <mliska@suse.cz>
36053 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36054 in case of a function with ifunc attribute.
36056 2017-02-03 Martin Liska <mliska@suse.cz>
36058 * cgraph.c (cgraph_node::dump): Dump function version info.
36059 * symtab.c (symtab_node::dump_base): Add missing new line.
36061 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36063 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36064 (ifcombine_ifandif): Use it.
36066 2017-02-03 Martin Liska <mliska@suse.cz>
36068 * doc/invoke.texi: Document default value for
36069 use-after-scope-direct-emission-threshold.
36071 2017-02-03 Martin Liska <mliska@suse.cz>
36073 PR tree-optimization/79339
36074 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36075 (format_floating): Likewise.
36077 2017-02-03 Martin Liska <mliska@suse.cz>
36080 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36082 (ipa_node_params_t::remove): Likewise.
36083 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36084 initialization from removed ipa_node_params_t::insert.
36085 (ipa_node_params::~ipa_node_params): Move from removed
36086 ipa_node_params_t::release.
36087 * symbol-summary.h (symbol_summary::m_released): New member.
36088 Do not release a summary twice. Do not allow to call finalizer
36089 for types of a summary that live in GGC memory.
36091 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36093 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36096 2017-02-02 Martin Sebor <msebor@redhat.com>
36098 PR middle-end/79275
36099 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36100 (format_string): Tighten up the range of output for non-constant
36101 strings and correct the expected range for wide non-constant strings.
36103 2017-02-02 Martin Sebor <msebor@redhat.com>
36105 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36107 PR middle-end/32003
36108 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36110 (-fdump-tree-@var): Add to index and document how to come up
36111 with pass-specific option and dump file names.
36112 (-fdump-passes): Clarify where to look for output.
36114 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36116 PR middle-end/77445
36117 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36118 statistics of the analyzed path; allow threading for speed when
36119 any of BBs along the path are optimized for speed.
36121 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36123 PR middle-end/78468
36124 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36125 settings of the virtual registers.
36128 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36130 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36131 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36134 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36136 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36137 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36139 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36141 * config/s390/s390.md: Add missing comments with the expanded
36143 * config/s390/vector.md: Likewise.
36144 * config/s390/vx-builtins.md: Likewise.
36146 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36149 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36150 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36151 conditions on a single line.
36153 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36155 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36156 __S390_VX__ to __VX__.
36158 2017-02-01 Andrew Pinski <apinski@cavium.com>
36160 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36161 stmt_info to record_stmt_cost.
36162 (vect_get_known_peeling_cost): Pass stmt_info if known to
36164 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36165 cpu_vector_cost field into
36166 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36167 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36168 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36169 splitting of scalar_stmt_cost and vec_stmt_cost.
36170 (thunderx_vector_cost): Likewise.
36171 (cortexa57_vector_cost): LIkewise.
36172 (exynosm1_vector_cost): Likewise.
36173 (xgene1_vector_cost): Likewise.
36174 (thunderx2t99_vector_cost): Improve after the splitting of the two
36176 (aarch64_builtin_vectorization_cost): Update for the splitting of
36177 scalar_stmt_cost and vec_stmt_cost.
36179 2017-02-01 Torvald Riegel <triegel@redhat.com>
36180 Richard Henderson <rth@redhat.com>
36182 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36183 conditional on existance of a fast atomic load.
36184 * optabs-query.c (can_atomic_load_p): New function.
36185 * optabs-query.h (can_atomic_load_p): Declare it.
36186 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36187 no fast atomic load is available for the particular size of access.
36188 (expand_atomic_compare_and_swap): Likewise.
36189 (expand_atomic_load): Likewise.
36190 (expand_atomic_store): Likewise.
36191 (expand_atomic_fetch_op): Likewise.
36192 * testsuite/lib/target-supports.exp
36193 (check_effective_target_sync_int_128): Remove x86 because it provides
36194 no fast atomic load.
36195 (check_effective_target_sync_int_128_runtime): Likewise.
36197 2017-02-01 Richard Biener <rguenther@suse.de>
36199 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36200 (graphite_transform_loops): Provide opt-info for optimized nests.
36201 * tree-parloop.c (parallelize_loops): Provide opt-info for
36202 parallelized loops.
36204 2017-02-01 Richard Biener <rguenther@suse.de>
36206 PR middle-end/79315
36207 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36208 was not set before.
36210 2017-02-01 Richard Biener <rguenther@suse.de>
36212 PR tree-optimization/71824
36213 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36214 Verify the loops are valid in the merged SESE region.
36215 (scop_detection::can_represent_loop_1): Check analyzing the
36216 evolution of the number of iterations in the region succeeds.
36218 2017-01-31 Ian Lance Taylor <iant@golang.org>
36220 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36221 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36223 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36225 PR preprocessor/79210
36226 * input.c (get_substring_ranges_for_loc): Replace line_width
36227 assertion with error-handling.
36229 2017-01-31 Richard Biener <rguenther@suse.de>
36231 PR tree-optimization/77318
36232 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36233 (create_pw_aff_from_tree): Take loop parameter.
36234 (add_condition_to_pbb): Pass loop of the condition to
36235 create_pw_aff_from_tree.
36237 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36239 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36240 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36242 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36246 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36248 (convert_int_to_float128): Likewise.
36249 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36250 (convert_int_to_float128): Likewise.
36251 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36252 (UNSPEC_IEEE128_CONVERT): Likewise.
36253 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36254 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36255 Use local variables for IBM extended format.
36256 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36257 (fix_trunc<mode>si2_fprs): Likewise.
36258 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36259 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36260 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36261 to know that we can now have integers of all sizes in vector
36263 (fix<uns>_<mode>di2_hw): Likewise.
36264 (float<uns>_<mode>si2_hw): Likewise.
36265 (fix_<mode>si2_hw): Likewise.
36266 (fixuns_<mode>si2_hw): Likewise.
36267 (float<uns>_<mode>di2_hw): Likewise.
36268 (float_<mode>di2_hw): Likewise.
36269 (float_<mode>si2_hw): Likewise.
36270 (floatuns_<mode>di2_hw): Likewise.
36271 (floatuns_<mode>si2_hw): Likewise.
36272 (xscvqp<su>wz_<mode>): Delete, no longer used.
36273 (xscvqp<su>dz_<mode>): Likewise.
36274 (xscv<su>dqp_<mode>): Likewise.
36275 (ieee128_mfvsrd_64bit): Likewise.
36276 (ieee128_mfvsrd_32bit): Likewise.
36277 (ieee128_mfvsrwz): Likewise.
36278 (ieee128_mtvsrw): Likewise.
36279 (ieee128_mtvsrd_64bit): Likewise.
36280 (ieee128_mtvsrd_32bit): Likewise.
36282 2017-01-31 Martin Liska <mliska@suse.cz>
36285 * ipa-prop.c (ipa_free_all_node_params): Call release method
36286 instead of ~sumbol_summary to not to trigger double times
36289 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36291 PR tree-optimization/71691
36292 * bitmap.h (class auto_bitmap): New.
36293 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36294 is_maybe_undefined instead of ssa_undefined_value_p.
36296 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36298 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36299 __S390_ARCH_LEVEL__ to __ARCH__.
36301 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36303 PR tree-optimization/79267
36304 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36305 if should_remove_lhs_p is true.
36307 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36310 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36311 (add_alignment_attribute): New.
36312 (base_type_die): Add alignment attribute.
36313 (subrange_type_die): Likewise.
36314 (modified_type_die): Likewise.
36315 (gen_array_type_die): Likewise.
36316 (gen_descr_array_type_die: Likewise.
36317 (gen_enumeration_type_die): Likewise.
36318 (gen_subprogram_die): Likewise.
36319 (gen_variable_die): Likewise.
36320 (gen_field_die): Likewise.
36321 (gen_ptr_to_mbr_type_die): Likewise.
36322 (gen_struct_or_union_type_die): Likewise.
36323 (gen_subroutine_type_die): Likewise.
36324 (gen_typedef_die): Likewise.
36325 (base_type_cmp): Compare alignment attribute.
36327 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36330 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36331 (setb_unsigned) New pattern for setb with CCUNS.
36332 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36333 subfc./subfe sequence to avoid overflow problems. Generate a
36334 shorter sequence with cmpld/setb for power9.
36335 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36336 for generating subfc. instruction.
36337 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36338 now uses this instruction.
36340 2017-01-30 Ian Lance Taylor <iant@google.com>
36343 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36344 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36346 2017-01-30 Martin Sebor <msebor@redhat.com>
36348 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36349 Move constant to the right of a relational operator.
36350 (get_mpfr_format_length, format_character, format_string): Ditto.
36351 (should_warn_p, maybe_warn): Same.
36353 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36355 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36358 * asan.c (get_translation_unit_decl): Remove function.
36359 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36361 2017-01-30 Martin Liska <mliska@suse.cz>
36363 PR gcov-profile/79259
36364 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36365 -fprofile-generate.
36367 2017-01-30 Martin Liska <mliska@suse.cz>
36370 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36371 Initialize variables with NULL value.
36373 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36376 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36378 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36380 2017-01-30 Richard Biener <rguenther@suse.de>
36382 PR tree-optimization/79276
36383 * tree-vrp.c (process_assert_insertions): Properly adjust common
36384 when removing a duplicate.
36386 2017-01-30 Richard Biener <rguenther@suse.de>
36388 PR tree-optimization/79256
36389 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36390 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36392 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36394 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36397 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36398 ("*r<noxa>sbg_<mode>_sll_bitmask")
36399 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36400 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36401 Use contiguous_bitmask_nowrap_operand.
36403 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36406 * config/rs6000/altivec.h (vec_xl): Revise #define.
36407 (vec_xst): Likewise.
36409 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36411 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36413 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36415 PR rtl-optimization/79194
36416 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36417 traps before call to bypass_conditional_jumps.
36419 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36421 PR tree-optimization/71374
36422 * lra-constraints.c (check_conflict_input_operands): New.
36423 (match_reload): Use it.
36425 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36428 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36429 account to calculate conflict_set.
36431 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36433 PR rtl-optimization/78559
36434 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36435 other_insn in combine.
36437 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36439 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36440 uint16_type_node for BT_UINT16.
36442 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36444 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36445 "RTL Tests" to menu.
36446 (GIMPLE Tests): New node.
36447 (RTL Tests): New node.
36449 2017-01-27 Richard Biener <rguenther@suse.de>
36451 PR tree-optimization/79245
36452 * tree-loop-distribution.c (distribute_loop): Apply cost
36453 modeling also to detected patterns.
36455 2017-01-27 Richard Biener <rguenther@suse.de>
36457 PR tree-optimization/71433
36458 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36459 (compare_assert_loc): New function.
36460 (process_assert_insertions): Sort and optimize assert locations
36461 to remove duplicates and push down identical assertions on
36462 edges to their destination block.
36464 2017-01-27 Richard Biener <rguenther@suse.de>
36466 PR tree-optimization/79244
36467 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36468 out SSA names even if abnormal.
36470 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36472 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36473 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36474 instead of MPFR_RNDN.
36476 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36479 * arm.c (arm_option_override): Don't call build_target_option_node
36480 until after doing all option overrides.
36481 (arm_valid_target_attribute_tree): Likewise.
36483 2017-01-27 Martin Liska <mliska@suse.cz>
36485 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36486 for {cd}tors and C++ {cd}tors.
36488 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36490 * config/s390/s390.md ("*setmem_long_and")
36491 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36493 2017-01-26 Martin Sebor <msebor@redhat.com>
36495 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36498 2017-01-26 Martin Sebor <msebor@redhat.com>
36500 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36501 HAVE_DFmode before using XFmode or DFmode.
36502 (parse_directive): Avoid using the z length modifier to avoid
36503 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36505 PR middle-end/78703
36506 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36507 to accept adjustment as an array.
36508 (get_int_range): New function.
36509 (struct directive): Make width and prec arrays.
36510 (directive::set_width, directive::set_precision): Call get_int_range.
36511 (format_integer, format_floating): Handle width and precision ranges.
36512 (format_string, parse_directive): Same.
36514 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36517 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36518 comdat_type_p set, just clone them, but keep the children in the
36522 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36523 which have direct callers with -fvar-tracking-assignments enabled
36525 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36526 inside of type units.
36528 2017-01-26 Martin Sebor <msebor@redhat.com>
36530 PR middle-end/78703
36531 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36533 (struct format_result): Replace number_chars, number_chars_min,
36534 and number_chars_max with a single member of struct result_range.
36536 (format_result::operator+=): Adjust.
36537 (struct fmtresult): Remove bounded. Handle likely and unlikely
36539 (fmtresult::adjust_for_width_or_precision): New function.
36540 (fmtresult:type_max_digits): New function.
36541 (bytes_remaining): Handle likely and unlikely counters.
36542 (min_bytes_remaining): Remove.
36543 (format_percent): Simplify.
36544 (format_integer, format_floating): Set likely and unlikely counters.
36545 (get_string_length, format_character, format_string): Same.
36546 (format_plain, should_warn_p): New function.
36547 (maybe_warn): Call should_warn_p. Update diagnostic messages
36548 and handle those for all directives, including plain strings.
36549 (format_directive): Handle likely and unlikely counters.
36550 Remove unnecessary quoting from diagnostics. Add an informational
36552 (add_bytes): Remove.
36553 (pass_sprintf_length::compute_format_length): Simplify.
36554 (try_substitute_return_value): Handle likely and unlikely counters.
36556 2017-01-26 Carl Love <cel@us.ibm.com>
36558 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36559 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36561 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36564 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36565 endianess for subregs into account.
36566 * lra-constraints.c (lra_constraints): Do risky transformations
36567 always on the first iteration.
36568 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36569 last_call_used_reg_set.
36570 (process_bb_lives): Define and use last_call_used_reg_set.
36571 * lra.c (lra): Always continue after lra_constraints on the first
36574 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36576 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36578 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36580 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36582 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36583 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36584 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36585 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36586 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36587 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36588 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36589 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36590 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36592 2017-01-26 Marek Polacek <polacek@redhat.com>
36595 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36596 for the third operand.
36598 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36600 PR middle-end/79236
36601 * omp-low.c (struct omp_context): Add simt_stmt field.
36602 (scan_omp_for): Return omp_context *.
36603 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36604 context to the _simt_ SIMD stmt.
36605 (lower_omp_for): For combined SIMD with sibling _simt_
36606 SIMD, make sure to use the same decls in _looptemp_
36607 clauses as in the sibling.
36609 2017-01-26 David Sherwood <david.sherwood@arm.com>
36611 PR middle-end/79212
36612 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36615 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36618 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36619 emit fld b; fld a; if possible.
36621 * brig-builtins.def: Update copyright years.
36622 * config/arm/arm_acle_builtins.def: Update copyright years.
36624 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36627 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36628 constraint instead of o for the stxsd instruction.
36630 2017-01-25 Carl Love <cel@us.ibm.com>
36632 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36633 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36635 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36637 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36639 2017-01-25 Richard Biener <rguenther@suse.de>
36641 PR tree-optimization/69264
36642 * target.def (vector_alignment_reachable): Improve documentation.
36643 * doc/tm.texi: Regenerate.
36644 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36646 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36647 earlier changes with respect to TYPE_USER_ALIGN.
36648 (vector_alignment_reachable_p): Likewise. Improve dumping.
36650 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36653 * config/arm/arm.md (xordi3): Force constant operand into a register
36656 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36658 * doc/invoke.texi (-fstore-merging): Correct default optimization
36659 levels at which it is enabled.
36660 (-O): Move -fstore-merging from list to...
36663 2017-01-25 Richard Biener <rguenther@suse.de>
36666 * omp-expand.c: Include debug.h.
36667 (expand_omp_taskreg): Make sure to generate early debug before
36668 outlining anything from a function.
36669 (expand_omp_target): Likewise.
36670 (grid_expand_target_grid_body): Likewise.
36672 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36675 * asan.c (get_translation_unit_decl): New function.
36676 (asan_add_global): Extract modules file name from globals
36677 TRANSLATION_UNIT_DECL name.
36679 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36682 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36683 for long calls with APCS frame and VFP.
36685 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36687 * cfg.c (original_copy_tables_initialized_p): New function.
36688 * cfg.h (original_copy_tables_initialized_p): New decl.
36689 * cfgrtl.c (relink_block_chain): Guard the call to
36690 free_original_copy_tables with a call to
36691 original_copy_tables_initialized_p.
36692 * cgraph.h (symtab_node::native_rtl_p): New decl.
36693 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36694 (symtab_node::needed_p): Don't assert for early assembly output
36695 for __RTL functions.
36696 (cgraph_node::finalize_function): Set "force_output" for __RTL
36698 (cgraph_node::analyze): Bail out early for __RTL functions.
36699 (analyze_functions): Update assertion to support __RTL functions.
36700 (cgraph_node::expand): Bail out early for __RTL functions.
36701 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36703 * function.h (struct function): Update comment for field
36705 * gimple-expr.c: Include "tree-pass.h".
36706 (gimple_has_body_p): Return false for __RTL functions.
36707 * Makefile.in (OBJS): Add run-rtl-passes.o.
36708 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36710 (gcc::pass_manager::get_clean_slate): New accessor.
36711 * passes.c: Include "insn-addr.h".
36712 (should_skip_pass_p): Add logging. Update logic for running
36713 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36714 property-provider override so it is only done for gimple passes.
36716 (skip_pass): New function.
36717 (execute_one_pass): Call skip_pass when skipping passes.
36718 * read-md.c (md_reader::read_char): Support filtering
36719 the input to a subset of line numbers.
36720 (md_reader::md_reader): Initialize fields
36721 m_first_line and m_last_line.
36722 (md_reader::read_file_fragment): New function.
36723 * read-md.h (md_reader::read_file_fragment): New decl.
36724 (md_reader::m_first_line): New field.
36725 (md_reader::m_last_line): New field.
36726 * read-rtl-function.c (function_reader::create_function): Only
36727 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36728 curr_properties. Set DECL_INITIAL to a dummy block.
36729 (read_rtl_function_body_from_file_range): New function.
36730 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36732 * run-rtl-passes.c: New file.
36733 * run-rtl-passes.h: New file.
36735 2017-01-24 Jeff Law <law@redhat.com>
36737 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36740 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36742 PR tree-optimization/79159
36743 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36744 (record_nonwrapping_iv): Improve boundary using above function if no
36745 value range information.
36747 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36748 Martin Jambor <mjambor@suse.cz>
36750 * brig-builtins.def: New file.
36751 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36752 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36753 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36754 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36755 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36756 * builtin-types.def (BT_INT8): New.
36757 (BT_INT16): Likewise.
36758 (BT_UINT8): Likewise.
36759 (BT_UINT16): Likewise.
36760 (BT_FN_ULONG): Likewise.
36761 (BT_FN_UINT_INT): Likewise.
36762 (BT_FN_UINT_ULONG): Likewise.
36763 (BT_FN_UINT_LONG): Likewise.
36764 (BT_FN_UINT_PTR): Likewise.
36765 (BT_FN_ULONG_PTR): Likewise.
36766 (BT_FN_INT8_FLOAT): Likewise.
36767 (BT_FN_INT16_FLOAT): Likewise.
36768 (BT_FN_UINT32_FLOAT): Likewise.
36769 (BT_FN_UINT16_FLOAT): Likewise.
36770 (BT_FN_UINT8_FLOAT): Likewise.
36771 (BT_FN_UINT64_FLOAT): Likewise.
36772 (BT_FN_UINT16_UINT32): Likewise.
36773 (BT_FN_UINT32_UINT16): Likewise.
36774 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36775 (BT_FN_INT_PTR_INT): Likewise.
36776 (BT_FN_UINT_PTR_UINT): Likewise.
36777 (BT_FN_LONG_PTR_LONG): Likewise.
36778 (BT_FN_ULONG_PTR_ULONG): Likewise.
36779 (BT_FN_VOID_UINT64_UINT64): Likewise.
36780 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36781 (BT_FN_INT8_INT8_INT8): Likewise.
36782 (BT_FN_INT16_INT16_INT16): Likewise.
36783 (BT_FN_INT_INT_INT): Likewise.
36784 (BT_FN_UINT_FLOAT_UINT): Likewise.
36785 (BT_FN_FLOAT_UINT_UINT): Likewise.
36786 (BT_FN_ULONG_UINT_UINT): Likewise.
36787 (BT_FN_ULONG_UINT_PTR): Likewise.
36788 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36789 (BT_FN_UINT_UINT_UINT): Likewise.
36790 (BT_FN_VOID_UINT_PTR): Likewise.
36791 (BT_FN_UINT_UINT_PTR: Likewise.
36792 (BT_FN_UINT32_UINT64_PTR): Likewise.
36793 (BT_FN_INT_INT_UINT_UINT): Likewise.
36794 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36795 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36796 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36797 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36798 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36799 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36800 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36801 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36802 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36803 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36804 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36805 * doc/frontends.texi: List BRIG FE.
36806 * doc/install.texi (Testing): Add BRIG tesring requirements.
36807 * doc/invoke.texi (Overall Options): Mention BRIG.
36808 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36810 2017-01-24 Richard Biener <rguenther@suse.de>
36812 PR translation/79208
36813 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36815 2017-01-24 Martin Jambor <mjambor@suse.cz>
36818 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36819 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36820 and known_contexts.
36822 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36824 PR middle-end/79123
36825 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36826 casts from signed to unsigned really don't have a range.
36828 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36830 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36831 GMP_RNDx for compatiblity.
36833 2017-01-24 Martin Liska <mliska@suse.cz>
36836 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36837 that would prevent us to call alloca with -1 as argument.
36839 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36841 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36842 Avoid -Wformat-security warning.
36844 2017-01-23 Andrew Pinski <apinski@cavium.com>
36846 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36849 2017-01-23 Martin Sebor <msebor@redhat.com>
36851 PR middle-end/78703
36852 * gimple-ssa-sprintf.c (warn_level): New global.
36853 (format_integer): Use it here and throughout the rest of the file.
36854 Use the same switch to compute sign as base.
36855 (maybe_warn): New function.
36856 (format_directive): Factor out warnings into maybe_warn.
36857 Add debugging output. Use warn_level.
36858 (add_bytes): Use warn_level.
36859 (pass_sprintf_length::compute_format_length): Add debugging output.
36860 (try_substitute_return_value): Same.
36861 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36863 PR middle-end/78703
36864 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36865 (struct fmtresult, format_integer, format_floating): Adjust.
36866 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36867 (get_string_length, format_string,format_directive): Same.
36868 (pass_sprintf_length::compute_format_length): Same.
36869 (try_substitute_return_value): Simplify slightly.
36871 PR middle-end/78703
36872 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36873 (fmtresult::operator+=): Outlined.
36874 (struct fmtresult): Add ctors.
36875 (struct conversion_spec): Rename...
36876 (struct directive): ...to this. Add and remove data members.
36877 (directive::set_width, directive::set_precision): New functions.
36878 (format_percent): Use fmtresult ctor.
36879 (get_width_and_precision): Remove.
36880 (format_integer): Make naming changes. Avoid computing width and
36882 (format_floating): Same. Adjust indentation.
36883 (format_character, format_none): New functions.
36884 (format_string): Moved character handling to format_character.
36885 (format_directive): Remove arguments, change return type.
36886 (parse_directive): New function.
36887 (pass_sprintf_length::compute_format_length): Move directive
36888 parsing to parse_directive.
36890 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36892 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36893 (assign_assembler_name_if_needed): ... this.
36894 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36895 (assign_assembler_name_if_needed): ... this.
36896 (free_lang_data_in_cgraph): Adjust callers.
36897 * cgraphunit.c (cgraph_node::analyze): Likewise.
36898 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36901 2017-01-23 Richard Biener <rguenther@suse.de>
36903 PR tree-optimization/79088
36904 PR tree-optimization/79188
36905 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36906 resetting loop bounds after last path deletion. Reset loop
36907 bounds of the target loop, make code match the comments.
36908 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36909 Make sure loops need no fixups.
36911 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36913 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36914 exponent support with double type for first argument.
36915 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36916 type returned by __builtin_vec_extract_sig,
36917 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36918 functions from "vector int" to "vector unsigned int" or from
36919 "vector long long int" to "vector unsigned long long int".
36920 Changed type returned by __builtin_vec_extract_exp,
36921 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36922 functions from "vector int" to "vector unsigned int" or from
36923 "vector long long int" to "vector unsigned long long int".
36924 Changed return type of __builtin_vec_test_data_class,
36925 __builtin_vec_test_data_class_sp, and
36926 __builtin_vec_test_data_class_dp from "vector int" to
36927 "vector bool int" or from "vector long long int" to "vector bool
36928 long long int" and changed second argument type from "unsigned
36929 int" to "int". Added new overloaded function forms "vector float
36930 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36931 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36932 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36933 double, vector unsigned long long int)" and "vector double
36934 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36935 long int)". Changed return type of
36936 __builtin_scalar_test_data_class and
36937 __builtin_scalar_test_data_class_sp and
36938 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36939 int" and changed second argument from "unsigned int" to "int".
36940 Changed type returned by __builtin_scalar_test_neg,
36941 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36942 from "int" to "bool int". Added new overloaded function form
36943 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36944 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36945 exponent double-precision with floating point first argument.
36946 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36947 documentation of scalar_test_data_class, scalar_test_neg,
36948 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36949 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36950 vec_test_data_class built-in functions to reflect refinements in
36951 their type signatures.
36953 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36955 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36957 (aarch64_elf_asm_destructor): Likewise.
36959 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36961 PR rtl-optimization/78634
36962 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36963 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36964 * ifcvt.c (noce_try_cmove): Add missing cost check.
36966 PR rtl-optimization/71724
36967 * combine.c (if_then_else_cond): Look for situations where it is
36968 beneficial to undo the work of one of the recursive calls.
36970 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36972 PR tree-optimization/70754
36973 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36974 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36975 combined stmt before it if not NULL.
36976 (combine_chains): Process refs reversely and compute dominance point
36979 2017-01-23 Martin Liska <mliska@suse.cz>
36981 PR tree-optimization/79196
36982 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36983 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36985 (strlen_optimize_stmt): Call the renamed function.
36987 2017-01-23 Michael Matz <matz@suse.de>
36989 PR tree-optimization/78384
36990 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36992 2017-01-23 Richard Biener <rguenther@suse.de>
36994 PR tree-optimization/79186
36995 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36996 both incoming edges before moving an assert.
36998 2017-01-23 Martin Jambor <mjambor@suse.cz>
37000 * ipa-prop.c (load_from_param_1): Removed.
37001 (load_from_unmodified_param): Bits from load_from_param_1 put back
37003 (load_from_param): Removed.
37004 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37005 with stmt. Reverted back to use of load_from_unmodified_param.
37007 2017-01-23 Martin Jambor <mjambor@suse.cz>
37010 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37011 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37012 field a pointer to garbage collected vector, mark lattices and
37013 ipcp_orig_node with GTY((skip)).
37014 (ipa_get_param_count): Adjust to descriptors being a pointer.
37015 (ipa_get_param): Likewise.
37016 (ipa_get_type): Likewise.
37017 (ipa_get_param_move_cost): Likewise.
37018 (ipa_set_param_used): Likewise.
37019 (ipa_get_controlled_uses): Likewise.
37020 (ipa_set_controlled_uses): Likewise.
37021 (ipa_is_param_used): Likewise.
37022 (ipa_node_params_t): Move into garbage collector. New methods insert
37024 (ipa_node_params_sum): Annotate wth GTY(()).
37025 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37027 (ipa_load_from_parm_agg): Adjust declaration.
37028 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37029 * ipa-profile.c (ipa_profile): Likewise.
37030 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37031 (ipa_populate_param_decls): Make descriptors parameter garbage
37033 (ipa_dump_param): Adjust to descriptors being a pointer.
37034 (ipa_alloc_node_params): Likewise.
37035 (ipa_initialize_node_params): Likewise.
37036 (load_from_param_1): Make descriptors parameter garbage collected.
37037 (load_from_unmodified_param): Likewise.
37038 (load_from_param): Likewise.
37039 (ipa_load_from_parm_agg): Likewise.
37040 (ipa_node_params::~ipa_node_params): Removed.
37041 (ipa_free_all_node_params): Remove call to delete operator.
37042 (ipa_node_params_t::insert): New.
37043 (ipa_node_params_t::remove): Likewise.
37044 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37045 copy known_csts and known_contexts vectors.
37046 (ipa_read_node_info): Adjust to descriptors being a pointer.
37047 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37049 (ipcp_transform_function): Make descriptors variable garbage
37052 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37054 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37055 * config/i386/avx512dqintrin.h: Ditto.
37056 * config/i386/avx512fintrin.h: Ditto.
37057 * config/i386/i386.c: Handle new builtins.
37058 * config/i386/i386-builtin.def: Add new builtins.
37059 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37060 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37062 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37063 Martin Liska <mliska@suse.cz>
37065 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37066 * asan.c (asan_expand_poison_ifn): Support stores and use
37067 appropriate ASAN report function.
37068 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37069 * internal-fn.def (ASAN_POISON_USE): Declare.
37070 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37071 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37072 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37073 ASAN_POISON calls w/o LHS.
37074 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37075 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37076 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37077 * gimplify.c (asan_poison_variables): Add attribute
37078 use_after_scope_memory to variables that really needs to live
37080 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37081 having the attribute.
37083 2017-01-23 Martin Liska <mliska@suse.cz>
37085 * asan.c (create_asan_shadow_var): New function.
37086 (asan_expand_poison_ifn): Likewise.
37087 * asan.h (asan_expand_poison_ifn): New declaration.
37088 * internal-fn.c (expand_ASAN_POISON): Likewise.
37089 * internal-fn.def (ASAN_POISON): New builtin.
37090 * sanopt.c (pass_sanopt::execute): Expand
37091 asan_expand_poison_ifn.
37092 * tree-inline.c (copy_decl_for_dup_finish): Make function
37094 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37095 * tree-ssa.c (is_asan_mark_p): New function.
37096 (execute_update_addresses_taken): Rewrite local variables
37097 (identified just by use-after-scope as addressable) into SSA.
37099 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37101 * doc/install.texi (Specific): opensource.apple.com uses https
37102 now. Remove trailing slash.
37104 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37106 * README.Portability: Remove note on an Irix compatibility issue.
37108 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37110 * gcov.c (INCLUDE_ALGORITHM): Define.
37111 (INCLUDE_VECTOR): Define.
37112 No longer include <vector> and <algorithm> directly.
37114 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37116 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37118 * doc/invoke.texi (Code Gen Options): Ditto.
37120 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37123 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37125 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37127 rtl-optimization/79125
37128 * cprop.c (local_cprop_pass): Handle cases where we make an
37129 unconditional trap.
37131 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37135 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37136 read from, for big endian.
37138 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37140 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37141 register pauth builtins for LP64 only.
37143 2017-01-20 Marek Polacek <polacek@redhat.com>
37146 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37149 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37151 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37153 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37154 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37155 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37157 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37160 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37161 in big-endian mode when they are not a single duplicated value.
37163 2017-01-20 Richard Biener <rguenther@suse.de>
37165 * BASE-VER: Bump to 7.0.1.
37167 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37169 * omp-low.c (omplow_simd_context): New struct. Use it...
37170 (lower_rec_simd_input_clauses): ...here and...
37171 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37172 references to idx, lane, max_vf, is_simt.
37174 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37176 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37179 2017-01-20 Martin Jambor <mjambor@suse.cz>
37181 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37182 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37184 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37185 (GTFILES): Rename hsa.c to hsa-common.c.
37186 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37187 * hsa-dump.c: Likewise.
37188 * hsa-gen.c: Likewise.
37189 * hsa-regalloc.c: Likewise.
37190 * ipa-hsa.c: Likewise.
37191 * omp-expand.c: Likewise.
37192 * omp-low.c: Likewise.
37193 * toplev.c: Likewise.
37195 2017-01-20 Marek Polacek <polacek@redhat.com>
37198 * doc/invoke.texi: Document -Wduplicated-branches.
37199 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37200 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37201 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37202 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37203 return 0 only when not OEP_LEXICOGRAPHIC.
37204 (fold_build_cleanup_point_expr): Use the expression
37205 location when building CLEANUP_POINT_EXPR.
37206 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37207 * tree.c (add_expr): Handle error_mark_node.
37209 2017-01-20 Martin Liska <mliska@suse.cz>
37212 * tree-profile.c (init_ic_make_global_vars): Do not call
37214 (gimple_init_gcov_profiler): Likewise.
37216 2017-01-20 Martin Liska <mliska@suse.cz>
37219 * cgraph.h (maybe_create_reference): Remove argument and
37221 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37223 * ipa-cp.c (create_specialized_node): Likewise.
37224 * symtab.c (symtab_node::maybe_create_reference): Handle
37225 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37227 2017-01-20 Martin Liska <mliska@suse.cz>
37229 * read-rtl-function.c (function_reader::create_function): Use
37230 build_decl instread of build_decl_stat.
37232 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37234 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37235 * config/i386/avx512dqintrin.h: Ditto.
37236 * config/i386/avx512fintrin.h: Ditto.
37237 * config/i386/i386-builtin-types.def: Add new types.
37238 * config/i386/i386.c: Handle new types.
37239 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37240 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37241 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37242 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37243 (__builtin_ia32_kshiftridi): New.
37244 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37246 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37250 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37251 define to rs6000_init_stack_protect_guard.
37252 (rs6000_init_stack_protect_guard): New function.
37254 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37255 Yunqiang Su <yunqiang.su@imgtec.com>
37257 * config.gcc (supported_defaults): Add madd4.
37258 (with_madd4): Add validation.
37259 (all_defaults): Add madd4.
37260 * config/mips/mips.opt (mmadd4): New option.
37261 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37263 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37265 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37266 (ISA_HAS_FUSED_MADD4): Likewise.
37267 * doc/invoke.texi (-mmadd4): Document the new option.
37268 * doc/install.texi (--with-madd4): Document the new option.
37270 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37272 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37273 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37274 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37275 (aarch64_init_pauth_hint_builtins): New.
37276 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37277 (aarch64_expand_builtin): Expand new builtins.
37279 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37281 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37282 * combine-stack-adj.c (no_unhandled_cfa): Handle
37283 REG_CFA_TOGGLE_RA_MANGLE.
37284 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37285 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37286 info for return address signing.
37287 (aarch64_expand_epilogue): Likewise.
37289 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37291 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37292 * config/aarch64/aarch64-protos.h
37293 (aarch64_return_address_signing_enabled): New declaration.
37294 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37296 (aarch64_expand_prologue): Sign return address before it's pushed onto
37298 (aarch64_expand_epilogue): Authenticate return address fetched from
37300 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37301 (aarch64_attributes): New function attributes for "sign-return-address".
37302 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37303 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37304 ("*do_return"): Generate combined instructions according to key index.
37305 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37306 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37308 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37309 * config/aarch64/aarch64.opt (msign-return-address=): New.
37310 * doc/extend.texi (AArch64 Function Attributes): Documents
37311 "sign-return-address=".
37312 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37314 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37316 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37317 overall option summary.
37319 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37321 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37322 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37323 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37324 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37326 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37328 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37329 -mpower9-minmax by default for -mcpu=power9.
37330 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37331 128-bit floating point.
37333 2017-01-20 Alan Modra <amodra@gmail.com>
37335 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37336 optimizing for size.
37338 2017-01-20 Alan Modra <amodra@gmail.com>
37341 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37342 for strcmp and strncmp from corresponding builtin decl.
37344 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37346 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37347 instead of i386/rtems-64.h.
37348 * config/i386/rtems-64.h: Remove.
37350 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37354 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37356 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37358 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37360 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37361 Change int to HOST_WIDE_INT.
37362 * config/aarch64/aarch64-protos.h
37363 (aarch64_simd_gen_const_vector_dup): Likewise.
37364 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37366 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37368 * langhooks-def.h (lhd_type_for_size): New decl.
37369 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37370 * langhooks.c (lhd_type_for_size): New function, taken from
37373 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37375 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37376 define_bypass for CR latency.
37377 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37378 (power9-alu2): Add define_bypass for CR latency.
37380 (power9-mul): Update insn latency.
37381 (power9-mul-compare): Update insn latency, bypass latency and remove
37384 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37386 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37388 * config/aarch64/aarch64.md
37389 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37390 aarch64_nopcrelative_literal_loads.
37391 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37393 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37395 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37396 TARGET_LOONGSON_3A.
37397 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37399 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37402 * config.gcc (supported_defaults): Add lxc1-sxc1.
37403 (with_lxc1_sxc1): Add validation.
37404 (all_defaults): Add lxc1-sxc1.
37405 * config/mips/mips.opt (mlxc1-sxc1): New option.
37406 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37408 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37409 __mips_no_lxc1_sxc1.
37410 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37411 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37412 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37414 2017-01-19 Richard Biener <rguenther@suse.de>
37416 PR tree-optimization/72488
37417 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37418 sure to restore SSA info.
37419 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37421 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37423 PR rtl-optimization/79121
37424 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37425 of the inner type when shifting an extended value.
37427 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37430 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37431 interposable aliases.
37433 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37436 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37437 Use the evmergelohi instruction.
37438 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37439 (mov_sitf_e500_subreg8_2_be): Likewise.
37440 (mov_sitf_e500_subreg12_2_le): Likewise.
37441 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37442 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37443 (mov_sitf_e500_subreg8_2_le): Likewise.
37444 (mov_sitf_e500_subreg12_2_be): Likewise.
37446 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37448 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37449 attribute from vecsimple to vecperm.
37450 (altivec_vbpermq2): Likewise.
37452 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37455 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37457 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37458 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37459 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37460 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37461 case where N arg is SIZE_MAX.
37462 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37463 (cmpstrsi): Add pattern.
37465 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37467 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37468 __builtin_vec_revb builtins.
37469 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37470 built-in functions to support generation of the ISA 3.0 XXBR<x>
37471 vector byte reverse instructions.
37472 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37473 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37474 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37475 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37476 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37477 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37478 (P9V_BUILTIN_VEC_REVB): Likewise.
37479 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37480 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37481 (p9_xxbrq_v16qi): Likewise.
37482 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37483 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37484 (p9_xxbrh_v8hi): Likewise.
37485 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37486 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37487 vec_revb built-in functions.
37489 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37491 PR rtl-optimization/78952
37492 * config/i386/i386.md (any_extract): New code iterator.
37493 (*insvqi_2): Use any_extract for source operand.
37494 (*insvqi_3): Use any_shiftrt for source operand.
37496 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37498 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37500 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37502 2017-01-18 Matthias Klose <doko@ubuntu.com>
37504 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37506 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37508 * config/rs6000/altivec.h (vec_bperm): Change #define.
37509 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37510 (altivec_vbpermq2): New define_insn.
37511 (altivec_vbpermd): Likewise.
37512 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37513 function interface.
37514 (VBPERMD): Likewise.
37515 (VBPERM): New polymorphic function interface.
37516 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37517 Add entries for P9V_BUILTIN_VEC_VBPERM.
37518 * doc/extend.texi: Add interfaces for vec_bperm.
37520 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37522 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37523 first letter of error messages.
37524 (s390_resolve_overloaded_builtin): Likewise.
37525 * config/s390/s390.c (s390_expand_builtin): Likewise.
37526 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37527 (s390_valid_target_attribute_inner_p): Likewise.
37528 * config/s390/s390.md ("tabort"): Likewise.
37530 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37532 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37533 (ISA_AVOID_DIV_HILO): New macro.
37534 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37535 (ISA_HAS_DDIV): Likewise.
37537 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37539 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37541 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37543 * doc/invoke.texi (fabi-version): Spelling fix.
37545 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37548 * doc/invoke.texi (fabi-version): Mention mangling fix for
37551 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37554 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37556 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37559 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37560 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37562 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37564 (stack_protect_test): Ditto.
37565 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37566 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37568 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37569 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37570 -mstack-protector-guard-offset=.
37571 (RS/6000 and PowerPC Options): Ditto.
37573 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37575 * config/i386/i386.h (MASK_CLASS_P): New define.
37576 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37577 there are no registers from different register sets also when
37578 mask registers are used. Update function comment.
37579 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37580 to (*k/*r) and (*k/*km) alternatives.
37582 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37584 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37585 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37586 (EH_RETURN_HANDLER_RTX): New define.
37587 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37588 Force frame pointer in EH return functions.
37589 (aarch64_expand_epilogue): Add barrier for eh_return.
37590 (aarch64_final_eh_return_addr): Remove.
37591 (aarch64_eh_return_handler_rtx): New function.
37592 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37594 (aarch64_eh_return_handler_rtx): New prototype.
37596 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37598 * config/rs6000/altivec.h (vec_rlmi): New #define.
37599 (vec_vrlnm): Likewise.
37600 (vec_rlnm): Likewise.
37601 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37602 (UNSPEC_VRLNM): Likewise.
37603 (VIlong): New mode iterator.
37604 (altivec_vrl<VI_char>mi): New define_insn.
37605 (altivec_vrl<VI_char>nm): Likewise.
37606 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37608 (VRLDNM): Likewise.
37609 (RLNM): New polymorphic function entry.
37610 (VRLWMI): New monomorphic function entry.
37611 (VRLDMI): Likewise.
37612 (RLMI): New polymorphic function entry.
37613 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37614 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37615 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37618 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37621 * dwarf2out.c (field_byte_offset): Restore the
37622 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37623 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37624 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37627 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37630 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37632 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37635 * dwarf2out.c (add_data_member_location_attribute): For constant
37636 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37637 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37638 DW_AT_byte_size attributes.
37640 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37642 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37643 after forcing to constant memory when the code model is medium.
37645 2017-01-17 Julia Koval <julia.koval@intel.com>
37648 * config/i386/avx512fintrin.h
37649 (_mm512_i32gather_ps): Change __addr type to void const*.
37650 (_mm512_mask_i32gather_ps): Ditto.
37651 (_mm512_i32gather_pd): Ditto.
37652 (_mm512_mask_i32gather_pd): Ditto.
37653 (_mm512_i64gather_ps): Ditto.
37654 (_mm512_mask_i64gather_ps): Ditto.
37655 (_mm512_i64gather_pd): Ditto.
37656 (_mm512_mask_i64gather_pd): Ditto.
37657 (_mm512_i32gather_epi32): Ditto.
37658 (_mm512_mask_i32gather_epi32): Ditto.
37659 (_mm512_i32gather_epi64): Ditto.
37660 (_mm512_mask_i32gather_epi64): Ditto.
37661 (_mm512_i64gather_epi32): Ditto.
37662 (_mm512_mask_i64gather_epi32): Ditto.
37663 (_mm512_i64gather_epi64): Ditto.
37664 (_mm512_mask_i64gather_epi64): Ditto.
37665 (_mm512_i32scatter_ps): Change __addr type to void*.
37666 (_mm512_mask_i32scatter_ps): Ditto.
37667 (_mm512_i32scatter_pd): Ditto.
37668 (_mm512_mask_i32scatter_pd): Ditto.
37669 (_mm512_i64scatter_ps): Ditto.
37670 (_mm512_mask_i64scatter_ps): Ditto.
37671 (_mm512_i64scatter_pd): Ditto.
37672 (_mm512_mask_i64scatter_pd): Ditto.
37673 (_mm512_i32scatter_epi32): Ditto.
37674 (_mm512_mask_i32scatter_epi32): Ditto.
37675 (_mm512_i32scatter_epi64): Ditto.
37676 (_mm512_mask_i32scatter_epi64): Ditto.
37677 (_mm512_i64scatter_epi32): Ditto.
37678 (_mm512_mask_i64scatter_epi32): Ditto.
37679 (_mm512_i64scatter_epi64): Ditto.
37680 (_mm512_mask_i64scatter_epi64): Ditto.
37681 * config/i386/avx512pfintrin.h
37682 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37683 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37684 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37685 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37686 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37687 (_mm512_prefetch_i32scatter_ps): Ditto.
37688 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37689 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37690 (_mm512_prefetch_i64scatter_pd): Ditto.
37691 (_mm512_prefetch_i64scatter_ps): Ditto.
37692 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37693 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37694 * config/i386/avx512vlintrin.h
37695 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37696 (_mm_mmask_i32gather_ps): Ditto.
37697 (_mm256_mmask_i32gather_pd): Ditto.
37698 (_mm_mmask_i32gather_pd): Ditto.
37699 (_mm256_mmask_i64gather_ps): Ditto.
37700 (_mm_mmask_i64gather_ps): Ditto.
37701 (_mm256_mmask_i64gather_pd): Ditto.
37702 (_mm_mmask_i64gather_pd): Ditto.
37703 (_mm256_mmask_i32gather_epi32): Ditto.
37704 (_mm_mmask_i32gather_epi32): Ditto.
37705 (_mm256_mmask_i32gather_epi64): Ditto.
37706 (_mm_mmask_i32gather_epi64): Ditto.
37707 (_mm256_mmask_i64gather_epi32): Ditto.
37708 (_mm_mmask_i64gather_epi32): Ditto.
37709 (_mm256_mmask_i64gather_epi64): Ditto.
37710 (_mm_mmask_i64gather_epi64): Ditto.
37711 (_mm256_i32scatter_ps): Change __addr type to void*.
37712 (_mm256_mask_i32scatter_ps): Ditto.
37713 (_mm_i32scatter_ps): Ditto.
37714 (_mm_mask_i32scatter_ps): Ditto.
37715 (_mm256_i32scatter_pd): Ditto.
37716 (_mm256_mask_i32scatter_pd): Ditto.
37717 (_mm_i32scatter_pd): Ditto.
37718 (_mm_mask_i32scatter_pd): Ditto.
37719 (_mm256_i64scatter_ps): Ditto.
37720 (_mm256_mask_i64scatter_ps): Ditto.
37721 (_mm_i64scatter_ps): Ditto.
37722 (_mm_mask_i64scatter_ps): Ditto.
37723 (_mm256_i64scatter_pd): Ditto.
37724 (_mm256_mask_i64scatter_pd): Ditto.
37725 (_mm_i64scatter_pd): Ditto.
37726 (_mm_mask_i64scatter_pd): Ditto.
37727 (_mm256_i32scatter_epi32): Ditto.
37728 (_mm256_mask_i32scatter_epi32): Ditto.
37729 (_mm_i32scatter_epi32): Ditto.
37730 (_mm_mask_i32scatter_epi32): Ditto.
37731 (_mm256_i32scatter_epi64): Ditto.
37732 (_mm256_mask_i32scatter_epi64): Ditto.
37733 (_mm_i32scatter_epi64): Ditto.
37734 (_mm_mask_i32scatter_epi64): Ditto.
37735 (_mm256_i64scatter_epi32): Ditto.
37736 (_mm256_mask_i64scatter_epi32): Ditto.
37737 (_mm_i64scatter_epi32): Ditto.
37738 (_mm_mask_i64scatter_epi32): Ditto.
37739 (_mm256_i64scatter_epi64): Ditto.
37740 (_mm256_mask_i64scatter_epi64): Ditto.
37741 (_mm_i64scatter_epi64): Ditto.
37742 (_mm_mask_i64scatter_epi64): Ditto.
37743 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37744 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37745 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37746 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37747 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37748 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37749 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37750 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37751 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37752 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37753 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37754 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37755 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37756 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37757 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37758 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37759 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37760 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37761 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37762 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37763 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37764 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37765 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37766 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37767 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37768 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37769 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37770 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37771 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37772 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37773 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37774 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37775 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37776 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37777 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37778 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37779 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37780 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37781 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37782 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37783 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37784 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37785 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37786 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37787 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37788 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37789 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37790 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37791 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37792 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37793 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37794 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37795 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37796 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37797 definitions accordingly.
37799 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37800 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37803 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37806 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37809 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37810 hard regs for inner regno.
37812 2017-01-17 Martin Liska <mliska@suse.cz>
37815 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37816 assumption and add comment.
37818 2017-01-17 Nathan Sidwell <nathan@acm.org>
37820 * ipa-visibility.c (localize_node): New function, broken out of ...
37821 (function_and_variable_visibility): ... here. Call it.
37823 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37825 PR middle-end/77445
37826 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37827 correctly set frequency of oudgoing edge.
37828 (duplicate_thread_path): Fix profile updating.
37830 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37833 * configure.ac: Add GCC_BASE_VER.
37834 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37835 version from BASE-VER file.
37836 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37837 (gcc.o): Depend on $(BASEVER).
37838 * common.opt (dumpfullversion): New option.
37839 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37840 * doc/invoke.texi: Document -dumpfullversion.
37841 * doc/install.texi: Document --with-gcc-major-version-only.
37842 * configure: Regenerated.
37844 2017-01-17 Richard Biener <rguenther@suse.de>
37846 PR tree-optimization/71433
37847 * tree-vrp.c (register_new_assert_for): Merge same asserts
37848 on all incoming edges.
37849 (process_assert_insertions_for): Handle insertions at the
37852 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37854 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37855 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37857 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37860 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37863 2017-01-17 Alan Modra <amodra@gmail.com>
37866 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37867 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37868 symbolic stack limit when pic.
37870 2017-01-16 Martin Sebor <msebor@redhat.com>
37872 PR tree-optimization/78608
37873 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37875 2017-01-16 Jeff Law <law@redhat.com>
37878 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37879 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37880 for several include directories that may be relative to sysroot.
37881 * config/i386/x-mingw32 (gplus_includedir): Define.
37882 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37883 (native_system_includedir): Likewise.
37884 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37885 override if TARGET_SYSTEM_ROOT is defined.
37886 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37888 PR tree-optimization/79090
37889 PR tree-optimization/33562
37890 PR tree-optimization/61912
37891 PR tree-optimization/77485
37892 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37893 and computed trims into the dump file.
37895 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37897 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37899 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37902 * gimplify.c (gimplify_init_constructor): If want_value and
37903 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37907 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37908 sequence. Formatting fixes.
37909 (doloop_optimize): Formatting fixes.
37912 * gcc.c (debug_level_greater_than_spec_func): New function.
37913 (static_spec_functions): Add debug-level-gt spec function.
37914 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37916 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37917 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37918 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37919 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37920 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37921 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37923 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37925 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37926 QImode fixups to general and mask registers only.
37928 2017-01-16 Carl Love <cel@us.ibm.com>
37930 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37931 for built-in functions
37932 vector signed char vec_nabs (vector signed char)
37933 vector signed short vec_nabs (vector signed short)
37934 vector signed int vec_nabs (vector signed int)
37935 vector signed long long vec_nabs (vector signed long long)
37936 vector float vec_nabs (vector float)
37937 vector double vec_nabs (vector double)
37938 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37940 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37941 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37942 * doc/extend.texi: Update the documentation file for the new built-in
37945 2017-01-16 Martin Sebor <msebor@redhat.com>
37947 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37950 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37952 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37953 UNSPEC_VSX__XXSPLTD to require special splat handling.
37955 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37958 * system.h: Poison strndup.
37960 2017-01-16 Alan Modra <amodra@gmail.com>
37963 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37966 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37968 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37970 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37972 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37973 call recog here. Assert that INSN_CODE (insn) is non-negative.
37975 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37978 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37980 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37981 in the currently scheduled RTL fragment.
37983 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37985 PR rtl-optimization/78751
37986 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37989 2017-01-14 Jeff Law <law@redhat.com>
37991 PR tree-optimization/79090
37992 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37993 variable length stores.
37994 (compute_trims): Delete dead assignment to *trim_tail.
37995 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37998 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38000 PR rtl-optimization/78626
38001 PR rtl-optimization/78727
38002 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38003 of a block, and split such blocks after everything else is finished.
38005 2017-01-14 Alan Modra <amodra@gmail.com>
38008 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38009 target legitimate_combined_insn.
38010 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38011 (rs6000_legitimate_combined_insn): New function.
38012 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38014 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38015 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38016 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38018 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38020 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38022 2017-01-13 Jeff Law <law@redhat.com>
38024 PR tree-optimization/33562
38025 PR tree-optimization/61912
38026 PR tree-optimization/77485
38027 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38029 (delete_dead_assignment): Likewise.
38030 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38031 statement to delete_dead_call and delete_dead_assignment.
38033 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38036 * substring-locations.c (format_warning_va): Strengthen case 1 so
38037 that both endpoints of the substring must be within the format
38038 range for just the substring to be printed.
38040 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38042 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38043 * config/i386/i386.c (ix86_target_string): Add missing options
38044 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38045 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38046 flags_other and ix86_target_other to flags2_other. Display unknown
38048 (ix86_valid_target_attribute_inner_p): Add missing options and
38049 reorder options by implied ISAs, as in ix86_target_string.
38051 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38053 * hash-table.h (hash_table::too_empty_p): New function.
38054 (hash_table::expand): Use it.
38055 (hash_table::traverse): Likewise.
38056 (hash_table::empty_slot): Use sizeof (value_type) instead of
38057 sizeof (PTR) to convert bytes to elements. Shrink the table
38058 if the current size is excessive for the current number of
38061 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38063 * ira-costs.c (record_reg_classes): Break from the inner loop
38064 early once alt_fail is known to be true. Update outer loop
38065 handling accordingly.
38067 2017-01-13 Jeff Law <law@redhat.com>
38069 * tree-ssa-dse.c (decrement_count): New function.
38070 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38071 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38072 when we know the partially dead statement is a mem* function.
38074 PR tree-optimization/61912
38075 PR tree-optimization/77485
38076 * tree-ssa-dse.c: Include expr.h.
38077 (maybe_trim_constructor_store): New function.
38078 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38080 PR tree-optimization/33562
38081 PR tree-optimization/61912
38082 PR tree-optimization/77485
38083 * doc/invoke.texi: Document new dse-max-object-size param.
38084 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38085 * tree-ssa-dse.c: Include params.h.
38086 (dse_store_status): New enum.
38087 (initialize_ao_ref_for_dse): New, partially extracted from
38089 (valid_ao_ref_for_dse, normalize_ref): New.
38090 (setup_live_bytes_from_ref, compute_trims): Likewise.
38091 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38092 (maybe_trim_partially_dead_store): Likewise.
38093 (maybe_trim_complex_store): Likewise.
38094 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38095 Track what bytes live from the original store. Return tri-state
38096 for dead, partially dead or live.
38097 (dse_dom_walker): Add constructor, destructor and new private members.
38098 (delete_dead_call, delete_dead_assignment): New extracted from
38100 (dse_optimize_stmt): Make a member of dse_dom_walker.
38101 Use initialize_ao_ref_for_dse.
38103 PR tree-optimization/33562
38104 PR tree-optimization/61912
38105 PR tree-optimization/77485
38106 * sbitmap.h (bitmap_count_bits): Prototype.
38107 (bitmap_clear_range, bitmap_set_range): Likewise.
38108 * sbitmap.c (bitmap_clear_range): New function.
38109 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38111 2017-01-13 Martin Liska <mliska@suse.cz>
38114 * function.c (set_cfun): Add new argument force.
38115 * function.h (set_cfun): Likewise.
38116 * ipa-inline-transform.c (inline_call): Use the function when
38117 strict alising from is dropped for function we inline to.
38119 2017-01-13 Richard Biener <rguenther@suse.de>
38121 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38122 for dumping GIMPLE INTEGER_CSTs.
38124 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38126 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38127 to 201112L since C++17.
38129 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38132 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38133 if -fsanitize=kernel-address is present.
38135 2017-01-13 Richard Biener <rguenther@suse.de>
38137 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38138 as _Literal ( type ) number in case usual suffixes do not
38139 preserve all information.
38141 2017-01-13 Richard Biener <rguenther@suse.de>
38143 PR tree-optimization/77283
38144 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38145 and ssa-iterators.h.
38146 (is_feasible_trace): Implement a cost model based on joiner
38149 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38152 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38153 char or short to __float128/_Float128 directly.
38155 2017-01-12 Martin Sebor <msebor@redhat.com>
38157 to -Wformat-overflow.
38158 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38159 (min_bytes_remaining): Same.
38160 (get_string_length): Same.
38161 (format_string): Same.
38162 (format_directive): Same.
38164 (pass_sprintf_length::handle_gimple_call): Same.
38166 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38168 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38169 info.nowrite calls with no lhs that can't throw. Return bool
38170 whether gsi_remove has been called or not.
38171 (pass_sprintf_length::handle_gimple_call): Return bool whether
38172 try_substitute_return_value called gsi_remove. Formatting fix.
38173 (pass_sprintf_length::execute): Don't use gsi_remove if
38174 handle_gimple_call returned true.
38177 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38178 be removed due to side-effects, don't remove following barrier nor
38179 turn the successor edge into fallthru edge.
38181 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38184 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38185 element-reversing loads and stores as not swappable.
38187 2017-01-12 Nathan Sidwell <nathan@acm.org>
38188 Nicolai Stange <nicstange@gmail.com>
38190 * combine.c (try_combine): Don't ignore result of overlap checking
38191 loop. Combine overlap & asm check into single loop.
38193 2017-01-12 Richard Biener <rguenther@suse.de>
38195 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38196 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38198 2017-01-12 Richard Biener <rguenther@suse.de>
38200 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38201 and TS_TARGET_OPTION directly derive from TS_BASE.
38202 * tree-core.h (tree_optimization_option): Derive from tree_base.
38203 (tree_target_option): Likewise.
38205 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38207 * config/i386/i386.c (memory_address_length): Increase len
38208 only when rip_relative_addr_p returns false.
38210 2017-01-11 Julia Koval <julia.koval@intel.com>
38212 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38213 (OPTION_MASK_ISA_SGX_SET): New.
38214 (ix86_handle_option): Handle OPT_msgx.
38215 * config.gcc: Added sgxintrin.h.
38216 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38217 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38218 * config/i386/i386.c (ix86_target_string): Add -msgx.
38220 (ix86_option_override_internal): Handle new options.
38221 (ix86_valid_target_attribute_inner_p): Add sgx.
38222 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38223 * config/i386/i386.opt: Add msgx.
38224 * config/i386/sgxintrin.h: New file.
38225 * config/i386/x86intrin.h: Add sgxintrin.h.
38227 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38230 * fold-const.c (maybe_nonzero_address): Return 1 for function
38232 (tree_single_nonzero_warnv_p): Don't handle function local objects
38236 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38239 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38242 * opts.c: Include "spellcheck.h"
38243 (struct string_fragment): New struct.
38244 (struct edit_distance_traits<const string_fragment &>): New
38246 (get_closest_sanitizer_option): New function.
38247 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38249 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38251 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38253 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38254 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38255 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38256 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38257 for initial die_offset if dwarf_split_debug_info.
38258 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38259 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38261 (output_skeleton_debug_sections): Formatting fix. Use
38262 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38263 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38265 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38267 * config/arm/cortex-a53.md: Add bypasses for
38268 cortex_a53_r2f_cvt.
38269 (cortex_a53_r2f): Only use for transfers.
38270 (cortex_a53_f2r): Likewise.
38271 (cortex_a53_r2f_cvt): Add reservation for conversions.
38272 (cortex_a53_f2r_cvt): Likewise.
38274 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38276 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38277 to all inlined functions, change static to extern.
38279 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38282 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38284 (arm_assemble_integer): Likewise.
38286 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38288 * config.gcc: Use new awk script to check CPU, FPU and architecture
38289 parameters for --with-... options.
38290 * config/arm/parsecpu.awk: New file
38291 * config/arm/arm-cpus.in: New file.
38292 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38294 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38296 * config/arm/t-arm: Update dependency rules.
38297 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38298 of processing .def files.
38299 * config/arm/genopt.sh: Deleted.
38300 * config/arm/gentune.sh: Deleted.
38301 * config/arm/arm-cores.def: Deleted.
38302 * config/arm/arm-arches.def: Deleted.
38303 * config/arm/arm-fpus.def: Deleted.
38304 * config/arm/arm-tune.md: Regenerated.
38305 * config/arm/arm-tables.opt: Regenerated.
38306 * config/arm/arm-cpu.h: New generated file.
38307 * config/arm/arm-cpu-data.h: New generated file.
38308 * config/arm/arm-cpu-cdata.h: New generated file.
38310 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38313 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38315 (input_varpool_node): Unpack dynamically_initialized bit.
38317 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38319 PR rtl-optimization/79032
38320 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38321 the alignment of the adjusted memory reference against that of MODE,
38322 instead of the alignment of the original memory reference.
38324 2017-01-11 Martin Jambor <mjambor@suse.cz>
38326 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38328 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38329 decorated functions.
38331 2017-01-11 Richard Biener <rguenther@suse.de>
38333 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38334 set range/nonnull info for PHI results. Do not set it on
38335 stmts marked for removal.
38337 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38339 * expr.c (store_field): In the bitfield case, fetch the return value
38340 from the registers before applying a single big-endian adjustment.
38341 Always do a final load for a BLKmode value not larger than a word.
38343 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38346 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38347 that we correctly handle column numbers greater than
38348 LINE_MAP_MAX_COLUMN_NUMBER.
38350 2017-01-10 Martin Sebor <msebor@redhat.com>
38352 PR middle-end/78245
38353 * gimple-ssa-sprintf.c (get_destination_size): Call
38354 {init,fini}object_sizes.
38355 * tree-object-size.c (addr_object_size): Adjust.
38356 (pass_through_call): Adjust.
38357 (pass_object_sizes::execute): Adjust.
38358 * tree-object-size.h (fini_object_sizes): Declare.
38360 2017-01-10 Martin Sebor <msebor@redhat.com>
38362 PR tree-optimization/78775
38363 * builtins.c (get_size_range): Move...
38364 * calls.c: ...to here.
38365 (alloc_max_size): Accept zero argument.
38366 (operand_signed_p): Remove.
38367 (maybe_warn_alloc_args_overflow): Call get_size_range.
38368 * calls.h (get_size_range): Declare.
38370 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38372 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38373 from TI's devices.csv file as of September 2016.
38374 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38376 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38378 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38379 * doc/invoke.texi: Likewise.
38380 * doc/md.texi: Likewise.
38381 * doc/objc.texi: Likewise.
38383 2017-01-10 Joshua Conner <joshconner@google.com>
38385 * config/arm/fuchsia-elf.h: New file.
38386 * config/fuchsia.h: New file.
38387 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38388 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38390 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38392 2016-01-10 Richard Biener <rguenther@suse.de>
38394 PR tree-optimization/79034
38395 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38396 Propagate out degenerate PHIs in the joiner.
38398 2017-01-10 Martin Liska <mliska@suse.cz>
38400 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38401 (sort_congruence_classes_by_decl_uid): Likewise.
38402 (sort_congruence_class_groups_by_decl_uid): Likewise.
38403 (sem_item_optimizer::merge_classes): Sort class, groups in these
38404 classes and members in the groups by DECL_UID of declarations.
38405 This would make merge operations stable.
38407 2017-01-10 Martin Liska <mliska@suse.cz>
38409 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38410 usage of m_classes_vec.
38411 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38412 (sem_item_optimizer::get_group_by_hash): Likewise.
38413 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38414 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38415 (sem_item_optimizer::verify_classes): Likewise.
38416 (sem_item_optimizer::process_cong_reduction): Likewise.
38417 (sem_item_optimizer::dump_cong_classes): Likewise.
38418 (sem_item_optimizer::merge_classes): Likewise.
38419 * ipa-icf.h (congruence_class_hash): Rename from
38420 congruence_class_group_hash. Remove declaration of m_classes_vec.
38422 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38424 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38425 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38426 * config.gcc: Add avx512vpopcntdqintrin.h.
38427 * config/i386/avx512vpopcntdqintrin.h: New.
38428 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38429 * config/i386/i386-builtin-types.def: Add new types.
38430 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38431 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38432 __builtin_ia32_vpopcountq_v8di_mask): New.
38433 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38434 __AVX512VPOPCNTDQ__.
38435 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38436 (PTA_AVX512VPOPCNTDQ): Define.
38437 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38438 TARGET_AVX512VPOPCNTDQ_P): Define.
38439 * config/i386/i386.opt: Add mavx512vpopcntdq.
38440 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38441 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38443 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38445 PR middle-end/77484
38446 * predict.def (PRED_CALL): Set to 67.
38448 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38450 * expr.c (store_field): In the bitfield case, if the value comes from
38451 a function call and is of an aggregate type returned in registers, do
38452 not modify the field mode; extract the value in all cases if the mode
38453 is BLKmode and the size is not larger than a word.
38455 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38458 * config/i386/cpuid.h: Fix undefined behavior.
38460 2017-01-04 Jeff Law <law@redhat.com>
38462 PR tree-optimization/79007
38463 PR tree-optimization/67955
38464 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38465 conservative for pt.null when flag_non_call_exceptions is on.
38467 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38469 PR translation/79019
38470 PR translation/79020
38471 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38472 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38473 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38475 * config/avr/avr.opt (maccumulate-args): Likewise.
38476 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38477 * common.opt (freport-bug): Likewise.
38478 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38479 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38480 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38481 translatable string.
38482 * config/i386/i386.c (function_value_32): Likewise.
38483 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38484 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38486 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38487 * common/config/msp430/msp430-common.c (msp430_handle_option):
38489 * symtab.c (symtab_node::verify_base): Likewise.
38490 * opts.c (set_debug_level): Likewise.
38491 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38492 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38493 missing whitespace to translatable strings.
38494 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38495 * config/sh/superh.h: Likewise.
38496 * config/i386/xopintrin.h: Likewise.
38497 * config/i386/znver1.md: Likewise.
38498 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38499 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38500 * double-int.h (struct double_int): Likewise.
38501 * double-int.c (div_and_round_double): Likewise.
38502 * wide-int.cc: Likewise.
38503 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38504 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38505 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38506 (crossjumps_occurred): ... this.
38507 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38510 PR tree-optimization/78899
38511 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38512 returning bool return struct loop *, NULL for failure and the new
38514 (versionable_outer_loop_p): Don't version outer loop if it has
38515 dont_vectorized bit set.
38516 (tree_if_conversion): When versioning outer loop, ensure
38517 tree_if_conversion is performed also on the inner loop of the
38518 non-vectorizable outer loop copy.
38519 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38520 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38521 prevent vectorization of it.
38522 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38523 the outer loop vectorization of the non-scalar version is attempted
38524 before vectorization of the inner loop in scalar version. If
38525 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38526 vectorization of its inner loop.
38527 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38528 has 2 inner loops, rename also on edges from bb whose single pred
38529 is outer_loop->header. Fix typo in function comment.
38531 2017-01-09 Martin Sebor <msebor@redhat.com>
38534 * asan.c (asan_emit_stack_protection): Increase local buffer size
38535 to avoid snprintf truncation warning.
38537 2017-01-09 Andrew Pinski <apinski@cavium.com>
38539 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38540 to reference thunderx2t99 for the tuning structure
38541 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38543 (thunderx2t99_extra_costs): This.
38544 * config/aarch64/aarch64-tune.md: Regenerate.
38545 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38546 (vulcan_addrcost_table): This.
38547 (vulcan_regmove_cost): Rename to ...
38548 (thunderx2t99_regmove_cost): This.
38549 (vulcan_vector_cost): Rename to ...
38550 (thunderx2t99_vector_cost): this.
38551 (vulcan_branch_cost): Rename to ...
38552 (thunderx2t99_branch_cost): This.
38553 (vulcan_tunings): Rename to ...
38554 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38555 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38557 2017-01-09 Martin Jambor <mjambor@suse.cz>
38561 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38562 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38563 (propagate_vr_accross_jump_function): Use the above function for all
38564 value range computations for pass-through jump functions and type
38565 converasion from explicit value range values.
38566 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38567 parameters from TYPE_ARG_TYPES.
38568 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38569 (ipa_write_node_info): Stream type of the actual argument.
38570 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38572 2017-01-09 Martin Liska <mliska@suse.cz>
38575 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38576 (lookup_compiler): Do not show error message with have_E.
38578 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38580 PR tree-optimization/78938
38581 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38582 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38583 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38584 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38587 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38589 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38592 2017-01-09 Richard Biener <rguenther@suse.de>
38594 PR tree-optimization/78997
38595 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38596 name condition properly.
38598 2017-01-09 Richard Biener <rguenther@suse.de>
38601 * dwarf2out.c (is_cxx): New overload with context.
38602 (is_naming_typedef_decl): Use it.
38604 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38606 * invoke.texi (Option Summary): Correct spacing in option lists
38607 and add line breaks to fix over-long lines.
38609 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38611 PR middle-end/17660
38613 * extend.texi (Common Variable Attributes): Add xref to GCC
38614 Internals manual to explain mode attribute keywords.
38616 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38619 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38620 and Preprocessor Options.
38621 (Options for Linking): Document -pthread here....
38622 (RS/6000 and PowerPC Options): ...not here.
38623 (Solaris 2 Options): ...or here.
38624 * doc/cppopts.texi: Document -pthread.
38626 2017-01-08 Martin Sebor <msebor@redhat.com>
38628 PR middle-end/77708
38629 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38630 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38631 New member functions.
38632 (format_directive): Used them.
38634 (pass_sprintf_length::handle_gimple_call): Same.
38635 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38636 to avoid truncation for any argument.
38637 (extract_affine_mul): Same.
38638 * tree.c (get_file_function_name): Same.
38640 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38642 PR middle-end/77484
38643 * predict.def (PRED_INDIR_CALL): Set to 86.
38645 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38647 PR preprocessor/54124
38648 * doc/cppopts.texi: Reformat -d subtable to list the full name
38649 of the options. Add cross-reference to the docs for the general
38650 compiler -d options.
38651 * doc/invoke.texi (Developer Options): Add cross-reference to the
38652 preprocessor-specific -d option documentation.
38654 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38656 PR preprocessor/13498
38657 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38658 redudant material, and reflect new command-line options.
38659 (System Headers): Likewise.
38661 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38663 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38664 -isystem, and -idirafter. Copy-edit.
38665 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38666 default for -ftrack-macro-expansion. Delete obsolete and
38667 badly-formatted implementation details about -fdebug-cpp output.
38668 * doc/cppwarnopts.texi: Copy-edit.
38670 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38673 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38674 that the transition from a max line width >= 1<<10 to narrower
38675 lines works correctly.
38677 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38679 * doc/options.texi (PerFunction): New.
38680 * opt-functions.awk (switch_flags): Map both Optimization and
38681 PerFunction to CL_OPTIMIZATION.
38682 * opth-gen.awk: Test for PerFunction flag along with
38684 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38685 it only when the latter is present. Skip those that don't in
38686 the hash function generator.
38687 * common.opt (fvar-tracking): Mark as PerFunction instead of
38689 (fvar-tracking-assignments): Likewise.
38690 (fvar-tracking-assignments-toggle): Likewise.
38691 (fvar-tracking-uninit): Likewise.
38693 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38695 PR translation/79018
38696 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38699 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38702 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38703 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38704 TARGET_LONG_JUMP_TABLE_OFFSETS.
38705 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38706 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38707 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38708 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38709 * config/m68k/m68k.md (tablejump expander): Likewise.
38710 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38711 TARGET_LONG_JUMP_TABLE_OFFSETS.
38712 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38713 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38715 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38716 David Holsgrove <david.holsgrove@xilinx.com>
38718 * common/config/microblaze/microblaze-common.c
38719 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38720 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38722 * config/microblaze/microblaze.c (microblaze_must_save_register)
38723 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38725 (microblaze_eh_return): New function.
38726 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38727 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38728 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38729 * config/microblaze/microblaze.md (eh_return): New pattern.
38731 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38733 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38734 GCC_DIAGNOSTIC_STRINGIFY): Define.
38736 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38738 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38740 * config/arm/arm.md (<mcrr>): New.
38742 * config/arm/arm.c (arm_arch5te): New.
38743 (arm_option_override): Set arm_arch5te.
38744 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38746 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38747 (arm_mcrr_qualifiers): ... this. New.
38748 (MRRC_QUALIFIERS): Define to...
38749 (arm_mrrc_qualifiers): ... this. New.
38750 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38752 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38753 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38754 (MRRCI, mrrc, MRRC): New.
38755 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38756 VUNSPEC_MRRC2): New.
38758 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38760 * config/arm/arm.md (<mcr>): New.
38762 * config/arm/arm.c (arm_coproc_builtin_available): Add
38763 support for mcr, mrc, mcr2 and mrc2.
38764 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38765 (arm_mcr_qualifiers): ... this. New.
38766 (MRC_QUALIFIERS): Define to ...
38767 (arm_mrc_qualifiers): ... this. New.
38768 (MCR_QUALIFIERS): Define to ...
38769 (arm_mcr_qualifiers): ... this. New.
38770 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38772 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38773 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38774 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38775 VUNSPEC_MRC2): New.
38777 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38779 * config/arm/arm.md (*ldc): New.
38783 * config/arm/arm.c (arm_coproc_builtin_available): Add
38784 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38785 (arm_coproc_ldc_stc_legitimate_address): New.
38786 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38787 'qualifier_const_pointer'.
38788 (LDC_QUALIFIERS): Define to...
38789 (arm_ldc_qualifiers): ... this. New.
38790 (STC_QUALIFIERS): Define to...
38791 (arm_stc_qualifiers): ... this. New.
38792 * config/arm/arm-protos.h
38793 (arm_coproc_ldc_stc_legitimate_address): New.
38794 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38795 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38796 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38797 stc2, stcl, stc2l): New.
38798 * config/arm/constraints.md (Uz): New.
38799 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38800 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38801 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38802 VUNSPEC_STC2L): New.
38804 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38806 * config/arm/arm.md (<cdp>): New.
38807 * config/arm/arm.c (neon_const_bounds): Rename this ...
38808 (arm_const_bounds): ... this.
38809 (arm_coproc_builtin_available): New.
38810 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38811 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38812 (CDP_QUALIFIERS): Define to...
38813 (arm_cdp_qualifiers): ... this. New.
38815 (arm_expand_builtin_args): Add case for 6 arguments.
38816 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38817 (arm_const_bounds): ... this.
38818 (arm_coproc_builtin_available): New.
38819 * config/arm/arm_acle.h (__arm_cdp): New.
38821 * config/arm/arm_acle_builtins.def (cdp): New.
38823 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38824 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38825 'arm_const_bounds'.
38826 * config/arm/types.md (coproc): New.
38827 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38828 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38829 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38830 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38832 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38834 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38835 (UBINOP_QUALIFIERS): New.
38837 (acle_builtin_data): New. Change comment.
38838 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38839 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38840 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38841 arm_acle_builtins.def.
38842 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38843 (arm_init_acle_builtins): New.
38844 (CRC32_BUILTIN): Remove.
38845 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38846 crc32cb, crc32ch and crc32cw.
38847 (arm_init_crc32_builtins): Remove.
38848 (arm_init_builtins): Use arm_init_acle_builtins rather
38849 than arm_init_crc32_builtins.
38850 (arm_expand_acle_builtin): New.
38851 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38852 * config/arm/arm_acle_builtins.def: New.
38854 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38856 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38857 (arm_builtin_datum): ... this.
38858 (arm_init_neon_builtin): Rename to ...
38859 (arm_init_builtin): ... this. Add a new parameters PREFIX
38860 and USE_SIG_IN_NAME.
38861 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38862 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38863 'arm_builtin_datum'.
38864 (arm_init_vfp_builtins): Likewise.
38865 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38866 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38867 (arm_expand_neon_args): Rename to ...
38868 (arm_expand_builtin_args): ... this. Rename builtin_arg
38869 enum values and differentiate between ARG_BUILTIN_MEMORY
38870 and ARG_BUILTIN_NEON_MEMORY.
38871 (arm_expand_neon_builtin_1): Rename to ...
38872 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38873 values, arm_expand_builtin_args and add bool parameter NEON.
38874 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38875 (arm_expand_vfp_builtin): Likewise.
38876 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38878 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38880 PR middle-end/77484
38881 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38882 * predict.c (tree_estimate_probability_bb): Reverse direction of
38883 polymorphic call predictor.
38885 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38887 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38888 (determine_pass_name_match): ...this new function and...
38889 (should_skip_pass_p): ...this new function.
38891 2017-01-06 Nathan Sidwell <nathan@acm.org>
38893 * ipa-visibility.c (function_and_variable_visibility): Reformat
38894 comments and long lines. Remove extrneous if.
38895 * symtab.c (symtab_node::make_decl_local): Fix code format.
38896 (symtab_node::set_section_for_node): Fix comment typo.
38898 2017-01-06 Martin Liska <mliska@suse.cz>
38901 * lra-constraints.c: Rename invariant to lra_invariant.
38902 * predict.c (set_even_probabilities): Initialize e to NULL.
38904 2017-01-05 Martin Sebor <msebor@redhat.com>
38906 PR tree-optimization/78910
38907 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38908 (format_integer): Correct off-by-one error in the handling
38909 of precision with negative numbers in signed conversions..
38911 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38913 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38915 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38917 PR tree-optimization/71016
38918 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38919 factor_out_conditional_conversion. Formatting fix.
38920 (factor_out_conditional_conversion): Add cond_stmt argument.
38921 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38922 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38925 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38927 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38928 read-rtl-function.o, and selftest-rtl.o.
38929 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38930 (selftest::aarch64_test_loading_full_dump): New function.
38931 (selftest::aarch64_run_selftests): New function.
38932 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38933 selftest::aarch64_run_selftests.
38934 * config/i386/i386.c
38935 (selftest::ix86_test_loading_dump_fragment_1): New function.
38936 (selftest::ix86_test_loading_call_insn): New function.
38937 (selftest::ix86_test_loading_full_dump): New function.
38938 (selftest::ix86_test_loading_unspec): New function.
38939 (selftest::ix86_run_selftests): Call the new functions.
38940 * emit-rtl.c (maybe_set_max_label_num): New function.
38941 * emit-rtl.h (maybe_set_max_label_num): New decl.
38942 * function.c (instantiate_decls): Guard call to
38943 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38944 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38946 * gensupport.c (gen_reader::gen_reader): Pass "false"
38947 for new "compact" param of rtx_reader.
38948 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38949 rather than an empty string for NULL strings.
38950 * read-md.c: Potentially include config.h rather than bconfig.h.
38951 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38952 (have_error): New global, copied from errors.c.
38953 (md_reader::read_name): Rename to...
38954 (md_reader::read_name_1): ...this, adding "out_loc" param,
38955 and converting "missing name or number" to returning false, rather
38957 (md_reader::read_name): Reimplement in terms of read_name_1.
38958 (md_reader::read_name_or_nil): New function.
38959 (md_reader::read_string): Handle "(nil)" by returning NULL.
38960 (md_reader::md_reader): Add new param "compact".
38961 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38962 (md_reader::read_file): New method.
38963 * read-md.h (md_reader::md_reader): Add new param "compact".
38964 (md_reader::read_file): New method.
38965 (md_reader::is_compact): New accessor.
38966 (md_reader::read_name): Convert return type from void to file_location.
38967 (md_reader::read_name_or_nil): New decl.
38968 (md_reader::read_name_1): New decl.
38969 (md_reader::m_compact): New field.
38970 (noop_reader::noop_reader): Pass "false" for new "compact" param
38972 (rtx_reader::rtx_reader): Add new "compact" param.
38973 (rtx_reader::read_rtx_operand): Make virtual and convert return
38974 type from void to rtx.
38975 (rtx_reader::read_until): New decl.
38976 (rtx_reader::handle_any_trailing_information): New virtual function.
38977 (rtx_reader::postprocess): New virtual function.
38978 (rtx_reader::finalize_string): New virtual function.
38979 (rtx_reader::m_in_call_function_usage): New field.
38980 (rtx_reader::m_reuse_rtx_by_id): New field.
38981 * read-rtl-function.c: New file.
38982 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38983 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38984 (selftest::verify_three_block_rtl_cfg): New decl.
38985 * read-rtl-function.h: New file.
38986 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38987 For host, include function.h, memmodel.h, and emit-rtl.h.
38988 (one_time_initialization): New function.
38989 (struct compact_insn_name): New struct.
38990 (compact_insn_names): New array.
38991 (find_code): Handle insn codes in compact dumps.
38992 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38993 (bind_subst_iter_and_attr): Likewise.
38994 (add_condition_to_string): Likewise.
38995 (add_condition_to_rtx): Likewise.
38996 (apply_attribute_uses): Likewise.
38997 (add_current_iterators): Likewise.
38998 (apply_iterators): Likewise.
38999 (initialize_iterators): Guard usage of apply_subst_iterator with
39000 #ifdef GENERATOR_FILE.
39001 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39002 (md_reader::read_mapping): Likewise.
39003 (add_define_attr_for_define_subst): Likewise.
39004 (add_define_subst_attr): Likewise.
39005 (read_subst_mapping): Likewise.
39006 (check_code_iterator): Likewise.
39007 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39009 (one_time_initialization): New function.
39010 (rtx_reader::read_until): New method.
39011 (read_flags): New function.
39012 (parse_reg_note_name): New function.
39013 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39014 Handle reuse_rtx ids.
39015 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39016 Add parsing support for RTL dumps, mirroring the special-cases in
39017 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39018 values, and calling handle_any_trailing_information.
39019 (rtx_reader::read_rtx_operand): Convert return type from void
39020 to rtx, returning return_rtx. Handle case 'e'. Call
39021 finalize_string on XSTR and XTMPL fields.
39022 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39023 "(nil)" values were omitted. Call the postprocess vfunc on the
39025 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39026 class ctor. Initialize m_in_call_function_usage. Call
39027 one_time_initialization.
39028 * rtl-tests.c (selftest::test_uncond_jump): Call
39029 set_new_first_and_last_insn.
39030 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39031 * selftest-rtl.c: New file.
39032 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39033 (selftest::get_insn_by_uid): New decl.
39034 * selftest-run-tests.c (selftest::run_tests): Call
39035 read_rtl_function_c_tests.
39036 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39037 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39040 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39042 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39043 operands in a special way. Assert that pos+len <= mode precision.
39045 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39047 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39048 3 argument Alias with unlimited for the negative form.
39049 (fno-vect-cost-model): Removed.
39051 2017-01-05 Martin Liska <mliska@suse.cz>
39053 * hsa-gen.c (gen_hsa_divmod): New function.
39054 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39056 2017-01-05 Martin Liska <mliska@suse.cz>
39059 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39062 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39064 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39065 small constant length operands.
39067 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39069 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39070 between loop iterations.
39072 2017-01-05 Martin Liska <mliska@suse.cz>
39075 * gimplify.c (gimplify_decl_expr): Compare to
39076 asan_poisoned_variables instread of checking flags.
39077 (gimplify_target_expr): Likewise.
39078 (gimplify_expr): Likewise.
39079 (gimplify_function_tree): Conditionally initialize
39080 asan_poisoned_variables.
39082 2017-01-04 Jeff Law <law@redhat.com>
39084 PR tree-optimizatin/78812
39085 * rtl.h (contains_mem_rtx_p): Prototype.
39086 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39087 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39088 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39089 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39090 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39092 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39094 * input.c (assert_char_at_range): Default-initialize actual_range.
39096 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39098 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39099 to match the caller.
39101 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39103 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39104 insns after final jump in test to emit dummy move.
39106 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39108 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39109 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39111 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39113 * multiple_target.c (create_dispatcher_calls): Init e_next.
39114 * tree-ssa-loop-split.c (split_loop): Init border.
39115 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39118 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39123 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39124 (altivec_register_operand): Do not return true if the operand
39125 contains a SUBREG mixing SImode and SFmode.
39126 (vsx_register_operand): Likewise.
39127 (vsx_reg_sfsubreg_ok): New predicate.
39128 (vfloat_operand): Do not return true if the operand contains a
39129 SUBREG mixing SImode and SFmode.
39130 (vint_operand): Likewise.
39131 (vlogical_operand): Likewise.
39132 (gpc_reg_operand): Likewise.
39133 (int_reg_operand): Likewise.
39134 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39135 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39136 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39138 (rs6000_emit_move_si_sf_subreg): New helper function.
39139 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39140 fixup SUBREGs involving SImode and SFmode.
39141 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39142 numbers for the new peephole2 optimization.
39143 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39144 the GLIBC math library that do AND/IOR/XOR operations on single
39145 precision floating point.
39146 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39147 target macros to say whether we need to avoid SUBREGs mixing
39149 (TARGET_ALLOW_SF_SUBREG): Likewise.
39150 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39151 (UNSPEC_SI_FROM_SF): Likewise.
39152 (iorxor): Change spacing.
39153 (and_ior_xor): New iterator for AND, IOR, and XOR.
39154 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39155 (movdi_from_sf_zero_ext): Likewise.
39156 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39157 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39158 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39159 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39160 (fms<mode>4): Likewise.
39161 (fnma<mode>4): Likewise.
39162 (fnms<mode>4): Likewise.
39163 (nfma<mode>4): Likewise.
39164 (nfms<mode>4): Likewise.
39166 2017-01-04 Marek Polacek <polacek@redhat.com>
39169 * doc/invoke.texi: Document -Wpointer-compare.
39171 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39173 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39176 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39177 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39180 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39183 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39184 documentation of the powerpc_popcntb_ok attribute.
39185 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39186 code to issue warning messages if a requested CPU configuration is
39187 not supported by the binary (assembler and loader) toolchain.
39188 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39189 made to define a built-in function that has been disabled.
39190 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39191 made to define a built-in function that has been disabled.
39192 (altivec_init_builtins): Add comment explaining why definition
39193 of the DST built-in functions is not preceded by an assertion
39194 check. Add assertions to prevent ICE if attempts are made to
39195 define an altivec predicate or an abs* built-in function that has
39197 (htm_init_builtins): Add comment explaining why definition of the
39198 htm built-in functions is not preceded by an assertion check.
39200 2017-01-04 Jeff Law <law@redhat.com>
39202 PR tree-optimizatin/67955
39203 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39204 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39205 the points-to solution does not include pt_null. Use DECL_PT_UID
39208 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39210 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39211 Use gen_int_mode instead of gen_lopwart for const_int operands.
39213 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39215 PR tree-optimization/71563
39216 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39217 out of range value - has low bits known to be zero.
39219 2017-01-04 Alan Modra <amodra@gmail.com>
39221 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39222 * configure: Regenerate.
39223 * config.in: Regenerate.
39225 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39228 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39229 a substring of the message, but strcmp with the whole message. Ifdef
39230 ENABLE_NLS, translate the message first using dgettext.
39232 2017-01-03 Jeff Law <law@redhat.com>
39234 PR tree-optimizatin/78856
39235 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39236 (mark_threaded_blocks): Remove code to truncate thread paths that
39237 cross multiple loop headers. Instead invalidate the cached loop
39238 iteration information and handle case of a thread path walking
39239 into an irreducible region.
39241 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39244 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39245 assertions. Add support for doing the signbit if the IEEE 128-bit
39246 floating point value is in a GPR.
39247 * config/rs6000/rs6000.md (Fsignbit): Delete.
39248 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39249 Update the length attribute if the value is in a GPR.
39250 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39251 the sign or zero extension instruction, since the value is always 0/1.
39252 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39255 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39256 extracting SImode to a GPR register so that we can generate a
39257 store, limit the vector to be in a traditional Altivec register
39258 for the vextuwrx instruction.
39260 2017-01-03 Ian Lance Taylor <iant@google.com>
39262 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39264 2017-01-03 Martin Sebor <msebor@redhat.com>
39266 PR tree-optimization/78696
39267 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39268 precision. Use MPFR for %f for greater fidelity. Correct handling
39270 (pass_sprintf_length::compute_format_length): Set width and precision
39271 specified by asrerisk to void_node for vararg functions.
39272 (try_substitute_return_value): Adjust dump output.
39274 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39276 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39278 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39280 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39281 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39282 -mlra/-mno-lra was passed to the compiler.
39284 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39286 PR rtl-optimization/65618
39287 * emit-rtl.c (try_split): Move initialization of "before" and
39288 "after" to just before the call to emit_insn_after_setloc.
39290 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39292 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39294 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39296 * dwarf2out.c (gen_enumeration_type_die): When
39297 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39299 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39301 PR tree-optimization/78965
39302 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39303 Change first argument from const call_info & to call_info &. For %n
39304 set info.nowrite to false.
39306 PR middle-end/78901
39307 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39308 possibly throwing calls.
39310 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39311 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39312 and fns handling, rather than in a separate case SSA_NAME.
39314 2017-01-02 Jeff Law <law@redhat.com>
39316 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39317 fixes for first_period and second_period variables.
39319 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39322 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39323 (*insvqi_1): New insn pattern.
39324 (*insvqi_1_mem_rex64): Ditto.
39325 (*insvqi_2): Ditto.
39326 (*insvqi_3): Rename from *insvqi.
39328 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39330 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39332 * doc/cfg.texi (Edges): Remove reference to Java.
39333 (Maintaining the CFG): Ditto.
39335 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39337 PR middle-end/77674
39338 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39339 transparent aliases.
39341 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39343 PR middle-end/77484
39344 * predict.def (PRED_CALL): Update hitrate.
39345 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39346 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39347 into direct/indirect/polymorphic variants.
39349 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39351 Update copyright years.
39353 * gcc.c (process_command): Update copyright notice dates.
39354 * gcov-dump.c (print_version): Ditto.
39355 * gcov.c (print_version): Ditto.
39356 * gcov-tool.c (print_version): Ditto.
39357 * gengtype.c (create_file): Ditto.
39358 * doc/cpp.texi: Bump @copying's copyright year.
39359 * doc/cppinternals.texi: Ditto.
39360 * doc/gcc.texi: Ditto.
39361 * doc/gccint.texi: Ditto.
39362 * doc/gcov.texi: Ditto.
39363 * doc/install.texi: Ditto.
39364 * doc/invoke.texi: Ditto.
39366 Copyright (C) 2017 Free Software Foundation, Inc.
39368 Copying and distribution of this file, with or without modification,
39369 are permitted in any medium without royalty provided the copyright
39370 notice and this notice are preserved.