1 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3 * doc/invoke.texi: Fix some typos.
5 2017-12-16 Martin Sebor <msebor@redhat.com>
7 PR tree-optimization/78918
8 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
9 * builtins.c (check_sizes): Rename...
10 (check_access): ...to this. Rename function arguments for clarity.
11 (check_memop_sizes): Adjust names.
12 (expand_builtin_memchr, expand_builtin_memcpy): Same.
13 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
14 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
15 (check_strncat_sizes, expand_builtin_strncat): Same.
16 (expand_builtin_strncpy, expand_builtin_memset): Same.
17 (expand_builtin_bzero, expand_builtin_memcmp): Same.
18 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
19 (maybe_emit_sprintf_chk_warning): Same.
20 (expand_builtin_strcpy): Adjust.
21 (expand_builtin_stpcpy): Same.
22 (expand_builtin_with_bounds): Detect out-of-bounds accesses
23 in pointer-checking forms of memcpy, memmove, and mempcpy.
24 (gcall_to_tree_minimal, max_object_size): Define new functions.
25 * builtins.h (max_object_size): Declare.
26 * calls.c (alloc_max_size): Call max_object_size instead of
27 hardcoding ssizetype limit.
28 (get_size_range): Handle new argument.
29 * calls.h (get_size_range): Add a new argument.
30 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
31 * doc/invoke.texi (-Wrestrict): Adjust, add example.
32 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
34 (gimple_fold_builtin_memory_chk): Same.
35 (gimple_fold_builtin_stxcpy_chk): New function.
36 * gimple-ssa-warn-restrict.c: New source.
37 * gimple-ssa-warn-restrict.h: New header.
38 * gimple.c (gimple_build_call_from_tree): Propagate location.
39 * passes.def (pass_warn_restrict): Add new pass.
40 * tree-pass.h (make_pass_warn_restrict): Declare.
41 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
43 (handle_builtin_strcat): Same.
44 (strlen_optimize_stmt): Rename...
45 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
46 stpncpy, strncpy, and their checking forms.
48 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
50 PR rtl-optimization/82849
51 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
52 and get_max_loop_iterations_int.
54 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
55 Alan Hayward <alan.hayward@arm.com>
56 David Sherwood <david.sherwood@arm.com>
58 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
59 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
60 size as a poly_uint64.
61 (mode_for_vector, mode_for_int_vector): Take the number of vector
62 elements as a poly_uint64.
63 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
64 size as a poly_uint64.
65 (mode_for_vector, mode_for_int_vector): Take the number of vector
66 elements as a poly_uint64.
68 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
69 Alan Hayward <alan.hayward@arm.com>
70 David Sherwood <david.sherwood@arm.com>
72 * machmode.h (MACRO_MODE): New macro.
73 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
74 * targhooks.c (default_libcall_value, default_secondary_reload)
75 (default_memory_move_cost, default_register_move_cost)
76 (default_class_max_nregs): Likewise.
78 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
79 Alan Hayward <alan.hayward@arm.com>
80 David Sherwood <david.sherwood@arm.com>
82 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
83 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
85 * genautomata.c (main): Likewise.
86 * genconditions.c (write_header): Likewise.
87 * genemit.c (main): Likewise.
88 * genextract.c (print_header): Likewise.
89 * genopinit.c (main): Likewise.
90 * genoutput.c (output_prologue): Likewise.
91 * genpeep.c (main): Likewise.
92 * genpreds.c (write_insn_preds_c): Likewise.
93 * genrecog.c (writer_header): Likewise.
94 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
95 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
96 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
97 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
98 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
99 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
100 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
101 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
102 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
103 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
104 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
105 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
106 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
107 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
108 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
109 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
110 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
111 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
112 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
113 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
114 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
115 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
116 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
117 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
118 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
119 * config/darwin.c (IN_TARGET_CODE): Likewise.
120 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
121 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
122 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
123 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
124 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
125 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
126 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
127 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
128 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
129 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
130 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
131 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
132 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
133 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
134 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
135 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
136 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
137 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
138 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
139 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
140 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
141 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
142 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
143 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
144 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
145 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
146 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
147 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
148 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
149 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
150 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
151 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
152 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
153 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
154 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
155 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
156 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
157 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
158 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
159 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
160 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
161 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
162 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
163 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
164 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
165 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
166 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
167 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
168 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
169 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
170 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
171 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
172 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
173 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
174 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
175 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
176 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
177 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
178 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
179 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
180 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
181 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
182 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
183 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
184 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
185 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
186 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
187 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
188 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
189 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
190 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
191 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
192 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
193 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
194 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
195 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
196 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
197 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
198 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
199 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
200 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
201 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
202 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
203 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
204 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
205 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
206 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
207 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
208 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
209 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
210 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
211 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
212 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
213 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
214 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
215 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
216 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
217 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
218 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
219 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
220 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
221 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
222 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
223 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
224 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
225 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
227 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
228 Alan Hayward <alan.hayward@arm.com>
229 David Sherwood <david.sherwood@arm.com>
231 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
234 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
235 Alan Hayward <alan.hayward@arm.com>
236 David Sherwood <david.sherwood@arm.com>
238 * doc/generic.texi (VEC_SERIES_EXPR): Document.
239 * doc/md.texi (vec_series@var{m}): Document.
240 * tree.def (VEC_SERIES_EXPR): New tree code.
241 * tree.h (build_vec_series): Declare.
242 * tree.c (build_vec_series): New function.
243 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
244 * tree-pretty-print.c (dump_generic_node): Likewise.
245 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
246 * tree-inline.c (estimate_operator_cost): Likewise.
247 * expr.c (expand_expr_real_2): Likewise.
248 * optabs-tree.c (optab_for_tree_code): Likewise.
249 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
250 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
251 * expmed.c (make_tree): Handle VEC_SERIES.
252 * optabs.def (vec_series_optab): New optab.
253 * optabs.h (expand_vec_series_expr): Declare.
254 * optabs.c (expand_vec_series_expr): New function.
255 * tree-vect-generic.c (expand_vector_operations_1): Check that
256 the operands also have vector type.
258 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
259 Alan Hayward <alan.hawyard@arm.com>
260 David Sherwood <david.sherwood@arm.com>
262 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
263 (VEC_COND_EXPR): Add missing @tindex.
264 * doc/md.texi (vec_duplicate@var{m}): Document.
265 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
266 * tree.c (build_vector_from_val): Add stubbed-out handling of
267 variable-length vectors, using VEC_DUPLICATE_EXPR.
268 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
269 * cfgexpand.c (expand_debug_expr): Likewise.
270 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
271 * tree-inline.c (estimate_operator_cost): Likewise.
272 * tree-pretty-print.c (dump_generic_node): Likewise.
273 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
274 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
275 (test_vec_duplicate_folding): New function.
276 (fold_const_c_tests): Call it.
277 * optabs.def (vec_duplicate_optab): New optab.
278 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
279 * optabs.h (expand_vector_broadcast): Declare.
280 * optabs.c (expand_vector_broadcast): Make non-static. Try using
282 * expr.c (store_constructor): Try using vec_duplicate_optab for
284 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
286 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
289 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
290 div/mod latencies a bit.
292 2017-12-15 Jeff Law <law@redhat.com>
294 PR tree-optimization/36550
295 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
296 (mark_threaded_blocks): Rewrite code to avoid block copying when
297 optimizing for size. Don't pessimize blocks which will be
298 copied, but all the statements will be dead.
300 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
302 PR tree-optimization/81165
303 * tree-ssa-threadupdate.c (uses_in_bb): New.
304 (estimate_threading_killed_stmts): New.
305 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
306 * tree-ssa-threadedge.c
307 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
310 2017-12-15 Jeff Law <law@redhat.com>
312 PR tree-optimization/83410
313 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
314 threads when parallelizing loops.
316 2017-12-15 Jakub Jelinek <jakub@redhat.com>
318 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
320 * config/alpha/alpha.c (vms_attribute_table): Swap
321 affects_type_identity and handler fields, adjust comments.
322 * config/mips/mips.c (mips_attribute_table): Likewise.
323 * config/visium/visium.c (visium_attribute_table): Likewise.
324 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
325 * config/microblaze/microblaze.c (microblaze_attribute_table):
327 * config/spu/spu.c (spu_attribute_table): Likewise.
328 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
329 * config/arc/arc.c (arc_attribute_table): Likewise.
330 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
331 * config/v850/v850.c (v850_handle_interrupt_attribute,
332 v850_handle_data_area_attribute): Formatting fixes.
333 (v850_attribute_table): Swap affects_type_identity and handler
334 fields, adjust comments.
335 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
336 * config/arm/arm.c (arm_attribute_table): Likewise.
337 * config/avr/avr.c (avr_attribute_table): Likewise.
338 * config/s390/s390.c (s390_attribute_table): Likewise.
339 * config/sh/sh.c (sh_attribute_table): Likewise.
340 * config/i386/i386.c (ix86_handle_cconv_attribute,
341 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
342 (ix86_attribute_table): Swap affects_type_identity and handler
343 fields, adjust comments.
344 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
345 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
346 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
347 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
348 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
349 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
350 * config/rx/rx.c (rx_attribute_table): Likewise.
351 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
352 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
353 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
354 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
355 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
356 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
357 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
358 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
359 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
360 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
361 * doc/plugins.texi (user_attr): Likewise. Add NULL for
363 * attribs.c (empty_attribute_table): Swap affects_type_identity and
365 (register_scoped_attributes, decl_attributes): Formatting fixes.
367 PR tree-optimization/83269
368 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
369 subtraction in arg0's type if type is signed and arg0 is unsigned.
373 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
374 convert? on @0 instead of convert. Check type of @1, not @0.
375 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
376 convert? on @0 instead of convert. Check type of @1, not @0.
377 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
378 only optimize if either both @1 and @2 types are narrower
379 precision, or both are wider or equal precision, and in the former
380 case only if both have undefined overflow.
382 2017-12-15 Richard Biener <rguenther@suse.de>
385 * internal-fn.def (IFN_NOP): Add.
386 * internal-fn.c (expand_NOP): Do nothing.
387 * lto-streamer-in.c (input_function): Instead of removing
388 sanitizer calls replace them with IFN_NOP calls.
390 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
391 Alan Hayward <alan.hayward@arm.com>
392 David Sherwood <david.sherwood@arm.com>
394 * dse.c (store_info, read_info_type): Replace begin and end with
396 (print_range): New function.
397 (set_all_positions_unneeded, any_positions_needed_p)
398 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
400 (record_store): Likewise. Optimize the case in which all positions
402 (get_stored_val): Replace read_begin and read_end with read_offset
404 (replace_read): Update call accordingly.
406 2017-12-15 Bin Cheng <bin.cheng@arm.com>
408 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
409 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
410 (loop_cand::loop_cand): Initialize above members.
411 (loop_cand::supported_operations): Delete.
412 (loop_cand::can_interchange_p): Inline above function.
413 (loop_cand::classify_simple_reduction): Record number of constant
414 initialized simple reductions.
415 (should_interchange_loops): New parameters. Check stmt cost of loops
417 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
418 Update call to should_interchange_loops.
419 (should_interchange_loop_nest): Update call to
420 should_interchange_loops.
422 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
425 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
426 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
427 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
429 2017-12-15 Julia Koval <julia.koval@intel.com>
431 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
432 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
433 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
434 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
435 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
437 2017-12-15 Julia Koval <julia.koval@intel.com>
439 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
440 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
441 * config/i386/sse.md (vaesenc_<mode>): New pattern.
442 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
443 _mm_aesenc_epi128): New intrinsics.
445 2017-12-15 Julia Koval <julia.koval@intel.com>
447 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
448 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
449 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
450 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
451 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
453 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
455 * gimple-ssa-strength-reduction.c (analyze_increments):
456 Distinguish replacement costs for constant strides from those for
459 2017-12-14 Jakub Jelinek <jakub@redhat.com>
461 * var-tracking.c (variable_tracking_main_1): Formatting fix.
463 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
465 * doc/invoke.texi: Document -Wcast-function-type.
466 * recog.h (stored_funcptr): Change signature.
467 * tree-dump.c (dump_node): Avoid warning.
468 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
470 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
473 * gimple-fold.c (get_range_strlen): Add the handling of non-member
476 2017-12-14 David Malcolm <dmalcolm@redhat.com>
478 PR tree-optimization/83312
479 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
480 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
481 "val" param, and to cope with arbitrary basic blocks.
482 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
483 handle NULL_TREE for "val", dropping "bb" param.
484 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
485 drop "bb" param. Handle NULL_TREE for "val".
486 (find_case_label_for_value): Make "switch_stmt" param const.
487 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
489 (vrp_prop::check_all_array_refs): Reimplement as...
490 (check_array_bounds_dom_walker::before_dom_children): ...this new
491 vfunc. Replace linear search through BB block list, excluding
492 those with non-executable in-edges via dominator walk.
494 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
496 * config/arm/arm.opt (mverbose-cost-dump): New option.
497 * config/arm/arm.c (arm_rtx_costs): Use it.
499 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
502 * reload1.c (emit_input_reload_insns): Skip debug markers.
504 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
506 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
510 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
511 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
512 * config/sh/sh.c: Skip debug insns besides notes.
513 * config/sh/sh.md: Likewise.
514 * config/sh/sh_treg_combine.cc: Likewise.
515 * config/sh/sync.md: Likewise.
517 2017-12-14 Tom de Vries <tom@codesourcery.com>
519 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
520 Add item for weak_undefined.
522 2017-12-14 Richard Biener <rguenther@suse.de>
524 PR tree-optimization/67842
525 * sese.h (bb_in_region): Remove #if 0'ed code.
527 2017-12-14 Richard Biener <rguenther@suse.de>
529 PR tree-optimization/83326
530 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
531 may_be_zero parameter and handle it by not marking the first
532 peeled copy as not exiting the loop.
533 (try_peel_loop): Likewise.
534 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
535 to handle the case of constant or zero iterations and perform
536 loop header copying on-the-fly.
538 2017-12-14 Richard Biener <rguenther@suse.de>
540 PR tree-optimization/83418
541 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
542 Instead of asserting we don't get unfolded comparisons deal with
545 2017-12-14 Jakub Jelinek <jakub@redhat.com>
548 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
549 basic blocks. Assert debug bind insns don't appear outside of bbs,
550 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
553 PR tree-optimization/83198
554 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
555 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
556 value if arg is a REAL_CST with incompatible type.
558 2017-12-14 Sudakshina Das <sudi.das@arm.com>
559 Bin Cheng <bin.cheng@arm.com>
562 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
564 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
567 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
569 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
570 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
572 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
573 Alan Hayward <alan.hayward@arm.com>
574 David Sherwood <david.sherwood@arm.com>
576 * poly-int.h: New file.
577 * poly-int-types.h: Likewise.
578 * coretypes.h: Include them.
579 (POLY_INT_CONVERSION): Define.
580 * target.def (estimated_poly_value): New hook.
581 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
582 * doc/tm.texi: Regenerate.
583 * doc/poly-int.texi: New file.
584 * doc/gccint.texi: Include it.
585 * doc/rtl.texi: Describe restrictions on subreg modes.
586 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
587 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
588 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
589 * targhooks.h (default_estimated_poly_value): Declare.
590 * targhooks.c (default_estimated_poly_value): New function.
591 * target.h (estimated_poly_value): Likewise.
592 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
593 (wi::unary_traits): Delete.
594 (wi::binary_traits::signed_shift_result_type): Define for
595 offset_int << HOST_WIDE_INT, etc.
596 (generic_wide_int::operator <<=): Define for all types and use
597 wi::lshift instead of <<.
598 (wi::hwi_with_prec): Add a default constructor.
599 (wi::ints_for): New class.
600 (operator <<): Define for all wide-int types.
601 (operator /): New function.
602 (operator %): Likewise.
603 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
604 (ASSERT_MAYBE_NE_AT): New macros.
606 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
607 Dominik Vogt <vogt@linux.vnet.ibm.com>
610 * emit-rtl.c (init_emit): Remove ??? comment.
611 * explow.c (get_dynamic_stack_size): Take known alignment of stack
612 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
613 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
614 alignment of 3 virtual registers to BITS_PER_WORD.
616 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
618 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
620 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
621 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
623 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
626 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
628 2017-12-13 Jakub Jelinek <jakub@redhat.com>
630 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
631 with the exception of debug begin stmt markers appear before
635 * final.c (rest_of_handle_final): Call variable_tracking_main only
636 if !flag_var_tracking.
638 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
639 Jakub Jelinek <jakub@redhat.com>
643 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
644 labels debug stmts that can only appear after labels.
646 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
648 PR rtl-optimization/82398
649 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
650 EXPR_USEFULNESS in priority comparison.
652 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
654 PR rtl-optimization/83393
655 * combine.c (move_deaths): If reg_stat points to a too new insn in
656 last_death, do not use it: find the proper insn instead.
658 2017-12-12 Jeff Law <law@redhat.com>
660 PR tree-optimization/83298
661 PR tree-optimization/83362
662 PR tree-optimization/83383
663 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
664 push_value_range a public interface. Add new argument to
665 record_ranges_from_stmt.
666 * gimple-ssa-evrp-analyze.c
667 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
668 Update comments. Handle recording temporary equivalences.
669 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
670 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
671 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
672 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
673 gimple-ssa-evrp-analyze.h.
674 (record_temporary_equivalences_from_phis): Add new argument. When
675 the PHI arg is an SSA_NAME, set the result's range to the range
677 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
679 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
680 Pass it down to children as needed.
681 (thread_outgoing_edges): Likewise.
682 (thread_across_edge): Likewise. Push/pop range state as needed.
683 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
685 2017-12-12 Julia Koval <julia.koval@intel.com>
687 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
688 (PTA_CANNONLAKE): Remove PTA_CLWB.
690 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
693 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
694 (vcondv2div2df): Likewise.
695 (vconduv2dfv2di): Likewise.
697 2017-12-12 Jakub Jelinek <jakub@redhat.com>
699 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
700 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
701 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
702 ATTR_MATHFN_FPROUNDING_ERRNO.
704 2017-12-12 Richard Biener <rguenther@suse.de>
706 PR tree-optimization/83385
707 * graphite-scop-detection.c (get_order, order): Remove.
708 (bb_to_rpo): New global.
710 (build_scops): Sort pbbs in RPO order.
712 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
714 * combine.c (simplify_set): Do not transform subregs to zero_extends
715 if the destination mode is a vector mode.
717 2017-12-12 Jakub Jelinek <jakub@redhat.com>
719 PR tree-optimization/83359
720 * tree-cfg.h (fold_loop_internal_call): Declare.
721 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
722 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
723 (find_loop_dist_alias): New function.
724 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
725 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
728 PR tree-optimization/80631
729 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
730 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
731 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
732 hardcoding zero as the value if COND_EXPR is never true. For
733 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
734 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
735 hardcoding MAX_EXPR as the reduction operation.
736 (is_nonwrapping_integer_induction): Allow negative step.
737 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
738 vect_create_epilog_for_reduction, if no value is suitable, don't
739 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
741 2017-12-12 Richard Biener <rguenther@suse.de>
743 PR tree-optimization/81889
744 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
745 range info from the non-wrapping IV instead of just the range
748 2017-12-12 Julia Koval <julia.koval@intel.com>
750 * config.gcc: Add vaesintrin.h.
751 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
753 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
754 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
756 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
757 * config/i386/immintrin.h: Include vaesintrin.h.
758 * config/i386/sse.md (vaesdec_<mode>): New pattern.
759 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
760 _mm_aesdec_epi128): New intrinsics.
762 2017-12-12 Julia Koval <julia.koval@intel.com>
764 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
765 OPTION_MASK_ISA_VAES_UNSET): New.
766 (ix86_handle_option): Handle -mvaes.
767 * config/i386/cpuid.h: Define bit_VAES.
768 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
769 * config/i386/i386-c.c (__VAES__): New.
770 * config/i386/i386.c (ix86_target_string): Add -mvaes.
771 (ix86_valid_target_attribute_inner_p): Ditto.
772 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
773 * config/i386/i386.opt: Add -mvaes.
774 * doc/invoke.texi: Ditto.
776 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
778 * debug.h (gcc_debug_hooks): Add inline_entry.
779 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
780 * debug.c (do_nothing_debug_hooks): Likewise.
781 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
782 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
783 (dwarf2_lineno_debug_hooks): Likewise.
785 * common.opt (gstatement-frontiers): New, setting
786 debug_nonbind_markers_p.
787 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
788 * toplev.c (process_options): Autodetect value for debug statement
790 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
791 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
793 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
794 markers. Integrate source bind into debug stmt expand loop.
795 (pass_expand::execute): Check debug marker limit. Avoid deep
796 TER and expand debug locations for debug bind insns only.
797 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
799 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
800 * final.c (reemit_insn_block_notes): Take current block from
801 nonbind markers. Declare note where it's first set.
802 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
803 begin stmt markers if enabled.
804 (notice_source_line): Handle nonbind markers. Fail if their
805 location is unknown or that of builtins.
806 (rest_of_handle_final): Convert begin stmt markers to notes if
807 var-tracking didn't run.
808 (rest_of_clean_state): Skip begin stmt markers.
809 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
811 * function.c (allocate_struct_function): Set begin_stmt_markers.
812 * function.h (struct function): Add debug_marker_count counter
813 and debug_nonbind_markers flag.
814 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
815 * gimple-low.c (lower_function_body): Adjust
816 debug_nonbind_markers.
817 (lower_stmt): Drop or skip gimple debug stmts.
818 (lower_try_catch): Skip debug stmts.
819 * gimple.c (gimple_build_debug_begin_stmt): New.
820 (gimple_copy): Increment debug_marker_count if copying one.
821 * gimple.h (gimple_build_debug_begin_stmt): Declare.
822 * gimplify.c (rexpr_location): New.
823 (rexpr_has_location): New.
824 (warn_switch_unreachable_r): Handle gimple debug stmts.
825 (shortcut_cond_r): Call expr_location.
827 (find_goto_label): New.
828 (shortcut_cond_expr): Call expr_has_location, expr_location, and
830 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
832 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
833 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
834 (LANG_HOOKS_INITIALIZER): ... this.
835 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
836 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
838 (update_ebb_live_info): Skip debug insn markers.
839 * lra.c (debug_insn_static_data): Rename to...
840 (debug_bind_static_data): ... this.
841 (debug_marker_static_data): New.
842 (lra_set_insn_recog_data): Select one of the above depending
844 (lra_update_isn_regno_info): Don't assume debug insns have
846 (push_insns): Skip debug insns.
847 * lto-streamer-in.c (input_function): Drop debug stmts
848 depending on active options. Adjust debug_nonbind_markers.
849 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
850 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
851 begin stmt marker notes.
852 (print_insn): Likewise.
853 * recog.c (extract_insn): Recognize rtl for debug markers.
854 * rtl.def (DEBUG_MARKER): New.
855 * tree-inline.c: Include params.h.
856 (remap_gimple_stmt): Handle nonbind markers.
857 (maybe_move_debug_stmts_to_successors): Likewise.
858 (copy_debug_stmt): Likewise.
859 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
860 markers regardless of no side effects.
861 (tsi_link_before): Don't update container's side effects when adding
863 (tsi_link_after): Likewise.
864 (expr_first): Skip begin stmt markers.
865 (expr_last): Likewise.
866 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
867 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
868 Disregard nonbind markers.
869 * tree.c (make_node_stat): Don't set side effects for begin stmt
871 (build1_stat): Likewise.
872 * tree.def (DEBUG_BEGIN_STMT): New.
873 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
874 * var-tracking.c (delete_debug_insns): Renamed to...
875 (delete_vta_debug_insns): ... this.
876 (reemit_marker_as_note): New.
877 (vt_initialize): Reemit markers.
878 (delete_vta_debug_insns): Likewise.
879 (vt_debug_insns_local): Reemit or delete markers.
880 (variable_tracking_main_1): Likewise.
881 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
882 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
883 (gimple_debug_nonbind_marker_p): New.
884 (gimple_build_debug_bind): Adjust.
885 (gimple_build_debug_begin_stmt): New.
886 * doc/invoke.texi (max-debug-marker-count): New param.
887 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
888 (debug_parameter_ref, debug_marker): New.
889 (NOTE_INSN_BEGIN_STMT): New.
890 (DEBUG_INSN): Describe begin stmt markers.
892 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
893 without debug insns, we wouldn't, but clean up debug insns
894 after a control flow insn nevertheless.
896 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
897 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
898 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
899 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
900 (rtl_tidy_fallthru_edge): Likewise.
901 (rtl_verify_fallthru): Likewise.
902 (rtl_verify_bb_layout): Likewise.
903 (skip_insns_after_block): Likewise.
904 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
905 * dwarf2out.c: Include print-rtl.h.
906 (dwarf2out_next_real_insn): New.
907 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
908 Dump debug binds in asm comments.
909 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
910 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
911 callers to use gsi_start_nondebug_bb instead.
912 (gsi_after_labels): Skip gimple debug stmts.
913 (gsi_start_nondebug): New.
914 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
915 (proper_loop_form_for_interchange): Adjust.
916 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
917 * gimple.h (gimple_seq_last_nondebug_stmt): New.
918 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
919 (collect_fallthrough_labels): Likewise.
920 (should_warn_for_implicit_fallthrough): Likewise.
921 (warn_implicit_fallthrough_r): Likewise.
922 (expand_FALLTHROUGH_r): Likewise.
923 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
924 (graphite_copy_stmts_from_block): Skip nonbind markers.
925 * haifa-sched.c (sched_extend_bb): Skip debug insns.
926 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
927 * jump.c (clean_barriers): Skip debug insns.
928 * omp-expand.c (expand_parallel_call): Skip debug insns.
929 (expand_task_call): Likewise.
930 (remove_exit_barrier): Likewise.
931 (expand_omp_taskreg): Likewise.
932 (expand_omp_for_init_counts): Likewise.
933 (expand_omp_for_generic): Likewise.
934 (expand_omp_for_static_nochunk): Likewise.
935 (expand_omp_for_static_chunk): Likewise.
936 (expand_omp_simd): Likewise.
937 (expand_omp_taskloop_for_outer): Likewise.
938 (expand_omp_taskloop_for_inner): Likewise.
939 (expand_oacc_for): Likewise.
940 (expand_omp_sections): Likewise.
941 (expand_omp_single): Likewise.
942 (expand_omp_synch): Likewise.
943 (expand_omp_atomic_load): Likewise.
944 (expand_omp_atomic_store): Likewise.
945 (expand_omp_atomic_fetch_op): Likewise.
946 (expand_omp_atomic_pipeline): Likewise.
947 (expand_omp_atomic_mutex): Likewise.
948 (expand_omp_target): Likewise.
949 (grid_expand_omp_for_loop): Likewise.
950 (grid_expand_target_grid_body): Likewise.
951 (build_omp_regions_1): Likewise.
952 * omp-low.c (check_combined_parallel): Skip debug stmts.
953 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
954 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
955 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
957 * tree-cfg.c (make_blobs_1): Skip debug stmts.
958 (make_edges): Likewise.
959 (cleanup_dead_labels): Likewise.
960 (gimple_can_merge_blocks_p): Likewise.
961 (stmt_starts_bb_p): Likewise.
962 (gimple_block_label): Likewise.
963 (gimple_redirect_edge_and_branch): Likewise.
964 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
966 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
968 * tree-pretty-print (print_declaration): Omit initializer in slim
970 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
972 (eliminate_unnecessary_stmts): Stabilize block removal order.
973 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
974 * var-tracking.c (get_first_insn): New.
975 (vt_emit_notes): Call it.
976 (vt_initialize): Walk any insns before the first BB.
977 (delete_debug_insns): Likewise.
979 * gimple.h (enum gimple_debug_subcode): Add
980 GIMPLE_DEBUG_BEGIN_STMT.
981 (gimple_debug_begin_stmt_p): New.
982 (gimple_debug_nonbind_marker_p): New.
983 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
984 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
985 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
986 * insn-notes.def (BEGIN_STMT): New.
987 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
988 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
989 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
990 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
991 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
992 (INSN_DEBUG_MARKER_KIND): New.
993 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
994 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
995 (INSN_VAR_LOCATION_PTR): New.
996 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
998 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
999 deps only if debug bind insns are enabled.
1000 (pass_expand::execute): Avoid deep TER and expand
1001 debug locations for debug bind insns only.
1002 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1003 debug stmts special handling down to debug bind stmts.
1004 * combine.c (try_combine): Narrow debug insns special handling
1005 down to debug bind insns.
1006 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1007 Narrow debug insns preexisting special handling down to debug
1009 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1010 handling down to debug bind insns.
1011 * function.c (instantiate_virtual_regs): Skip debug markers,
1012 adjust handling of debug binds.
1013 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1014 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1015 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1016 insns to debug bind insns.
1017 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1018 special handling of debug stmts to debug bind stmts.
1019 * ipa-split.c (split_function): Likewise.
1020 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1021 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1023 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1024 * regrename.c (build_def_use): Likewise.
1025 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1026 (pass_cprop_hardreg): Narrow special casing of debug insns to
1028 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1029 * reload1.c (reload): Likewise.
1030 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1031 casing of debug stmts to debug bind stmts.
1032 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1033 * ssa-iterators.h (num_imm_uses): Likewise.
1034 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1035 debug stmts to debug bind stmts.
1036 * tree-inline.c (tree_function_versioning): Narrow special casing
1037 of debug stmts to debug bind stmts.
1038 * tree-loop-distribution.c (generate_loops_for_partition):
1039 Narrow special casing of debug stmts to debug bind stmts.
1040 * tree-sra.c (analyze_access_subtree): Narrow special casing
1041 of debug stmts to debug bind stmts.
1042 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1043 stmts to debug bind stmts.
1044 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1045 casing of debug stmts to debug bind stmts.
1046 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1047 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1048 casing of debug stmts to debug bind stmts.
1049 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1051 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1052 debug stmts to debug bind stmts.
1053 (gimple_replace_ssa_lhs): Likewise.
1054 (insert_debug_temp_for_var_def): Likewise.
1055 (insert_debug_temps_for_defs): Likewise.
1056 (reset_debug_uses): Likewise.
1057 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1058 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1059 (adjust_debug_stmts): Likewise.
1060 (adjust_phi_and_debug_stmts): Likewise.
1061 (vect_do_peeling): Likewise.
1062 * tree-vect-loop.c (vect_transform_loop): Likewise.
1063 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1064 * var-tracking.c (adjust_mems): Narrow special casing of debug
1065 insns to debug bind insns.
1066 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1067 (compute_bb_dataflow, vt_find_locations): Likewise.
1068 (vt_expand_loc, emit_notes_for_changes): Likewise.
1069 (vt_init_cfa_base): Likewise.
1070 (vt_emit_notes): Likewise.
1071 (vt_initialize): Likewise.
1072 (vt_finalize): Likewise.
1074 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1075 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1076 (next_nonnote_nondebug_insn_bb): New.
1077 (prev_nonnote_nondebug_insn_bb): New.
1078 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1079 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1080 (prev_nonnote_nondebug_insn_bb): Declare.
1081 (next_nonnote_nondebug_insn_bb): Declare.
1082 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1083 * cfgrtl.c (get_last_bb_insn): Likewise.
1084 * lra.c (push_insns): Likewise.
1086 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1089 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1091 * selftest.c (selftest::test_runner::test_runner): New ctor.
1092 (selftest::test_runner::~test_runner): New dtor.
1093 * selftest.h (class selftest::test_runner): New class.
1095 2017-12-11 Carl Love <cel@us.ibm.com>
1097 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1098 vec_extract_fp32_from_shortl]): Add #defines.
1099 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1100 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1101 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1102 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1103 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1104 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1105 * doc/extend.texi: Add documentation for the added builtins.
1107 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1109 PR rtl-optimization/80693
1110 PR rtl-optimization/81019
1111 PR rtl-optimization/81020
1112 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1113 are not mentioned in i3. Place the REG_UNUSED note on i2,
1114 possibly modified to REG_DEAD, if it did not originate in i3.
1116 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1118 * recog.c (store_data_bypass_p_1): New function.
1119 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1120 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1122 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1124 PR rtl-optimization/83361
1125 * ifcvt.c (if_convert): Call fixup_partitions.
1127 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1129 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1130 early folding of splat_u{8,16,32}.
1132 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1134 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1135 output_operand_lossage first argument with capital letter.
1136 (aarch64_override_options): Don't start error and sorry first argument
1137 with capital letter.
1139 2017-12-11 Andi Kleen <ak@linux.intel.com>
1141 PR gcov-profile/83355
1142 * auto-profile.c (string_table::get_index_by_decl): Don't
1143 recurse when abstract origin points to itself.
1145 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1147 PR tree-optimization/83320
1148 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1149 (prune_datarefs_not_in_loop): Ditto.
1151 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1153 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1155 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1157 PR tree-optimization/83337
1158 * gimple-loop-interchange.cc (compute_access_stride): Handle
1159 bitfield DRs properly.
1161 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1163 PR tree-optimization/83338
1164 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1165 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1166 vector of unsigned integers to vector of signed integers.
1168 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1170 PR rtl-optimization/83317
1171 * lra-constraints.c (process_address_1): Add insn code check.
1173 2017-12-08 Michael Matz <matz@suse.de>
1175 Fix PR tree-optimization/83323
1176 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1177 head-controlled loops and loop BBs.
1178 * common.opt (funroll-and-jam): Remove, instead ...
1179 (floop-unroll-and-jam): ... reuse this option.
1180 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1181 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1182 (-floop-unroll-and-jam): ... this option.
1184 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1186 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1187 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1188 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1190 2017-12-08 Richard Biener <rguenther@suse.de>
1193 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1194 handle accesses outside of zero-sized vars.
1196 2017-12-08 Martin Jambor <mjambor@suse.cz>
1198 PR tree-optimization/83141
1199 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1200 test for MEM_REFs implicitely changing types with padding. Remove
1202 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1204 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1206 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1208 * config/avr/avr.c (avr_attribute_table): Likewise.
1209 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1210 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1211 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1212 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1213 of struct attribute_spec.
1214 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1216 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1219 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1221 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1224 * profile-count.c (profile_count::from_gcov_type): Move from
1225 profile-count.h; handle overflow.
1226 * profile-count.h (profile_count::from_gcov_type): Move offline.
1228 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1230 PR rtl-optimization/83304
1231 * combine.c (move_deaths): If we do not know where a register died,
1234 2017-12-08 Richard Biener <rguenther@suse.de>
1236 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1237 Provide -fopt-info-loop feedback when we interchange in a nest.
1239 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1241 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1242 for armv6 ARM CPU IDs.
1244 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1246 * common/config/arm/arm-common.c: Include <algorithm>.
1247 (INCLUDE_VECTOR): Define.
1248 (compare_opt_names): New function.
1249 (arm_rewrite_selected_arch): Only strip out extensions that can be
1250 expressed through -mfpu. Sort the remaining extensions
1253 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1255 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1256 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1257 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1258 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1260 2017-06-08 Tristan Gingold <gindold@adacore.com>
1263 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1264 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1266 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1268 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1269 of struct attribute_spec.
1271 2017-12-08 Julia Koval <julia.koval@intel.com>
1273 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1274 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1275 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1276 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1277 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1278 _mm_maskz_dpwssds_epi32): Ditto.
1280 2017-12-08 Richard Biener <rguenther@suse.de>
1282 PR tree-optimization/81303
1283 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1284 conditions try to create a comparison vector type matching
1285 the data vector type.
1286 (vectorizable_condition): Adjust.
1287 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1288 Leave invariant conditions alone in case we can vectorize those.
1290 2017-12-08 Julia Koval <julia.koval@intel.com>
1292 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1293 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1294 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1295 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1296 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1298 2017-12-08 Julia Koval <julia.koval@intel.com>
1300 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1301 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1302 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1303 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1304 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1305 _mm_maskz_dpbusds_epi32): New intrinsics.
1307 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1309 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1311 (ld<bh>io_signed): Likewise.
1312 (st<bhw>io): Likewise.
1313 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1314 SMALL_INT12 constant integer operand.
1316 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1318 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1320 (nios2_large_constant_allowed): ...this. Adjust uses.
1321 (nios2_plus_symbolic_constant_p): Rename to...
1322 (nios2_plus_large_constant_p): ...this. Adjust uses.
1323 (nios2_legitimate_address_p): Correct CONST_INT handling.
1324 (nios2_symbolic_memory_operand_p): Rename to...
1325 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1326 (nios2_large_constant_p): Check for large constant integers too.
1327 (nios2_split_large_constant): Handle constant integers.
1328 (nios2_split_symbolic_memory_operand): Rename to...
1329 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1330 (nios2_legitimize_constant_address): Handle constant integers.
1331 (r0rel_constant_p): Handle small constant integers.
1332 (nios2_print_operand_address): Handle r0-relative integer addresses.
1333 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1334 * config/nios2/nios2.md: Adjust for renamed functions.
1336 2017-12-07 Andrew Waterman <andrew@sifive.com>
1338 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1339 (TARGET_HAVE_SRODATA_SECTION): New define.
1340 (riscv_select_section): New function.
1342 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1343 Alexander Monakov <amonakov@ispras.ru>
1344 Jakub Jelinek <jakub@redhat.com>
1347 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1349 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1351 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1352 a bool success value. Don't call output_operand_lossage here.
1353 (aarch64_print_ldpstp_address): Return a bool success value.
1354 (aarch64_print_operand_address): Call output_addr_const if
1355 aarch64_print_address_internal fails.
1356 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1357 'y'; call output_operand_lossage instead. Call output_operand_lossage
1358 if aarch64_print_ldpstp_address fails.
1360 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1362 * tree-vector-builder.h
1363 (tree_vector_builder::binary_encoded_nelts): Declare.
1364 * tree-vector-builder.c
1365 (tree_vector_builder::binary_encoded_nelts): New function.
1366 * fold-const.c (negate_expr_p): Likewise.
1367 (operand_equal_p, fold_checksum_tree): Likewise.
1368 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1369 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1370 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1371 (uniform_vector_p): Likewise.
1372 * varasm.c (const_hash_1, compare_constant): Likewise.
1373 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1374 (valid_lattice_transition): Operate directly on the VECTOR_CST
1376 * ipa-icf.c: Include tree-vector-builder.h.
1377 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1378 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1380 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1382 * tree.c (build_vector): Delete.
1383 * tree.h (build_vector): Make static and move into the self-testing
1386 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1388 * vector-builder.h (vector_builder::derived): New const overload.
1389 (vector_builder::elt): New function.
1390 * tree-vector-builder.h (tree_vector_builder::type): New function.
1391 (tree_vector_builder::apply_step): Declare.
1392 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1394 * gimple-fold.h (tree_vector_builder): Declare.
1395 (gimple_build_vector): Take a tree_vector_builder instead of a
1396 type and vector of elements.
1397 * gimple-fold.c (gimple_build_vector): Likewise.
1398 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1400 (get_initial_defs_for_reduction): Likewise.
1401 (vectorizable_induction): Likewise.
1403 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1405 * tree-vector-builder.h
1406 (tree_vector_builder::new_binary_operation): Declare.
1407 * tree-vector-builder.c
1408 (tree_vector_builder::new_binary_operation): New function.
1409 * fold-const.c (fold_relational_const): Use it.
1410 (const_binop): Likewise. Check that both input vectors have
1411 the same number of elements, thus excluding things like WIDEN_SUM.
1412 Check whether it is possible to operate directly on the encodings
1415 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1417 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1418 new_unary_operation, operating only on the encoded elements.
1419 (const_unop): Likewise.
1420 (exact_inverse): Likewise.
1421 (distributes_over_addition_p): New function.
1422 (const_binop): Use tree_vector_builder and new_unary_operation
1423 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1424 on the encoded elements unless the encoding is strided and the
1425 operation does not distribute over addition.
1426 (fold_convert_const): Use tree_vector_builder and
1427 new_unary_operation. Operate only on the encoded elements
1428 for truncating integer conversions, or for non-stepped encodings.
1430 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1432 * config/sparc/sparc.c: Include tree-vector-builder.h.
1433 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1434 * expmed.c: Include tree-vector-builder.h.
1435 (make_tree): Use tree_vector_builder instead of build_vector.
1436 * fold-const.c: Include tree-vector-builder.h.
1437 (const_binop): Use tree_vector_builder instead of build_vector.
1438 (const_unop): Likewise.
1439 (native_interpret_vector): Likewise.
1440 (fold_vec_perm): Likewise.
1441 (fold_ternary_loc): Likewise.
1442 * gimple-fold.c: Include tree-vector-builder.h.
1443 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1445 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1446 (simplify_vector_constructor): Use tree_vector_builder instead
1448 * tree-vect-generic.c: Include tree-vector-builder.h.
1449 (add_rshift): Use tree_vector_builder instead of build_vector.
1450 (expand_vector_divmod): Likewise.
1451 (optimize_vector_constructor): Likewise.
1452 * tree-vect-loop.c: Include tree-vector-builder.h.
1453 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1454 of build_vector. Explicitly use a stepped encoding for
1456 * tree-vect-slp.c: Include tree-vector-builder.h.
1457 (vect_get_constant_vectors): Use tree_vector_builder instead
1459 (vect_transform_slp_perm_load): Likewise.
1460 (vect_schedule_slp_instance): Likewise.
1461 * tree-vect-stmts.c: Include tree-vector-builder.h.
1462 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1463 (vect_gen_perm_mask_any): Likewise.
1464 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1465 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1467 (build_vector_from_val): Likewise. Explicitly use a duplicate
1470 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1472 * doc/generic.texi (VECTOR_CST): Describe new representation of
1474 * vector-builder.h: New file.
1475 * tree-vector-builder.h: Likewise.
1476 * tree-vector-builder.c: Likewise.
1477 * Makefile.in (OBJS): Add tree-vector-builder.o.
1478 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1479 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1480 (tree_vector): Change the number of elements to
1481 vector_cst_encoded_nelts.
1482 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1483 (VECTOR_CST_ELTS): Delete.
1484 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1485 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1486 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1487 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1488 (VECTOR_CST_ENCODED_ELT): Likewise.
1489 (vector_cst_encoded_nelts): New function.
1490 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1491 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1492 (vector_cst_int_elt, vector_cst_elt): Declare.
1493 * tree.c: Include tree-vector-builder.h.
1494 (tree_code_size): Abort if passed VECTOR_CST.
1495 (tree_size): Update for new VECTOR_CST layout.
1496 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1497 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1498 (build_vector): Use tree_vector_builder.
1499 (vector_cst_int_elt, vector_cst_elt): New functions.
1500 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1501 encoded elements and then create the vector in the canonical form.
1502 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1503 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1504 (tree_c_tests): Call test_vector_cst_patterns.
1505 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1507 (hash_tree): Likewise.
1508 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1509 (streamer_write_tree_header): Likewise.
1510 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1511 (streamer_alloc_tree): Likewise. Update call to make_vector.
1512 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1514 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1516 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1517 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1518 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1520 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1521 Richard Biener <rguenther@suse.de>
1523 PR tree-optimization/81303
1524 * Makefile.in (gimple-loop-interchange.o): New object file.
1525 * common.opt (floop-interchange): Reuse the option from graphite.
1526 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1527 -floop-interchange and mention it for -O3.
1528 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1529 * gimple-loop-interchange.cc: New file.
1530 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1531 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1532 * passes.def (pass_linterchange): New pass.
1533 * timevar.def (TV_LINTERCHANGE): New time var.
1534 * tree-pass.h (make_pass_linterchange): New declaration.
1535 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1536 interchange. Record IV before/after increment in new parameters.
1537 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1538 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1540 (check_reduction_path): ...New function here.
1541 * tree-vectorizer.h (check_reduction_path): New declaration.
1543 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1546 PR rtl-optimization/80818
1547 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1548 always early clobbered.
1549 * lra-lives.c (process_bb_lives): Check input hard regs for early
1550 clobbered non-operand hard reg.
1552 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1555 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1556 types_compatible_p, just that TYPE_MODE is the same.
1558 2017-12-07 Martin Sebor <msebor@redhat.com>
1561 * attribs.c (empty_attribute_table): Initialize new member of
1562 struct attribute_spec.
1563 (decl_attributes): Add argument. Handle mutually exclusive
1564 combinations of attributes.
1565 (selftests::test_attribute_exclusions): New function.
1566 (selftests::attribute_c_tests): Ditto.
1567 * attribs.h (decl_attributes): Add default argument.
1568 * selftest.h (attribute_c_tests): Declare.
1569 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1570 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1572 * doc/extend.texi (Common Function Attributes): Update const and pure.
1573 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1574 of struct attribute_spec.
1575 * config/arc/arc.c (arc_attribute_table): Same.
1576 * config/arm/arm.c (arm_attribute_table): Same.
1577 * config/avr/avr.c ( avr_attribute_table): Same.
1578 * config/bfin/bfin.c (bfin_attribute_table): Same.
1579 * config/cr16/cr16.c (cr16_attribute_table): Same.
1580 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1581 * config/h8300/h8300.c (h8300_attribute_table): Same.
1582 * config/i386/i386.c (ix86_attribute_table): Same.
1583 * config/ia64/ia64.c (ia64_attribute_table): Same.
1584 * config/m32c/m32c.c (m32c_attribute_table): Same.
1585 * config/m32r/m32r.c (m32r_attribute_table): Same.
1586 * config/m68k/m68k.c (m68k_attribute_table): Same.
1587 * config/mcore/mcore.c (mcore_attribute_table): Same.
1588 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1589 * config/mips/mips.c (mips_attribute_table): Same.
1590 * config/msp430/msp430.c (msp430_attribute_table): Same.
1591 * config/nds32/nds32.c (nds32_attribute_table): Same.
1592 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1593 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1594 * config/rl78/rl78.c (rl78__attribute_table): Same.
1595 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1596 * onfig/rx/rx.c (rx_attribute_table): Same.
1597 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1598 * config/sh/sh.c (sh_attribute_table): Same.
1599 * config/sparc/sparc.c (sparc_attribute_table): Same.
1600 * config/spu/spu.c (spu_attribute_table): Same.
1601 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1602 * config/v850/v850.c (v850_attribute_table): Same.
1603 * config/visium/visium.c (visium_attribute_table): Same.
1605 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1608 * config/arm/arm.c (INCLUDE_STRING): Define.
1609 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1610 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1613 2017-12-07 Michael Matz <matz@suse.de>
1615 Add unroll and jam pass
1617 * gimple-loop-jam.c: New file.
1618 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1619 * common.opt (funroll-and-jam): New option.
1620 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1621 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1622 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1623 * passes.def: Add pass_loop_jam.
1624 * timevar.def (TV_LOOP_JAM): Add.
1625 * tree-pass.h (make_pass_loop_jam): Declare.
1626 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1627 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1628 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1629 to flow_loop_tree_node_add.
1630 (duplicate_subloops, copy_loops_to): Append to sibling list.
1631 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1632 * doc/invoke.texi (-funroll-and-jam): Document new option.
1633 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1635 2017-12-07 Richard Biener <rguenther@suse.de>
1637 PR tree-optimization/83296
1638 PR tree-optimization/67769
1639 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1640 flow sensitive info in an unrelated BB.
1641 (value_replacement): Use reset_flow_sensitive_info.
1642 (minmax_replacement): Reset flow sensitive info on the def
1643 we move. Do not reset flow sensitive info in the whole BB
1644 we move the stmt to.
1645 (abs_replacement): Likewise.
1647 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1651 rs6000_cpu to the given -mcpu=, or to the default processor.
1653 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1655 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1656 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1657 instead of rs6000_cpu_attr.
1658 (rs6000_adjust_cost): Ditto.
1659 (is_microcoded_insn): Ditto.
1660 (rs6000_adjust_priority): Ditto.
1661 (rs6000_issue_rate): Ditto.
1662 (rs6000_use_sched_lookahead): Ditto.
1663 (rs6000_use_sched_lookahead_guard): Ditto.
1664 (rs6000_sched_reorder): Ditto.
1665 (force_new_group): Ditto.
1666 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1667 (group_ending_nop): Ditto.
1669 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1671 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1672 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1673 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1674 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1676 (rs6000_reassociation_width): Ditto.
1677 (rs6000_emit_epilogue): Ditto.
1678 (rs6000_adjust_cost): Ditto.
1679 (is_microcoded_insn): Ditto.
1680 (is_cracked_insn): Ditto.
1681 (rs6000_adjust_priority): Ditto.
1682 (rs6000_sched_reorder): Ditto.
1683 (rs6000_sched_reorder2): Ditto.
1684 (insn_must_be_first_in_group): Ditto.
1685 (insn_must_be_last_in_group): Ditto.
1686 (rs6000_register_move_cost): Ditto.
1687 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1690 2017-12-07 Julia Koval <julia.koval@intel.com>
1692 * config.gcc: Add vaesintrin.h.
1693 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1695 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1696 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1698 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1699 * config/i386/immintrin.h: Include vaesintrin.h.
1700 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1701 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1702 _mm_aesdec_epi128): New intrinsics.
1704 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1706 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1707 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1708 c-family/c-spellcheck.cc.
1709 (best_macro_match::best_macro_match): Likewise.
1711 (struct edit_distance_traits<cpp_hashnode *>): Move to
1712 c-family/c-spellcheck.h.
1713 (class best_macro_match): Likewise.
1715 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1717 PR tree-optimization/83293
1718 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1719 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1720 might insert into empty bb.
1723 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1724 simplify for plus with :c added, and pointer_plus without that.
1725 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1726 with undefined overflow and the conversion is not widening,
1727 perform negation in utype and only convert to type afterwards.
1728 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1729 simplify for plus with :c added, and pointer_plus without that.
1730 If type is integral with undefined overflow and the conversion is
1731 not widening, perform minus in utype and only convert to type
1732 afterwards. Move the last pointer_diff_expr simplify into the
1735 2017-12-06 Martin Sebor <msebor@redhat.com>
1737 PR tree-optimization/82646
1738 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1739 strncpy, not maxlen.
1741 2017-12-06 Martin Sebor <msebor@redhat.com>
1743 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1746 PR tree-optimization/83075
1747 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1748 strncat/strncpy don't change length of source string.
1750 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1753 2017-11-29 Martin Aberg <maberg@gaisler.com>
1755 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1756 to prevent b2bst errata sequence.
1757 (sqrtdf2_fix): Likewise.
1759 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1761 PR tree-optimization/81945
1762 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1763 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1764 to dest_cfun has orig_loop_num set, either remap it to the new
1765 loop number if the loop got moved too, or clear it.
1767 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1769 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1771 (thunderx2t99_nothing): New insn reservation.
1772 (thunderx2t99_mrs): New insn reservation.
1773 (thunderx2t99_multiple): New insn reservation.
1774 (thunderx2t99_alu_basi): Add bfx to reservation.
1775 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1777 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1780 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1782 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1784 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1785 tree-ssa-pre.c::remove_dead_inserted_code.
1786 * tree-ssa-dce.h: New file.
1787 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1788 (remove_dead_inserted_code): Move and rename to function
1789 tree-ssa-dce.c::simple_dce_from_worklist.
1790 (pass_pre::execute): Update use.
1792 2017-12-05 Richard Biener <rguenther@suse.de>
1794 PR tree-optimization/83277
1795 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1796 to code-gen liveout vars.
1798 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1800 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1801 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1802 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1805 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1807 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1809 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1810 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1813 2017-12-05 Richard Biener <rguenther@suse.de>
1815 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1816 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1817 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1818 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1820 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1822 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1824 2017-12-05 Martin Liska <mliska@suse.cz>
1825 Jakub Jelinek <jakub@redhat.com>
1827 * doc/invoke.texi: Document the options.
1828 * flag-types.h (enum sanitize_code): Add
1829 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1830 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1831 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1832 * opts.c: Define new sanitizer options.
1833 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1834 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1836 2017-12-05 Julia Koval <julia.koval@intel.com>
1838 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1839 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1840 (ix86_handle_option): Handle -mavx512vnni.
1841 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1842 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1843 * config/i386/i386-c (__AVX512VNNI__): New.
1844 * config/i386/i386.c (ix86_target_string): Handle new option.
1845 (ix86_valid_target_attribute_inner_p): Handle new option.
1846 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1847 * config/i386/i386.opt (mavx512vnni): New option.
1849 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1852 * athlon.md: Disable for generic.
1853 * haswell.md: Enable for generic.
1854 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1855 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1857 (ix86_adjust_cost): Move generic to haswell path.
1859 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1861 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1862 instead of MEM_P in a couple more places. Fix formatting issues.
1864 2017-12-04 Jim Wilson <jimw@sifive.com>
1866 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1867 instead of GP_REG_LAST-1.
1868 (riscv_adjust_libcall_cfi_prologue): Likewise.
1869 (riscv_adjust_libcall_cri_epilogue): Likewise.
1870 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1873 2017-12-04 Luis Machado <luis.machado@linaro.org>
1875 * ipa-pure-const.c (check_decl): Add missing newline.
1876 (state_from_flags): Likewise.
1878 2017-12-04 Jeff Law <law@redhat.com>
1880 PR tree-optimizatin/78496
1881 * gimple-ssa-evrp-analyze.h
1882 (evrp_range_analyzer::get_vr_values): Simplify.
1883 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1884 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1885 and gimple-ssa-evrp-analyze.h.
1886 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1887 (simplify_stmt_for_jump_threading): Copy a blob of code from
1888 tree-vrp.c to use ranges to simplify statements.
1889 (dom_opt_dom_walker::before_dom_children): Call
1890 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1891 (dom_opt_dom_walker::after_dom_children): Similarly for
1892 evrp_range_analyzer::leave.
1893 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1896 * gimple-ssa-evrp-analyze.c
1897 (evrp_range_analyzer::extract_range_from_stmt): Always use
1898 vr_values::update_value_range so preexisting range info is
1899 medged with new range info, even if the new range is VR_VARYING.
1901 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1903 * combine.c: Adjust comment.
1904 (use_crosses_set_p): Delete.
1905 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1906 (try_combine): Ditto.
1908 2017-12-04 Richard Biener <rguenther@suse.de>
1910 PR tree-optimization/83255
1911 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1912 Re-add zero-iteration check.
1914 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1916 PR rtl-optimization/83245
1917 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1918 hard registers as earlyclobber, also if not in an asm.
1920 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1924 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1927 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1928 rs6000_cpu based on cpu_index, not tune_index.
1930 2017-12-04 Richard Biener <rguenther@suse.de>
1932 PR tree-optimization/83238
1933 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1934 code match comment, rejecting invalid SESE regions.
1936 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1938 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1939 require base operand is a REG_POINTER prior to reload on targets
1940 with non-equivalent space registers.
1942 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1944 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1945 (update_specialized_profile): Fix updating of counts.
1946 (perhaps_add_new_callers): Likewise.
1948 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1951 * x86-tune.def: Remove obsolette FIXMEs.
1952 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1953 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1954 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1956 (X86_TUNE_PAD_RETURNS): Disable for generic.
1958 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1960 PR tree-optimization/83170
1961 PR tree-optimization/83241
1962 * gimple-ssa-store-merging.c
1963 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1964 gimple_vuse (ins_stmt) in case it has changed.
1965 (imm_store_chain_info::output_merged_store): Likewise.
1967 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1971 * tree-cfg.c (pass_warn_function_return::execute): Handle
1972 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1973 with BUILTINS_LOCATION.
1977 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1978 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1979 instead of DECL_MODE.
1981 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1982 last argument to rtx pointer.
1983 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1984 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1985 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1986 depending on the chosen ISAs.
1987 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1988 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1989 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1991 * config/i386/sse.md (mov<mode>_internal): Likewise.
1992 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1994 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1996 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1998 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2000 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2001 parameters from prototype.
2002 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2003 parameters. Don't print a comment.
2004 (emit_fusion_gpr_load): Adjust.
2005 (emit_fusion_load_store): Adjust.
2006 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2007 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2008 comment on the second line.
2010 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2013 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2014 rs6000_cpu based on cpu_index, not tune_index.
2016 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2018 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2019 which_alternative instead of which_alternative + 1.
2020 (output_asm_insn): Print an extra tab if the template is short.
2022 2017-12-01 Jim Wilson <jimw@sifive.com>
2024 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2026 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2027 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2029 * doc/tm.texi: Regenerate.
2031 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2034 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2035 whether we can allocate pseudos before trying to fix an address.
2036 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2037 memory address is indexed or indirect.
2038 (floatuns_<mode>si2_hw2): Likewise.
2040 2017-12-01 Jason Merrill <jason@redhat.com>
2042 * Makefile.in (TAGS): Add c-family/*.cc.
2044 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2046 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2047 (call_value_insn): Likewise.
2048 (sibcall_insn): Likewise.
2049 (sibcall_value_insn): Likewise.
2050 (movsi_aarch64): Likewise.
2051 (movdi_aarch64): Likewise.
2052 (add_losym_): Likewise.
2053 (ldr_got_small_): Likewise.
2054 (ldr_got_small_sidi): Likewise.
2055 (ldr_got_small_28k_): Likewise.
2056 (ldr_got_small_28k_sidi): Likewise.
2057 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2058 Move output_addr_const to symbolic case. Add error check.
2060 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2062 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2063 * config/s390/s390.md: Disable MVC merging peephole if it would
2064 disable operand forwarding.
2065 (new peephole2): Split MVCs if it would turn them into up to 2
2068 2017-12-01 Richard Biener <rguenther@suse.de>
2070 PR tree-optimization/83232
2071 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2072 detection of same access. Instead of breaking the group here
2073 do not consider the duplicate. Add comment explaining real fix.
2075 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2077 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2079 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2081 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2083 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2085 * function.h (struct function): Remove cilk_frame_decl,
2086 is_cilk_function and calls_cilk_spawn fields.
2087 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2089 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2090 cilk_elemental field.
2091 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2092 * target.def: Adjust comment.
2093 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2094 Don't test cilk_elemental.
2096 PR tree-optimization/83233
2097 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2098 bswap_stat name for the struct.
2101 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2102 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2103 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2104 SWITCH_BREAK_LABEL_P set on the label.
2105 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2106 added for default case if it was missing and not all cases covered.
2107 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2108 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2110 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2111 to avoid -Wimplicit-fallthrough warning.
2112 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2113 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2115 PR tree-optimization/83221
2116 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2118 (init_reassoc): Formatting fix.
2121 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2122 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2126 * tree-cfg.c: Include asan.h.
2127 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2129 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2131 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2133 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2135 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2136 vpcmpeqd instruction.
2138 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2140 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2141 registers type for 128bit mode.
2143 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2145 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2147 * gengtype-state.c (read_state_pair): Likewise.
2148 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2149 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2150 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2151 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2152 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2153 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2154 * builtins.c (fold_builtin_3): Likewise.
2155 * graphite-scop-detection.c
2156 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2157 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2160 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2163 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2164 parameter. Return true if ignore_decl_rtl_set_p is true and other
2165 conditions are satisfied.
2166 * asan.h (asan_protect_global): Add new parameter.
2167 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2168 to asan_protect_global calls.
2170 2017-11-30 Jim Wilson <jimw@sifive.com>
2172 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2173 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2174 -mstrict-align, add info on default value. Delete redundant lines for
2175 -mabi. Add missing -mexplicit-relocs docs.
2177 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2179 * config/arc/arc.md (trap): New pattern.
2181 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2183 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2184 instruction to end into a delay slot.
2185 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2186 can be placed into a delay slot against reg_note.
2188 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2190 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2191 labels number of usages.
2193 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2195 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2197 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2199 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2202 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2203 multiplication by power of 2 constant into two shifts + comparison.
2205 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2208 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2210 2017-11-30 Richard Biener <rguenther@suse.de>
2212 PR tree-optimization/83202
2213 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2214 allow_peel argument and guard peeling.
2215 (canonicalize_loop_induction_variables): Likewise.
2216 (canonicalize_induction_variables): Pass false.
2217 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2218 peeling from cunrolli.
2220 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2222 * combine.c (try_combine): Print a message to dump file whenever
2223 I0, I1, or I2 cannot be combined into I3.
2225 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2227 PR rtl-optimization/83156
2228 PR rtl-optimization/82621
2229 * combine.c (try_combine): Don't split an I2 if one of the dests is
2230 set again before I3. Allow unused dests.
2232 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2234 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2236 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2238 PR rtl-optimization/80818
2239 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2240 recursively. Use insn code for clobber.
2241 (lra_set_insn_recog_data): Pass the new arg to
2242 collect_non_operand_hard_regs.
2243 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2245 (lra_update_insn_regno_info): Pass insn to
2246 add_regs_to_insn_regno_info.
2248 2017-11-29 Jim Wilson <jimw@sifive.com>
2249 Andrew Waterman <andrew@sifive.com>
2251 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2252 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2253 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2254 use SINGLE_SHIFT_COST.
2255 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2256 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2258 2017-11-29 Julia Koval <julia.koval@intel.com>
2260 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2261 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2262 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2263 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2264 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2265 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2266 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2267 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2268 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2269 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2270 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2271 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2272 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2273 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2274 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2275 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2276 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2277 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2278 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2279 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2280 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2281 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2282 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2283 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2284 __builtin_ia32_vpshldv_v2di_mask,
2285 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2286 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2287 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2289 2017-11-29 Julia Koval <julia.koval@intel.com>
2291 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2292 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2293 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2294 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2295 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2296 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2297 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2298 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2299 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2300 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2301 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2302 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2303 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2304 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2305 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2306 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2307 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2308 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2309 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2310 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2311 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2313 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2315 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2316 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2319 2017-11-29 Martin Aberg <maberg@gaisler.com>
2321 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2322 to prevent b2bst errata sequence.
2323 (sqrtdf2_fix): Likewise.
2325 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2327 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2328 (div_sqrt_insn_p): New function.
2329 (sparc_do_work_around_errata): Insert NOP instructions to
2330 prevent sequences that could trigger the TN-0013 errata for
2331 certain LEON3 processors.
2332 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2333 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2334 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2335 (in_branch_delay): Prevent div and sqrt in delay slot if
2337 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2339 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2341 * config/sparc/sparc.c (atomic_insn_p): New function.
2342 (sparc_do_work_around_errata): Insert NOP instructions to
2343 prevent sequences that could trigger the TN-0010 errata for
2345 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2346 instruction referable in atomic_insns_p.
2348 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2350 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2351 (atomic_compare_and_swap_leon3_1): Likewise.
2354 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2356 * config/sparc/sparc.c (fpop_insn_p): New function.
2357 (sparc_do_work_around_errata): Insert NOP instructions to
2358 prevent sequences that could trigger the TN-0012 errata for
2360 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2361 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2362 (in_branch_annul_delay): Prevent floating-point instructions
2363 in delay slot of annulled integer branch.
2365 2017-11-29 Richard Biener <rguenther@suse.de>
2367 PR tree-optimization/83202
2368 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2370 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2371 nodes vectorized to the same stmts multiple times.
2372 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2373 (vect_analyze_slp_instance): Adjust.
2374 (scalar_stmts_to_slp_tree_map_t): New typedef.
2375 (vect_schedule_slp_instance): Add a map recording the SLP node
2376 representing the vectorized stmts for a set of scalar stmts.
2377 Avoid code-generating redundancies.
2378 (vect_schedule_slp): Allocate map and pass it down.
2380 2017-11-29 Nathan Sidwell <nathan@acm.org>
2383 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2386 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2388 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2389 cases for printing LDP/STP memory addresses.
2390 (aarch64_print_address_internal): Renamed from
2391 aarch64_print_operand_address, added parameter, add Pmode check.
2392 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2393 (aarch64_print_operand_address): Indirect to
2394 aarch64_print_address_internal.
2395 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2396 'y' operand output specifier.
2398 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2401 * tree.c (build_simple_mem_ref_loc): Handle
2402 get_addr_base_and_unit_offset returning a MEM_REF.
2405 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2408 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2411 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2413 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2415 (layout::move_to_column): Likewise.
2417 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2420 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2421 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2423 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2424 Martin Jambor <mjambor@suse.cz>
2427 * tree.h (expr_type_first_operand_type_p): Declare
2428 * tree.c (expr_type_first_operand_type_p): New function.
2429 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2430 (ipa_value_from_jfunc): Adjust declaration.
2431 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2432 Use it as result type for arithmetics, unless it is NULL in which case
2433 be more conservative.
2434 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2435 ipa_get_jf_pass_through_result.
2436 (propagate_vals_across_pass_through): Likewise.
2437 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2438 is to propagate_vals_across_pass_through.
2439 (propagate_constants_across_call): Pass PARM_TYPE to
2440 propagate_scalar_across_jump_function.
2441 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2442 ipa_value_from_jfunc.
2443 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2444 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2445 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2446 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2447 target_type, pass it to ipa_value_from_jfunc.
2448 (update_indirect_edges_after_inlining): Pass parameter type to
2449 try_make_edge_direct_simple_call.
2451 2017-11-28 Jeff Law <law@redhat.com>
2453 * gimple-ssa-evrp-analyze.c
2454 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2455 refine ranges if scev_initialized_p returns true.
2456 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2458 2017-11-28 Julia Koval <julia.koval@intel.com>
2460 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2461 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2462 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2463 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2464 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2465 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2466 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2467 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2468 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2469 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2470 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2471 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2472 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2473 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2474 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2475 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2476 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2477 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2478 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2479 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2480 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2481 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2483 2017-11-28 Julia Koval <julia.koval@intel.com>
2485 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2486 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2487 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2488 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2489 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2490 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2491 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2492 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2493 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2494 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2495 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2496 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2497 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2498 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2499 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2500 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2501 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2502 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2503 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2504 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2505 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2506 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2507 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2508 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2509 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2510 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2511 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2512 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2513 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2514 __builtin_ia32_vpshld_v2di_mask): New builtins.
2515 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2516 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2518 2017-11-28 Richard Biener <rguenther@suse.de>
2520 PR tree-optimization/80776
2521 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2523 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2525 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2526 If the incoming edge is an effective fallthru because the other
2527 edge only reaches a __builtin_unreachable () then record ranges
2528 derived from the controlling condition in SSA info.
2529 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2530 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2532 2017-11-28 Olivier Hainque <hainque@adacore.com>
2534 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2535 on mingw build hosts.
2537 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2540 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2542 2017-11-28 Richard Biener <rguenther@suse.de>
2545 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2546 copies generated from memcpy use a character array as reference
2549 2017-11-28 Julia Koval <julia.koval@intel.com>
2550 Sebastian Peryt <sebastian.peryt@intel.com>
2552 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2553 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2554 cilk-common.o, cilk.h, cilk-common.c): Remove.
2556 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2557 * builtins.c (is_builtin_name): Remove cilkplus condition.
2558 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2559 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2560 cilk-builtins.def, cilkplus.def): Remove.
2561 * cif-code.def (CILK_SPAWN): Remove.
2562 * cilk-builtins.def: Delete.
2563 * cilk-common.c: Ditto.
2565 * cilkplus.def: Ditto.
2566 * config/darwin.h (fcilkplus): Delete.
2567 * cppbuiltin.c: Ditto.
2568 * doc/extend.texi: Remove cilkplus doc.
2569 * doc/generic.texi: Ditto.
2570 * doc/invoke.texi: Ditto.
2571 * doc/passes.texi: Ditto.
2572 * gcc.c (fcilkplus): Remove.
2573 * gengtype.c (cilk.h): Remove.
2574 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2576 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2578 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2580 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2581 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2582 cilkplus conditions.
2583 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2584 inline_read_section): Ditto.
2585 * ipa-inline-analysis.c (cilk.h): Remove.
2586 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2587 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2588 append_linker_options): Remove condition for fcilkplus.
2589 * lto/lto-lang.c (cilk.h): Remove.
2590 (lto_init): Remove condition for fcilkplus.
2591 * omp-expand.c (expand_cilk_for_call): Delete.
2592 (expand_omp_taskreg, expand_omp_for_static_chunk,
2593 expand_omp_for): Remove cilkplus
2595 (expand_cilk_for): Delete.
2596 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2597 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2598 execute_lower_omp, diagnose_sb_0): Ditto.
2599 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2600 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2601 * tree-nested.c: Ditto.
2602 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2603 (dump_generic_node): Ditto.
2604 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2605 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2607 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2609 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2611 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2612 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2613 for vector mode and !TARGET_SIMD.
2615 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2617 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2619 * tree.h (SWITCH_LABELS): Remove.
2620 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2621 assert SWITCH_BODY is non-NULL.
2622 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2624 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2626 PR tree-optimization/80788
2627 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2628 has TREE_OVERFLOW set, call drop_tree_overflow.
2630 2017-11-28 Richard Biener <rguenther@suse.de>
2632 PR tree-optimization/83158
2633 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2635 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2638 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2639 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2641 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2643 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2644 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2646 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2649 * varasm.c (compare_constant): Take the mode of the constants into
2650 account when comparing floating point constants.
2652 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2654 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2655 from explicit instantiation of debug_helper.
2656 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2658 2017-11-27 Richard Biener <rguenther@suse.de>
2660 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2663 2017-11-27 Richard Biener <rguenther@suse.de>
2665 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2666 (type_hash_canon): Also clear int_cst_hash_table entry for
2668 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2670 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2672 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2673 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2674 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2676 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2678 * hash-map.h (gt_cleare_cache): Avoid UB.
2680 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2682 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2683 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2684 (decide_unroll_runtime_iterations): Likewise.
2685 (decide_unroll_stupid): Likewise.
2687 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2690 * config/i386/i386.md: Add a loop with incssp.
2692 2017-11-27 Martin Jambor <mjambor@suse.cz>
2694 PR tree-optimization/81248
2695 * tree-sra.c (splice_param_accesses): Remove size check.
2696 (decide_one_param_reduction): Fix size check.
2697 * gimple-pretty-print.c (dump_profile): Silence warning.
2698 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2700 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2703 * dbxout.c (lastlineno): New variable.
2704 (dbx_debug_hooks): Use dbxout_switch_text_section as
2705 switch_text_section debug hook.
2706 (dbxout_function_end): Switch to current_function_section
2707 rather than function_section. If crtl->has_bb_partition,
2708 output just one N_FUN, depending on in_cold_section_p.
2709 (dbxout_source_line): Remember last lineno in lastlineno.
2710 (dbxout_switch_text_section): New function.
2711 (dbxout_function_decl): Adjust dbxout_block caller.
2712 (dbx_block_with_cold_children): New function.
2713 (dbxout_block): Return true if any LBRAC/RBRAC have been
2714 emitted. Use dbx_block_with_cold_children at depth == 0
2715 in second partition. Add PARENT_BLOCKNUM argument, pass
2716 it optionally adjusted to children. Output LBRAC/RBRAC
2717 around recursive call only if the block is in the current
2718 partition, if not and anything was output, emit empty
2720 * final.c (final_scan_insn): Compute cold_function_name
2721 before calling switch_text_section debug hook. Call
2722 that hook even if dwarf2out_do_frame if not emitting
2726 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2727 TREE_READONLY decls.
2729 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2731 PR rtl-optimization/82488
2732 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2734 2017-11-26 Julia Koval <julia.koval@intel.com>
2736 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2738 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2741 2017-11-26 Julia Koval <julia.koval@intel.com>
2743 * config/i386/driver-i386.c (host_detect_local_cpu):
2744 Detect skylake-avx512.
2746 2017-11-26 Julia Koval <julia.koval@intel.com>
2748 * config.gcc: Add -march=cannonlake.
2749 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2750 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2751 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2752 (PTA_CANNONLAKE): New.
2753 (processor_target_table): Add cannonlake.
2754 (ix86_option_override_internal): Ditto.
2755 (fold_builtin_cpu): Ditto.
2756 (get_builtin_code_for_version): Handle cannonlake.
2757 (M_INTEL_COREI7_CANNONLAKE): New.
2758 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2759 * doc/invoke.texi: Add -march=cannonlake.
2761 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2763 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2764 (try_init_one_plugin): Alternative implementation for MinGW.
2765 * Makefile.in (plugin_implib): New.
2766 (gengtype-lex.c): Fix broken AIX workaround.
2767 * configure: Regenerate.
2768 * doc/plugins.texi: Document support for MinGW.
2770 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2772 PR rtl-optimization/81553
2773 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2774 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2775 is a shift where C1 has different mode than the whole shift, use C1's
2776 mode for MULT rather than the shift's mode.
2779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2780 builtins not enabled in the currently selected ISA.
2782 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2784 PR tree-optimization/71026
2785 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2786 (insert_reciprocals): Change to insert reciprocals before a division
2787 by a square and to insert the square of a reciprocal.
2788 (execute_cse_reciprocals_1): Change to consider division by a square.
2789 (register_division_in): Add importance parameter.
2791 2017-11-24 Richard Biener <rguenther@suse.de>
2793 PR tree-optimization/82402
2794 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2795 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2797 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2799 * match.pd (0-ptr): New transformation.
2801 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2804 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2805 when skipping checking.
2807 2017-11-24 Richard Biener <rguenther@suse.de>
2809 PR tree-optimization/83128
2810 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2811 (vn_reference_lookup_3): Likewise.
2813 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2816 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2817 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2820 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2821 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2822 check for the latter with gimple_call_builtin_p. Do not handle
2823 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2825 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2827 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2830 2017-11-23 Julia Koval <julia.koval@intel.com>
2832 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2833 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2834 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2835 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2836 _mm512_maskz_expandloadu_epi16): New intrinsics.
2837 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2838 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2839 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2840 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2841 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2842 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2843 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2844 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2845 _mm256_maskz_expandloadu_epi8): New intrinsics.
2846 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2847 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2848 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2849 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2850 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2851 * config/i386/sse.md (VI248_VLBW): New iterator.
2852 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2854 2017-11-23 Julia Koval <julia.koval@intel.com>
2856 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2857 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2858 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2859 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2860 _mm512_mask_compressstoreu_epi16): New.
2861 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2862 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2863 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2864 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2865 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2866 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2867 _mm256_mask_compressstoreu_epi8): New.
2868 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2869 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2870 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2871 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2872 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2873 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2874 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2875 __builtin_ia32_compresshi128_mask,
2876 __builtin_ia32_compressstoreuqi512_mask,
2877 __builtin_ia32_compressstoreuhi512_mask,
2878 __builtin_ia32_compressstoreuqi256_mask,
2879 __builtin_ia32_compressstoreuqi128_mask,
2880 __builtin_ia32_compressstoreuhi256_mask,
2881 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2882 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2884 (ix86_expand_special_args_builtin): Handle new types.
2885 (s4fma_expand): Handle new builtin array.
2886 * config/i386/immintrin.h: Include new headers.
2887 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2888 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2890 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2893 fixes. Declare temp and g variables at the top in order to avoid
2894 {} in most of the cases.
2896 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2898 * match.pd (ptr-0): New transformation.
2900 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2902 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2903 (struct addr_mode_cost_table): New.
2904 (struct tune_params): Add field addr_mode_costs.
2905 * config/arm/arm.c (generic_addr_mode_costs): New.
2906 (arm_slowmul_tune): Initialise addr_mode_costs field.
2907 (arm_fastmul_tune): Likewise.
2908 (arm_strongarm_tune): Likewise.
2909 (arm_xscale_tune): Likewise.
2910 (arm_9e_tune): Likewise.
2911 (arm_marvell_pj4_tune): Likewise.
2912 (arm_v6t2_tune): Likewise.
2913 (arm_cortex_tune): Likewise.
2914 (arm_cortex_a8_tune): Likewise.
2915 (arm_cortex_a7_tune): Likewise.
2916 (arm_cortex_a15_tune): Likewise.
2917 (arm_cortex_a35_tune): Likewise.
2918 (arm_cortex_a53_tune): Likewise.
2919 (arm_cortex_a57_tune): Likewise.
2920 (arm_exynosm1_tune): Likewise.
2921 (arm_xgene1_tune): Likewise.
2922 (arm_cortex_a5_tune): Likewise.
2923 (arm_cortex_a9_tune): Likewise.
2924 (arm_cortex_a12_tune): Likewise.
2925 (arm_cortex_a73_tune): Likewise.
2926 (arm_v7m_tune): Likewise.
2927 (arm_cortex_m7_tune): Likewise.
2928 (arm_v6m_tune): Likewise.
2929 (arm_fa726te_tune): Likewise.
2930 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2932 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2934 * config/arm/arm.c (arm_mem_costs): New function.
2935 (arm_rtx_costs_internal): Use arm_mem_costs.
2937 2017-11-23 Mark Wielaard <mark@klomp.org>
2939 * dwarf2out.c (init_sections_and_labels): Use generation to create
2940 unique ranges_section_label and ranges_base_label. Return generation.
2941 (output_rnglists): Add generation argument. Use generation to create
2942 unique ranges labels.
2943 (dwarf2out_finish): Get generation from init_sections_and_labels
2944 and pass generation to output_rnglists.
2946 2017-11-23 Mike Stump <mikestump@comcast.net>
2947 Eric Botcazou <ebotcazou@adacore.com>
2949 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2950 * cfgloop.h (struct loop): Add unroll field.
2951 * function.h (struct function): Add has_unroll bitfield.
2952 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2953 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2954 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2955 (pass_rtl_unroll_loops::gate): Likewise.
2956 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2957 for which loop->unroll==1.
2958 (decide_unroll_constant_iterations): Use note for consistency and
2959 take loop->unroll into account. Return early if loop->unroll is set.
2960 Fix thinko in existing test.
2961 (decide_unroll_runtime_iterations): Use note for consistency and
2962 take loop->unroll into account.
2963 (decide_unroll_stupid): Likewise.
2964 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2965 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2966 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2968 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2969 (print_loop): Print loop->unroll if set.
2970 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2971 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2972 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2974 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2975 loop->unroll is set and smaller than the trip count. Otherwise bypass
2976 entirely the heuristics if loop->unroll is set. Remove dead note.
2977 Fix off-by-one bug in other note.
2978 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2979 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2981 (tree_unroll_loops_completely): Make static.
2982 (pass_complete_unroll::execute): Use correct type for variable.
2983 (pass_complete_unrolli::execute): Fix formatting.
2984 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2986 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2988 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2989 enable when TARGET_PREFER_AVX128 is set.
2991 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2993 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2995 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2996 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2998 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3000 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3001 frequencies. Fix estimation of aggregate parameters.
3003 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3005 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3008 2017-11-23 Tom de Vries <tom@codesourcery.com>
3010 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3012 2017-11-23 Tom de Vries <tom@codesourcery.com>
3014 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3017 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3020 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3021 sibcall_value_pcrel_fdpic): Use local variable instead of
3023 (calli_tbr_rel): Add missing operand 2.
3024 (call_valuei_tbr_rel): Add missing operand 3.
3026 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3029 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3030 bitpos/bitsize covering the whole destination, use store_expr only if
3031 the complex mode is the same. Otherwise, use expand_normal and if
3032 it returns CONCAT, subreg each part separately instead of trying to
3033 subreg the whole result.
3035 2017-11-23 Richard Biener <rguenther@suse.de>
3037 PR tree-optimization/23094
3038 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3039 come here from walking over backedges in the first iteration.
3040 (vn_reference_lookup_3): Skip clobbers that store the same value.
3042 2017-11-23 Richard Biener <rguenther@suse.de>
3044 PR tree-optimization/81403
3045 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3046 a block we need a leader relative to.
3047 (phi_translate_1): For nary processing require a leader from
3048 get_representative_for given we run expression simplification
3049 using match-and-simplify. Remove previous fix.
3051 2017-11-22 Jeff Law <law@redhat.com>
3053 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3054 Use new method allocate_value_range rather than accessing the
3055 vrp_value_range_pool data member directly.
3056 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3057 to use extract_range_from_stmt method to avoid need for
3058 extract_range_from_assignment method.
3059 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3060 method rather than setting values_propgated data member directly.
3061 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3062 and values propagated data members and extract_range_from_assignment
3063 method. Reorder private data members to conform to standards.
3064 Add new methods set_lattice_propagation_complete and
3065 allocate_value_range.
3067 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3069 PR rtl-optimization/83030
3070 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3071 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3072 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3073 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3074 CROSSING_JUMP_P flag is already set before setting it.
3075 * cfgrtl.c (fixup_partition_crossing): Likewise.
3076 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3079 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3081 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3082 Handle the case where both arguments are using gen_const_vec_series.
3084 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3087 * pretty-print.c (pp_format): Move quoting implementation to
3088 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3089 to pass address of "quote" local.
3090 (pp_begin_quote): New function.
3091 (pp_end_quote): New function.
3092 * pretty-print.h (printer_fn): Convert penultimate param from bool
3094 (pp_begin_quote): New decl.
3095 (pp_end_quote): New decl.
3096 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3097 param from bool to bool *.
3098 * tree-diagnostic.h (default_tree_printer): Likewise.
3100 2017-11-22 Jeff Law <law@redhat.com>
3102 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3103 Initialize vr_values.
3104 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3105 vr_values via vr_values class instance rather than delegators.
3106 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3107 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3108 (evrp_range_analyzer::push_value_range): Likewise.
3109 (evrp_range_analyzer::pop_value_range): Likewise.
3110 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3111 most delegators. Those remaining are exposed as public interfaces.
3112 Make vr_values a pointer and private.
3113 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3115 (evrp_range_analyzer::get_vr_value): New method.
3116 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3117 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3118 (class evrp_dom_walker): Attach evrp_folder class, initialize
3119 it in the ctor. Remove temporary delegators.
3120 (evrp_dom_walker::before_dom_children): Call methods in attached
3121 evrp_range_analyzer class via class instance pointer. Use
3122 free value_range_constant_singleton to remove need for
3123 op_with_constant_singleton_value delegator method. Do not
3124 create a vrp_prop class instance for every call! Narrow
3125 scope of a couple variables.
3126 (evrp_dom_walker::cleanup): Call methods in attached
3127 evrp_range_analyzer class via class instance pointer.
3128 * vr-values.h (class vr_values): Privatize many methods and
3131 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3133 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3134 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3135 * expr.c (expand_expr_real_2): Likewise.
3136 * fold-const.c (const_unop): Likewise.
3137 * optabs-tree.c (optab_for_tree_code): Likewise.
3138 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3139 * tree-inline.c (estimate_operator_cost): Likewise.
3140 * tree-pretty-print.c (dump_generic_node): Likewise.
3141 (op_code_prio): Likewise.
3142 (op_symbol_code): Likewise.
3143 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3144 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3145 * internal-fn.c (direct_internal_fn_optab): New function.
3146 (direct_internal_fn_array, direct_internal_fn_supported_p
3147 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3148 * fold-const-call.c (fold_const_reduction): New function.
3149 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3151 * tree-vect-loop.c: Include internal-fn.h.
3152 (reduction_code_for_scalar_code): Rename to...
3153 (reduction_fn_for_scalar_code): ...this and return an internal
3155 (vect_model_reduction_cost): Take an internal_fn rather than
3157 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3159 (vectorizable_reduction): Use internal functions rather than tree
3160 codes for the reduction operation. Update calls to the functions
3162 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3163 Use calls to internal functions rather than REDUC tree codes.
3164 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3166 2017-11-22 Olivier Hainque <hainque@adacore.com>
3168 * config/vxworks.c (vxworks_override_options): Pick default
3169 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3170 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3171 DWARF_GNAT_ENCODINGS_DEFAULT.
3172 * config/vxworksae.h: Likewise.
3174 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3176 PR tree-optimization/83104
3177 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3180 2017-11-22 Marek Polacek <polacek@redhat.com>
3181 H.J. Lu <hongjiu.lu@intel.com>
3182 Jason Merrill <jason@redhat.com>
3187 * calls.c (initialize_argument_information): Call
3188 warn_parameter_passing_abi target hook.
3189 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3190 argument onto stack.
3191 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3192 * common.opt: Update -fabi-version description.
3193 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3194 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3196 (ix86_is_empty_record): New function.
3197 (ix86_warn_parameter_passing_abi): New function.
3198 (TARGET_EMPTY_RECORD_P): Redefine.
3199 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3200 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3201 * doc/tm.texi: Regenerated.
3202 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3203 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3204 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3205 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3206 instead of int_size_in_bytes.
3207 * expr.c (copy_blkmode_to_reg): Likewise.
3208 * function.c (aggregate_value_p): Return 0 for empty types.
3209 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3211 (locate_and_pad_parm): Call arg size_in_bytes instead
3213 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3214 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3215 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3217 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3218 (std_gimplify_va_arg_expr): Skip empty records. Call
3219 arg_size_in_bytes instead size_in_bytes.
3220 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3221 * tree-core.h (tree_type_common): Add empty_flag.
3222 (tree_decl_common): Update comments.
3223 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3225 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3226 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3228 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3229 * tree.c (default_is_empty_type): New function.
3230 (default_is_empty_record): New function.
3231 (arg_int_size_in_bytes): New function.
3232 (arg_size_in_bytes): New function.
3233 (get_ultimate_context): New function.
3234 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3235 TRANSLATION_UNIT_WARN_EMPTY_P.
3236 (default_is_empty_record, arg_int_size_in_bytes,
3237 arg_size_in_bytes, get_ultimate_context): Declare.
3239 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3241 * config/arm/arm.c (cmse_clear_registers): New function.
3242 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3243 code by call to cmse_clear_registers.
3244 (cmse_nonsecure_entry_clear_before_return): Likewise.
3246 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3248 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3249 (vdot_s32, vdotq_s32): New.
3250 (vdot_lane_u32, vdotq_lane_u32): New.
3251 (vdot_lane_s32, vdotq_lane_s32): New.
3254 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3257 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3258 for unsigned values with fewer HWIs than the precision.
3259 (test_overflow): New function.
3260 (wide_int_cc_tests): Call it.
3262 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3264 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3265 has the same mode class as Pmode.
3267 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3269 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3270 padding_bits_to_clear_ptr.
3271 (cmse_nonsecure_entry_clear_before_return): Likewise.
3273 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3275 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3276 auto_sbitap instead of integer bitfield to control register needing
3279 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3281 PR tree-optimization/83044
3282 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3283 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3284 that need the upper bound. Subtract offset from
3285 get_addr_base_and_unit_offset only if positive and subtract it
3286 before division by eltsize rather than after it.
3289 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3290 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3291 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3295 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3296 expand_binop with *mul_widen_optab, make sure at least one of the
3297 operands doesn't have VOIDmode.
3300 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3302 PR rtl-optimization/82044
3303 PR tree-optimization/82042
3304 * dse.c (record_store): Check for overflow.
3305 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3306 add_wild_read instead of clear_rhs_from_active_local_stores on
3307 overflow and log it into dump_file.
3309 2017-11-22 Richard Biener <rguenther@suse.de>
3311 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3312 fake edges to exit when looking for a place to insert.
3313 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3315 (insert_into_preds_of_block): Commit edge insertion immediately,
3316 assert that doesn't require new BBs.
3317 (fini_pre): Release expressions.
3318 (pass_pre::execute): Shuffle things around a bit, if the fn
3319 is too large do not compute AVAIL either as this is really the
3322 2017-11-22 Richard Biener <rguenther@suse.de>
3324 PR tree-optimization/83089
3325 * tree-if-conv.c (pass_if_conversion::execute): If anything
3326 changed reset SCEV and free the number of iteration estimates.
3328 2017-11-21 Martin Sebor <msebor@redhat.com>
3330 PR tree-optimization/82945
3331 * calls.h (warn_nonstring_bound): Remove unused function.
3333 2017-11-21 Martin Sebor <msebor@redhat.com>
3335 PR tree-optimization/82945
3336 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3337 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3338 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3340 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3341 * calls.h (get_attr_nonstring_decl): Declare new function.
3342 * doc/extend.texi (attribute nonstring): Update.
3343 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3344 get_attr_nonstring_decl and handle it.
3345 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3346 detection of nul-termination.
3347 (strlen_to_stridx): Change to a pointer.
3348 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3349 (pass_strlen::execute): Same.
3351 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3353 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3354 for the new option -mprefer-vector-width=[none|128|256|512].
3355 * config/i386/i386.c (ix86_target_string): remove old style options
3356 -mprefer-avx256 and make -mprefer-avx128 as alias.
3357 (ix86_option_override_internal): Apply defaults for the
3358 -mprefer-vector-width=[128|256] option.
3359 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3360 Implement macros to work with -mprefer-vector-width=.
3361 * config/i386/i386.opt: Implemented option
3362 -mprefer-vector-width=[none|128|256|512].
3363 * doc/invoke.texi: Documentation for
3364 -mprefer-vector-width=[none|128|256|512].
3366 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3368 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3370 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3372 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3374 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3376 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3378 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3379 * expr.c (expand_expr_real_2): Likewise.
3380 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3381 fold_binary_loc): Likewise.
3382 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3383 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3384 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3385 MINUS_EXPR transformations.
3386 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3387 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3388 * tree-inline.c (estimate_operator_cost): Likewise.
3389 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3390 op_symbol_code): Likewise.
3391 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3392 * vr-values.c (extract_range_from_binary_expr): Likewise.
3393 * varasm.c (initializer_constant_valid_p_1): Likewise.
3394 * tree.def: New tree code POINTER_DIFF_EXPR.
3396 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3398 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3399 integer suffix to movbe mnemonic.
3400 (*bswaphi2_movbe): Ditto.
3401 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3403 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3406 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3407 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3408 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3410 2017-11-21 Martin Liska <mliska@suse.cz>
3412 * tree-inline.c (expand_call_inline): Remove not needed
3415 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3416 Jakub Jelinek <jakub@redhat.com>
3419 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3420 Remove static keyword from f variable.
3422 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3424 PR tree-optimization/83086
3425 * gimple-ssa-store-merging.c
3426 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3427 rather than n.base_addr.
3429 2017-11-21 Martin Liska <mliska@suse.cz>
3431 PR rtl-optimization/82044
3432 PR tree-optimization/82042
3433 * dse.c (check_mem_read_rtx): Check for overflow.
3435 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3437 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3440 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3442 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3443 and its corresponding call arg location note.
3445 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3446 Andrew Burgess <andrew.burgess@embecosm.com>
3448 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3450 (arc_return_slot_offset): Likewise.
3451 (arc_eh_return_address_location): New declaration.
3452 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3453 (MUST_SAVE_REGISTER): Add exception handler case.
3454 (MUST_SAVE_RETURN_ADDR): Likewise.
3455 (arc_frame_pointer_required): Likewise.
3456 (arc_frame_pointer_needed): New function.
3457 (arc_compute_frame_size): Changed.
3458 (arc_expand_prologue): Likewise.
3459 (arc_expand_epilogue): Likewise.
3460 (arc_initial_elimination_offset): Likewise.
3461 (arc_return_slot_offset): Delete.
3462 (arc_eh_return_address_location): New function.
3463 (arc_builtin_setjmp_frame_value): Likewise.
3464 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3465 (EH_RETURN_STACKADJ_RTX): Define.
3466 (EH_RETURN_HANDLER_RTX): Likewise.
3467 * config/arc/arc.md (eh_return): Delete.
3469 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3471 * print-tree.h (debug_vec_tree): Remove prototype.
3472 * gdbinit.in (pvt): Remove macro.
3474 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3476 PR tree-optimization/83047
3477 * gimple-ssa-store-merging.c
3478 (imm_store_chain_info::output_merged_store): If the loads with the
3479 same vuse are in different basic blocks, for load_gsi pick a load
3480 location that is dominated by the other loads.
3483 * config/i386/i386.c (ix86_memmodel_check): Start
3484 -Winvalid-memory-model diagnostics with lowercase letter.
3487 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3488 set in_cold_section_p to the partition containing loc_list->first.
3489 When seeing loc_list->last_before_switch node, update secname and
3490 perform range_across_switch second partition handling only after that.
3493 * run-rtl-passes.c: Include debug.h.
3494 (run_rtl_passes): Call debug_hooks->assembly_start.
3495 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3499 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3500 OPTAB_DIRECT in calls to expand_simple_binop.
3502 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3505 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3507 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3510 * spellcheck.h (best_match::blithely_get_best_candidate): New
3513 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3515 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3516 with lower case letter.
3518 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3520 * config/i386/i386.md (bswaphi2): New expander.
3521 (*bswaphi2_movbe): New insn pattern.
3522 (bswaphi -> rorhi pepehole2): New peephole pattern.
3524 2017-11-20 Jeff Law <law@redhat.com>
3526 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3527 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3528 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3529 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3530 gimple-ssa-evrp-analyze.h.
3532 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3533 set BB_VISITED here.
3534 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3536 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3538 * tree-predcom.c: Add general comment on Store-Store chains.
3539 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3541 (get_chain_last_ref_at): Rename into...
3542 (get_chain_last_write_at): ...this.
3543 (get_chain_last_write_before_load): New function.
3544 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3545 CT_STORE_STORE when write reference is added.
3546 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3547 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3548 (initialize_root_vars_store_elim_1): Ditto.
3549 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3550 definition is created.
3551 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3552 chain by replacing it with dominant stored value.
3554 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3556 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3558 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3560 * vr-values.c (extract_range_from_binary_expr): Use a full range
3563 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3565 * config/arm/arm.md (R4_REGNUM): Define constant.
3566 (nonsecure_call_internal): Remove r4 clobber.
3567 (nonsecure_call_value_internal): Likewise.
3568 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3569 clobber and resequence match_operands.
3570 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3571 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3572 (nonsecure_call_value_reg_thumb2): Likewise.
3574 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3576 PR tree-optimization/78821
3577 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3578 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3579 to the address of the base rather than the base itself.
3580 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3581 (find_bswap_or_nop_finalize): New function.
3582 (find_bswap_or_nop): Use it.
3583 (bswap_replace): Return a tree rather than bool, change first
3584 argument from gimple * to gimple_stmt_iterator, allow inserting
3585 into an empty sequence, allow ins_stmt to be NULL - then emit
3586 all stmts into gsi. Fix up MEM_REF address gimplification.
3587 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3588 (struct store_immediate_info): Add N and INS_STMT non-static
3590 (store_immediate_info::store_immediate_info): Initialize them
3591 from newly added ctor args.
3592 (merged_store_group::apply_stores): Formatting fixes. Sort by
3594 (stmts_may_clobber_ref_p): For stores call also
3595 refs_anti_dependent_p.
3596 (gather_bswap_load_refs): New function.
3597 (imm_store_chain_info::try_coalesce_bswap): New method.
3598 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3599 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3600 (imm_store_chain_info::output_merged_store): Fail if number of
3601 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3602 and NOP_EXPR rhs_code.
3603 (pass_store_merging::process_store): Compute n and ins_stmt, if
3604 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3605 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3607 (pass_store_merging::execute): Calculate dominators.
3609 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3610 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3611 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3612 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3613 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3614 class pass_optimize_bswap, bswap_replace,
3615 pass_optimize_bswap::execute): Moved to ...
3616 * gimple-ssa-store-merging.c: ... this file.
3617 Include optabs-tree.h.
3618 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3619 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3620 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3621 anonymous namespace, remove static keywords.
3622 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3623 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3625 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3628 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3630 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3632 * vec.h (debug_helper): New function.
3633 (DEFINE_DEBUG_VEC): New macro.
3634 * hash-set.h (debug_helper): New function.
3635 (DEFINE_DEBUG_HASH_SET): New macro.
3636 * cfg.c (debug_slim (edge)): New function.
3637 Call DEFINE_DEBUG_VEC for edges.
3638 Call DEFINE_DEBUG_HASH_SET for edges.
3639 * cfghooks.c (debug_slim (basic_block)): New function.
3640 Call DEFINE_DEBUG_VEC for basic blocks.
3641 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3642 * print-tree.c (debug_slim): New function to handle trees.
3643 Call DEFINE_DEBUG_VEC for trees.
3644 Call DEFINE_DEBUG_HASH_SET for trees.
3645 (debug (vec<tree, va_gc>) &): Remove.
3646 (debug (<vec<tree, va_gc>) *): Remove.
3647 * print-rtl.c (debug_slim): New function to handle const_rtx.
3648 Call DEFINE_DEBUG_VEC for rtx_def.
3649 Call DEFINE_DEBUG_VEC for rtx_insn.
3650 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3651 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3652 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3653 (debug (vec<rtx_insn *> *ptr): Remove.
3654 (debug_insn_vector): Remove.
3655 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3657 2017-11-20 Tom de Vries <tom@codesourcery.com>
3659 PR rtl-optimization/82020
3660 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3661 IF_THEN_ELSE condition.
3663 2017-11-19 Jeff Law <law@redhat.com>
3665 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3666 of degenerates resulting from ignoring an edge.
3668 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3671 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3673 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3676 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3677 for uninitialied counts.
3679 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3682 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3684 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3687 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3689 2017-11-19 Tom de Vries <tom@codesourcery.com>
3691 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3694 2017-11-19 Tom de Vries <tom@codesourcery.com>
3696 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3699 2017-11-19 Tom de Vries <tom@codesourcery.com>
3701 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3704 2017-11-19 Tom de Vries <tom@codesourcery.com>
3706 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3708 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3710 2017-11-19 Tom de Vries <tom@codesourcery.com>
3712 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3714 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3715 semicolon after MCORE_EXPORT_NAME call.
3717 2017-11-19 Tom de Vries <tom@codesourcery.com>
3720 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3721 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3723 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3725 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3726 (cgraph_edge::clone): Cleanup updating of profile.
3727 * ipa-cp.c (update_profiling_info): Likewise.
3728 * ipa-inline-transform.c (inline_transform): Likewise.
3729 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3730 * ipa-split.c (execute_split_functions): Do not split when function
3732 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3733 * profile-count.c (profile_count::dump): Add global0.
3734 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3736 (profile_count::to_sreal_scale): Likewise.
3737 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3738 (profile_count::combine_with_ipa_count): New function.
3739 * profile-count.h (profile_guessed_global0adjusted): New.
3740 (profile_count::adjusted_zero): New.
3741 (profile_count::global0adjusted): New.
3742 (profile_count::combine_with_ipa_count): New.
3743 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3744 correct profile of return block of split functions.
3745 (copy_cfg_body): Remove unused profile_count.
3746 (copy_body): Likewise.
3747 (expand_call_inline): Update.
3748 (tree_function_versioning): Update.
3750 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3752 * hash-set.h (hash_set::empty): New.
3753 * tree-ssa-threadbackward.h: Delete.
3754 * tree-ssa-threadbackward.c (class thread_jumps): New.
3755 Move max_threaded_paths into class.
3756 (fsm_find_thread_path): Remove arguments that are now in class.
3757 (profitable_jump_thread_path): Rename to...
3758 (thread_jumps::profitable_jump_thread_path): ...this.
3759 (convert_and_register_jump_thread_path): Rename to...
3760 (thread_jumps::convert_and_register_current_path): ...this.
3761 (check_subpath_and_update_thread_path): Rename to...
3762 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3763 (register_jump_thread_path_if_profitable): Rename to...
3764 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3765 (handle_phi): Rename to...
3766 (thread_jumps::handle_phi): ...this.
3767 (handle_assignment): Rename to...
3768 (thread_jumps::handle_assignment): ...this.
3769 (fsm_find_control_statement_thread_paths): Rename to...
3770 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3771 (find_jump_threads_backwards): Rename to...
3772 (thread_jumps::find_jump_threads_backwards): ...this.
3773 Initialize path local data.
3774 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3775 from within thread_jumps class.
3776 (pass_early_thread_jumps::execute): Same.
3778 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3780 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3782 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3784 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3785 to canonical location.
3787 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3790 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3792 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3794 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3796 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3797 rcrt1.o%s/grcrt1.o%s for -static-pie.
3799 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3801 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3802 ix86_shift_rotate_cost): Break out from ...
3803 (ix86_rtx_costs): ... here.
3804 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3807 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3809 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3810 when entry block was promoted unlikely.
3811 (estimate_bb_frequencies): Increase frequency scale.
3812 * profile-count.h (profile_count): Export precision info.
3814 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3816 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3817 disturbin profile of entry block.
3819 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3821 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3824 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3826 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3827 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3829 2017-11-17 Jeff Law <law@redhat.com>
3831 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3832 from evrp_dom_walker class. Various methods moved into new class.
3833 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3834 (evrp_range_analyzer::enter): New method.
3835 (evrp_range_analyzer::leave): New method.
3836 (evrp_dom_walker): Remove delegators no longer needed by this class.
3837 Replace vr_values data member with evrp_range_analyzer
3839 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3840 method extracted from evrp_dom_walker::before_dom_children.
3841 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3842 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3844 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3845 Add private copy constructor and move assignment operators.
3846 Privatize methods and class data where trivially possible.
3847 (evrp_dom_walker::cleanup): New function, extracted from
3848 execute_early_vrp. Simplify access to class data.
3850 * vr-values.h (get_output_for_vrp): Prototype.
3851 * vr-values.c (get_output_for_vrp): New function extracted from
3852 vrp_visit_assignment_or_call and extract_range_from_stmt.
3853 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3855 2017-11-17 Luis Machado <luis.machado@linaro.org>
3857 * config/aarch64/aarch64.c
3858 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3859 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3860 tune_params::AUTOPREFETCHER_WEAK.
3862 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3865 * config/arm/arm.c (arm_valid_target_attribute_rec):
3866 Parse "arch=" and "+<ext>".
3867 (arm_valid_target_attribute_tree): Re-init global options.
3868 (arm_option_override): Make non-static.
3869 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3870 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3871 (__ARM_FEATURE_CRC32): Support undef.
3872 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3873 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3875 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3877 * gdbinit.in (break-on-diagnostic): New command.
3879 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3881 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3882 return to indirect jump for EH return if control-flow
3883 protection is enabled. Change explicit 'false' argument in
3884 pro_epilogue_adjust_stack with a value of
3886 * config/i386/i386.md (simple_return_indirect_internal):
3887 Remove SImode restriction to support 64-bit.
3889 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3891 * combine.c (added_notes_insn): New.
3892 (try_combine): Handle added_notes_insn like added_links_insn.
3893 Rewrite return value code.
3894 (distribute_notes): Set added_notes_insn to the earliest insn we added
3897 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3899 PR rtl-optimization/82621
3900 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3901 dest of one of those SETs is unused.
3903 2017-11-17 Richard Biener <rguenther@suse.de>
3906 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3907 * tree-pretty-print.c (dump_generic_node): Handle
3908 annot_expr_parallel_kind.
3909 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3910 * gimplify.c (gimple_boolify): Likewise.
3912 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3914 * config.gcc (extra_headers): Add cet.h for x86 targets.
3915 * config/i386/cet.h: New file.
3916 * doc/install.texi: Add --enable-cet/--disable-cet.
3918 2017-11-17 Richard Biener <rguenther@suse.de>
3920 PR tree-optimization/83017
3921 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3922 (gen_parallel_loop): Properly count iterations.
3923 (parallelize_loops): Handle loop->can_be_parallel independent
3924 of flag_loop_parallelize_all. Make static profitability test match
3926 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3927 * invoke.texi (parloops-min-per-thread): Document.
3929 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3931 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3932 upstreaming review comments.
3934 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3936 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3937 with fast unaligned access.
3938 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3940 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3942 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3943 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3945 2017-11-17 Richard Biener <rguenther@suse.de>
3947 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3948 folding of references.
3950 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3953 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3954 of replacing call to strncmp with corresponding call to strcmp when
3957 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3959 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3960 option prefer-avx256 for skylake-avx512 configuration.
3961 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3962 (get_builtin_code_for_version): Ditto.
3964 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3965 Monk Chiang <sh.chiang04@gmail.com>
3967 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3968 (FIXED_REGISTERS): Reserve more register numbers.
3969 (CALL_USED_REGISTERS): Likewise.
3970 (REG_ALLOC_ORDER): Likewise.
3971 (REG_CLASS_CONTENTS): Likewise.
3972 (REGISTER_NAMES): Likewise.
3974 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3975 Kito Cheng <kito.cheng@gmail.com>
3977 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3979 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3981 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3983 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3984 parameter list for vec_splats.
3986 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3988 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3990 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3992 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3995 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3996 with the required version of automake 1.11.6.
3998 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4000 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4002 (DU_C2_3_power9): Correct reservation combinations.
4003 (FP_DIV_power9, VEC_DIV_power9): New.
4004 (power9-alu): Split out rotate/shift...
4005 (power9-rot): ...to here, correct dispatch resource.
4006 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4008 (power9-fp): Correct latency.
4009 (power9-sdiv): Add div/sqrt resource.
4010 (power9-ddiv): Correct latency, add div/sqrt resource.
4011 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4012 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4014 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4016 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4018 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4019 switch statement mapping KF built-ins to TF built-ins if we don't
4020 have the proper ISA 3.0 assembler support.
4022 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4024 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4025 (lower_emutls_function_body): Do not compute it.
4027 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4029 * ipa-split.c (split_bb_info): Turn time to sreal.
4030 (split_point): Likewise.
4031 (dump_split_point): Likewise.
4032 (fine_split_points): Likewise.
4033 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4035 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4037 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4040 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4042 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4044 (expensive_function_p): Remove useless assert.
4045 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4047 2017-11-16 Martin Sebor <msebor@redhat.com>
4049 PR tree-optimization/82588
4050 PR tree-optimization/82583
4051 * tree-vrp.c (check_array_ref): Handle flexible array members,
4052 string literals, and inner indices.
4053 (search_for_addr_array): Add detail to diagnostics.
4055 2017-11-16 Nathan Sidwell <nathan@acm.org>
4059 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4060 (SET_DECL_ASSEMBLER_NAME): Forward to
4061 overwrite_decl_assembler_name.
4062 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4063 (overwrite_decl_assembler_name): Declare.
4064 * tree.c (overwrite_decl_assembler_name): New.
4065 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4066 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4067 (LANG_HOOKS_INITIALIZER): Add it.
4068 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4069 * langhooks.c (lhd_set_decl_assembler_name): Use
4070 SET_DECL_ASSEMBLER_NAME.
4071 (lhd_overwrite_decl_assembler_name): Default implementation.
4073 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4074 Jackson Woodruff <jackson.woodruff@arm.com>
4076 PR tree-optimization/71026
4077 * match.pd: Canonicalize constant multiplies in division.
4079 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4081 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4082 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4083 Remove OPT_fomit_frame_pointer entry.
4084 * common/config/arc/arc-common.c: Likewise.
4085 * common/config/arm/arm-common.c: Likewise.
4086 * common/config/avr/avr-common.c: Likewise.
4087 * common/config/c6x/c6x-common.c: Likewise.
4088 * common/config/cr16/cr16-common.c: Likewise.
4089 * common/config/cris/cris-common.c: Likewise.
4090 * common/config/epiphany/epiphany-common.c: Likewise.
4091 * common/config/fr30/fr30-common.c: Likewise.
4092 * common/config/frv/frv-common.c: Likewise.
4093 * common/config/ia64/ia64-common.c: Likewise.
4094 * common/config/iq2000/iq2000-common.c: Likewise.
4095 * common/config/lm32/lm32-common.c: Likewise.
4096 * common/config/m32r/m32r-common.c: Likewise.
4097 * common/config/mcore/mcore-common.c: Likewise.
4098 * common/config/microblaze/microblaze-common.c: Likewise.
4099 * common/config/mips/mips-common.c: Likewise.
4100 * common/config/mmix/mmix-common.c: Likewise.
4101 * common/config/mn10300/mn10300-common.c: Likewise.
4102 * common/config/nios2/nios2-common.c: Likewise.
4103 * common/config/pa/pa-common.c: Likewise.
4104 * common/config/pdp11/pdp11-common.c: Likewise.
4105 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4106 * common/config/riscv/riscv-common.c: Likewise.
4107 * common/config/rs6000/rs6000-common.c: Likewise.
4108 * common/config/rx/rx-common.c: Likewise.
4109 * common/config/s390/s390-common.c: Likewise.
4110 * common/config/sh/sh-common.c: Likewise.
4111 * common/config/sparc/sparc-common.c: Likewise.
4112 * common/config/tilegx/tilegx-common.c: Likewise.
4113 * common/config/tilepro/tilepro-common.c: Likewise.
4114 * common/config/v850/v850-common.c: Likewise.
4115 * common/config/visium/visium-common.c: Likewise.
4116 * common/config/xstormy16/xstormy16-common.c: Likewise.
4117 * common/config/xtensa/xtensa-common.c: Likewise.
4118 * invoke.texi (-fomit-frame-pointer): Update documentation.
4120 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4122 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4124 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4126 * cfg.c (scale_bbs_frequencies_int,
4127 cale_bbs_frequencies_gcov_type): Remove.
4128 * cfg.h (scale_bbs_frequencies_int,
4129 cale_bbs_frequencies_gcov_type): Remove.
4131 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4133 * tree-ssa-loop-manip.c
4134 (scale_dominated_blocks_in_loop): Update to profile counts.
4135 (tree_transform_and_unroll_loop): Likewise.
4137 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4139 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4140 scale_bbs_frequencies_int.
4142 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4144 * final.c (compute_alignments): Use counts rather than frequencies.
4146 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4148 * cfgloopanal.c: Include sreal.h
4149 (average_num_loop_insns): Use counts and sreal for accounting.
4151 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4153 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4156 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4157 Kito Cheng <kito.cheng@gmail.com>
4159 * config/nds32/constraints.md: Provide more constraints.
4160 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4161 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4162 support constraints usage.
4164 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4166 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4168 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4169 Kito Cheng <kito.cheng@gmail.com>
4171 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4172 * config/nds32/nds32.opt: Refine the layout.
4173 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4174 TARGET_EXT_STRING): Support new options.
4175 * config/nds32/nds32.h: Likewise.
4176 * config/nds32/nds32.md: Likewise.
4177 * config/nds32/nds32-predicates.c: Likewise.
4178 * config/nds32/constraints.md: Likewise.
4179 * common/config/nds32/nds32-common.c: Likewise.
4181 2017-11-16 Julia Koval <julia.koval@intel.com>
4184 * config/i386/gfniintrin.h: Add sse check.
4185 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4187 2017-11-16 Julia Koval <julia.koval@intel.com>
4189 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4190 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4191 (ix86_handle_option): Handle -mavx512vbmi2.
4192 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4193 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4194 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4195 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4196 (ix86_valid_target_attribute_inner_p): Ditto.
4197 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4198 * config/i386/i386.opt (mavx512vbmi2): New option.
4199 * doc/invoke.texi: Add new option.
4201 2017-11-16 Julia Koval <julia.koval@intel.com>
4203 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4204 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4205 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4206 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4207 _mm512_gf2p8mul_epi8): New intrinsics.
4208 * config/i386/i386-builtin-types.def
4209 (V64QI_FTYPE_V64QI_V64QI): New type.
4210 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4211 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4212 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4213 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4214 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4215 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4217 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4219 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4220 explicitly as a stream of bytes.
4222 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4224 * config/rs6000/altivec.h (vec_xst_be): New #define.
4225 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4226 and externalize from *altivec_vperm_<mode>_internal.
4227 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4229 (XL_BE_V8HI): Likewise.
4230 (XL_BE_V4SI): Likewise.
4231 (XL_BE_V4SI): Likewise.
4232 (XL_BE_V2DI): Likewise.
4233 (XL_BE_V4SF): Likewise.
4234 (XL_BE_V2DF): Likewise.
4235 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4236 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4237 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4238 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4239 VSX_BUILTIN_VEC_XST_BE.
4240 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4241 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4243 (altivec_init_builtins): Replace conditional calls to def_builtin
4244 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4245 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4246 with unconditional calls. Remove calls to def_builtin for
4247 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4248 __builtin_vec_xst_be.
4249 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4250 to define_expand, and add alternate RTL generation for P8.
4251 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4252 vsx_ld_elemrev_v8hi.
4253 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4254 add alternate RTL generation for P8.
4255 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4256 vsx_ld_elemrev_v16qi.
4257 (vsx_st_elemrev_v8hi): Convert define_insn
4258 to define_expand, and add alternate RTL generation for P8.
4259 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4260 vsx_st_elemrev_v8hi.
4261 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4262 add alternate RTL generation for P8.
4263 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4264 vsx_st_elemrev_v16qi.
4266 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4269 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4270 TARGET_AVX512ER check.
4271 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4272 neither -mzeroupper nor -mno-zeroupper is used and
4273 TARGET_EMIT_VZEROUPPER is set.
4274 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4275 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4277 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4279 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4280 folding of vector compares.
4281 (fold_build_vec_cmp): New helper function.
4282 (fold_compare_helper): New helper function.
4283 (builtin_function_type): Add compare builtins to the list of functions
4284 having unsigned arguments. Cosmetic updates to comment indentation.
4285 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4286 the not+eq combination.
4288 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4290 PR tree-optimization/82726
4291 PR tree-optimization/70754
4292 * tree-predcom.c (order_drefs_by_pos): New function.
4293 (combine_chains): Move code setting has_max_use_after to...
4294 (try_combine_chains): ...here. New parameter. Sort combined chains
4295 according to position information.
4296 (tree_predictive_commoning_loop): Update call to above function.
4297 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4299 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4301 PR tree-optimization/82726
4303 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4305 PR tree-optimization/70754
4306 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4307 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4308 combined stmt before it if not NULL.
4309 (combine_chains): Process refs reversely and compute dominance point
4313 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4315 PR tree-optimization/79663
4316 * tree-predcom.c (combine_chains): Process refs in reverse order
4317 only for ZERO length chains, and add explaining comment.
4319 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4321 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4322 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4323 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4325 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4327 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4329 2017-11-15 Martin Liska <mliska@suse.cz>
4331 * tree-cfg.c (pass_warn_function_return::execute):
4332 Compare warn_return_type for greater than zero.
4334 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4338 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4339 to return true on Xeon and not on Xeon Phi.
4340 (ix86_check_avx256_register): Changed to ...
4341 (ix86_check_avx_upper_register): ... this. Add extra check for
4342 VALID_AVX512F_REG_OR_XI_MODE.
4343 (ix86_avx_u128_mode_needed): Changed
4344 ix86_check_avx256_register to ix86_check_avx_upper_register.
4345 (ix86_check_avx256_stores): Changed to ...
4346 (ix86_check_avx_upper_stores): ... this. Changed
4347 ix86_check_avx256_register to ix86_check_avx_upper_register.
4348 (ix86_avx_u128_mode_after): Changed
4349 avx_reg256_found to avx_upper_reg_found. Changed
4350 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4351 (ix86_avx_u128_mode_entry): Changed
4352 ix86_check_avx256_register to ix86_check_avx_upper_register.
4353 (ix86_avx_u128_mode_exit): Ditto.
4354 * config/i386/i386.h: (host_detect_local_cpu): New define.
4356 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4358 * config/arm/xgene1.md (xgene1): Split into automatons
4359 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4360 (xgene1_f_load): Adjust reservations and/or types.
4361 (xgene1_f_store): Likewise.
4362 (xgene1_load_pair): Likewise.
4363 (xgene1_store_pair): Likewise.
4364 (xgene1_fp_load1): Likewise.
4365 (xgene1_load1): Likewise.
4366 (xgene1_store1): Likewise.
4367 (xgene1_move): Likewise.
4368 (xgene1_alu): Likewise.
4369 (xgene1_simd): Likewise.
4370 (xgene1_bfm): Likewise.
4371 (xgene1_neon_load1): Likewise.
4372 (xgene1_neon_store1): Likewise.
4373 (xgene1_neon_logic): Likewise.
4374 (xgene1_neon_st1): Likewise.
4375 (xgene1_neon_ld1r): Likewise.
4376 (xgene1_alu_cond): Added.
4377 (xgene1_shift_reg): Likwise.
4378 (xgene1_bfx): Likewise.
4379 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4381 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4384 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4386 (can_widen_mult_without_libcall): New function.
4387 (expand_mul_overflow): If only checking unsigned mul overflow,
4388 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4389 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4390 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4391 (expand_DIVMOD): Formatting fix.
4392 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4393 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4394 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4396 PR tree-optimization/82977
4397 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4398 constructed temporary to strlen_to_stridx.put.
4400 2017-11-15 Martin Liska <mliska@suse.cz>
4402 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4403 * configure: Regenerate.
4405 2017-11-15 Martin Liska <mliska@suse.cz>
4408 * config/sh/sh-mem.cc: Use proper probability for
4411 2017-11-14 Jeff Law <law@redhat.com>
4413 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4414 the red zone for stack_clash_protection_final_dynamic_probe targets
4415 when the total dynamic stack size is zero bytes.
4417 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4418 blocks is post order.
4420 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4422 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4423 * final.c (rest_of_clean_state): Set it for the
4424 -fcompare-debug dump.
4425 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4426 class when TDF_COMPARE_DEBUG is set.
4428 * dwarf2out.c (gen_producer_string): Discard
4431 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4434 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4435 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4436 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4437 (__TGMATH_REAL_2_3): Remove macros.
4438 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4439 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4441 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4443 (remquo): Define using __TGMATH_REAL_3.
4445 2017-11-14 Jeff Law <law@redhat.com>
4447 * vr-values.c: New file with contents extracted from tree-vrp.c.
4448 * Makefile.in (OBJS): Add vr-values.o
4449 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4450 (set_value_range, set_and_canonicalize_value_range): Likewise.
4451 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4452 (value_range_constant_singleton, symbolic_range_p): Likewise.
4453 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4454 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4455 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4456 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4457 (find_case_label_range, find_case_label_index): Likewise.
4458 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4459 (range_int_cst_singleton_p, value_inside_range): Likewise.
4460 (get_single_symbol): Likewise.
4461 (switch_update): Move structure definition here.
4462 (to_remove_edges, to_update_switch_stmts): Provide externs.
4463 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4464 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4465 (vrp_val_is_min, set_value_range): Likewise.
4466 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4467 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4468 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4469 (range_is_nonnull, range_int_cst_p): Likewwise.
4470 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4471 (get_single_symbol, operand_less_p): Likewise
4472 (compare_values_warnv, compare_values): Likewise.
4473 (value_inside_range, value_range_constant_singleton): Likewise.
4474 (zero_nonzero_bitgs_from_vr): Likewise.
4475 (extract_range_from_binary_expr_1): Likewise.
4476 (overflow_comparison_p): Likewise.
4477 (to_remove_edges, to_update_switch_stmts): Likewise.
4478 (find_case_label-index, find_case_label_range): Likewise.
4479 (switch_update, set_value_range_to_nonnegative): Remove.
4480 (set_value_range_to_truthvalue): Likewise.
4481 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4482 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4483 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4484 (find_case_label_ranges, test_for_singularity): Likewise.
4485 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4486 (x_vr_values): Move to its remaining use site.
4488 2017-11-10 Jeff Law <law@redhat.com>
4490 * vr-values.h (VR_INITIALIZER): Move #define here.
4491 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4492 * Makefile.in (OBJS): Add tree-evrp.o
4493 * tree-vrp.h (assert_info): Move structure definition here.
4494 (set_value_range_to_varying): Prototype.
4495 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4496 (infer_value_range, register_edge_assert_for): Likewise.
4497 (stmt_interesting_for_vrp): Likewise.
4498 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4499 (set_value_range_to_varying): No longer static.
4500 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4501 (infer_value_range, register_edge_assert_for): Likewise.
4503 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4505 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4506 generate the XXBRD instruction.
4508 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4509 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4510 as being compatible if -mabi=ieeelongdouble.
4511 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4512 to setup float128 built-ins with hardware support.
4513 (BU_FLOAT128_HW_2): Likewise.
4514 (BU_FLOAT128_HW_3): Likewise.
4515 (BU_FLOAT128_HW_VSX_1): Likewise.
4516 (BU_FLOAT128_HW_VSX_2): Likewise.
4517 (scalar_extract_expq): Change float128 built-in functions to
4518 accommodate having both KFmode and TFmode functions. Use the
4519 KFmode variant as the default.
4520 (scalar_extract_sigq): Likewise.
4521 (scalar_test_neg_qp): Likewise.
4522 (scalar_insert_exp_q): Likewise.
4523 (scalar_insert_exp_qp): Likewise.
4524 (scalar_test_data_class_qp): Likewise.
4525 (sqrtf128_round_to_odd): Delete processing the round to odd
4526 built-in functions as special built-in functions, and define them
4527 as float128 built-ins. Use the KFmode variant as the default.
4528 (truncf128_round_to_odd): Likewise.
4529 (addf128_round_to_odd): Likewise.
4530 (subf128_round_to_odd): Likewise.
4531 (mulf128_round_to_odd): Likewise.
4532 (divf128_round_to_odd): Likewise.
4533 (fmaf128_round_to_odd): Likewise.
4534 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4535 support for KFmode and TFmode xststdcqp calls.
4536 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4537 point, switch the built-in handlers for the get/set float128
4538 exponent, get float128 mantissa, float128 test built-ins, and the
4539 float128 round to odd built-in functions. Eliminate creating the
4540 float128 round to odd built-in functions as special built-ins.
4541 (rs6000_init_builtins): Eliminate special creation of the float128
4542 round to odd built-in functions.
4543 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4544 function insns to support both TFmode and KFmode variants.
4545 (xsxsigqp_<mode>): Likewise.
4546 (xsiexpqpf_<mode>): Likewise.
4547 (xsiexpqp_<mode>): Likewise.
4548 (xststdcqp_<mode>): Likewise.
4549 (xststdcnegqp_<mode>): Likewise.
4550 (xststdcqp_<mode>): Likewise.
4552 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4554 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4555 unused path_in_freq_ptr parameter.
4556 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4558 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4560 * ipa-inline.c (edge_badness): Dump sreal frequency.
4561 (compute_inlined_call_time): Match natural implementaiton ...
4562 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4563 forgotten division by CGRAPH_FREQ_BASE.
4565 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4567 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4568 Solaris 11. Update comment.
4569 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4571 * config/sol2.h (STARTFILE_SPEC): Likewise.
4572 * configure: Regenerate.
4574 2017-11-14 Carl Love <cel@us.ibm.com>
4576 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4577 le_ and be_ prefixes to swap* variables. Remove
4578 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4581 2017-11-14 Jason Merrill <jason@redhat.com>
4583 Support GTY((cache)) on hash_map.
4584 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4585 (ggc_cache_remove): Override it instead of ggc_mx.
4586 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4587 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4588 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4589 (simple_cache_map_traits): Override maybe_mx.
4590 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4591 (hash_map): Friend gt_cleare_cache.
4592 (gt_cleare_cache): New.
4593 * tree.h (tree_cache_traits): New hash_map traits class.
4594 (tree_cache_map): New typedef.
4596 2017-11-14 Richard Biener <rguenther@suse.de>
4598 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4599 paramter and handling.
4600 (cleanup_control_flow_bb): Likewise.
4601 (cleanup_control_flow_pre): New helper performing a DFS walk
4602 to call cleanup_control_flow_bb in PRE order.
4603 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4604 via cleanup_control_flow_pre.
4606 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4608 * config/aarch64/aarch64-simd.md
4609 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4610 (*aarch64_simd_bsl<mode>_alt): Likewise.
4611 (aarch64_simd_bsldi_internal): New.
4612 (aarch64_simd_bsldi_alt): Likewise.
4614 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4616 * tracer.c (better_p): Do not compare frequencies.
4617 * reg-stack.c (better_edge): Likewise.
4618 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4621 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4623 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4624 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4625 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4626 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4627 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4628 * predict.c (maybe_hot_frequency_p): Inline to ...
4629 (maybe_hot_count_p): ... here; rewrite to counts.
4630 (counts_to_freqs): Rename to ...
4631 (update_max_bb_count): ... this one.
4632 (expensive_function_p): Use counts.
4633 (estimate_bb_frequencies): Update.
4634 (rebuild_frequencies): Update.
4635 * predict.h (counts_to_freqs): Rename to ...
4636 (update_max_bb_count): ... this one.
4637 * profile.c (compute_branch_probabilities): Add debug info
4638 * tree-inline.c (expand_call_inline): Update debug info.
4639 (optimize_inline_calls): Use update_max_bb_count..
4640 (tree_function_versioning): Use update_max_bb_count..
4641 * value-prof.c (gimple_value_profile_transformations):
4642 Do not use update_max_bb_count.
4644 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4646 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4647 always use frequencies.
4649 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4651 * bb-reorder.c: Remove frequencies from comments.
4652 (better_edge_p): Use profile counts.
4653 (find_traces): Dump profile counts.
4654 (rotate_loop): Use profile counts.
4655 (find_traces_1_round): Likewise.
4656 (connect_better_edge_p): Use counts instead of probabilities for
4658 (copy_bb_p): Drop early check for non-0 frequency.
4659 (sanitize_hot_paths): Update comments.
4661 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4663 * ipa-split.c (struct split_point): Add count.
4664 (consider_split): Do not compute incoming frequency; compute incoming
4665 count and store it to split_point.
4666 (split_function): Set count of the call to split part correctly.
4668 2017-11-13 Carl Love <cel@us.ibm.com>
4670 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4672 2017-11-13 Tom Tromey <tom@tromey.com>
4674 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4676 2017-11-13 Carl Love <cel@us.ibm.com>
4678 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4679 Add support for builtins:
4680 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4681 vector {un,}signed {char,int,short},
4682 vector {un,}signed {char,int,short}) arguments.
4683 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4684 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4685 Add BU_P9V_AV_2 expansions for the builtins.
4686 * config/rs6000/altivec.h (vec_first_match_index,
4687 vec_first_mismatch_index, vec_first_match_or_eos_index,
4688 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4689 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4690 new extern declaration.
4691 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4692 * config/rs6000/vsx.md (first_match_index_<mode>,
4693 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4694 first_mismatch_or_eos_index_<mode>): Add define expand.
4695 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4696 * doc/extend.texi: Update the built-in documenation file for the new
4699 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4701 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4702 into the min/max operations for _Float<N> and _Float<N>X types.
4704 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4707 * dwarf2out.c (do_eh_frame): New static variable.
4708 (dwarf2out_begin_prologue): Set it.
4709 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4711 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4713 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4715 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4717 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4720 * debug.h (dwarf2out_do_eh_frame): Declare.
4721 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4722 (dwarf2out_do_frame): Use it.
4723 (dwarf2out_do_cfi_asm): Likewise.
4724 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4725 (dwarf2out_assembly_start): Likewise.
4726 (dwarf2out_begin_prologue): Fix comment.
4727 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4728 if the target needs either debug or unwind DWARF2 info.
4729 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4730 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4731 -ftrapping-math, -ftrapv and -fwrapv.
4733 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4735 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4736 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4737 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4738 (estimate_edge_size_and_time): Likewise.
4739 (ipa_merge_fn_summary_after_inlining): Likewise.
4740 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4741 (compute_uninlined_call_time): Use sreal_frequency.
4742 (compute_inlined_call_time): Likewise.
4743 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4744 * profile-count.c: Include sreal.h.
4745 (profile_count::to_sreal_scale): New.
4746 * profile-count.h: Forward declare sreal.
4747 (profile_count::to_sreal_scale): Declare.
4749 2017-11-13 Nathan Sidwell <nathan@acm.org>
4751 * diagnostic.c (maybe_line_and_column): New.
4752 (diagnostic_get_location_text): Use it.
4753 (diagnostic_report_current_module): Likewise.
4754 (test_diagnostic_get_location_text): Add tests.
4756 2017-11-13 Luis Machado <luis.machado@linaro.org>
4758 * doc/md.texi (Specifying processor pipeline description): Fix
4759 incorrect latency for the div instruction example.
4761 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4763 PR tree-optimization/78821
4764 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4765 that bit_not_p is the same.
4766 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4767 (split_group): Count precisely bit_not_p bits in each statement.
4768 (invert_op): New function.
4769 (imm_store_chain_info::output_merged_store): Use invert_op to
4770 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4771 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4773 2017-11-13 Martin Liska <mliska@suse.cz>
4775 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4776 (struct source_info): Likewise.
4777 (add_branch_counts): Likewise.
4778 (add_line_counts): Likewise.
4779 (function_summary): Likewise.
4780 (output_intermediate_line): Likewise.
4781 (generate_results): Likewise.
4783 2017-11-13 Martin Liska <mliska@suse.cz>
4785 * gcov.c (struct block_info): Remove typedef for block_t.
4786 (struct line_info): Likewise.
4787 (line_info::has_block): Likewise.
4788 (EXIT_BLOCK): Likewise.
4789 (unblock): Likewise.
4790 (circuit): Likewise.
4791 (get_cycles_count): Likewise.
4792 (process_file): Likewise.
4793 (read_graph_file): Likewise.
4794 (solve_flow_graph): Likewise.
4795 (find_exception_blocks): Likewise.
4796 (add_line_counts): Likewise.
4797 (accumulate_line_info): Likewise.
4798 (output_line_details): Likewise.
4800 2017-11-13 Martin Liska <mliska@suse.cz>
4802 * gcov.c (struct arc_info): Remove typedef for arc_t.
4803 (struct line_info): Likewise.
4804 (add_branch_counts): Likewise.
4805 (output_branch_count): Likewise.
4806 (function_info::~function_info): Likewise.
4807 (circuit): Likewise.
4808 (output_intermediate_line): Likewise.
4809 (read_graph_file): Likewise.
4810 (solve_flow_graph): Likewise.
4811 (find_exception_blocks): Likewise.
4812 (add_line_counts): Likewise.
4813 (accumulate_line_info): Likewise.
4814 (output_line_details): Likewise.
4815 (output_function_details): Likewise.
4817 2017-11-13 Martin Liska <mliska@suse.cz>
4819 * gcov.c (struct function_info): Remove typedef for function_t.
4820 (struct source_info): Likewise.
4821 (source_info::get_functions_at_location): Likewise.
4822 (solve_flow_graph): Likewise.
4823 (find_exception_blocks): Likewise.
4824 (add_line_counts): Likewise.
4825 (output_intermediate_file): Likewise.
4826 (process_file): Likewise.
4827 (generate_results): Likewise.
4828 (release_structures): Likewise.
4829 (read_graph_file): Likewise.
4830 (read_count_file): Likewise.
4831 (accumulate_line_counts): Likewise.
4832 (output_lines): Likewise.
4834 2017-11-13 Martin Liska <mliska@suse.cz>
4836 * gcov.c (function_info::function_info): Remove num_counts
4837 and add vector<gcov_type>.
4838 (function_info::~function_info): Use the vector.
4839 (process_file): Likewise.
4840 (read_graph_file): Likewise.
4841 (read_count_file): Likewise.
4842 (solve_flow_graph): Likewise.
4844 2017-11-13 Martin Liska <mliska@suse.cz>
4846 * gcov.c (function_info::is_artificial): New function.
4847 (process_file): Erase all artificial early.
4848 (generate_results): Skip as all artificial are already
4851 2017-11-13 Martin Liska <mliska@suse.cz>
4853 * gcov.c (read_graph_file): Store to global vector of functions.
4854 (read_count_file): Iterate the vector.
4855 (process_file): Likewise.
4856 (generate_results): Likewise.
4857 (release_structures): Likewise.
4859 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4861 PR tree-optimization/82954
4862 * gimple-ssa-store-merging.c
4863 (imm_store_chain_info::coalesce_immediate_stores): If
4864 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4866 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4868 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4869 Upddate call to ENDIAN_LANE_N.
4870 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4871 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4872 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4873 and use aarch64_endian_lane_rtx.
4874 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4876 2017-11-12 Tom de Vries <tom@codesourcery.com>
4878 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4880 2017-11-12 Tom de Vries <tom@codesourcery.com>
4882 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4884 2017-11-12 Tom de Vries <tom@codesourcery.com>
4886 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4888 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4889 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4890 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4892 2017-11-11 Martin Sebor <msebor@redhat.com>
4895 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4898 * prefic.c (translate_name): Replace strncpy with memcpy to
4899 avoid -Wstringop-truncation.
4901 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4903 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4905 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4907 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4909 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4911 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4914 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4916 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4917 enable generating XXBRH if the value is in a vector register.
4918 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4919 value is in a vector register.
4920 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4921 register to register bswap64's instead of doing the GPR sequence
4922 used on previous machines.
4923 (bswapdi2_xxbrd): New insn.
4924 (bswapdi2_reg): Disallow on ISA 3.0.
4925 (register to register bswap64 splitter): Do not split the insn on
4926 ISA 3.0 systems that use XXBRD.
4928 2017-11-10 Martin Sebor <msebor@redhat.com>
4931 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4933 (find_subframework_file): Same.
4935 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4937 * auto-profile.c (afdo_indirect_call): Drop frequency.
4938 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4939 (cgraph_node::create_edge): Drop frequency argument.
4940 (cgraph_node::create_indirect_edge): Drop frequency argument.
4941 (cgraph_edge::make_speculative): Drop frequency arguments.
4942 (cgraph_edge::resolve_speculation): Do not update frequencies
4943 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4944 (cgraph_node::dump): Check consistency in IPA mode.
4945 (cgraph_edge::maybe_hot_p): Use IPA counter.
4946 (cgraph_edge::verify_count_and_frequency): Rename to ...
4947 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4948 (cgraph_node::verify_node): Update.
4949 * cgraph.h (struct cgraph_edge): Drop frequency.
4950 (cgraph_edge::frequency): New function.
4951 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4953 (cgraph_edge::rebuild_edges): Likewise.
4954 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4955 (duplicate_thunk_for_node): Do not pass frequency.
4956 (cgraph_node::create_clone): Scale only counts.
4957 (cgraph_node::create_virtual_clone): Do not pass frequency.
4958 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4959 (cgraph_node::create_version_clone): Do not pass frequency.
4960 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4961 (cgraph_node::expand_thunk): Do not pass frequency.
4962 (cgraph_node::create_wrapper): Do not pass frequency.
4963 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4965 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4966 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4967 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4968 (gather_caller_stats): Use frequency function.
4969 (ipcp_cloning_candidate_p): Use frequency function.
4970 (ipcp_propagate_stage): Use frequency function.
4971 (get_info_about_necessary_edges): Use frequency function.
4972 (update_profiling_info): Update only IPA profile.
4973 (update_specialized_profile): Use frequency functoin.
4974 (perhaps_add_new_callers): Update only IPA profile.
4975 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4976 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4977 (dump_ipa_call_summary): Use frequency function.
4978 (estimate_edge_size_and_time): Use frequency function.
4979 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4980 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4981 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4982 (update_noncloned_counts): ... ths one; scale counts only.
4983 (clone_inlined_nodes): Do not scale frequency.
4984 (inline_call): Do not pass frequency.
4985 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4986 (compute_inlined_call_time): Use IPA profile.
4987 (want_inline_small_function_p): Use IPA profile.
4988 (want_inline_self_recursive_call_p): Use IPA profile.
4989 (edge_badness): Use IPA profile.
4990 (lookup_recursive_calls): Use IPA profile.
4991 (recursive_inlining): Do not pass frequency.
4992 (resolve_noninline_speculation): Do not update frequency.
4993 (inline_small_functions): Collect max of IPA profile.
4994 (dump_overall_stats): Dump IPA porfile.
4995 (dump_inline_stats): Dump IPA porfile.
4996 (ipa_inline): Collect IPA stats.
4997 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4998 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4999 (ipa_propagate_frequency): Use frequency function.
5000 (ipa_profile): Cleanup.
5001 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5002 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5003 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5004 (input_node): Do not stream frequency.
5005 (input_edge): Do not stream frequency.
5006 (merge_profile_summaries): Scale only IPA profiles.
5007 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5008 * predict.c (drop_profile): Do not recompute frequency.
5009 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5010 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5011 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5012 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5013 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5014 * tree-inline.c (copy_bb): Do not scale frequency.
5015 (expand_call_inline): Do not scale frequency.
5016 (tree_function_versioning): Do not scale frequency.
5017 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5019 2017-11-10 Julia Koval <julia.koval@intel.com>
5021 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5022 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5023 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5024 (_mm256_maskz_gf2p8affine_epi64_epi8)
5025 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5026 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5027 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5028 (__builtin_ia32_vgf2p8affineqb_v32qi)
5029 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5030 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5032 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5036 (arm_option_override): Refactor.
5037 (arm_option_reconfigure_globals): New.
5038 (arm_options_perform_arch_sanity_checks): New.
5039 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5041 (arm_options_perform_arch_sanity_checks): Likewise
5043 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5045 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5046 (power9-qpmul): New.
5047 * rs6000/rs6000.md ("type" attr): Add qmul.
5048 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5049 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5050 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5052 2017-11-10 Martin Sebor <msebor@redhat.com>
5055 * builtins.c (compute_objsize): Handle arrays that
5056 compute_builtin_object_size likes to fail for. Make extern.
5057 * builtins.h (compute_objsize): Declare.
5058 (check_strncpy_sizes): New function.
5059 (expand_builtin_strncpy): Call check_strncpy_sizes.
5060 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5061 -Wstringop-truncation.
5062 (gimple_fold_builtin_strncat): Same.
5063 * gimple.c (gimple_build_call_from_tree): Set call location.
5064 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5065 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5066 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5067 (handle_builtin_strlen): Use strlen_to_stridx.
5068 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5070 Use strlen_to_stridx.
5071 (pass_strlen::execute): Release strlen_to_stridx.
5072 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5073 (-Wstringop-truncation): Document new option.
5075 2017-11-10 Martin Liska <mliska@suse.cz>
5077 PR gcov-profile/82702
5078 * gcov.c (main): Handle intermediate files in a different
5080 (get_gcov_intermediate_filename): New function.
5081 (output_gcov_file): Remove support of intermediate files.
5082 (generate_results): Allocate intermediate file.
5083 (release_structures): Clean-up properly fn_end.
5084 (output_intermediate_file): Start iterating with line 1.
5086 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5088 PR tree-optimization/82929
5089 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5090 ops_swapped_p non-static data member.
5091 (store_immediate_info::store_immediate_info): Clear it.
5092 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5093 ops set ops_swapped_p.
5094 (count_multiple_uses): Handle ops_swapped_p.
5096 2017-11-10 Martin Liska <mliska@suse.cz>
5098 * coverage.c (coverage_init): Stream information about
5099 support of has_unexecuted_blocks.
5100 * doc/gcov.texi: Document that.
5101 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5102 * gcov.c (read_graph_file): Likewise.
5103 (output_line_beginning): Fix a small issue with
5106 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5108 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5109 reference of trivial component.
5111 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5114 * gimple-ssa-store-merging.c
5115 (pass_store_merging::terminate_all_aliasing_chains): For
5116 gimple_store_p stmts also call refs_output_dependent_p.
5118 PR rtl-optimization/82913
5119 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5122 2017-11-09 Jeff Law <law@redhat.com>
5124 * vr-values.h: New file with vr_values class.
5125 * tree-vrp.c: Include vr-values.h
5126 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5127 data objects into the vr_values class.
5128 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5129 (get_value_range): Make it a member function within vr_values class.
5130 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5131 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5132 (op_with_constant_singleton_value_range): Likewise.
5133 (extract_range_for_var_from_comparison_expr): Likewise.
5134 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5135 (extract_range_from_binary_expr): Likewise.
5136 (extract_range_from_unary_expr): Likewise.
5137 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5138 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5139 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5140 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5141 (compare_name_with_value, compare_names): Likewise.
5142 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5143 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5144 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5145 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5146 (extract_range_from_phi_node): Likewise.
5147 (simplify_truth_ops_using_ranges): Likewise.
5148 (simplify_div_or_mod_using_ranges): Likewise.
5149 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5150 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5151 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5152 (simplify_float_conversion_using_ranges): Likewise.
5153 (simplify_internal_call_using_ranges): Likewise.
5154 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5155 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5156 poitner via x_vr_values for calls into gimple folder.
5157 (vrp_initialize_lattice): Make this the vr_values ctor.
5158 (vrp_free_lattice): Make this the vr_values dtor.
5159 (set_vr_value): New function.
5160 (class vrp_prop): Add vr_values data member. Add various member
5161 functions as well as member functions that delegate to vr_values.
5162 (check_array_ref): Make a member function within vrp_prop class.
5163 (search_for_addr_array, vrp_initialize): Likewise.
5164 (vrp_finalize): Likewise. Revamp to avoid direct access to
5165 vr_value, values_propagated, etc.
5166 (check_array_bounds): Extract vrp_prop class instance pointer from
5167 walk info structure. Use it to call member functions.
5168 (check_all_array_refs): Make a member function within vrp_prop class.
5169 Smuggle class instance pointer via walk info structure.
5170 (x_vr_values): New local static.
5171 (vrp_valueize): Use x_vr_values to get class instance.
5172 (vr_valueize_1): Likewise.
5173 (class vrp_folder): Add vr_values data member. Add various member
5174 functions as well as member functions that delegate to vr_values.
5175 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5176 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5177 class instance from vr_values. Use it to call member functions.
5178 (vrp_dom_walker): Add vr_values data member.
5179 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5180 instance via x_vr_values.
5181 (identify_jump_threads): Accept vr_values as argument. Store
5182 it into the walker structure.
5183 (evrp_dom_walker): Add vr_values class data member. Add various
5185 (evrp_dom_walker::try_find_new_range): Use vr_values data
5186 member to access the memory allocator.
5187 (evrp_dom_walker::before_dom_children): Store vr_values class
5188 instance into the vrp_folder class.
5189 (evrp_dom_walker::push_value_range): Rework to avoid direct
5190 access to num_vr_values and vr_value.
5191 (evrp_dom_walker::pop_value_range): Likewise.
5192 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5193 Use vr_values to get to dump_all_value_ranges member function.
5194 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5195 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5196 Pass vr_values class instance down to identify_jump_threads.
5197 Remove call to vrp_free_lattice.
5198 (debug_all_value_ranges): Remove.
5200 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5201 (vrp_folder): Likewise.
5203 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5204 Get it from the existing bitmap instead.
5205 (vrp_intersect_ranges_1): Likewise.
5207 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5209 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5211 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5212 set bit_not_p to it.
5213 (imm_store_chain_info::coalesce_immediate_stores): Break group
5214 if bit_not_p is different.
5215 (count_multiple_uses, split_group,
5216 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5217 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5218 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5219 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5220 to store_immediate_info ctor.
5222 2017-11-09 Jim Wilson <jimw@sifive.com>
5224 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5225 (scan_prog_file): Likewise.
5227 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5229 * bb-reorder.c (max_entry_frequency): Remove.
5230 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5231 connect_traces, push_to_next_round_p): Remove prototypes.
5232 (find_traces_1_round): Use counts only.
5233 (push_to_next_round_p): Likewise.
5234 (find_traces): Likewise.
5235 (rotate_loop): Likewise.
5236 (find_traces_1_round): Likewise.
5237 (connect_traces): Likewise.
5238 (edge_order): Likewise.
5240 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5242 * config/arm/arm.c (output_return_instruction): Add comments to
5243 indicate requirement for cmse_nonsecure_entry return to account
5244 for the size of clearing instruction output here.
5245 (thumb_exit): Likewise.
5246 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5247 return in hardfloat mode.
5249 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5251 * config/rs6000/rs6000.c (machine_function): Add a bool,
5252 "toc_is_wrapped_separately".
5253 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5254 if it wasn't explicitly set or unset, we are optimizing for speed, and
5255 doing separate shrink-wrapping.
5256 (rs6000_get_separate_components): Enable the TOC component if
5257 saving the TOC register in the prologue.
5258 (rs6000_components_for_bb): Handle the TOC component.
5259 (rs6000_emit_prologue_components): Store the TOC register where needed.
5260 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5261 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5263 2017-11-09 Martin Jambor <mjambor@suse.cz>
5265 * ipa-param-manipulation.c: New file.
5266 * ipa-param-manipulation.h: Likewise.
5267 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5268 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5269 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5270 (ipa_parm_adjustment): Likewise.
5271 (ipa_parm_adjustment_vec): Likewise.
5272 (ipa_get_vector_of_formal_parms): Moved declaration to
5273 ipa-param-manipulation.h.
5274 (ipa_get_vector_of_formal_parm_types): Likewise.
5275 (ipa_modify_formal_parameters): Likewise.
5276 (ipa_modify_call_arguments): Likewise.
5277 (ipa_combine_adjustments): Likewise.
5278 (ipa_dump_param_adjustments): Likewise.
5279 (ipa_modify_expr): Likewise.
5280 (ipa_get_adjustment_candidate): Likewise.
5281 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5282 ipa-param-manipulation.c.
5283 (ipa_get_vector_of_formal_parm_types): Likewise.
5284 (ipa_modify_formal_parameters): Likewise.
5285 (ipa_modify_call_arguments): Likewise.
5286 (ipa_modify_expr): Likewise.
5287 (get_ssa_base_param): Likewise.
5288 (ipa_get_adjustment_candidate): Likewise.
5289 (index_in_adjustments_multiple_times_p): Likewise.
5290 (ipa_combine_adjustments): Likewise.
5291 (ipa_dump_param_adjustments): Likewise.
5292 * tree-sra.c: Also include ipa-param-manipulation.h
5293 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5296 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5297 Alan Hayward <alan.hayward@arm.com>
5298 David Sherwood <david.sherwood@arm.com>
5300 * doc/sourcebuild.texi (vect_masked_store): Document.
5302 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5303 Alan Hayward <alan.hayward@arm.com>
5304 David Sherwood <david.sherwood@arm.com>
5306 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5308 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5309 Alan Hayward <alan.hayward@arm.com>
5310 David Sherwood <david.sherwood@arm.com>
5312 * doc/sourcebuild.texi (vect_variable_length): Document.
5314 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5315 Alan Hayward <alan.hayward@arm.com>
5316 David Sherwood <david.sherwood@arm.com>
5318 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5320 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5321 Alan Hayward <alan.hayward@arm.com>
5322 David Sherwood <david.sherwood@arm.com>
5324 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5326 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5327 Alan Hayward <alan.hayward@arm.com>
5328 David Sherwood <david.sherwood@arm.com>
5330 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5331 previously undocumented selectors.
5332 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5334 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5336 * doc/rtl.texi (const_vector): Say that elements can be
5337 const_wide_ints too.
5338 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5339 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5340 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5341 * optabs.c (expand_vector_broadcast): Likewise.
5343 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5344 Alan Hayward <alan.hayward@arm.com>
5345 David Sherwood <david.sherwood@arm.com>
5347 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5348 scaled index even if the unscaled address was invalid.
5349 Don't increase the complexity of using a scale in that case.
5351 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5352 Alan Hayward <alan.hayward@arm.com>
5353 David Sherwood <david.sherwood@arm.com>
5355 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5356 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5357 UNITS_PER_WORD bytes.
5358 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5359 based on REGMODE_NATURAL_SIZE of the inner mode.
5360 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5361 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5362 * expmed.c (lowpart_bit_field_p): Divide the value up into
5363 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5364 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5365 whether something is likely to occupy more than one register.
5367 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5370 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5371 profile_count::adjust_for_ipa_scaling.
5372 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5373 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5375 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5377 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5379 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5380 (split_group): Add total_orig and total_new arguments, estimate the
5381 number of statements related to the store group without store merging
5382 and with store merging.
5383 (imm_store_chain_info::output_merged_store): Adjust split_group
5384 callers, punt if estimated number of statements with store merging
5385 is not smaller than estimated number of statements without it.
5387 (handled_load): Remove has_single_use checks.
5388 (pass_store_merging::process_store): Likewise.
5390 2017-11-09 Richard Biener <rguenther@suse.de>
5392 PR tree-optimization/82902
5393 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5395 2017-11-09 Martin Liska <mliska@suse.cz>
5398 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5401 2017-11-09 Martin Liska <mliska@suse.cz>
5403 PR tree-optimization/82669
5404 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5406 2017-11-09 Martin Liska <mliska@suse.cz>
5408 PR gcov-profile/48463
5409 * coverage.c (coverage_begin_function): Output also end locus
5410 of a function and information whether the function is
5412 * gcov-dump.c (tag_function): Parse and print the information.
5413 * gcov.c (INCLUDE_MAP): Add include.
5414 (INCLUDE_SET): Likewise.
5415 (struct line_info): Move earlier in the source file because
5416 of vector<line_info> in function_info structure.
5417 (line_info::line_info): Likewise.
5418 (line_info::has_block): Likewise.
5419 (struct source_info): Add new member index.
5420 (source_info::get_functions_at_location): New function.
5421 (function_info::group_line_p): New function.
5422 (output_intermediate_line): New function.
5423 (output_intermediate_file): Use the mentioned function.
5424 (struct function_start): New.
5425 (struct function_start_pair_hash): Likewise.
5426 (process_file): Add code that identifies group functions.
5427 Assign lines either to global or function scope.
5428 (generate_results): Skip artificial functions.
5429 (find_source): Assign index for each source file.
5430 (read_graph_file): Read new flag artificial and end_line.
5431 (add_line_counts): Assign it either to global of function scope.
5432 (accumulate_line_counts): Isolate core of the function to
5433 accumulate_line_info and call it for both function and global
5435 (accumulate_line_info): New function.
5436 (output_line_beginning): Fix GNU coding style.
5437 (print_source_line): New function.
5438 (output_line_details): Likewise.
5439 (output_function_details): Likewise.
5440 (output_lines): Iterate both source (global) scope and function
5442 (struct function_line_start_cmp): New class.
5443 * doc/gcov.texi: Reflect changes in documentation.
5445 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5448 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5449 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5450 and similarly for not instead of neg.
5452 2017-11-08 Andi Kleen <ak@linux.intel.com>
5454 * config/i386/i386.opt: Add -mforce-indirect-call.
5455 * config/i386/predicates.md: Check for flag_force_indirect_call.
5456 * doc/invoke.texi: Document -mforce-indirect-call
5458 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5460 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5462 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5463 config/riscv/riscv.c (predict.h): New include.
5464 (riscv_slow_unaligned_access_p): No longer static.
5465 (riscv_block_move_straight): Add require.
5466 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5468 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5471 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5472 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5473 nonimmediate_operand predicate for operand 1 instead of
5476 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5478 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5480 * config/aarch64/constraints.md (Uml): New constraint.
5481 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5484 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5486 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5487 of two vec_duplicates into a vec_concat.
5489 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5491 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5492 Simplify vec_merge of vec_duplicate and vec_concat.
5493 * config/aarch64/constraints.md (Utq): New constraint.
5494 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5497 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5499 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5500 Simplify vec_merge of vec_duplicate and const_vector.
5501 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5503 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5504 mode iterator. Update predicate on operand 1 to
5505 handle non-const_vec constants. Delete constraints.
5506 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5508 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5510 PR tree-optimization/78821
5511 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5513 (store_operand_info::store_operand_info): Initialize it to false.
5514 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5515 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5516 store in the group, and if chain_info is non-NULL, to ignore altogether
5518 (compatible_load_p): Fail if bit_not_p does not match.
5519 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5520 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5521 (pass_store_merging::process_store): Adjust
5522 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5523 call terminate_all_aliasing_chains newly when adding a store into
5524 a chain with non-NULL chain_info.
5526 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5528 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5530 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5532 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5534 (aarch64_layout_frame): Initialise emit_frame_chain.
5535 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5536 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5538 2017-11-08 Martin Liska <mliska@suse.cz>
5540 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5541 of gimple_call_internal_p.
5543 2017-11-07 Tom de Vries <tom@codesourcery.com>
5545 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5547 2017-11-07 Tom de Vries <tom@codesourcery.com>
5549 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5552 2017-11-08 Martin Liska <mliska@suse.cz>
5555 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5557 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5559 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5561 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5564 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5565 (*cmp<mode>_ccz_1): New insn with $k alternative.
5568 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5569 RTX_COMM_COMPARE as commutative as well.
5570 (ix86_binary_operator_ok): Formatting fix.
5571 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5572 *<code><mode>3<mask_name><round_saeonly_name>,
5573 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5574 *<s>mul<mode>3_highpart<mask_name>,
5575 *vec_widen_umult_even_v16si<mask_name>,
5576 *vec_widen_umult_even_v8si<mask_name>,
5577 *vec_widen_umult_even_v4si<mask_name>,
5578 *vec_widen_smult_even_v16si<mask_name>,
5579 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5580 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5581 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5582 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5583 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5584 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5585 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5586 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5587 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5588 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5589 ix86_binary_operator_ok. Formatting fixes.
5590 (*<plusminus_insn><mode>3<mask_name><round_name>,
5591 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5594 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5596 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5597 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5598 eq and ne if TARGET_ISEL.
5599 (cmp): New code_iterator.
5600 (UNS, UNSU_, UNSIK): New code_attrs.
5601 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5602 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5604 ("eq<mode>3"): ... this.
5605 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5607 ("ne<mode>3"): ... this.
5609 2017-11-07 Julia Koval <julia.koval@intel.com>
5612 * common/config/i386/i386-common.c
5613 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5614 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5615 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5616 * config/i386/i386.opt: Ditto.
5617 * config/i386/i386.c (ix86_target_string): Ditto.
5618 (ix86_option_override_internal): Ditto.
5619 (ix86_init_mpx_builtins): Move MPX to args2.
5620 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5621 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5622 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5623 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5624 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5625 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5626 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5628 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5631 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5632 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5633 (zero-extendsidi peephole2): Remove peephole.
5635 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5638 * stor-layout.c: Include attribs.h.
5639 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5640 explicit lookup of "aligned" attribute.
5642 2017-11-07 Andrew Waterman <andrew@sifive.com>
5644 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5645 (riscv_expand_block_move): Likewise.
5646 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5648 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5649 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5650 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5651 (riscv_adjust_block_mem): Likewise.
5652 (riscv_block_move_loop): Likewise.
5653 (riscv_expand_block_move): Likewise.
5654 * config/riscv/riscv.md (movmemsi): New pattern.
5656 2017-11-07 Michael Clark <michaeljclark@mac.com>
5658 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5659 (MUSL_DYNAMIC_LINKER): Likewise.
5661 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5663 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5664 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5667 2017-11-07 Richard Biener <rguenther@suse.de>
5669 * match.pd: Fix build.
5671 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5672 Jackson Woodruff <jackson.woodruff@arm.com>
5674 PR tree-optimization/71026
5675 * match.pd: Canonicalize negate in division.
5677 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5680 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5682 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5684 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5685 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5687 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5689 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5690 non-scalar integral types.
5691 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5692 (-(A-B), -(~A)): New transformations.
5694 2017-11-07 Tom de Vries <tom@codesourcery.com>
5696 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5697 semicolon after "do {} while (0)".
5698 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5699 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5700 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5701 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5702 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5704 2017-11-07 Tom de Vries <tom@codesourcery.com>
5706 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5707 after "do {} while (0)".
5708 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5709 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5710 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5711 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5712 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5714 2017-11-07 Tom de Vries <tom@codesourcery.com>
5717 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5719 (arm_rtx_costs_internal): Add missing semicolon after
5720 HANDLE_NARROW_SHIFT_ARITH call.
5722 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5724 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5725 disable isel if it was not set explicitly.
5727 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5729 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5730 (add_type_attribute) likewise.
5732 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5734 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5735 of struct ix86_frame.
5736 (ix86_initial_elimination_offset): Likewise.
5737 (ix86_expand_split_stack_prologue): Likewise.
5739 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5741 * tree-vrp.h (enum value_range_type): Update stale comment.
5743 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5744 Alan Hayward <alan.hayward@arm.com>
5745 David Sherwood <david.sherwood@arm.com>
5747 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5748 (aarch64_expand_vec_perm_const): Take the number of units too.
5749 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5750 (aarch64_expand_vec_perm_const): Likewise.
5751 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5752 (vec_perm<mode>): Update accordingly.
5754 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5755 Alan Hayward <alan.hayward@arm.com>
5756 David Sherwood <david.sherwood@arm.com>
5758 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5759 Take the number of units too.
5760 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5761 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5762 but check for a vector mode before rather than after the call.
5763 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5764 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5765 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5766 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5767 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5768 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5769 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5770 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5771 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5772 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5773 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5774 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5775 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5776 (aarch64_sqdmull2_n<mode>): Update accordingly.
5778 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5779 Alan Hayward <alan.hayward@arm.com>
5780 David Sherwood <david.sherwood@arm.com>
5782 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5783 the number of units too.
5784 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5785 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5786 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5787 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5788 (vec_store_lanesxi<mode>): Update accordingly.
5790 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5791 Alan Hayward <alan.hayward@arm.com>
5792 David Sherwood <david.sherwood@arm.com>
5794 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5795 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5796 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5797 of units rather than the mode.
5798 * config/aarch64/iterators.md (nunits): New mode attribute.
5799 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5800 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5801 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5802 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5803 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5804 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5805 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5806 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5807 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5808 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5809 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5810 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5811 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5812 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5813 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5814 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5815 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5816 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5817 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5818 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5819 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5820 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5821 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5822 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5823 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5824 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5825 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5826 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5827 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5828 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5829 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5830 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5831 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5832 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5833 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5834 (aarch64_simd_vec_setv2di): Likewise.
5836 2017-11-06 Carl Love <cel@us.ibm.com>
5838 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5840 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5841 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5843 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5845 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5846 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5847 Add power 8 macro expansions.
5848 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5849 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5850 power 8 instructions. (VSX_XXBR): Add iterator.
5852 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5854 * config/arm/arm.md (predicable_short_it): Change default to "no",
5855 improve documentation, remove uses that are identical to the default.
5856 (enabled_for_depr_it): Rename to enabled_for_short_it.
5857 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5858 Remove default uses.
5859 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5860 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5861 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5862 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5864 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5867 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5868 float128 helper macros, which are no longer used after deleting
5869 the old 'q' built-in functions, and moving the round to odd
5870 built-in functions to being special built-in functions.
5871 (BU_FLOAT128_2): Likewise.
5872 (BU_FLOAT128_1_HW): Likewise.
5873 (BU_FLOAT128_2_HW): Likewise.
5874 (BU_FLOAT128_3_HW): Likewise.
5875 (FABSQ): Delete old 'q' built-in functions.
5876 (COPYSIGNQ): Likewise.
5877 (SQRTF128_ODD): Move round to odd built-in functions to be
5878 special built-in functions, so that we can handle
5879 -mabi=ieeelongdouble.
5880 (TRUNCF128_ODD): Likewise.
5881 (ADDF128_ODD): Likewise.
5882 (SUBF128_ODD): Likewise.
5883 (MULF128_ODD): Likewise.
5884 (DIVF128_ODD): Likewise.
5885 (FMAF128_ODD): Likewise.
5886 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5887 built-in names to 'f128'.
5888 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5889 old 'q' built-in functions, as the machine independent code for
5890 'f128' built-in functions handles this.
5891 (rs6000_expand_builtin): Add expansion for float128 round to odd
5892 functions, keying off on -mabi=ieeelongdouble of whether to use
5893 the KFmode or TFmode variant.
5894 (rs6000_init_builtins): Initialize the _Float128 round to odd
5896 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5897 _Float128 'q' built-in functions are now mapped into the new
5898 'f128' built-in functions.
5900 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5902 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5903 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5905 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5908 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5909 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5910 * toplev.c: Include "ipa-fnsummary.h".
5911 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5913 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5915 PR tree-optimization/82838
5916 * gimple-ssa-store-merging.c
5917 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5918 on a separate gimple_seq which is then appended to seq.
5920 2017-11-06 Jeff Law <law@redhat.com>
5923 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5924 (get_probe_interval): New functions.
5925 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5926 (ix86_adjust_stack_and_probe): Likewise.
5927 (output_adjust_stack_and_probe): Likewise.
5928 (ix86_emit_probe_stack_range): Likewise.
5929 (ix86_expand_prologue): Likewise.
5931 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5933 PR tree-optimization/82816
5934 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5935 if the modes of the two types are the same.
5936 (convert_plusminus_to_widen): Likewise.
5938 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5940 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5942 (usadv16qi): New define_expand.
5943 (usadv8hi): New define_expand.
5945 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5948 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5950 (inline_transform): Likewise.
5951 * predict.c (counts_to_freqs): Remove useless conditional.
5952 * profile-count.h (profile_count::apply_scale): Move sanity check.
5953 * tree-inline.c (copy_bb): Always scale.
5954 (copy_cfg_body): Likewise.
5956 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5959 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5961 (*cmp_ite0): Add enabled_for_depr_it attribute.
5962 (*cmp_ite1): Likewise.
5964 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5966 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5969 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5971 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5972 return the value by pointer.
5973 (extract_range_from_multiplicative_op_1): Update accordingly.
5974 Return as soon as an operation fails.
5976 2017-11-05 Tom de Vries <tom@codesourcery.com>
5979 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5980 (DEF_SANITIZER_BUILTIN): ... here.
5981 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5982 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5984 2017-11-05 Tom de Vries <tom@codesourcery.com>
5987 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5989 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5990 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5991 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5993 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5994 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5995 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5997 2017-11-05 Tom de Vries <tom@codesourcery.com>
6000 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6003 2017-11-04 Michael Clark <michaeljclark@mac.com>
6005 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6006 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6008 (*addsi3_extended): Likewise.
6009 (*addsi3_extended2): Likewise.
6010 (<optab>si3): Likewise.
6011 (<optab>di3): Likewise.
6012 (<optab><mode>3): Likewise.
6013 (<*optabe>si3_internal): Likewise.
6014 (zero_extendqi<SUPERQI:mode>2): Likewise.
6015 (*add<mode>hi3): Likewise.
6016 (*xor<mode>hi3): Likewise.
6017 (<optab>di3): Likewise.
6018 (*<optab>si3_extend): Likewise.
6019 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6020 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6021 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6023 2017-11-04 Andrew Waterman <andrew@sifive.com>
6025 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6026 TARGET_STRICT_ALIGN based upon -mtune argument.
6028 2017-11-04 Andrew Waterman <andrew@sifive.com>
6030 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6032 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6034 * config/i386/i386.c (choose_basereg): Use optional scratch
6035 register and add assertion.
6036 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6037 needed, and don't allocate stack.
6038 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6039 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6040 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6042 2017-11-03 Jeff Law <law@redhat.com>
6044 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6045 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6046 to probe at the start of a noreturn function.
6048 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6050 PR tree-optimization/78821
6051 * gimple-ssa-store-merging.c: Update the file comment.
6052 (MAX_STORE_ALIAS_CHECKS): Define.
6053 (struct store_operand_info): New type.
6054 (store_operand_info::store_operand_info): New constructor.
6055 (struct store_immediate_info): Add rhs_code and ops data members.
6056 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6057 and op1r arguments to the ctor, initialize corresponding data members.
6058 (struct merged_store_group): Add load_align_base and load_align
6060 (merged_store_group::merged_store_group): Initialize them.
6061 (merged_store_group::do_merge): Update them.
6062 (merged_store_group::apply_stores): Pick the constant for
6063 encode_tree_to_bitpos from one of the two operands, or skip
6064 encode_tree_to_bitpos if neither operand is a constant.
6065 (class pass_store_merging): Add process_store method decl. Remove
6066 bool argument from terminate_all_aliasing_chains method decl.
6067 (pass_store_merging::terminate_all_aliasing_chains): Remove
6068 var_offset_p argument and corresponding handling.
6069 (stmts_may_clobber_ref_p): New function.
6070 (compatible_load_p): New function.
6071 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6072 if there is overlap and rhs_code is not INTEGER_CST. For
6073 non-overlapping stores terminate group if rhs is not mergeable.
6074 (get_alias_type_for_stmts): Change first argument from
6075 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6076 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6077 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6079 (get_location_for_stmts): Change first argument from
6080 auto_vec<gimple *> & to vec<gimple *> &.
6081 (struct split_store): Remove orig_stmts data member, add orig_stores.
6082 (split_store::split_store): Create orig_stores rather than orig_stmts.
6083 (find_constituent_stmts): Renamed to ...
6084 (find_constituent_stores): ... this. Change second argument from
6085 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6086 to info structures rather than the statements.
6087 (split_group): Rename ALLOW_UNALIGNED argument to
6088 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6089 it. Adjust find_constituent_stores caller.
6090 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6091 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6092 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6093 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6094 (mem_valid_for_store_merging): New function.
6095 (handled_load): New function.
6096 (pass_store_merging::process_store): New method.
6097 (pass_store_merging::execute): Use process_store method. Adjust
6098 terminate_all_aliasing_chains caller.
6100 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6102 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6103 Return true for more constants, symbols and label references.
6104 (aarch64_valid_floating_const): Remove unused function.
6106 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6109 * config/aarch64/aarch64.c (aarch64_layout_frame):
6110 Undo forcing of LR at bottom of frame.
6112 2017-11-03 Jeff Law <law@redhat.com>
6115 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6116 for int_registers_saved.
6118 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6119 extracted from tree-ssa-dom.c.
6120 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6121 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6122 (record_equivalences_from_incoming_edge): Add additional argument
6123 to single_pred_edge_ignoring_loop_edges call.
6124 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6125 (uncprop_dom_walker::before_dom_children): Add additional argument
6126 to single_pred_edge_ignoring_loop_edges call.
6127 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6128 single_pred_edge_ignoring_loop_edges rather than open coding.
6129 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6131 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6133 * match.pd (-(-A)): Rewrite.
6135 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6137 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6138 (rs6000_emit_int_cmove): New declaration.
6139 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6140 (rs6000_emit_sISEL): Delete.
6141 (rs6000_emit_int_cmove): Make non-static.
6142 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6143 instead of rs6000_emit_sISEL.
6145 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6147 * asan.c (create_cond_insert_point): Maintain profile.
6148 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6150 * basic-block.h (struct basic_block_def): Remove frequency.
6151 (EDGE_FREQUENCY): Use to_frequency
6152 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6154 (find_traces): Update to use to_frequency.
6155 (find_traces_1_round): Likewise; use only IPA counts.
6156 (bb_to_key): Likewise.
6157 (connect_traces): Use IPA counts only.
6158 (copy_bb_p): Update to use to_frequency.
6159 (fix_up_crossing_landing_pad): Likewise.
6160 (sanitize_hot_paths): Likewise.
6161 * bt-load.c (basic_block_freq): Likewise.
6162 * cfg.c (init_flow): Set count_max to uninitialized.
6163 (check_bb_profile): Remove frequencies; check counts.
6164 (dump_bb_info): Do not dump frequencies.
6165 (update_bb_profile_for_threading): Update counts only.
6166 (scale_bbs_frequencies_int): Likewise.
6167 (MAX_SAFE_MULTIPLIER): Remove.
6168 (scale_bbs_frequencies_gcov_type): Update counts only.
6169 (scale_bbs_frequencies_profile_count): Update counts only.
6170 (scale_bbs_frequencies): Update counts only.
6171 * cfg.h (struct control_flow_graph): Add count-max.
6172 (update_bb_profile_for_threading): Update prototype.
6173 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6174 (find_many_sub_basic_blocks): Likewise.
6175 * cfgcleanup.c (try_forward_edges): Likewise.
6176 (try_crossjump_to_edge): Likewise.
6177 * cfgexpand.c (expand_gimple_cond): Likewise.
6178 (expand_gimple_tailcall): Likewise.
6179 (construct_init_block): Likewise.
6180 (construct_exit_block): Likewise.
6181 * cfghooks.c (verify_flow_info): Check consistency of counts.
6182 (dump_bb_for_graph): Do not dump frequencies.
6183 (split_block_1): Do not update frequencies.
6184 (split_edge): Do not update frequencies.
6185 (make_forwarder_block): Do not update frequencies.
6186 (duplicate_block): Do not update frequencies.
6187 (account_profile_record): Do not update frequencies.
6188 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6189 for global heuristics.
6190 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6191 (expected_loop_iterations_unbounded): Use counts only.
6192 * cfgloopmanip.c (scale_loop_profile): Simplify.
6193 (create_empty_loop_on_edge): Simplify
6195 (duplicate_loop_to_header_edge): Simplify
6196 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6197 (update_br_prob_note): Take care of removing note when profile
6199 (relink_block_chain): Do not dump frequency.
6200 (rtl_account_profile_record): Use to_frequency.
6201 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6202 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6203 (cgraph_update_edges_for_call_stmt_node): Likewise.
6204 (cgraph_edge::verify_count_and_frequency): Update.
6205 (cgraph_node::verify_node): Temporarily disable frequency verification.
6206 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6207 to_cgraph_frequency.
6208 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6209 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6211 (cgraph_node::expand_thunk): Update profile.
6212 * except.c (dw2_build_landing_pads): Do not update frequency.
6213 * final.c (compute_alignments): Use to_frequency.
6214 (dump_basic_block_info): Do not dump frequency.
6215 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6216 (dump_gimple_bb_header): Do not dump frequency.
6217 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6219 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6220 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6221 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6222 (init_before_recovery): Do not update frequency.
6223 (sched_create_recovery_edges): Do not update frequency.
6224 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6225 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6226 (ipa_cp_c_finalize): Set max_count to uninitialized.
6227 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6228 (param_change_prob): Use counts.
6229 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6231 * ipa-split.c (consider_split): Use to_frequency.
6232 (split_function): Use to_frequency.
6233 * ira-build.c (loop_compare_func): Likewise.
6234 (mark_loops_for_removal): Likewise.
6235 (mark_all_loops_for_removal): Likewise.
6236 * loop-doloop.c (doloop_modify): Do not update frequency.
6237 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6239 * lto-streamer-in.c (input_function): Update count_max.
6240 * omp-expand.c (expand_omp_taskreg): Update count_max.
6241 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6242 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6243 (maybe_hot_count_p): Use ipa counts only.
6244 (maybe_hot_bb_p): Simplify.
6245 (maybe_hot_edge_p): Simplify.
6246 (probably_never_executed): Do not take frequency argument.
6247 (probably_never_executed_bb_p): Do not pass frequency.
6248 (probably_never_executed_edge_p): Likewise.
6249 (combine_predictions_for_bb): Check that profile is nonzero.
6250 (propagate_freq): Do not set frequency.
6251 (drop_profile): Simplify.
6252 (counts_to_freqs): Simplify.
6253 (expensive_function_p): Use to_frequency.
6254 (propagate_unlikely_bbs_forward): Simplify.
6255 (determine_unlikely_bbs): Simplify.
6256 (estimate_bb_frequencies): Add hack to silence graphite issues.
6257 (compute_function_frequency): Use ipa counts.
6258 (pass_profile::execute): Update.
6259 (rebuild_frequencies): Use counts only.
6260 (force_edge_cold): Use counts only.
6261 * profile-count.c (profile_count::dump): Dump new count types.
6262 (profile_count::differs_from_p): Check compatiblity.
6263 (profile_count::to_frequency): New function.
6264 (profile_count::to_cgraph_frequency): New function.
6265 * profile-count.h (struct function): Declare.
6266 (enum profile_quality): Add profile_guessed_local and
6267 profile_guessed_global0.
6268 (class profile_proability): Decrease number of bits to 29;
6269 update from_reg_br_prob_note and to_reg_br_prob_note.
6270 (class profile_count: Update comment; decrease number of bits
6271 to 61. Check compatibility.
6272 (profile_count::compatible_p): New private member function.
6273 (profile_count::ipa_p): New member function.
6274 (profile_count::operator<): Handle global zero correctly.
6275 (profile_count::operator>): Handle global zero correctly.
6276 (profile_count::operator<=): Handle global zero correctly.
6277 (profile_count::operator>=): Handle global zero correctly.
6278 (profile_count::nonzero_p): New member function.
6279 (profile_count::force_nonzero): New member function.
6280 (profile_count::max): New member function.
6281 (profile_count::apply_scale): Handle IPA scalling.
6282 (profile_count::guessed_local): New member function.
6283 (profile_count::global0): New member function.
6284 (profile_count::ipa): New member function.
6285 (profile_count::to_frequency): Declare.
6286 (profile_count::to_cgraph_frequency): Declare.
6287 * profile.c (OVERLAP_BASE): Delete.
6288 (compute_frequency_overlap): Delete.
6289 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6290 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6291 * sched-ebb.c (rank): Use counts only.
6292 * shrink-wrap.c (handle_simple_exit): Use counts only.
6293 (try_shrink_wrapping): Use counts only.
6294 (place_prologue_for_one_component): Use counts only.
6295 * tracer.c (find_best_predecessor): Use to_frequency.
6296 (find_trace): Use to_frequency.
6297 (tail_duplicate): Use to_frequency.
6298 * trans-mem.c (expand_transaction): Do not update frequency.
6299 * tree-call-cdce.c: Do not update frequency.
6300 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6301 (gimple_merge_blocks): Likewise.
6302 (gimple_split_edge): Likewise.
6303 (gimple_duplicate_sese_region): Likewise.
6304 (gimple_duplicate_sese_tail): Likewise.
6305 (move_sese_region_to_fn): Likewise.
6306 (gimple_account_profile_record): Likewise.
6307 (insert_cond_bb): Likewise.
6308 * tree-complex.c (expand_complex_div_wide): Likewise.
6309 * tree-eh.c (lower_resx): Update profile.
6310 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6312 (initialize_cfun): Do not initialize frequencies
6313 (freqs_to_counts): Delete.
6314 (copy_cfg_body): Ignore count parameter.
6315 (copy_body): Update.
6316 (expand_call_inline): Update count_max.
6317 (optimize_inline_calls): Update count_max.
6318 (tree_function_versioning): Update count_max.
6319 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6320 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6322 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6323 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6324 (try_peel_loop): Likewise.
6325 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6327 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6328 (tree_transform_and_unroll_loop): Do not use frequencies
6329 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6330 Use reliable prediction only.
6331 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6332 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6333 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6334 probability scaling.
6335 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6336 not update frequency
6337 (any_remaining_duplicated_blocks): Likewise.
6338 (update_profile): Likewise.
6339 (estimated_freqs_path): Delete.
6340 (freqs_to_counts_path): Delete.
6341 (clear_counts_path): Delete.
6342 (ssa_fix_duplicate_block_edges): Likewise.
6343 (duplicate_thread_path): Likewise.
6344 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6345 * tree-tailcall.c (decrease_profile): Do not update frequency.
6346 (eliminate_tail_call): Likewise.
6347 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6348 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6349 (optimize_mask_stores): Likewise.
6350 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6351 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6352 (ubsan_expand_ptr_ifn): Update profile.
6353 * value-prof.c (gimple_ic): Simplify.
6354 * value-prof.h (gimple_ic): Update prototype.
6355 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6356 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6358 (want_inline_self_recursive_call_p): Likewise.
6359 (resolve_noninline_speculation): Only cummulate defined counts.
6360 (inline_small_functions): Use nonzero_p.
6361 (ipa_inline): Do not access freed node.
6363 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6365 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6366 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6368 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6370 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6371 non-legitimate address.
6373 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6375 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6376 (*lt0_<mode>di, *lt0_<mode>si): New.
6378 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6380 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6381 TARGET_PAIRED_FLOAT.
6383 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6384 Jim Wilson <jim.wilson@linaro.org>
6386 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6387 * config/aarch64/aarch64-tune.md: Regenerated.
6388 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6389 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6391 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6393 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6394 (arc_expand_prologue): Restore blink for millicode.
6395 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6397 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6400 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6401 gen_vec_duplicate after forcing the scalar into a register.
6403 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6405 * combine (try_combine): Print the insns input to try_combine to the
6408 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6411 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6412 Remove second argument from aarch64_process_target_attr call.
6413 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6415 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6417 (aarch64_handle_attr_arch): Remove second argument.
6418 (aarch64_handle_attr_cpu): Ditto.
6419 (aarch64_handle_attr_tune): Ditto.
6420 (aarch64_handle_attr_isa_flags): Ditto.
6421 (aarch64_process_one_target_attr): Ditto.
6422 (aarch64_process_target_attr): Ditto.
6423 (aarch64_option_valid_attribute_p): Remove second argument.
6424 on aarch64_process_target_attr call.
6426 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6428 * diagnostic.c: Include "selftest-diagnostic.h".
6429 (selftest::assert_location_text): New function.
6430 (selftest::test_diagnostic_get_location_text): New function.
6431 (selftest::diagnostic_c_tests): Call it.
6433 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6435 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6436 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6437 (class selftest::test_diagnostic_context): Move to...
6438 * selftest-diagnostic.c: New file.
6439 * selftest-diagnostic.h: New file.
6441 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6443 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6444 offset range for FT32B.
6445 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6446 * config/ft32/ft32.md: Add TARGET_NOPM.
6447 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6448 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6450 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6452 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6454 2017-11-02 Jeff Law <law@redhat.com>
6456 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6457 virtual keyword on FINAL OVERRIDE members.
6459 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6460 virtuals together. Add virtual destructor.
6461 (substitute_and_fold_engine): Similarly.
6463 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6465 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6467 2017-11-02 Richard Biener <rguenther@suse.de>
6469 PR tree-optimization/82795
6470 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6472 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6474 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6476 (gcc_GAS_CHECK_FEATURE): Remove.
6477 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6478 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6479 * configure: Regenerate.
6481 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6483 * config/arc/arc.c (hwloop_optimize): Account for empty
6486 2017-11-02 Richard Biener <rguenther@suse.de>
6489 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6490 Truncate ARRAY_REF index and element size.
6492 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6494 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6496 2017-11-01 Jeff Law <law@redhat.com>
6498 * tree-ssa-ccp.c (ccp_folder): New class derived from
6499 substitute_and_fold_engine.
6500 (ccp_folder::get_value): New member function.
6501 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6502 (ccp_fold_stmt): Remove prototype.
6503 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6504 * tree-ssa-copy.c (copy_folder): New class derived from
6505 substitute_and_fold_engine.
6506 (copy_folder::get_value): Renamed from get_value.
6507 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6508 * tree-vrp.c (vrp_folder): New class derived from
6509 substitute_and_fold_engine.
6510 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6511 (vrp_folder::get_value): New member function.
6512 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6513 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6514 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6515 provide a class interface to folder/substitute routines.
6516 (ssa_prop_fold_stmt_fn): Remove typedef.
6517 (ssa_prop_get_value_fn): Likewise.
6518 (subsitute_and_fold): Remove prototype.
6519 (replace_uses_in): Likewise.
6520 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6521 Renamed from replace_uses_in. Call the virtual member function
6522 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6523 (substitute_and_fold_dom_walker): Remove initialization of
6524 data member entries for calbacks. Add substitute_and_fold_engine
6525 member and initialize it.
6526 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6527 member functions for get_value, replace_phi_args_in c
6528 replace_uses_in, and fold_stmt calls.
6529 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6530 substitute_and_fold. Remove assert. Update ctor call.
6532 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6533 (ssa_prop_visit_phi_fn): Likewise.
6534 (class ssa_propagation_engine): New class to provide an interface
6536 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6538 (ssa_prop_visit_phi): Likewise.
6539 (ssa_propagation_engine::simulate_stmt): Moved into class.
6540 Call visit_phi/visit_stmt from the class rather than via
6541 file scoped static variables.
6542 (ssa_propagation_engine::simulate_block): Moved into class.
6543 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6544 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6545 set file scoped statics for the visit_stmt/visit_phi callbacks.
6546 * tree-complex.c (complex_propagate): New class derived from
6547 ssa_propagation_engine.
6548 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6549 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6550 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6552 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6553 ssa_propagation_engine.
6554 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6555 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6556 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6557 * tree-ssa-copy.c (copy_prop): New class derived from
6558 ssa_propagation_engine.
6559 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6560 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6561 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6562 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6563 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6564 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6565 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6567 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6569 PR rtl-optimization/82778
6570 PR rtl-optimization/82597
6571 * compare-elim.c (struct comparison): Add in_a_setter field.
6572 (find_comparison_dom_walker::before_dom_children): Remove killed
6573 bitmap and df_simulate_find_defs call, instead walk the defs.
6574 Compute last_setter and initialize in_a_setter. Merge definitions
6575 with first initialization for a few variables.
6576 (try_validate_parallel): Use insn_invalid_p instead of
6577 recog_memoized. Return insn rather than just the pattern.
6578 (try_merge_compare): Fix up comment. Don't uselessly test if
6579 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6581 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6585 Alan Hayward <alan.hayward@arm.com>
6586 David Sherwood <david.sherwood@arm.com>
6588 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6589 aarch64_hard_regno_nregs to get the number of registers
6592 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6593 Alan Hayward <alan.hayward@arm.com>
6594 David Sherwood <david.sherwood@arm.com>
6596 * config/aarch64/constraints.md (Upl): Rename to...
6598 * config/aarch64/aarch64.md
6599 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6602 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6603 Alan Hayward <alan.hayward@arm.com>
6604 David Sherwood <david.sherwood@arm.com>
6606 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6607 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6610 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6611 Alan Hayward <alan.hayward@arm.com>
6612 David Sherwood <david.sherwood@arm.com>
6614 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6615 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6616 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6618 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6619 of a single element.
6620 * config/aarch64/iterators.md: Add a comment above the permute
6621 unspecs to say that they are generated directly by
6622 aarch64_expand_vec_perm_const.
6623 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6625 2017-11-01 Nathan Sidwell <nathan@acm.org>
6627 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6629 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6631 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6632 and medany code models, and describe what they do.
6634 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6636 Revert accidental duplicate:
6638 * combine.c (can_change_dest_mode): Reject changes in
6639 REGMODE_NATURAL_SIZE.
6641 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6643 PR rtl-optimization/64682
6644 PR rtl-optimization/69567
6645 PR rtl-optimization/69737
6646 PR rtl-optimization/82683
6647 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6648 register mentioned in the note, drop the note, unless it came from I3,
6649 in which case it should go to I3 again.
6651 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6653 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6654 and return false if not.
6655 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6657 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6659 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6660 range is known to be empty.
6662 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6663 Alan Hayward <alan.hayward@arm.com>
6664 David Sherwood <david.sherwood@arm.com>
6666 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6667 and CONST_VECTOR_NUNITS instead of computing the number of units from
6668 the byte sizes of the vector and element.
6669 (simplify_binary_operation_1): Likewise.
6670 (simplify_const_binary_operation): Likewise.
6671 (simplify_ternary_operation): Likewise.
6673 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6674 Alan Hayward <alan.hayward@arm.com>
6675 David Sherwood <david.sherwood@arm.com>
6677 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6678 (int_mem_offset): ...this new function.
6679 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6680 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6681 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
6689 (compute_costs, find_decomposable_subregs, decompose_register)
6690 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6691 (resolve_clobber, dump_choices): Use it.
6693 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6694 Alan Hayward <alan.hayward@arm.com>
6695 David Sherwood <david.sherwood@arm.com>
6697 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * alias.c (find_base_value, find_base_term): Only process integer
6704 truncations. Check the precision rather than the size.
6706 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6707 Alan Hayward <alan.hayward@arm.com>
6708 David Sherwood <david.sherwood@arm.com>
6710 * machmode.h (is_narrower_int_mode): New function
6711 * optabs.c (expand_float, expand_fix): Use it.
6712 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6714 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6715 Alan Hayward <alan.hayward@arm.com>
6716 David Sherwood <david.sherwood@arm.com>
6718 * rtl.h (narrower_subreg_mode): New function.
6719 * ira-color.c (update_costs_from_allocno): Use it.
6721 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6722 Alan Hayward <alan.hayward@arm.com>
6723 David Sherwood <david.sherwood@arm.com>
6725 * optabs-query.h (convert_optab_p): New function, split out from...
6726 (convert_optab_handler): ...here.
6727 (widening_optab_handler): Delete.
6728 (find_widening_optab_handler): Remove permit_non_widening parameter.
6729 (find_widening_optab_handler_and_mode): Likewise. Provide an
6730 override that operates on mode class wrappers.
6731 * optabs-query.c (widening_optab_handler): Delete.
6732 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6733 parameter. Assert that the two modes are the same class and that
6734 the "from" mode is narrower than the "to" mode. Use
6735 convert_optab_handler instead of widening_optab_handler.
6736 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6737 instead of widening_optab_handler.
6738 * expr.c (expand_expr_real_2): Update calls to
6739 find_widening_optab_handler.
6740 * optabs.c (expand_widen_pattern_expr): Likewise.
6741 (expand_binop_directly): Take the insn_code as a parameter.
6742 (expand_binop): Only call find_widening_optab_handler for
6743 conversion optabs; use optab_handler otherwise. Update calls
6744 to find_widening_optab_handler and expand_binop_directly.
6745 Use convert_optab_handler instead of widening_optab_handler.
6746 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6747 find_widening_optab_handler and use scalar_mode rather than
6749 (convert_plusminus_to_widen): Likewise.
6751 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6752 Alan Hayward <alan.hayward@arm.com>
6753 David Sherwood <david.sherwood@arm.com>
6755 * machmode.h (fixed_size_mode): New class.
6756 * rtl.h (get_pool_mode): Return fixed_size_mode.
6757 * gengtype.c (main): Add fixed_size_mode.
6758 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6759 (get_raw_arg_mode): Likewise.
6760 * doc/tm.texi: Regenerate.
6761 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6762 * targhooks.c (default_get_reg_raw_mode): Likewise.
6763 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6764 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6765 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6766 (msp430_get_raw_result_mode): Likewise.
6767 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6768 * dbxout.c (dbxout_parms): Require fixed-size modes.
6769 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6770 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6771 * omp-low.c (lower_oacc_reductions): Likewise.
6772 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6773 (simplify_subreg): Update accordingly.
6774 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6775 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6776 that aren't fixed-size.
6777 (get_pool_mode): Return a fixed_size_mode.
6778 (output_constant_pool_2): Take a fixed_size_mode.
6780 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6781 Alan Hayward <alan.hayward@arm.com>
6782 David Sherwood <david.sherwood@arm.com>
6784 * doc/rtl.texi (vec_series): Document.
6785 (const): Say that the operand can be a vec_series.
6786 * rtl.def (VEC_SERIES): New rtx code.
6787 * rtl.h (const_vec_series_p_1): Declare.
6788 (const_vec_series_p): New function.
6789 * emit-rtl.h (gen_const_vec_series): Declare.
6790 (gen_vec_series): Likewise.
6791 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6792 (gen_vec_series): Likewise.
6793 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6794 * simplify-rtx.c (simplify_unary_operation): Handle negations
6796 (simplify_binary_operation_series): New function.
6797 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6798 (test_vector_ops_series): New function.
6799 (test_vector_ops): Call it.
6800 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6801 gen_const_vec_series.
6802 (altivec_lvsr): Likewise.
6803 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6806 Alan Hayward <alan.hayward@arm.com>
6807 David Sherwood <david.sherwood@arm.com>
6809 * doc/rtl.texi (const): Update description of address constants.
6810 Say that vector constants are allowed too.
6811 * common.md (E, F): Use CONSTANT_P instead of checking for
6813 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6814 checking for CONST_VECTOR.
6815 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6817 * expr.c (expand_expr_real_2): Check for vector modes instead
6818 of checking for CONST_VECTOR.
6819 * rtl.h (const_vec_p): New function.
6820 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6821 (unwrap_const_vec_duplicate): Handle them here too.
6823 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6824 David Malcolm <dmalcolm@redhat.com>
6825 Alan Hayward <alan.hayward@arm.com>
6826 David Sherwood <david.sherwood@arm.com>
6828 * rtl.h (vec_duplicate_p): New function.
6829 * selftest-rtl.c (assert_rtx_eq_at): New function.
6830 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6831 (assert_rtx_eq_at): Declare.
6832 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6833 * selftest-run-tests.c (selftest::run_tests): Call it.
6834 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6835 (simplify_unary_operation_1): Recursively handle vector duplicates.
6836 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6838 (simplify_subreg): Handle subregs of vector duplicates.
6839 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6840 (selftest::simplify_rtx_c_tests): New functions.
6842 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6843 Alan Hayward <alan.hayward@arm.com>
6844 David Sherwood <david.sherwood@arm.com>
6846 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6847 (gen_vec_duplicate): Likewise.
6848 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6850 (gen_const_vector): ...here.
6851 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6852 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6853 whose elements are all equal.
6854 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6855 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6856 (simplify_relational_operation): Likewise.
6857 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6859 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6860 (aarch64_expand_vector_init): Likewise.
6861 * config/arm/arm.c (neon_vdup_constant): Likewise.
6862 (neon_expand_vector_init): Likewise.
6863 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6864 (arm_block_set_unaligned_vect): Likewise.
6865 (arm_block_set_aligned_vect): Likewise.
6866 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6867 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6868 (expand_vec_perm_even_odd_pack): Likewise.
6869 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6870 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6871 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6872 gen_const_vec_duplicate.
6873 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6874 * config/mips/mips.c (mips_gen_const_int_vector): Use
6875 gen_const_vec_duplicate.
6876 (mips_expand_vector_init): Use CONST0_RTX.
6877 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6878 (define_split): Use gen_const_vec_duplicate.
6879 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6880 (define_split): Use gen_const_vec_duplicate.
6881 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6882 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6883 * config/spu/spu.c (spu_const): Likewise.
6885 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6886 Alan Hayward <alan.hayward@arm.com>
6887 David Sherwood <david.sherwood@arm.com>
6889 * combine.c (can_change_dest_mode): Reject changes in
6890 REGMODE_NATURAL_SIZE.
6892 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6894 * configure.ac (--enable-libssp): New.
6895 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6896 trying to determine target-specific default. Adjust indentation.
6897 * configure: Regenerated.
6898 * doc/install.texi (Configuration): Expand --disable-libssp
6901 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6903 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6906 2017-10-31 Martin Jambor <mjambor@suse.cz>
6909 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6911 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6913 * auto-profile.c (autofdo_source_profile::read): Use
6914 UNKNOWN_LOCATION rather than 0.
6915 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6916 (warning_at): ...this overload.
6917 (warning_at_rich_loc_n): Rename to...
6918 (warning_n): ...this overload.
6919 (error_at_rich_loc): Rename to...
6920 (error_at): ...this overload.
6921 (pedwarn_at_rich_loc): Rename to...
6922 (pedwarn): ...this overload.
6923 (permerror_at_rich_loc): Rename to...
6924 (permerror): ...this overload.
6925 (inform_at_rich_loc): Rename to...
6926 (inform): ...this overload.
6927 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6928 (diagnostic_n_impl_richloc): Rename to...
6929 (diagnostic_n_impl): ...this rich_location *-based decl.
6930 (inform_at_rich_loc): Rename to...
6931 (inform): ...this, and add an assertion.
6932 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6933 (warning_at_rich_loc): Rename to...
6934 (warning_at): ...this, and add an assertion.
6935 (warning_at_rich_loc_n): Rename to...
6936 (warning_n): ...this, and add an assertion.
6937 (warning_n): Update location_t-based implementation for removal of
6938 location_t-based diagnostic_n_impl.
6939 (pedwarn_at_rich_loc): Rename to...
6940 (pedwarn): ...this, and add an assertion.
6941 (permerror_at_rich_loc): Rename to...
6942 (permerror): ...this, and add an assertion.
6943 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6944 (error_at_rich_loc): Rename to...
6945 (error_at): ...this, and add an assertion.
6946 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6947 (driver::do_spec_on_infiles): Likewise.
6948 * substring-locations.c (format_warning_va): Update for renaming
6949 of inform_at_rich_loc.
6951 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6953 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6954 _Float<N>X built-in functions so that the variant without the
6955 "__builtin_" prefix is only enabled for the GNU C and Objective C
6956 languages when they are in non-strict ANSI/ISO mode.
6957 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6958 * target.def (floatn_builtin_p): Add a target hook to control
6959 whether _Float<N> and _Float<N>X built-in functions without the
6960 "__builtin_" prefix are enabled, and return true for C and
6961 Objective C in the default hook. Include langhooks.h in
6963 * targhooks.h (default_floatn_builtin_p): Likewise.
6964 * targhooks.c (default_floatn_builtin_p): Likewise.
6965 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6966 floatn_builtin_p target hook.
6967 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6969 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6970 Eric Botcazou <ebotcazou@adacore.com>
6972 PR rtl-optimization/81803
6973 * lra-constraints.c (curr_insn_transform): Also reload the whole
6974 register for a strict subreg no wider than a word if this is for
6975 a WORD_REGISTER_OPERATIONS target.
6977 2017-10-31 Jason Merrill <jason@redhat.com>
6979 * gdbinit.in: Skip over inlines from timevar.h.
6981 2017-10-31 Martin Liska <mliska@suse.cz>
6983 * doc/gcov.texi: Document new option.
6984 * gcov.c (print_usage): Likewise print it.
6985 (process_args): Support the argument.
6986 (format_count): New function.
6987 (format_gcov): Use the function.
6989 2017-10-31 Martin Liska <mliska@suse.cz>
6991 * gcov.c (struct name_map): do not use typedef.
6992 Define operator== and operator<.
6993 (name_search): Remove.
6994 (name_sort): Remove.
6995 (main): Do not allocate names.
6996 (process_file): Add vertical space.
6997 (generate_results): Use std::find.
6998 (release_structures): Do not release memory.
6999 (find_source): Use std::find.
7001 2017-10-31 Martin Liska <mliska@suse.cz>
7003 * gcov.c (struct line_info): Remove it's typedef.
7004 (line_info::line_info): Add proper ctor.
7005 (line_info::has_block): Do not use a typedef.
7006 (struct source_info): Do not use typedef.
7007 (circuit): Likewise.
7008 (get_cycles_count): Likewise.
7009 (output_intermediate_file): Iterate via vector iterator.
7010 (add_line_counts): Use std::vector methods.
7011 (accumulate_line_counts): Likewise.
7012 (output_lines): Likewise.
7014 2017-10-31 Martin Liska <mliska@suse.cz>
7016 * gcov.c (struct source_info): Remove typedef.
7017 (source_info::source_info): Add proper ctor.
7018 (accumulate_line_counts): Use struct, not it's typedef.
7019 (output_gcov_file): Likewise.
7020 (output_lines): Likewise.
7021 (main): Do not allocate an array.
7022 (output_intermediate_file): Use size of vector container.
7023 (process_file): Resize the vector.
7024 (generate_results): Do not preallocate, use newly added vector
7026 (release_structures): Do not release sources.
7027 (find_source): Use vector methods.
7028 (add_line_counts): Do not use typedef.
7030 2017-10-31 Martin Liska <mliska@suse.cz>
7032 * doc/gcov.texi: Document that.
7033 * gcov.c (add_line_counts): Mark lines with a non-executed
7035 (output_line_beginning): Handle such lines.
7036 (output_lines): Pass new argument.
7037 (output_intermediate_file): Print it in intermediate format.
7039 2017-10-31 Martin Liska <mliska@suse.cz>
7041 * color-macros.h: New file.
7042 * diagnostic-color.c: Factor out color related to macros to
7044 * doc/gcov.texi: Document -k option.
7045 * gcov.c (INCLUDE_STRING): Include string.h.
7046 (print_usage): Add -k option.
7047 (process_args): Parse it.
7048 (pad_count_string): New function.
7049 (output_line_beginning): Likewise.
7050 (DEFAULT_LINE_START): New macro.
7051 (output_lines): Support color output.
7053 2017-10-31 Martin Liska <mliska@suse.cz>
7055 PR gcov-profile/82633
7056 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7057 their interaction with GCOV infrastructure.
7058 * configure.ac: Add -fkeep-{inline,static}-functions to
7060 * configure: Regenerate.
7062 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7065 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7067 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7070 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7071 into a register if it does not fit into an immediate offset field.
7073 2017-10-31 Olivier Hainque <hainque@adacore.com>
7075 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7077 2017-10-31 Julia Koval <julia.koval@intel.com>
7079 * config.gcc: Add gfniintrin.h.
7080 * config/i386/gfniintrin.h: New.
7081 * config/i386/i386-builtin-types.def
7082 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7083 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7084 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7085 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7086 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7087 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7088 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7089 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7090 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7091 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7092 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7093 * config/i386/immintrin.h: Include gfniintrin.h.
7094 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7096 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7098 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7100 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7102 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7103 (arm_ashldi3_1bit): Remove pattern.
7104 (ashrdi3): Remove shift by 1 expansion.
7105 (arm_ashrdi3_1bit): Remove pattern.
7106 (lshrdi3): Remove shift by 1 expansion.
7107 (arm_lshrdi3_1bit): Remove pattern.
7108 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7109 cost of ashldi3 by 1.
7110 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7111 (<shift>di3_neon): Likewise.
7113 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7115 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7116 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7117 and (*aarch64_simd_mov<VQ:mode>).
7118 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7119 pattern alternative.
7120 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7121 attributes to match pattern alternative.
7123 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7125 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7126 * config/rs6000/emmintrin.h: New file.
7127 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7129 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7131 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7132 * (movdi_vfp_cortexa8): Remove pattern.
7134 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7136 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7138 (Specific, alpha*-dec-osf5.1): Remove.
7139 (Specific, mips-sgi-irix5): Remove.
7140 (Specific, mips-sgi-irix6): Remove.
7142 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7145 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7146 arguments to clear_bit_region_be.
7148 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7150 * gimplify.c: Include memmodel.h.
7152 2017-10-30 Martin Jambor <mjambor@suse.cz>
7154 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7155 condition whether loop should be executed at all.
7157 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7160 gimple folding of vec_madd() intrinsics.
7161 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7162 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7163 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7165 2017-10-30 Richard Biener <rguenther@suse.de>
7167 PR tree-optimization/82762
7169 2017-10-23 Richard Biener <rguenther@suse.de>
7171 PR tree-optimization/82129
7173 2017-08-01 Richard Biener <rguenther@suse.de>
7175 PR tree-optimization/81181
7176 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7177 (compute_antic): ... end of iteration here.
7179 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7181 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7182 -std=iso9899:2017 and -std=gnu17.
7183 * doc/standards.texi (C Language): Document C17 support.
7184 * doc/cpp.texi (Overview): Mention -std=c17.
7185 (Standard Predefined Macros): Document C11 and C17 values of
7186 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7187 * doc/extend.texi (Inline): Do not list individual options for
7188 standards newer than C99.
7189 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7191 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7194 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7196 * asan.c (asan_finish_file): Align asan globals array by shadow
7199 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7202 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7203 (struct store_immediate_info): Add bitregion_start and bitregion_end
7205 (store_immediate_info::store_immediate_info): Add brs and bre
7206 arguments and initialize bitregion_{start,end} from those.
7207 (struct merged_store_group): Add bitregion_start, bitregion_end,
7208 align_base and mask fields. Drop unnecessary struct keyword from
7209 struct store_immediate_info. Add do_merge method.
7210 (clear_bit_region_be): Use memset instead of loop storing zeros.
7211 (merged_store_group::do_merge): New method.
7212 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7213 stores as long as the surrounding bitregions have no gaps.
7214 (merged_store_group::merge_overlapping): Use do_merge.
7215 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7216 is byte aligned, rather than requiring that start and width are
7217 byte aligned. Drop unnecessary struct keyword from
7218 struct store_immediate_info. Allocate and populate also mask array.
7219 Make start of the arrays relative to bitregion_start rather than
7220 start and size them according to bitregion_{end,start} difference.
7221 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7222 struct store_immediate_info.
7223 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7224 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7225 struct keyword from struct store_immediate_info.
7226 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7227 between stores as long as the surrounding bitregions have no gaps.
7229 (struct split_store): Add orig non-static data member.
7230 (split_store::split_store): Initialize orig to false.
7231 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7232 if there is exactly a single original stmt. Change stmts argument
7233 to pointer from reference, if NULL, don't push anything to it. Add
7234 first argument, use it to optimize skipping over orig stmts that
7235 are known to be before bitpos already. Simplify.
7236 (split_group): Return unsigned int count how many stores are or
7237 would be needed rather than a bool. Add allow_unaligned argument.
7238 Change split_stores argument from reference to pointer, if NULL,
7239 only do a dry run computing how many stores would be produced.
7240 Rewritten algorithm to use both alignment and misalign if
7241 !allow_unaligned and handle bitfield stores with gaps.
7242 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7243 from bitregion_start instead of start. Compute allow_unaligned
7244 here, if true, do 2 split_group dry runs to compute which one
7245 produces fewer stores and prefer aligned if equal. Punt if
7246 new count is bigger or equal than original before emitting any
7247 statements, rather than during that. Remove no longer needed
7248 new_ssa_names tracking. Replace num_stmts with
7249 split_stores.length (). Use 32-bit stack allocated entries
7250 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7251 if possible. Handle bitfields with gaps.
7252 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7253 Compute bitregion_{start,end} for the stores and construct
7254 store_immediate_info with that. Formatting fixes.
7256 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7259 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7260 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7262 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7264 * gimplify.c: Include tm_p.h.
7266 * common.opt (gcoff): Re-add as ignored option.
7267 (gcoff1, gcoff2, gcoff3): Likewise.
7269 * Makefile.in (OBJS): Delete sdbout.o.
7270 (GTFILES): Delete $(srcdir)/sdbout.c.
7271 * debug.h: Delete sdb_debug_hooks.
7272 * final.c: Delete sdbout.h include.
7273 (final_scan_insn): Delete SDB_DEBUG check.
7274 (rest_of_clean_state): Likewise.
7275 * output.h: Delete sdb_begin_function_line.
7278 * toplev.c: Delete sdbout.h include.
7279 (process_options): Delete SDB_DEBUG check.
7280 * tree-core.h (tree_type_common): Delete pointer field of
7282 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7283 TYPE_SYMTAB_POINTER.
7284 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7285 (TYPE_SYMTAB_IS_POINTER): Delete.
7286 (TYPE_SYMTAB_IS_DIE): Renumber.
7287 * xcoffout.c: Refer to former sdbout.c file.
7288 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7290 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7291 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7292 (Debugging Options): Delete -gcoff.
7293 (-gstabs): Delete SDB reference.
7295 (-gcoff@var{level}): Delete.
7296 * doc/passes.texi (Debugging information output): Delete SDB and
7297 sdbout.c references.
7298 * doc/tm.texi: Regenerate.
7299 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7300 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7302 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7303 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7304 (SDB_DEBUGGING_INFO): Delete.
7305 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7306 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7307 * target.def (output_source_filename): Delete COFF reference.
7309 * common.opt (gcoff): Delete.
7310 (gxcoff+): Update Negative chain.
7311 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7313 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7314 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7315 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7317 (expand_function_start): Change sdb reference to past tense.
7318 (expand_function_end): Change sdb reference to past tense.
7319 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7320 * opts.c (debug_type_names): Delete coff entry.
7321 (common_handle_option): Delete OPT_gcoff case.
7322 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7324 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7325 * config/cris/cris.h: Delete SDB reference in comment.
7326 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7327 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7328 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7329 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7331 (ix86_expand_prologue): Likewise.
7332 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7333 * config/ia64/ia64.h: Likewise.
7334 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7335 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7337 * config/mmix/mmix.h: Likewise.
7338 * config/nds32/nds32.c: Likewise.
7339 * config/stormy/storym16.h: Likewise.
7340 * config/visium/visium.h: Likewise.
7341 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7343 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7345 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7346 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7347 FRAME_POINTER_REGNUM point at high end of local var area.
7349 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7351 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7352 Move comment around. Do not reset best_edge for a copiable
7353 destination if the copy would cause a partition change.
7354 (better_edge_p): Remove redundant check.
7356 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7358 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7360 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7362 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7363 for math functions that have _Float<N> and _Float<N>X variants.
7364 (mathfn_built_in_2): Add support for math functions that have
7365 _Float<N> and _Float<N>X variants.
7366 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7367 (expand_builtin_mathfn_ternary): Add support for fma with
7368 _Float<N> and _Float<N>X variants.
7369 (expand_builtin): Likewise.
7370 (fold_builtin_3): Likewise.
7371 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7372 create math function _Float<N> and _Float<N>X variants as external
7374 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7375 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7376 the __builtin_ prefix and if not strict ansi, without the prefix.
7377 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7378 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7379 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7380 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7381 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7382 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7383 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7384 function signatures for fma _Float<N> and _Float<N>X variants.
7385 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7386 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7387 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7388 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7389 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7390 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7391 * gencfn-macros.c (print_case_cfn): Add support for math functions
7392 that have _Float<N> and _Float<N>X variants.
7393 (print_define_operator_list): Likewise.
7394 (fltfn_suffixes): Likewise.
7396 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7397 for math functions that have _Float<N> and _Float<N>X variants.
7398 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7399 and _Float<N>X variants.
7400 (COPYSIGN): Likewise.
7403 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7404 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7406 (integer_valued_read_call_p): Likewise.
7407 * fold-const-call.c (fold_const_call_ss): Likewise.
7408 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7409 _Float<N> and _Float<N>X variants.
7410 (fold_const_call_ssss): Add support for fma _Float<N> and
7411 _Float<N>X variants.
7412 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7413 support for copysign and fma _Float<N> and _Float<N>X variants.
7414 (backprop::process_builtin_call_use): Likewise.
7415 * tree-call-cdce.c (can_test_argument_range); Add support for
7416 sqrt _Float<N> and _Float<N>X variants.
7417 (edom_only_function): Likewise.
7418 (get_no_error_domain): Likewise.
7419 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7420 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7421 copysign _Float<N> and _Float<N>X variants.
7422 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7423 handled by machine independent code.
7424 (FMAF128): Likewise.
7425 * doc/cpp.texi (Common Predefined Macros): Document defining
7426 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7427 fma _Float<N> and _Float<N>X variants.
7429 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7432 * config/i386/i386-modes.def (CCFPU): Remove definition.
7433 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7434 (ix86_cc_modes_compatible): Ditto.
7435 (ix86_expand_carry_flag_compare): Ditto.
7436 (ix86_expand_int_movcc): Ditto.
7437 (ix86_expand_int_addcc): Ditto.
7438 (ix86_reverse_condition): Ditto.
7439 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7440 Return true/false for unordered/ordered fp comparisons.
7441 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7442 (ix86_prepare_fp_compare_args): Update for rename.
7443 (ix86_expand_fp_compare): Update for rename. Generate unordered
7444 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7445 (ix86_expand_sse_compare_and_jump): Ditto.
7446 * config/i386/predicates.md (fcmov_comparison_operator):
7447 Remove CCFPU mode handling.
7448 (ix86_comparison_operator): Ditto.
7449 (ix86_carry_flag_operator): Ditto.
7450 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7451 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7452 (*cmpu<mode>_cc_i387): Ditto.
7453 (FPCMP): Remove mode iterator.
7454 (unord): Remove mode attribute.
7455 (unord_subst): New define_subst transformation
7456 (unord): New define_subst attribute.
7458 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7459 (*cmpi<unord>xf_i387): Ditto.
7460 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7461 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7462 using unord_subst transformation.
7463 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7464 (round_saeonly): Also handle CCFP mode.
7465 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7466 Remove UNSPEC_SAHF unspec handling.
7468 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7470 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7472 2017-10-27 Jeff Law <law@redhat.com>
7474 * gimple-ssa-sprintf.c: Include domwalk.h.
7475 (class sprintf_dom_walker): New class, derived from dom_walker.
7476 (sprintf_dom_walker::before_dom_children): New function.
7477 (struct call_info): Moved into sprintf_dom_walker class
7478 (compute_formath_length, handle_gimple_call): Likewise.
7479 (sprintf_length::execute): Call the dominator walker rather
7480 than walking the statements.
7482 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7483 gimple statement locations.
7484 (check_array_bounds): Corresponding changes. Get the statement's
7485 location directly from wi->stmt.
7487 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7490 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7492 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7494 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7495 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7497 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7500 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7501 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7502 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7503 maybe_get_pool_constant.
7504 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7507 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7509 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7510 2.26 caveat. Update gas and gld versions.
7511 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7514 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7516 * cgraph.h (set_malloc_flag): Declare.
7517 * cgraph.c (set_malloc_flag_1): New function.
7518 (set_malloc_flag): Likewise.
7519 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7520 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7522 (read_ipa_call_summary): Add support for reading is_return_callee.
7523 (write_ipa_call_summary): Stream is_return_callee.
7524 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7525 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7526 ipa-prop.h, ipa-fnsummary.h.
7527 (pure_const_names): Change to static.
7528 (malloc_state_e): Define.
7529 (malloc_state_names): Define.
7530 (funct_state_d): Add field malloc_state.
7531 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7532 (check_retval_uses): New function.
7533 (malloc_candidate_p): Likewise.
7534 (analyze_function): Add support for malloc attribute.
7535 (pure_const_write_summary): Stream malloc_state.
7536 (pure_const_read_summary): Add support for reading malloc_state.
7537 (dump_malloc_lattice): New function.
7538 (propagate_malloc): New function.
7539 (warn_function_malloc): New function.
7540 (ipa_pure_const::execute): Call propagate_malloc and
7541 ipa_free_fn_summary.
7542 (pass_local_pure_const::execute): Add support for malloc attribute.
7543 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7544 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7546 2017-10-27 Martin Liska <mliska@suse.cz>
7548 PR gcov-profile/82457
7549 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7550 for fork-like functions to be properly instrumented.
7552 2017-10-27 Richard Biener <rguenther@suse.de>
7555 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7556 info when we redirected EH.
7558 2017-10-26 Michael Collison <michael.collison@arm.com>
7560 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7562 (<optab>_trunchf<GPI:mode>2: New pattern.
7563 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7564 * config/aarch64/iterators.md (wv): New mode attribute.
7565 (vf, VF): New mode attributes.
7566 (vgp, VGP): New mode attributes.
7567 (s): Update attribute with SImode and DImode prefixes.
7569 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7571 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7572 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7573 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7574 (nios2_option_overide): Initialize it. Don't allow R0-relative
7575 addressing with PIC.
7576 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7577 (nios2_symbolic_constant_p): Likewise.
7578 (nios2_legitimate_address_p): Likewise.
7579 (nios2_r0rel_section_name_p): New.
7580 (nios2_symbol_ref_in_r0rel_data_p): New.
7581 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7582 (r0rel_constant_p): New.
7583 (nios2_print_operand_address): Handle r0rel_constant_p.
7584 (nios2_cdx_narrow_form_p): Likewise.
7585 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7586 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7587 (Nios II Options): Document -mr0rel-sec.
7589 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7591 * config/nios2/nios2.c: Include xregex.h.
7592 (nios2_gprel_sec_regex): New.
7593 (nios2_option_overide): Initialize it. Don't allow GP-relative
7594 addressing with PIC.
7595 (nios2_small_section_name_p): Check for regex match.
7596 * config/nios2/nios2.opt (mgprel-sec=): New option.
7597 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7598 (Nios II Options): Document -mgprel-sec.
7600 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7602 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7604 2017-10-26 Tom de Vries <tom@codesourcery.com>
7606 PR tree-optimization/82707
7607 * gimple.c (gimple_copy): Fix unsharing of
7608 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7610 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7612 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7613 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7614 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7615 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7616 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7617 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7618 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7619 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7620 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7621 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7622 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7623 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7624 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7625 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7626 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7627 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7628 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7630 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7632 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7634 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7635 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7636 warning to rs6000.c. Remove the Undocumented flag, since it has
7638 (-mabi=ibmlongdouble): Likewise.
7639 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7640 already set, set the default format for long double.
7641 (rs6000_debug_reg_global): Print whether long double is IBM or
7643 (rs6000_option_override_internal): Rework setting long double
7644 format. Only warn if the user is changing the long double default
7645 and they did not use -Wno-psabi.
7646 * doc/invoke.texi (PowerPC options): Update the documentation for
7647 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7649 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7650 Alan Hayward <alan.hayward@arm.com>
7651 David Sherwood <david.sherwood@arm.com>
7653 * rtl.h (wider_subreg_mode): New function.
7654 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7655 rather than an unsigned int *.
7656 * ira-color.c (regno_max_ref_width): Replace with...
7657 (regno_max_ref_mode): ...this new variable.
7658 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7659 Use wider_subreg_mode.
7660 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7661 rather than an unsigned int *.
7662 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7663 (process_alt_operands): Likewise.
7664 (invariant_p): Likewise.
7665 * lra-spills.c (assign_mem_slot): Likewise.
7666 (add_pseudo_to_slot): Likewise.
7667 * lra.c (collect_non_operand_hard_regs): Likewise.
7668 (add_regs_to_insn_regno_info): Likewise.
7669 * reload1.c (regno_max_ref_width): Replace with...
7670 (regno_max_ref_mode): ...this new variable.
7671 (reload): Update accordingly. Update call to
7672 ira_sort_regnos_for_alter_reg.
7673 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7674 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7675 (scan_paradoxical_subregs): Likewise.
7677 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7679 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7680 (aarch64_frame): Add emit_frame_chain boolean.
7681 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7682 Move eh_return case to aarch64_layout_frame.
7683 (aarch64_layout_frame): Initialize emit_frame_chain.
7684 (aarch64_expand_prologue): Use emit_frame_chain.
7686 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7688 * config/aarch64/aarch64.c (aarch64_layout_frame):
7689 Ensure LR is always stored at the bottom of the callee-saves.
7690 Remove rarely used frame layout which saves callee-saves at top of
7691 frame, so the store of LR can be used as a valid probe in all cases.
7693 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7695 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7696 Improve unaligned TImode/TFmode base/offset split.
7698 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7699 Alan Hayward <alan.hayward@arm.com>
7700 David Sherwood <david.sherwood@arm.com>
7702 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7703 * combine.c (find_single_use_1): Likewise.
7704 (expand_field_assignment): Likewise.
7705 (move_deaths): Likewise.
7706 * lra-constraints.c (simplify_operand_subreg): Likewise.
7707 (curr_insn_transform): Likewise.
7708 * lra.c (collect_non_operand_hard_regs): Likewise.
7709 (add_regs_to_insn_regno_info): Likewise.
7710 * rtlanal.c (reg_referenced_p): Likewise.
7711 (covers_regno_no_parallel_p): Likewise.
7713 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7715 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7716 Don't print any bits outside the precision of the value.
7717 * wide-int.cc (test_printing): Add some new tests.
7719 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7721 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7722 supports -xbrace_comment option.
7723 * configure: Regenerate.
7724 * config.in: Regenerate.
7725 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7726 (ASM_CPU_SPEC): Use it.
7728 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7730 * target.def (static_rtx_alignment): New hook.
7731 * targhooks.h (default_static_rtx_alignment): Declare.
7732 * targhooks.c (default_static_rtx_alignment): New function.
7733 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7734 * doc/tm.texi: Regenerate.
7735 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7736 instead of targetm.constant_alignment. Remove call to
7738 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7739 (cris_preferred_mininum_alignment): New function, split out from...
7740 (cris_constant_alignment): ...here.
7741 (cris_static_rtx_alignment): New function.
7742 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7744 (ix86_constant_alignment): ...here.
7745 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7746 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7747 (mmix_static_rtx_alignment): New function.
7748 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7749 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7751 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7754 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7755 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7757 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7759 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7760 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7761 mark as a sign-extending load.
7762 (local_pic_load_u): Define.
7764 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7767 * fold-const.c (operand_equal_for_comparison_p): Also return true
7768 if ARG0 is a simple variant of ARG1 with narrower precision.
7769 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7771 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7773 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7775 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7776 scatter_static, scatter_per_elt.
7777 * x86-tune-costs.h: Add new cost entries.
7779 2017-10-25 Richard Biener <rguenther@suse.de>
7781 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7782 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7783 class pass_fre): Move to ...
7784 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7785 class pass_fre): ... here and adjust for statistics.
7787 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7790 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7791 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7792 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7793 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7796 2017-10-25 Richard Biener <rguenther@suse.de>
7798 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7799 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7800 eliminate_push_avail, eliminate_insert): Move inside...
7801 (class eliminate_dom_walker): ... this class in preparation
7803 (fini_eliminate): Remove by merging with ...
7804 (eliminate): ... this function. Adjust for class changes.
7805 (pass_pre::execute): Remove fini_eliminate call.
7806 (pass_fre::execute): Likewise.
7808 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7811 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7812 (VPERMI2, VPERMI2I): New mode iterators.
7813 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7814 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7816 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7817 mode iterator. Remove 3 old define_insn patterns.
7818 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7819 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7820 VPERMI2 mode iterator, remove the other two expanders.
7821 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7822 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7823 instructions, remove the other two patterns.
7824 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7825 mode iterator, remove the other two patterns.
7826 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7827 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7828 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7829 and adjust argument order accordingly.
7830 (ix86_expand_vec_perm): Adjust caller.
7831 (expand_vec_perm_1): Likewise.
7832 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7833 (expand_vec_perm_vpermt2_vpshub2): ... this.
7834 (ix86_expand_vec_perm_const_1): Adjust caller.
7835 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7838 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7839 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7840 (vec_shl_<mode>): Remove unused expander.
7841 (avx512bw_<shift_insn><mode>3): New define_insn.
7842 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7843 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7845 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7848 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7851 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7853 PR rtl-optimization/82396
7854 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7855 (autopref_multipass_init): Simplify initialization.
7856 (autopref_rank_data): Simplify sort order.
7857 * gcc/sched-int.h (autopref_multipass_data_): Remove
7858 multi_mem_insn_p, min_offset and max_offset.
7860 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7863 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7864 Check special value of flag_omit_frame_pointer.
7865 (aarch64_can_eliminate): Likewise.
7866 (aarch64_override_options_after_change_1): Simplify handling of
7867 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7869 2017-10-24 Richard Biener <rguenther@suse.de>
7871 PR tree-optimization/82697
7872 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7873 zero for conditional load and unconditional store.
7875 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7877 * doc/install.texi: Document bootstrap-cet.
7879 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7882 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7883 ENDBR instruction at function entrance if function is only
7886 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7889 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7890 patterns to better describe from which operation the CF is computed.
7891 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7892 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7893 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7894 is 0, use _0 suffixed expanders instead of emitting a comparison
7897 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7899 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7900 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7902 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7905 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7906 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7907 * loop-iv.c (iv_get_reaching_def): Likewise.
7908 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7909 variable is promoted and the partition contains undefined values.
7911 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7913 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7915 (nios2_address_cost): Define.
7916 (nios2_legitimize_address): Recognize (exp + constant) directly.
7917 (TARGET_ADDRESS_COST): Define.
7919 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7921 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7922 (nios2_symbolic_memory_operand_p): Declare.
7923 (nios2_split_large_constant): Declare.
7924 (nios2_split_symbolic_memory_operand): Declare.
7925 * config/nios2/nios2.c: Adjust includes.
7926 (nios2_symbolic_constant_allowed): New.
7927 (nios2_symbolic_constant_p): New.
7928 (nios2_plus_symbolic_constant_p): New.
7929 (nios2_valid_addr_expr_p): Recognize addresses involving
7931 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7932 (nios2_symbolic_memory_operand_p): New.
7933 (nios2_large_constant_p): New.
7934 (nios2_split_large_constant): New.
7935 (nios2_split_plus_large_constant): New.
7936 (nios2_split_symbolic_memory_operand): New.
7937 (nios2_legitimize_address): Code refactoring. Handle addresses
7938 involving symbolic constants.
7939 (nios2_emit_move_sequence): Likewise.
7940 (nios2_print_operand): Improve error output.
7941 (nios2_print_operand_address): Handle LO_SUM.
7942 (nios2_cdx_narrow_form_p): Likewise.
7943 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7944 operands involving symbolic constants.
7945 (movhi_internal, movsi_internal): Likewise.
7946 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7947 (extendhisi2, extendqi<mode>2): Likewise.
7949 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7951 * tree-pass.h (PROP_rtl_split_insns): Define.
7952 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7954 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7956 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7958 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7961 * target.def (const_not_ok_for_debug_p): Default to
7962 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7963 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7964 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7965 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7966 which targetm.const_not_ok_for_debug_p returned true.
7967 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7969 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7971 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7972 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7973 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7975 (ix86_const_not_ok_for_debug_p): New function.
7976 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7977 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7979 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7982 * system.h: Conditionally include "unique-ptr.h" if
7983 INCLUDE_UNIQUE_PTR is defined.
7984 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7985 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7987 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7989 * config/rl78/rl78.md: New define_expand "subdi3".
7991 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7994 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7995 DF_REF_INSN if DF_REF_INSN_INFO is false.
7997 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7999 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8000 xmm_move instead of sse_move.
8001 (sse_store_index): New function.
8002 (ix86_register_move_cost): Be more sensible about mismatch stall;
8003 model AVX moves correctly; make difference between sse->integer and
8005 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8006 moves; make difference between SSE and AVX.
8007 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8008 and zmm_move. Increase size of sse load and store tables;
8009 add unaligned load and store tables; add ssemmx_to_integer.
8010 * x86-tune-costs.h: Update all entries according to real
8011 move latencies from Agner Fog's manual and chip documentation.
8013 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8016 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8017 * config/i386/constraints.md (Wf): New constraint.
8018 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8019 (cmp<dwi>_doubleword): Removed.
8020 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8021 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8022 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8023 expand with cmp<dwi>_doubleword. For LTU and GEU use
8024 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8026 * common.opt (gcolumn-info): Enable by default.
8027 * doc/invoke.texi (gcolumn-info): Document new default.
8029 2017-10-23 Richard Biener <rguenther@suse.de>
8031 PR tree-optimization/82672
8032 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8033 Fold the stmt if we propagated into it.
8035 2017-10-23 Richard Biener <rguenther@suse.de>
8037 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8038 (bitmap_remove_expr_from_set): ... this. All callers call this
8039 for non-constant values.
8040 (bitmap_set_subtract): Rename to...
8041 (bitmap_set_subtract_expressions): ... this. Adjust and
8043 (bitmap_set_contains_value): Remove superfluous check.
8044 (bitmap_set_replace_value): Inline into single caller ...
8045 (bitmap_value_replace_in_set): ... here and simplify.
8046 (dependent_clean): Merge into ...
8047 (clean): ... this using an overload. Adjust.
8048 (prune_clobbered_mems): Adjust.
8049 (compute_antic_aux): Likewise.
8050 (compute_partial_antic_aux): Likewise.
8052 2017-10-23 Richard Biener <rguenther@suse.de>
8054 PR tree-optimization/82129
8056 2017-08-01 Richard Biener <rguenther@suse.de>
8058 PR tree-optimization/81181
8059 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8060 (compute_antic): ... end of iteration here.
8062 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8064 * target.def (starting_frame_offset): New hook.
8065 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8066 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8067 * doc/tm.texi.in: Regenerate.
8068 * hooks.h (hook_hwi_void_0): Declare.
8069 * hooks.c (hook_hwi_void_0): New function.
8070 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8071 STARTING_FRAME_OFFSET.
8072 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8073 * reload1.c (reload): Likewise.
8074 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8075 instead of STARTING_FRAME_OFFSET.
8076 * function.c (try_fit_stack_local): Likewise.
8077 (assign_stack_local_1): Likewise
8078 (instantiate_virtual_regs): Likewise.
8079 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8080 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8081 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8082 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8083 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8084 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8085 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8086 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8087 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8088 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8089 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8090 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8091 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8092 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8093 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8094 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8095 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8096 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8097 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8098 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8099 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8100 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8101 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8102 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8103 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8104 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8105 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8106 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8107 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8108 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8109 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8110 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8111 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8112 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8113 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8114 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8115 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8116 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8117 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8118 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8119 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8120 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8121 return a HOST_WIDE_INT.
8122 (avr_builtin_setjmp_frame_value): Use it instead of
8123 STARTING_FRAME_OFFSET.
8124 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8125 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8126 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8128 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8129 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8130 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8131 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8132 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8133 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8134 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8135 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8136 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8137 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8138 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8139 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8141 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8142 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8143 * config/mips/mips.c (mips_compute_frame_info): Refer to
8144 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8145 (mips_starting_frame_offset): New function.
8146 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8147 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8148 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8149 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8150 and return a HOST_WIDE_INT.
8151 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8152 (mmix_initial_elimination_offset): Refer to
8153 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8154 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8155 * config/pa/pa.c (pa_starting_frame_offset): New function.
8156 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8157 (pa_expand_prologue): Likewise.
8158 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8159 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8160 !FRAME_GROWS_DOWNWARD handling to...
8161 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8162 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8163 !FRAME_GROWS_DOWNWARD handling to...
8164 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8165 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8166 !FRAME_GROWS_DOWNWARD handling to...
8167 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8168 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8170 (rs6000_starting_frame_offset): New function.
8171 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8172 !FRAME_GROWS_DOWNWARD handling to...
8173 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8174 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8175 !FRAME_GROWS_DOWNWARD handling to...
8176 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8177 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8178 !FRAME_GROWS_DOWNWARD handling to...
8179 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8180 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8181 (rs6000_starting_frame_offset): New function.
8182 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8183 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8184 * config/vax/vax.c (vax_starting_frame_offset): New function.
8185 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8186 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8187 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8188 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8189 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8190 * system.h (STARTING_FRAME_OFFSET): Poison.
8192 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8194 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8195 SCALAR_TYPE_MODE instead of TYPE_MODE.
8197 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8198 Alan Hayward <alan.hayward@arm.com>
8199 David Sherwood <david.sherwood@arm.com>
8201 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8203 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8204 Alan Hayward <alan.hayward@arm.com>
8205 David Sherwood <david.sherwood@arm.com>
8207 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8209 2017-10-23 Richard Biener <rguenther@suse.de>
8211 PR tree-optimization/82129
8212 * tree-ssa-pre.c (bitmap_set_and): Remove.
8213 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8214 canonicalizing expressions in the set to those with lowest
8215 ID rather than taking that from the first edge.
8217 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8219 * combine.c (rtx_equal_for_field_assignment_p): Use
8220 byte_lowpart_offset.
8222 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8223 Alan Hayward <alan.hayward@arm.com>
8224 David Sherwood <david.sherwood@arm.com>
8226 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8227 to a SUBREG_PROMOTED_VAR.
8229 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8230 Alan Hayward <alan.hayward@arm.com>
8231 David Sherwood <david.sherwood@arm.com>
8233 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8234 (expand_debug_source_expr): Likewise.
8235 * combine.c (combine_simplify_rtx): Likewise.
8236 * cse.c (fold_rtx): Likewise.
8237 * optabs.c (expand_float): Likewise.
8238 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8239 (simplify_binary_operation_1): Likewise.
8241 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8242 Alan Hayward <alan.hayward@arm.com>
8243 David Sherwood <david.sherwood@arm.com>
8245 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8246 (record_promoted_value): Likewise.
8247 * expr.c (expand_expr_real_2): Likewise.
8248 * ree.c (update_reg_equal_equiv_notes): Likewise.
8249 (combine_set_extension): Likewise.
8250 * rtlanal.c (low_bitmask_len): Likewise.
8251 * simplify-rtx.c (neg_const_int): Likewise.
8252 (simplify_binary_operation_1): Likewise.
8254 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8255 Alan Hayward <alan.hayward@arm.com>
8256 David Sherwood <david.sherwood@arm.com>
8258 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8259 * regcprop.c (maybe_mode_change): Likewise.
8260 * reload1.c (alter_reg): Likewise.
8262 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8264 * inchash.h (inchash::hash::add_wide_int): New function.
8265 * lto-streamer-out.c (hash_tree): Use it.
8267 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8269 * inchash.h (inchash::hash::add_wide_int): Rename to...
8270 (inchash::hash::add_hwi): ...this.
8271 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8272 (polymorphic_call_target_hasher::hash): Likewise.
8273 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8274 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8275 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8276 * lto-streamer-out.c (hash_tree): Likewise.
8277 * optc-save-gen.awk: Likewise.
8278 * tree.c (add_expr): Likewise.
8280 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8283 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8284 for ordered inequality comparisons even with TARGET_IEEE_FP.
8286 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8289 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8290 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8291 Expand with cmp<dwi>_doubleword.
8293 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8295 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8296 List CET intrinsics.
8297 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8298 specific to -fcf-protection option.
8300 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8303 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8304 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8305 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8306 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8307 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8308 (extra_objs): Add cet.o for Linux/x86 targets.
8309 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8310 * config/i386/cet.c: New file.
8311 * config/i386/cetintrin.h: Likewise.
8312 * config/i386/t-cet: Likewise.
8313 * config/i386/cpuid.h (bit_SHSTK): New.
8314 (bit_IBT): Likewise.
8315 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8316 pass IBT and SHSTK bits.
8317 * config/i386/i386-builtin-types.def
8318 (VOID_FTYPE_UNSIGNED_PVOID): New.
8319 (VOID_FTYPE_UINT64_PVOID): Likewise.
8320 * config/i386/i386-builtin.def: Add CET intrinsics.
8321 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8322 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8323 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8324 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8326 * config/i386/i386.c (rest_of_insert_endbranch): New.
8327 (pass_data_insert_endbranch): Likewise.
8328 (pass_insert_endbranch): Likewise.
8329 (make_pass_insert_endbranch): Likewise.
8330 (ix86_notrack_prefixed_insn_p): Likewise.
8331 (ix86_target_string): Add -mibt, -mshstk flags.
8332 (ix86_option_override_internal): Add flag_cf_protection
8334 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8335 (ix86_print_operand): Add 'notrack' prefix output.
8336 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8337 (ix86_expand_builtin): Expand CET intrinsics.
8338 (x86_output_mi_thunk): Add 'endbranch' instruction.
8339 * config/i386/i386.h (TARGET_IBT): New.
8340 (TARGET_IBT_P): Likewise.
8341 (TARGET_SHSTK): Likewise.
8342 (TARGET_SHSTK_P): Likewise.
8343 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8344 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8345 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8346 (builtin_setjmp_setup): New pattern.
8347 (builtin_longjmp): Likewise.
8348 (rdssp<mode>): Likewise.
8349 (incssp<mode>): Likewise.
8350 (saveprevssp): Likewise.
8351 (rstorssp): Likewise.
8352 (wrss<mode>): Likewise.
8353 (wruss<mode>): Likewise.
8354 (setssbsy): Likewise.
8355 (clrssbsy): Likewise.
8356 (nop_endbr): Likewise.
8357 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8359 * config/i386/immintrin.h: Include <cetintrin.h>.
8360 * config/i386/linux-common.h
8361 (file_end_indicate_exec_stack_and_cet): New prototype.
8362 (TARGET_ASM_FILE_END): New.
8364 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8366 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8367 latencies instead of having separate table; make difference between
8368 integer and float costs.
8369 * i386.h (processor_costs): Remove scalar_stmt_cost,
8370 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8371 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8373 * x86-tune-costs.h: Remove entries which has been removed in
8374 procesor_costs from all tables; make cond_taken_branch_cost
8375 and cond_not_taken_branch_cost COST_N_INSNS based.
8377 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8379 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8381 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8383 * config/i386/i386.md (isa): Remove fma_avx512f.
8384 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8385 <avx512>_fmadd_<mode>_mask3<round_name>,
8386 <avx512>_fmsub_<mode>_mask<round_name>,
8387 <avx512>_fmsub_<mode>_mask3<round_name>,
8388 <avx512>_fnmadd_<mode>_mask<round_name>,
8389 <avx512>_fnmadd_<mode>_mask3<round_name>,
8390 <avx512>_fnmsub_<mode>_mask<round_name>,
8391 <avx512>_fnmsub_<mode>_mask3<round_name>,
8392 <avx512>_fmaddsub_<mode>_mask<round_name>,
8393 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8394 <avx512>_fmsubadd_<mode>_mask<round_name>,
8395 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8396 (*vec_widen_umult_even_v16si<mask_name>,
8397 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8398 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8400 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8402 * extend.texi: Add 'nocf_check' documentation.
8403 * gimple.texi: Add second parameter to
8404 gimple_build_call_from_tree.
8405 * invoke.texi: Add -fcf-protection documentation.
8406 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8408 2017-10-20 Richard Biener <rguenther@suse.de>
8410 PR tree-optimization/82473
8411 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8412 the largest input type.
8414 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8416 * c-attribs.c (handle_nocf_check_attribute): New function.
8417 (c_common_attribute_table): Add 'nocf_check' handling.
8418 * gimple-parser.c: Add second argument NULL to
8419 gimple_build_call_from_tree.
8420 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8421 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8423 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8424 * common.opt: Add fcf-protection flag.
8425 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8426 * flag-types.h: Add enum cf_protection_level.
8427 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8428 Add 'nocf_check' attribute propagation to gimple call.
8429 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8430 (gimple_build_call_from_tree): Update prototype.
8431 (gimple_call_nocf_check_p): New function.
8432 (gimple_call_set_nocf_check): Likewise.
8433 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8434 * ipa-icf.c: Add nocf_check attribute in statement hash.
8435 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8436 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8437 * toplev.c (process_options): Add flag_cf_protection handling.
8439 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8441 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8443 2017-10-20 Richard Biener <rguenther@suse.de>
8445 PR tree-optimization/82603
8446 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8447 remove false predicated stores.
8449 2017-10-20 Richard Biener <rguenther@suse.de>
8451 * graphite-isl-ast-to-gimple.c
8452 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8453 Remove return value and simplify, dump copied stmt after lhs
8455 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8456 Reduce dump verbosity.
8457 (gsi_insert_earliest): Likewise.
8458 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8459 * graphite.c (print_global_statistics): Adjust dumping.
8460 (print_graphite_scop_statistics): Likewise.
8461 (print_graphite_statistics): Do not dump loops here.
8462 (graphite_transform_loops): But here.
8464 2017-10-20 Nicolas Roche <roche@adacore.com>
8466 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8467 * configure: Regenerate.
8469 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8472 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8473 functions when optimizing replace GIMPLE_RETURN stmts with
8474 calls to __builtin_unreachable ().
8477 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8478 for -fsanitize=thread link of executables.
8479 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8480 link of executables.
8483 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8485 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8486 define_insns for logical vector shifts to use VI248_AVX512BW
8487 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8488 condition, useless isa and prefix attributes. Change the first
8489 2 of these define_insns to ...
8490 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8491 define_insn for avx512vl.
8492 (<shift_insn><mode>3): ... and this, new define_insn without
8493 masking for non-avx512vl.
8496 * config/i386/sse.md (*andnot<mode>3,
8497 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8498 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8499 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8500 not applied use empty suffix even for TARGET_AVX512VL.
8501 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8502 is applied, supply evex,evex or evex,evex,evex instead of just
8505 2017-10-20 Julia Koval <julia.koval@intel.com>
8507 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8508 (OPTION_MASK_ISA_GFNI_UNSET): New.
8509 (ix86_handle_option): Handle OPT_mgfni.
8510 * config/i386/cpuid.h (bit_GFNI): New.
8511 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8512 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8513 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8514 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8515 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8516 * config/i386/i386.opt: Add mgfni.
8518 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8520 * config/msp430/msp430.c (msp430_option_override): Disable
8521 -fdelete-null-pointer-checks.
8522 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8524 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8526 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8527 of x87 and SSE instructions.
8529 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8531 * asan.c (create_cond_insert_point): Do not update edge count.
8532 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8533 (afdo_propagate_circuit): Likewise.
8534 (afdo_calculate_branch_prob): Likewise.
8535 (afdo_annotate_cfg): Likewise.
8536 * basic-block.h (struct edge_def): Remove count.
8537 (edge_def::count): New accessor.
8538 * bb-reorder.c (rotate_loop): Update.
8539 (find_traces_1_round): Update.
8540 (connect_traces): Update.
8541 (sanitize_hot_paths): Update.
8542 * cfg.c (unchecked_make_edge): Update.
8543 (make_single_succ_edge): Update.
8544 (check_bb_profile): Update.
8545 (dump_edge_info): Update.
8546 (update_bb_profile_for_threading): Update.
8547 (scale_bbs_frequencies_int): Update.
8548 (scale_bbs_frequencies_gcov_type): Update.
8549 (scale_bbs_frequencies_profile_count): Update.
8550 (scale_bbs_frequencies): Update.
8551 * cfganal.c (connect_infinite_loops_to_exit): Update.
8552 * cfgbuild.c (compute_outgoing_frequencies): Update.
8553 (find_many_sub_basic_blocks): Update.
8554 * cfgcleanup.c (try_forward_edges): Update.
8555 (try_crossjump_to_edge): Update
8556 * cfgexpand.c (expand_gimple_cond): Update
8557 (expand_gimple_tailcall): Update
8558 (construct_exit_block): Update
8559 * cfghooks.c (verify_flow_info): Update
8560 (redirect_edge_succ_nodup): Update
8561 (split_edge): Update
8562 (make_forwarder_block): Update
8563 (duplicate_block): Update
8564 (account_profile_record): Update
8565 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8566 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8567 * cfgloopmanip.c (scale_loop_profile): Update.
8569 (lv_adjust_loop_entry_edge): Update.
8570 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8571 (force_nonfallthru_and_redirect): Update.
8572 (purge_dead_edges): Update.
8573 (rtl_flow_call_edges_add): Update.
8574 * cgraphunit.c (init_lowered_empty_function): Update.
8575 (cgraph_node::expand_thunk): Update.
8576 * gimple-pretty-print.c (dump_probability): Update.
8577 (dump_edge_probability): Update.
8578 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8579 * haifa-sched.c (sched_create_recovery_edges): Update.
8580 * hsa-gen.c (convert_switch_statements): Update.
8581 * ifcvt.c (dead_or_predicable): Update.
8582 * ipa-inline-transform.c (inline_transform): Update.
8583 * ipa-split.c (split_function): Update.
8584 * ipa-utils.c (ipa_merge_profiles): Update.
8585 * loop-doloop.c (add_test): Update.
8586 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8587 * lto-streamer-in.c (input_cfg): Update.
8588 (input_function): Update.
8589 * lto-streamer-out.c (output_cfg): Update.
8590 * modulo-sched.c (sms_schedule): Update.
8591 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8592 * predict.c (maybe_hot_edge_p): Update.
8593 (unlikely_executed_edge_p): Update.
8594 (probably_never_executed_edge_p): Update.
8595 (dump_prediction): Update.
8596 (drop_profile): Update.
8597 (propagate_unlikely_bbs_forward): Update.
8598 (determine_unlikely_bbs): Update.
8599 (force_edge_cold): Update.
8600 * profile.c (compute_branch_probabilities): Update.
8601 * reg-stack.c (better_edge): Update.
8602 * shrink-wrap.c (handle_simple_exit): Update.
8603 * tracer.c (better_p): Update.
8604 * trans-mem.c (expand_transaction): Update.
8605 (split_bb_make_tm_edge): Update.
8606 * tree-call-cdce.c: Update.
8607 * tree-cfg.c (gimple_find_sub_bbs): Update.
8608 (gimple_split_edge): Update.
8609 (gimple_duplicate_sese_region): Update.
8610 (gimple_duplicate_sese_tail): Update.
8611 (gimple_flow_call_edges_add): Update.
8612 (insert_cond_bb): Update.
8613 (execute_fixup_cfg): Update.
8614 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8615 * tree-complex.c (expand_complex_div_wide): Update.
8616 * tree-eh.c (lower_resx): Update.
8617 (unsplit_eh): Update.
8618 (cleanup_empty_eh_move_lp): Update.
8619 * tree-inline.c (copy_edges_for_bb): Update.
8620 (freqs_to_counts): Update.
8621 (copy_cfg_body): Update.
8622 * tree-ssa-dce.c (remove_dead_stmt): Update.
8623 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8624 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8625 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8626 (unloop_loops): Update.
8627 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8628 * tree-ssa-loop-split.c (connect_loops): Update.
8629 (split_loop): Update.
8630 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8631 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8632 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8633 * tree-ssa-reassoc.c (branch_fixup): Update.
8634 * tree-ssa-tail-merge.c (replace_block_by): Update.
8635 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8636 (compute_path_counts): Update.
8637 (update_profile): Update.
8638 (recompute_probabilities): Update.
8639 (update_joiner_offpath_counts): Update.
8640 (estimated_freqs_path): Update.
8641 (freqs_to_counts_path): Update.
8642 (clear_counts_path): Update.
8643 (ssa_fix_duplicate_block_edges): Update.
8644 (duplicate_thread_path): Update.
8645 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8646 (case_bit_test_cmp): Update.
8647 (collect_switch_conv_info): Update.
8648 (gen_inbound_check): Update.
8649 (do_jump_if_equal): Update.
8650 (emit_cmp_and_jump_insns): Update.
8651 * tree-tailcall.c (decrease_profile): Update.
8652 (eliminate_tail_call): Update.
8653 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8654 (vect_do_peeling): Update.
8655 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8656 * ubsan.c (ubsan_expand_null_ifn): Update.
8657 (ubsan_expand_ptr_ifn): Update.
8658 * value-prof.c (gimple_divmod_fixed_value): Update.
8659 (gimple_mod_pow2): Update.
8660 (gimple_mod_subtract): Update.
8661 (gimple_ic): Update.
8662 (gimple_stringop_fixed_value): Update.
8664 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8667 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8669 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8671 PR rtl-optimization/82395
8672 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8673 based on non_spilled_static_chain_regno_p.
8675 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8677 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8678 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8679 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8681 2017-10-19 Martin Sebor <msebor@redhat.com>
8683 PR tree-optimization/82596
8684 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8686 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8688 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8689 (get_mem_refs_of_builtin_call): Likewise.
8690 * builtins.c (expand_builtin_apply): Adjust call to
8691 allocate_dynamic_stack_space.
8692 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8693 the third argument to allocate_dynamic_stack_space, otherwise -1.
8694 (expand_builtin): Deal with all alloca variants.
8695 (is_inexpensive_builtin): Likewise.
8696 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8697 * calls.c (special_function_p): Deal with all alloca variants.
8698 (initialize_argument_information): Adjust call to
8699 allocate_dynamic_stack_space.
8700 (expand_call): Likewise.
8701 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8702 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8703 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8704 use it for the stack usage computation.
8705 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8706 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8707 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8708 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8709 (in_loop_p): Remove first argument and useless check.
8710 (pass_walloca::execute): Remove useless test and adjust call to above.
8711 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8712 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8713 (gimplify_call_expr): Deal with all alloca variants.
8714 * hsa-gen.c (gen_hsa_alloca): Likewise.
8715 (gen_hsa_insns_for_call): Likewise.
8716 * ipa-pure-const.c (special_builtin_state): Likewise.
8717 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8718 * tree-object-size.c (alloc_object_size): Likewise.
8719 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8720 (call_may_clobber_ref_p_1): Likewise.
8721 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8722 (ccp_fold_stmt): Likewise.
8723 (optimize_stack_restore): Likewise.
8724 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8725 (mark_all_reaching_defs_necessary_1): Likewise.
8726 (propagate_necessity): Likewise.
8727 (eliminate_unnecessary_stmts): Likewise.
8728 * tree.c (build_common_builtin_nodes): Build
8729 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8730 (build_alloca_call_expr): New function.
8731 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8732 (CASE_BUILT_IN_ALLOCA): Likewise.
8733 (build_alloca_call_expr): Declare.
8734 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8736 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8739 * dwarf2out.c (new_die_raw): New static inline function.
8740 (new_die): Use it to create the DIE.
8741 (add_AT_external_die_ref): Likewise.
8742 (clone_die): Likewise.
8743 (clone_as_declaration): Likewise.
8744 (dwarf2out_vms_debug_main_pointer): Likewise.
8745 (base_type_die): Likewise. Remove early return for corner cases.
8746 Do not call add_pubtype on the DIE here.
8747 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8748 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8749 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8750 native order exists for base types, attach the DIE manually and call
8751 add_pubtype on it. Do not equate a reverse order DIE to the type.
8753 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8755 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8756 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8758 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8760 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8762 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8763 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8764 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8765 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8766 * opts.c (sanitizer_opts): Add builtin.
8767 * ubsan.c (instrument_builtin): New function.
8768 (pass_ubsan::execute): Call it.
8769 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8770 * doc/invoke.texi: Document -fsanitize=builtin.
8772 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8773 builtins, store max (log2 (align), 0) into uchar field instead of
8774 align into uptr field.
8775 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8776 store uchar 0 field instead of uptr 0 field.
8777 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8778 instead of passing one address of struct with 2 locations pass
8779 two addresses of structs with 1 location each.
8780 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8781 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8782 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8783 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8784 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8785 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8786 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8787 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8789 2017-10-19 Martin Liska <mliska@suse.cz>
8792 * file-find.c (remove_prefix): Remove.
8793 * file-find.h (remove_prefix): Likewise.
8794 * gcc-ar.c: Remove smartness of lookup.
8796 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8798 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8799 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8800 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8802 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8805 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8806 (setcc + and to xor + setcc): New peephole2.
8808 2017-10-19 Tom de Vries <tom@codesourcery.com>
8810 * doc/sourcebuild.texi (Test Directives, Variants of
8811 dg-require-support): Add dg-require-stack-size.
8813 2017-10-19 Martin Liska <mliska@suse.cz>
8816 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8817 that have a large alignment.
8818 (gimplify_target_expr): Likewise.
8820 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8822 PR rtl-optimization/82602
8823 * ira.c (rtx_moveable_p): Return false for volatile asm.
8825 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8828 * config/i386/i386-modes.def (CCGZ): New CC mode.
8829 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8830 * config/i386/predicates.md (ix86_comparison_operator):
8832 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8833 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8834 with double-word subtraction.
8835 (put_condition_code): Handle CCGZmode.
8837 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8839 * wide-int.cc (debug (const wide_int &)): New.
8840 (debug (const wide_int *)): New.
8841 (debug (const widest_int &)): New.
8842 (debug (const widest_int *)): New.
8844 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8847 * lra-constraints.c (curr_insn_transform): Use non-input operand
8848 instead of output one for matched reload.
8850 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8852 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8853 (tree-ssa-loop-ivopts.h): New header file.
8854 (struct builtin_info): New fields.
8855 (classify_builtin_1): Compute and record base and offset parts for
8856 memset builtin partition by calling strip_offset.
8857 (offset_cmp, fuse_memset_builtins): New functions.
8858 (finalize_partitions): Fuse adjacent memset partitions by calling
8860 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8861 Expose the interface.
8862 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8864 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8866 PR tree-optimization/82574
8867 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8868 that data reference must be executed exactly once per iteration
8869 against the outermost loop in nest.
8870 (classify_partition): Update call to above function.
8872 2017-10-18 Richard Biener <rguenther@suse.de>
8874 PR tree-optimization/82591
8875 * graphite.c (graphite_transform_loops): Move code gen message
8877 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8878 Here. Handle scop_to_isl_ast failing.
8879 (scop_to_isl_ast): Limit the number of ISL operations.
8881 2017-10-18 Richard Biener <rguenther@suse.de>
8883 * graphite-isl-ast-to-gimple.c
8884 (translate_isl_ast_to_gimple::set_rename): Simplify.
8885 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8886 (graphite_copy_stmts_from_block): ... here.
8887 (copy_bb_and_scalar_dependences): Simplify.
8888 (add_parameters_to_ivs_params): Canonicalize.
8889 (generate_entry_out_of_ssa_copies): Simplify.
8890 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8891 by passing in ISL dimension.
8892 (parameter_index_in_region_1): Rename to ...
8893 (parameter_index_in_region): ... this.
8894 (extract_affine): Adjust assert, pass down parameter index.
8895 (add_param_constraints): Use range-info when available.
8896 (build_scop_context): Adjust.
8897 * sese.c (new_sese_info): Adjust.
8898 (free_sese_info): Likewise.
8899 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8900 Remove unused typedefs.
8901 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8903 2017-10-18 Martin Liska <mliska@suse.cz>
8905 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8907 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8909 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8910 (s390_sched_init): Do not reset s390_sched_state if we entered the
8911 current basic block via a fallthru edge and all others are unlikely.
8913 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8915 * config/s390/s390.c (NUM_SIDES): New variable.
8916 (LONGRUNNING_THRESHOLD): New variable.
8917 (LATENCY_FACTOR): New variable.
8918 (s390_sched_score): Decrease score for long-running instructions on
8920 (s390_sched_variable_issue): Perform bookkeeping for long-running
8923 2017-10-18 Richard Biener <rguenther@suse.de>
8925 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8926 Simplify with removal of the parameter rename map.
8927 (set_rename): Likewise.
8928 (should_copy_to_new_region): Likewise.
8929 (graphite_copy_stmts_from_block): Likewise.
8930 (copy_bb_and_scalar_dependences): Remove initialization of
8931 unused copied_bb_map.
8933 (copy_internal_parameters): Likewise.
8934 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8935 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8936 Use INTEGRAL_TYPE_P.
8937 (parameter_index_in_region_1): Rename to ...
8938 (assign_parameter_index_in_region): ... this. Assert we have
8939 a parameter we handle.
8940 (scan_tree_for_params): Adjust.
8941 * sese.h (parameter_rename_map_t): Remove.
8942 (struct sese_info_t): Remove unused parameter_rename_map and
8943 copied_bb_map members.
8944 * sese.c (new_sese_info): Adjust.
8945 (free_sese_info): Likewise.
8947 2017-10-18 Martin Liska <mliska@suse.cz>
8950 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8951 on an abnormal edge.
8953 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8955 * doc/invoke.texi (ffunction-sections and fdata-sections):
8958 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8960 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8961 the use statement can throw internally.
8963 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8965 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8966 any RTX present on the RHS of a SET.
8967 * compare-elim.c (try_eliminate_compare): Restore comment.
8969 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8971 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8972 may be also called on tcc_type nodes.
8973 * langhooks.c (lhd_tree_size): Likewise.
8975 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8977 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8978 format_warning_at_substring.
8979 (maybe_warn): Convert source_range * param to a location_t. Pass
8980 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8981 (format_directive): Remove code to extract source_ranges and
8982 source_range * in favor of just a location_t.
8983 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8985 * substring-locations.c (format_warning_va): Convert
8986 source_range * param to a location_t.
8987 (format_warning_at_substring): Likewise.
8988 * substring-locations.h (format_warning_va): Likewise.
8989 (format_warning_at_substring): Likewise.
8991 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8993 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8995 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8996 and scatter/gather ops.
8998 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8999 vec_gather_load and vec_scatter_store.
9000 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9001 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9002 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9003 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9004 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9005 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9007 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9009 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9010 Detect FTST instruction and handle its register pops. Only pop
9011 second operand if can_pop_second_op is true.
9012 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9013 set can_pop_second_op to false in the compare_for_stack_reg call.
9015 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9016 output_fp_compare for stack register operands.
9017 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9018 instructions here. Do not emit stack register pops here. Assert
9019 that FCOMPP pops next to top stack register. Rewrite function.
9021 2017-10-17 Nathan Sidwell <nathan@acm.org>
9024 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9025 use DECL_ASSEMBLER_NAME_RAW.
9028 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9031 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9032 Wilco Dijkstra <wilco.dijkstra@arm.com>
9034 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9035 converstion to Pmode from the buf_addr.
9037 2017-10-17 Richard Biener <rguenther@suse.de>
9039 * graphite-dependences.c (scop_get_reads_and_writes): Change
9040 output parameters to references.
9042 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9044 PR 71026/tree-optimization
9045 * fold-const.c (distribute_real_division): Removed.
9046 (fold_binary_loc): Remove calls to distribute_real_divison.
9048 2017-10-17 Richard Biener <rguenther@suse.de>
9050 * graphite-scop-detection.c
9051 (scop_detection::stmt_has_simple_data_refs_p): Always use
9052 the full nest as region.
9053 (try_generate_gimple_bb): Likewise.
9054 * sese.c (scalar_evolution_in_region): Simplify now that
9055 SCEV can handle instantiation in regions.
9056 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9057 in the non-loop part of a function if requested.
9059 2017-10-17 Richard Biener <rguenther@suse.de>
9061 PR tree-optimization/82563
9062 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9064 (graphite_regenerate_ast_isl): Call it.
9065 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9067 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9069 PR tree-optimization/82549
9070 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9071 Formatting fixes. Instead of calling make_bit_field_ref with negative
9074 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9076 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9077 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9078 _mm_maskz_reduce_ss): New.
9079 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9080 __builtin_ia32_reducess_mask): Ditto..
9081 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9082 * config/i386/sse.md (reduces<mode>): Renamed to ...
9083 (reduces<mode><mask_scalar_name>): ... this.
9084 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9086 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9087 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9089 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9091 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9092 * selftest-run-tests.c (selftest::run_tests): Call
9093 selftest::unique_ptr_tests_cc_tests.
9094 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9095 * unique-ptr-tests.cc: New file.
9097 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9100 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9102 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9103 (make_hard_regno_born, make_hard_regno_dead): Update
9104 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9106 2017-10-16 Jeff Law <law@redhat.com>
9108 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9110 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9112 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9114 2017-10-16 Olivier Hainque <hainque@adacore.com>
9116 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9117 with_cpu if we were configured for an e500v2 target cpu name.
9119 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9121 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9122 * doc/invoke.texi: Document +nodsp as a valid extension for
9125 2017-10-16 Martin Liska <mliska@suse.cz>
9127 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9128 (test_set_range): Likewise.
9129 (test_range_functions): Rename to ...
9130 (test_bit_in_range): ... this.
9131 (sbitmap_c_tests): Add new test.
9133 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9135 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9137 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9138 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9140 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9142 * config/aarch64/aarch64-builtins.c
9143 (aarch64_types_quadopu_lane_qualifiers): New.
9144 (TYPES_QUADOPU_LANE): New.
9145 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9146 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9147 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9148 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9149 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9150 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9151 (Vdottype, DOTPROD): New.
9152 (sur): Add SDOT and UDOT.
9154 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9156 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9157 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9158 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9160 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9161 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9162 Enable TARGET_DOTPROD.
9163 (cortex-a75.cortex-a55): Likewise.
9164 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9166 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9168 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9169 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9171 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9173 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9174 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9175 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9176 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9177 * config/arm/types.md (neon_dot, neon_dot_q): New.
9178 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9180 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9182 * config/arm/arm.h (TARGET_DOTPROD): New.
9183 * config/arm/arm.c (arm_arch_dotprod): New.
9184 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9185 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9186 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9187 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9188 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9189 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9190 * doc/invoke.texi (armv8.2-a): Document dotprod
9192 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9194 * i386.c (ix86_vec_cost): New function.
9195 (ix86_rtx_costs): Handle vector operations better.
9196 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9197 * x86-tune-costs.h: Add new costs to all tables.
9199 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9201 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9203 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9204 divsd, sqrtss and sqrtsd
9205 * x86-tune-costs.h: Add new entries to all costs.
9206 (znver1_cost): Fix to match real instruction latencies.
9208 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9209 Michael Collison <michael.collison@arm.com>
9211 * compare-elim.c: Include emit-rtl.h.
9212 (can_merge_compare_into_arith): New function.
9213 (try_validate_parallel): Likewise.
9214 (try_merge_compare): Likewise.
9215 (try_eliminate_compare): Call the above when no previous clobber
9217 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9220 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9224 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9225 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9229 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9230 to be any operand_equal_p operands. For & (B - 1) require
9231 B to be power of 2. Recognize
9232 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9234 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9237 * optabs.c (expand_memory_blockage): Fix call of
9238 targetm.have_memory_blockage.
9240 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9243 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9244 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9245 objects to extra_objs instead of overwriting it.
9247 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9249 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9250 Use any_fp_register_operand as operand[3] predicate. Simplify
9251 equality test for operands[2] and operands[4] memory location.
9252 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9253 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9254 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9255 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9256 any_fp_register_operand as operand[1] predicate. Simplify
9257 equality test for operands[0] and operands[3] memory location.
9258 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9259 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9260 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9262 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9264 * target-insns.def: Add memory_blockage.
9265 * optabs.c (expand_memory_blockage): New function.
9266 (expand_asm_memory_barrier): Rename ...
9267 (expand_asm_memory_blockage): ... to this.
9268 (expand_mem_thread_fence): Call expand_memory_blockage
9269 instead of expand_asm_memory_barrier.
9270 (expand_mem_singnal_fence): Ditto.
9271 (expand_atomic_load): Ditto.
9272 (expand_atomic_store): Ditto.
9273 * doc/md.texi (Standard Pattern Names For Generation):
9274 Document memory_blockage instruction pattern.
9276 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9278 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9279 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9280 * config/rl78/rl78.md: New define_expand "adddi3".
9282 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9284 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9287 2017-10-13 Jeff Law <law@redhat.com>
9289 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9291 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9294 * internal-fn.c (expand_mul_overflow): If both operands have
9295 the same highpart of -1 or 0 and the topmost bit of lowpart
9296 is different, overflow is if res <= 0 rather than res < 0.
9298 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9300 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9301 TARGET_P9_VECTOR code for unaligned_load case.
9303 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9305 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9307 2017-10-13 Nathan Sidwell <nathan@acm.org>
9309 * tree-core.h (tree_contains_struct): Make bool.
9310 * tree.c (tree_contains_struct): Likewise.
9311 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9312 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9313 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9314 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9315 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9317 2017-10-13 Richard Biener <rguenther@suse.de>
9319 * graphite-isl-ast-to-gimple.c
9320 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9321 parameters and dominance check.
9322 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9323 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9324 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9325 Do not update SSA form here or do intermediate IL verification.
9326 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9327 (graphite_initialize): Remove check on the number of loops in
9328 the function and inline into graphite_transform_loops.
9329 (graphite_finalize): Inline into graphite_transform_loops.
9330 (graphite_transform_loops): Perform SSA update and IL verification
9332 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9334 2017-10-13 Richard Biener <rguenther@suse.de>
9336 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9337 graphite_expression_type_precision): Avoid global constructor
9339 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9340 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9341 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9342 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9343 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9344 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9346 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9349 * config/i386/i386.h (ix86_red_zone_size): New.
9350 * config/i386/i386.md (push peephole2s): Replace
9351 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9353 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9354 Alan Hayward <alan.hayward@arm.com>
9355 David Sherwood <david.sherwood@arm.com>
9357 * combine.c (can_change_dest_mode): Reject changes in
9358 REGMODE_NATURAL_SIZE.
9360 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9361 Alan Hayward <alan.hayward@arm.com>
9362 David Sherwood <david.sherwood@arm.com>
9364 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9365 (expand_debug_source_expr): Likewise.
9366 * combine.c (combine_simplify_rtx): Likewise.
9367 * cse.c (fold_rtx): Likewise.
9368 * fwprop.c (canonicalize_address): Likewise.
9369 * targhooks.c (default_shift_truncation_mask): Likewise.
9371 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9372 Alan Hayward <alan.hayward@arm.com>
9373 David Sherwood <david.sherwood@arm.com>
9375 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9376 (widened_mode): Likewise.
9377 (expand_unop): Likewise.
9378 * ree.c (transform_ifelse): Likewise.
9379 (merge_def_and_ext): Likewise.
9380 (combine_reaching_defs): Likewise.
9381 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9383 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9384 Alan Hayward <alan.hayward@arm.com>
9385 David Sherwood <david.sherwood@arm.com>
9387 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9388 * combine.c (gen_lowpart_for_combine): Likewise.
9389 * dwarf2out.c (rtl_for_decl_location): Likewise.
9390 * final.c (alter_subreg): Likewise.
9391 * rtlhooks.c (gen_lowpart_general): Likewise.
9392 (gen_lowpart_if_possible): Likewise.
9394 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9395 Alan Hayward <alan.hayward@arm.com>
9396 David Sherwood <david.sherwood@arm.com>
9398 * calls.c (expand_call): Use subreg_lowpart_offset.
9399 * cse.c (cse_insn): Likewise.
9400 * regcprop.c (copy_value): Likewise.
9401 (copyprop_hardreg_forward_1): Likewise.
9403 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9406 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9407 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9408 =Q constraints to +Q and into insn condition add check
9409 that operands[0] and operands[1] are equal.
9410 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9411 =Q constraints to +Q and into insn condition add check
9412 that operands[0] is equal to either operands[1] or operands[2].
9415 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9416 instead of handling MINUS_EXPR twice (once for each argument),
9417 canonicalize operand order and handle just once, use rtype where
9418 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9421 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9422 any values of __C while still being pattern recognizable as a simple
9425 2017-10-13 Richard Biener <rguenther@suse.de>
9427 PR tree-optimization/82451
9429 2017-10-02 Richard Biener <rguenther@suse.de>
9431 PR tree-optimization/82355
9432 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9433 a mapping for the enclosing loop but avoid generating one for
9435 (copy_bb_and_scalar_dependences): Remove premature codegen
9436 error on PHIs in blocks duplicated into multiple places.
9437 * graphite-scop-detection.c
9438 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9439 in the region use it as loop and nest to analyze the DR in.
9440 (try_generate_gimple_bb): Likewise.
9441 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9442 (add_loop_constraints): For blocks in a loop not in the region
9443 create a dimension with a single iteration.
9444 * sese.h (gbb_loop_at_index): Remove assert.
9446 * cfgloop.c (loop_preheader_edge): For the loop tree root
9447 return the single successor of the entry block.
9448 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9449 Reset the SCEV hashtable and niters.
9450 * graphite-scop-detection.c
9451 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9452 assert that we only have POLYNOMIAL_CHREC that vary in loops
9453 contained in the region.
9454 (scop_detection::graphite_can_represent_expr): Adjust.
9455 (scop_detection::stmt_has_simple_data_refs_p): For loops
9456 not in the region set loop to NULL. The nest is now the
9457 entry edge to the region.
9458 (try_generate_gimple_bb): Likewise.
9459 * sese.c (scalar_evolution_in_region): Adjust for
9460 instantiate_scev change.
9461 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9462 Make nest parameter the edge into the region.
9463 (create_data_ref): Likewise.
9464 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9465 entry edge into a region and adjust instantiate_scev calls.
9466 (create_data_ref): Likewise.
9467 (graphite_find_data_references_in_stmt): Likewise.
9468 (find_data_references_in_stmt): Pass the loop preheader edge
9469 from the nest argument.
9470 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9471 parameter the edge into the region.
9472 (instantiate_parameters): Use the loop preheader edge as entry.
9473 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9475 (get_instantiated_value_entry): Make instantiate_below parameter
9476 the edge into the region.
9477 (instantiate_scev_name): Likewise. Adjust dominance checks,
9478 when we cannot use loop-based instantiation instantiate by
9479 walking use-def chains.
9480 (instantiate_scev_poly): Adjust.
9481 (instantiate_scev_binary): Likewise.
9482 (instantiate_scev_convert): Likewise.
9483 (instantiate_scev_not): Likewise.
9484 (instantiate_array_ref): Remove.
9485 (instantiate_scev_3): Likewise.
9486 (instantiate_scev_2): Likewise.
9487 (instantiate_scev_1): Likewise.
9488 (instantiate_scev_r): Do not blindly handle N-operand trees.
9489 Do not instantiate array-refs. Handle all constants and invariants.
9490 (instantiate_scev): Make instantiate_below parameter
9491 the edge into the region.
9492 (resolve_mixers): Use the loop preheader edge for the region
9493 parameter to instantiate_scev_r.
9494 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9496 2017-10-13 Richard Biener <rguenther@suse.de>
9498 PR tree-optimization/82525
9499 * graphite-isl-ast-to-gimple.c
9500 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9502 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9503 Fail code generation when we cannot represent the isl integer.
9504 (binary_op_to_tree): Elide modulo operations that are no-ops
9505 in the type we code generate. Remove now superfluous code
9508 2017-10-13 Richard Biener <rguenther@suse.de>
9510 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9511 (scop_detection::harmful_loop_in_region): Remove premature
9512 IV type restriction.
9513 (scop_detection::graphite_can_represent_scev): We can handle
9514 pointer IVs just fine.
9516 2017-10-13 Alan Modra <amodra@gmail.com>
9518 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9519 "Clobbers and Scratch Registers". Add paragraph on
9520 alternative to clobbers for scratch registers and OpenBLAS
9523 2017-10-13 Alan Modra <amodra@gmail.com>
9525 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9526 example of a memory input for a string of known length. Move
9527 commentary out of table. Add a number of new examples
9528 covering array memory inputs.
9530 2017-10-12 Martin Liska <mliska@suse.cz>
9532 PR tree-optimization/82493
9533 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9534 (test_range_functions): New function.
9535 (sbitmap_c_tests): Likewise.
9536 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9537 * selftest.h (sbitmap_c_tests): New function.
9539 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9541 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9543 * config/rs6000/amo.h: Fix spacing issue.
9545 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9548 * config/i386/i386.md (*ashl<mode>3_mask_1,
9549 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9550 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9553 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9555 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9556 (profile_probability): Set max_probability
9557 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9559 * profile-count.c (profile_probability::differs_from_p): Do not
9560 rely on max_probaiblity == 10000
9562 2017-10-12 Jeff Law <law@redhat.com>
9564 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9567 2017-10-12 Martin Sebor <msebor@redhat.com>
9571 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9572 (handle_alias_pairs): Call it.
9573 * common.opt (-Wattribute-alias): New option.
9574 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9575 * doc/invoke.texi (-Wattribute-alias): Document.
9577 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9580 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9582 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9584 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9585 (make_hard_regno_born, make_hard_regno_dead): Update
9586 bb_killed_pseudos and bb_gen_pseudos.
9588 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9590 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9592 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9594 * config/alpha/alpha.c (alpha_split_conditional_move):
9595 Use std::swap instead of manually swapping.
9596 (alpha_stdarg_optimize_hook): Ditto.
9597 (alpha_canonicalize_comparison): Ditto.
9599 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9601 * tree-loop-distribution.c (struct builtin_info): New struct.
9602 (struct partition): Refactor fields into struct builtin_info.
9603 (partition_free): Free struct builtin_info.
9604 (build_size_arg_loc, build_addr_arg_loc): Delete.
9605 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9606 information from struct builtin_info.
9607 (find_single_drs): New function refactored from classify_partition.
9608 Also moved builtin validity checks to this function.
9609 (compute_access_range, alloc_builtin): New functions.
9610 (classify_builtin_st, classify_builtin_ldst): New functions.
9611 (classify_partition): Refactor code into functions find_single_drs,
9612 classify_builtin_st and classify_builtin_ldst.
9613 (distribute_loop): Don't do runtime alias check when distributing
9615 (find_seed_stmts_for_distribution): New function.
9616 (pass_loop_distribution::execute): Refactor code finding seed
9617 stmts into above function. Support distribution for the innermost
9618 two-level loop nest. Adjust dump information.
9620 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9622 * tree-loop-distribution.c: Adjust the general comment.
9623 (NUM_PARTITION_THRESHOLD): New macro.
9624 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9625 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9626 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9627 in call to build_partition_graph.
9628 (finalize_partitions): New parameter. Make loop distribution more
9629 conservative by fusing more partitions.
9630 (distribute_loop): Don't do runtime alias check in case of loop nest
9632 (find_seed_stmts_for_distribution): New function.
9633 (prepare_perfect_loop_nest): New function.
9634 (pass_loop_distribution::execute): Refactor code finding seed stmts
9635 and loop nest into above functions. Support loop nest distribution.
9636 Adjust dump information accordingly.
9638 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9640 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9641 and set PTYPE_SEQUENTIAL for merged partition.
9643 2017-10-12 Richard Biener <rguenther@suse.de>
9645 PR tree-optimization/69728
9647 2017-09-19 Richard Biener <rguenther@suse.de>
9649 PR tree-optimization/69728
9650 * graphite-sese-to-poly.c (schedule_error): New global.
9651 (add_loop_schedule): Handle empty domain by failing the
9653 (build_original_schedule): Handle schedule_error.
9655 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9656 domain by returning an unchanged schedule.
9658 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9660 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9661 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9663 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9665 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9668 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9671 * expr.c (store_field): Don't optimize away bitsize == 0 store
9672 from CALL_EXPR with addressable return type.
9674 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9676 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9677 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9678 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9679 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9680 TARGET_ISEL instead of TARGET_ISEL<sel>.
9682 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9684 * config/rs6000/rs6000.c
9685 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9687 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9689 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9691 (reg_or_cint_operand): Fix comment.
9692 (reg_or_zero_operand): New predicate.
9693 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9694 * config/rs6000/rs6000.c (output_isel): Delete.
9695 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9696 instead of reg_or_cint_operand. Output instruction directly (not via
9698 (isel_unsigned_<mode>): Ditto.
9699 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9700 gpc_reg_operand. Add an instruction alternative for this. Output
9701 instruction directly.
9702 (*isel_reversed_unsigned_<mode>): Ditto.
9704 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9706 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9707 (TARGET_CANONICALIZE_COMPARISON): Define.
9709 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9712 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9713 Check whether the dest is REG before adding REG_EQUIV note.
9715 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9718 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9720 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9721 (make_hard_regno_born, make_hard_regno_dead): Update
9722 bb_killed_pseudos and bb_gen_pseudos.
9724 2017-10-11 Nathan Sidwell <nathan@acm.org>
9726 * incpath.h (enum incpath_kind): Name enum, prefix values.
9727 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9728 * incpath.c (heads, tails): Use INC_MAX.
9729 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9730 (merge_include_chains, split_quote_chain,
9731 register_include_chains): Update incpath_kind names.
9732 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9733 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9735 (add_framework_path, darwin_register_objc_includes): Likewise.
9736 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9738 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9740 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9741 Do not use float_operator operator predicate.
9742 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9743 * config/i386/predicates.md (float_operator): Remove predicate.
9745 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9747 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9748 (*jccxf_i387): Ditto.
9749 (*jcc<mode>_i387): Ditto.
9750 (*jccu<mode>_i387): Ditto.
9751 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9752 (*jcc_*_i387 splitters): Remove.
9753 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9754 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9755 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9758 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9760 * profile-count.h (slow_safe_scale_64bit): New function.
9761 (safe_scale_64bit): New inline.
9762 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9763 * profile-count.c: Include wide-int.h
9764 (slow_safe_scale_64bit): New.
9766 2017-10-11 Nathan Sidwell <nathan@acm.org>
9768 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9769 HAS_DECL_ASSEMBLER_NAME_P.
9770 * gimple-expr.c (gimple_decl_printable_name: Check
9771 HAS_DECL_ASSEMBLER_NAME_P too.
9772 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9773 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9774 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9775 * passes.c (rest_of_decl_compilation): Check
9776 HAS_DECL_ASSEMBLER_NAME_P too.
9777 * recog.c (verify_changes): Likewise.
9778 * tree-pretty-print.c (dump_decl_name): Likewise.
9779 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9781 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9782 (SET_DECL_ASSEMBLER_NAME): Use it.
9783 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9784 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9785 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9787 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9789 * config.gcc (i386, x86_64): Add extra objects.
9790 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9791 (ix86_min_insn_size): Declare.
9792 (ix86_issue_rate): Declare.
9793 (ix86_adjust_cost): Declare.
9794 (ia32_multipass_dfa_lookahead): Declare.
9795 (ix86_macro_fusion_p): Declare.
9796 (ix86_macro_fusion_pair_p): Declare.
9797 (ix86_bd_has_dispatch): Declare.
9798 (ix86_bd_do_dispatch): Declare.
9799 (ix86_core2i7_init_hooks): Declare.
9800 (ix86_atom_sched_reorder): Declare.
9801 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9802 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9803 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9804 (rip_relative_addr_p): Rename to ...
9805 (ix86_rip_relative_addr_p): ... this one; export.
9806 (memory_address_length): Update.
9807 (ix86_issue_rate): Move to x86-tune-sched.c.
9808 (ix86_flags_dependent): Move to x86-tune-sched.c.
9809 (ix86_agi_dependent): Move to x86-tune-sched.c.
9810 (exact_dependency_1): Move to x86-tune-sched.c.
9811 (exact_store_load_dependency): Move to x86-tune-sched.c.
9812 (ix86_adjust_cost): Move to x86-tune-sched.c.
9813 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9814 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9815 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9816 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9817 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9818 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9819 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9820 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9821 (min_insn_size): Rename to ...
9822 (ix86_min_insn_size): ... this one; export.
9823 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9824 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9825 (core2i7_first_cycle_multipass_backtrack): Move to
9826 x86-tune-sched-core.c.
9827 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9828 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9829 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9830 (ix86_avoid_jump_mispredicts): Update.
9831 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9832 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9833 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9834 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9835 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9836 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9837 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9838 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9839 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9840 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9841 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9842 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9843 (BIG): Move to ix86-tune-sched-bd.c.
9844 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9845 (enum insn_path): Move to ix86-tune-sched-bd.c.
9846 (get_mem_group): Move to ix86-tune-sched-bd.c.
9847 (is_cmp): Move to ix86-tune-sched-bd.c.
9848 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9849 (is_branch): Move to ix86-tune-sched-bd.c.
9850 (is_prefetch): Move to ix86-tune-sched-bd.c.
9851 (init_window): Move to ix86-tune-sched-bd.c.
9852 (allocate_window): Move to ix86-tune-sched-bd.c.
9853 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9854 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9855 (process_end_window): Move to ix86-tune-sched-bd.c.
9856 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9857 (find_constant): Move to ix86-tune-sched-bd.c.
9858 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9859 (has_immediate): Move to ix86-tune-sched-bd.c.
9860 (get_insn_path): Move to ix86-tune-sched-bd.c.
9861 (get_insn_group): Move to ix86-tune-sched-bd.c.
9862 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9863 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9864 (add_insn_window): Move to ix86-tune-sched-bd.c.
9865 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9866 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9867 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9868 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9869 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9870 (do_dispatch): Move to ix86-tune-sched-bd.c.
9871 (has_dispatch): Move to ix86-tune-sched-bd.c.
9872 * config/i386/t-i386: Add new object files.
9873 * config/i386/x86-tune-costs.h: New file.
9874 * config/i386/x86-tune-sched-atom.c: New file.
9875 * config/i386/x86-tune-sched-bd.c: New file.
9876 * config/i386/x86-tune-sched-core.c: New file.
9877 * config/i386/x86-tune-sched.c: New file.
9879 2017-10-11 Liu Hao <lh_mouse@126.com>
9881 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9882 the generic version below instead.
9883 (should_colorize): Recognize Windows consoles as terminals
9885 * pretty-print.c [__MINGW32__] (write_all): New function.
9886 [__MINGW32__] (find_esc_head): Likewise.
9887 [__MINGW32__] (find_esc_terminator): Likewise.
9888 [__MINGW32__] (eat_esc_sequence): Likewise.
9889 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9891 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9894 2017-10-11 Richard Biener <rguenther@suse.de>
9896 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9897 Properly call analyze_scalar_evolution with the loop of the stmt.
9899 2017-10-11 Richard Biener <rguenther@suse.de>
9901 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9902 * tree-core.h (tree_base): Add chrec_var union member.
9903 * tree.h (CHREC_VAR): Remove.
9904 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9905 * tree-chrec.h (build_polynomial_chrec): Adjust.
9906 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9907 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9909 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9911 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9912 * match.pd: ... here.
9913 ((T) X == (T) Y): Relax condition.
9915 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9917 PR tree-optimization/82472
9918 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9920 (break_alias_scc_partitions): Update postorder number.
9922 2017-10-11 Martin Liska <mliska@suse.cz>
9925 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9927 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9929 2017-10-11 Martin Liska <mliska@suse.cz>
9931 * print-rtl.c (print_insn): Move declaration of idbuf
9932 to same scope as name.
9934 2017-10-11 Martin Liska <mliska@suse.cz>
9939 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9942 2017-10-11 Martin Liska <mliska@suse.cz>
9945 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9948 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9950 * config.gcc (default_gnu_indirect_function): Default to yes for
9951 arm*-*-linux* with glibc.
9953 2017-10-11 Richard Biener <rguenther@suse.de>
9955 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9956 default-defs and types we do not want to analyze.
9957 (interpret_loop_phi): Replace unreachable code with an assert.
9958 (compute_scalar_evolution_in_loop): Remove and inline ...
9959 (analyze_scalar_evolution_1): ... here, replacing condition with
9960 what makes the intent clearer. Remove handling of cases
9961 get_scalar_evolution now handles.
9963 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9965 PR rtl-optimization/81434
9966 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9967 comment for main loop. In sched_group_found if, also add checks for
9968 pass and min_cost_group.
9970 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9972 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9973 (rs6000_insn_cost): New function.
9974 * config/rs6000/rs6000.md (cost): New attribute.
9976 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9977 H.J. Lu <hongjiu.lu@intel.com>
9981 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9982 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9983 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9984 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9985 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9986 OPTION_MASK_ISA_MMX and some other bit set require both
9987 mmx and the ISAs without the mmx bit.
9988 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9989 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9990 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9991 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9992 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9993 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9994 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9995 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9996 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9997 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9998 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9999 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10000 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10001 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10002 Add OPTION_MASK_ISA_MMX.
10004 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10006 * config.gcc (armv7*-*-freebsd*): New target.
10007 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10009 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10011 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10012 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10015 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10017 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10018 a template parameter.
10019 (WIDE_INT_REF_FOR): Update accordingly.
10020 * tree.h (wi::int_traits <const_tree>): Delete.
10021 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10022 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10023 (wi::tree_to_wide_ref): New typedef.
10024 (wi::to_wide): New function.
10025 * calls.c (get_size_range): Use wi::to_wide when operating on
10026 trees as wide_ints.
10027 * cgraph.c (cgraph_node::create_thunk): Likewise.
10028 * config/i386/i386.c (ix86_data_alignment): Likewise.
10029 (ix86_local_alignment): Likewise.
10030 * dbxout.c (stabstr_O): Likewise.
10031 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10032 * expr.c (const_vector_from_tree): Likewise.
10033 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10034 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10035 (fold_negate_expr_1, int_const_binop_1, const_binop)
10036 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10037 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10038 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10039 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10040 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10041 (fold_not_const, round_up_loc): Likewise.
10042 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10043 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10044 (alloca_call_type): Likewise.
10045 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10046 * godump.c (go_output_typedef): Likewise.
10047 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10048 * internal-fn.c (get_min_precision): Likewise.
10049 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10050 * ipa-polymorphic-call.c
10051 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10052 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10053 (ipa_modify_call_arguments): Likewise.
10054 * match.pd: Likewise.
10055 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10056 * print-tree.c (print_node_brief, print_node): Likewise.
10057 * stmt.c (expand_case): Likewise.
10058 * stor-layout.c (layout_type): Likewise.
10059 * tree-affine.c (tree_to_aff_combination): Likewise.
10060 * tree-cfg.c (group_case_labels_stmt): Likewise.
10061 * tree-data-ref.c (dr_analyze_indices): Likewise.
10062 (prune_runtime_alias_test_list): Likewise.
10063 * tree-dump.c (dequeue_and_dump): Likewise.
10064 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10065 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10066 * tree-pretty-print.c (dump_generic_node): Likewise.
10067 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10068 (simple_iv_with_niters): Likewise.
10069 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10070 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10071 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10072 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10073 (refine_value_range_using_guard, number_of_iterations_ne_max)
10074 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10075 (get_cst_init_from_scev, record_nonwrapping_iv)
10076 (scev_var_range_cant_overflow): Likewise.
10077 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10078 * tree-ssa-pre.c (compute_avail): Likewise.
10079 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10080 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10081 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10082 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10083 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10084 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10085 (dump_case_nodes, try_switch_expansion): Likewise.
10086 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10087 (vect_do_peeling): Likewise.
10088 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10089 * tree-vect-stmts.c (vectorizable_load): Likewise.
10090 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10091 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10092 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10093 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10094 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10095 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10096 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10097 (evrp_dom_walker::before_dom_children): Likewise.
10098 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10099 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10100 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10101 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10102 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10103 (get_range_pos_neg): Likewise.
10104 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10105 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10106 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10107 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10108 * config/avr/avr.c (avr_fold_builtin): Likewise.
10109 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10110 * config/msp430/msp430.c (msp430_attr): Likewise.
10111 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10112 * config/powerpcspe/powerpcspe-c.c
10113 (altivec_resolve_overloaded_builtin): Likewise.
10114 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10115 (rs6000_expand_ternop_builtin): Likewise.
10116 * config/rs6000/rs6000-c.c
10117 (altivec_resolve_overloaded_builtin): Likewise.
10118 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10119 (rs6000_expand_ternop_builtin): Likewise.
10120 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10122 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10124 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10125 when copying loop nest with only one inner loop.
10127 2017-10-10 Richard Biener <rguenther@suse.de>
10129 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10130 blocks if SCEV is active.
10131 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10133 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10134 (scev_initialize): Assert we are not yet initialized.
10136 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10138 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10139 inner loop's exit stmt by making it always exit the loop, otherwise
10140 we would generate an infinite empty loop.
10142 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10144 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10145 renaming variables in new preheader if it's deleted.
10147 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10149 * tree-loop-distribution.c (struct partition): Remove unused field
10150 loops of the structure.
10151 (partition_alloc, partition_free): Ditto.
10152 (build_rdg_partition_for_vertex): Ditto.
10154 2017-10-09 Jeff Law <law@redhat.com>
10156 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10157 return type to match prototype and documentation.
10159 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10161 * config/rs6000/rs6000.c (processor_costs): Move to ...
10162 * config/rs6000/rs6000.h: ... here.
10163 (rs6000_cost): Declare.
10165 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10167 * except.c (setjmp_fn): New global variable.
10168 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10169 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10170 if DONT_USE_BUILTIN_SETJMP is defined.
10172 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10174 * target.def (insn_cost): New hook.
10175 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10176 * doc/tm.texi: Regenerate.
10177 * rtlanal.c (insn_cost): Use the new hook.
10179 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10181 * combine.c (combine_validate_cost): Compute the new insn_cost,
10182 not just pattern_cost.
10183 (try_combine): Adjust comment.
10185 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10187 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10189 * combine.c (uid_insn_cost): Adjust comment.
10190 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10192 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10193 * dse.c (find_shift_sequence): Ditto.
10194 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10195 (bb_valid_for_noce_process_p): Use pattern_cost.
10196 * rtl.h (insn_rtx_cost): Delete.
10197 (pattern_cost): New prototype.
10198 (insn_cost): New prototype.
10199 * rtlanal.c (insn_rtx_cost): Rename to...
10200 (pattern_cost): ... this.
10203 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10205 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10206 (*jcc<mode>_0_r_i387): Ditto.
10207 (*jccxf_r_i387): Ditto.
10208 (*jcc<mode>_r_i387): Ditto.
10209 (*jccu<mode>_r_i387): Ditto.
10210 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10211 (*jcc): Rename from *jcc_1.
10213 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10215 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10216 deferred rescans after the lvx/stvx recombination pre-pass.
10218 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10220 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10221 memory operation instruction support.
10222 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10223 (rs6000-ibm-aix[789]*): Likewise.
10224 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10225 Document new functions.
10227 2017-10-09 Richard Biener <rguenther@suse.de>
10229 PR tree-optimization/82397
10230 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10231 equality only for semantically equal trees.
10233 2017-10-09 Richard Biener <rguenther@suse.de>
10235 PR tree-optimization/82449
10236 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10237 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10238 allow constant addresses.
10239 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10242 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10244 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10247 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10250 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10253 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10256 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10258 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10261 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10262 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10264 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10266 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10267 (WI_BINARY_PREDICATE_RESULT): Likewise.
10268 (wi::binary_traits::operator_result): New type.
10269 (wi::binary_traits::predicate_result): Likewise.
10270 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10271 (generic_wide_int::operator==, generic_wide_int::operator!=)
10272 (generic_wide_int::operator&, generic_wide_int::and_not)
10273 (generic_wide_int::operator|, generic_wide_int::or_not)
10274 (generic_wide_int::operator^, generic_wide_int::operator+
10275 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10277 (operator~, unary operator-, operator==, operator!=, operator&)
10278 (operator|, operator^, operator+, binary operator-, operator*): New
10280 * expr.c (get_inner_reference): Use wi::bit_and_not.
10281 * fold-const.c (fold_binary_loc): Likewise.
10282 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10283 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10284 (bit_value_binop): Likewise.
10285 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10286 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10287 (extract_range_from_binary_expr_1): Likewise.
10288 (masked_increment): Likewise.
10289 (simplify_bit_ops_using_ranges): Likewise.
10291 2017-10-09 Martin Jambor <mjambor@suse.cz>
10294 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10295 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10296 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10298 (hsa_fixup_mov_insn_type): New function.
10299 (hsa_op_with_type::get_in_type): Use it.
10300 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10301 immediates in an assert.
10302 (hsa_op_with_type::extend_int_to_32bit): New method.
10303 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10304 types. Convert to dest type if necessary.
10305 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10306 (reg_for_gimple_ssa): Pass false as min32int to
10307 hsa_type_for_scalar_tree_type.
10308 (gen_hsa_addr): Fixup type when creating addresable temporary.
10309 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10310 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10311 necessary. Call hsa_fixup_mov_insn_type.
10312 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10313 extend operands and convert to dest type if necessary.
10314 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10315 to dest type if necessary.
10316 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10317 if conversion nt necessary and size matches.
10318 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10319 to dest type if necessary.
10320 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10321 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10323 (gen_hsa_clrsb): Likewise.
10324 (gen_hsa_ffs): Likewise.
10325 (gen_hsa_divmod): Extend operands and convert to dest type if
10327 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10329 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10331 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10332 Remove empty default arguments. Use a brace block as output
10334 (conditional return): Ditto.
10336 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10337 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10339 (group_ending_nop): Ditto.
10340 (doloop_end): Ditto.
10341 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10342 (splitters for those): Ditto.
10344 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10346 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10347 a conditional jump (and the compare for it) so that pc_rtx is the
10349 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10350 for the deleted and renamed ctr<mode>_internal[234] patterns.
10351 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10352 Delete second conditional return pattern.
10353 (ctr<mode>_internal2): Delete this second bdnz pattern.
10354 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10355 (ctr<mode>_internal4): Delete this second bdz pattern.
10357 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10359 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10360 (always_initialized_rtx_for_ssa_name_p): New predicate.
10361 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10362 (finish_out_of_ssa): Free new field of SA.
10363 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10364 * tree-ssa-coalesce.c: Include tree-ssa.h.
10365 (get_parm_default_def_partitions): Remove extern keyword.
10366 (get_undefined_value_partitions): New function.
10367 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10368 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10369 uninitialized bits.
10370 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10372 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10374 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10376 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10378 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10379 for targets that preffer 128bit.
10381 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10383 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10385 2017-10-08 Olivier Hainque <hainque@adacore.com>
10387 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10388 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10390 (thumb_set_return_address): Likewise.
10392 2017-10-08 Olivier Hainque <hainque@adacore.com>
10394 * common/config/arm/arm-common.c (arm_except_unwind_info):
10395 Handle DWARF2_UNWIND_INFO.
10397 2017-10-07 Michael Collison <michael.collison@arm.com>
10399 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10402 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10404 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10405 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10406 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10407 defined, force the creation of a new block for a dispatch label.
10409 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10411 * invoke.texi (Wsuggest-attribute=cold): Document.
10412 * common.opt (Wsuggest-attribute=cold): New
10413 * ipa-pure-const.c (warn_function_cold): New function.
10414 * predict.c (compute_function_frequency): Use it.
10415 * predict.h (warn_function_cold): Declare.
10417 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10419 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10422 2017-10-06 Martin Liska <mliska@suse.cz>
10424 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10425 keyword for member functions.
10426 (struct sanopt_tree_couple): New struct.
10427 (struct sanopt_tree_couple_hash): New function.
10428 (struct sanopt_ctx): Add new hash_map.
10429 (has_dominating_ubsan_ptr_check): New function.
10430 (record_ubsan_ptr_check_stmt): Likewise.
10431 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10432 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10433 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10435 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10438 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10439 aarch64_simd_valid_immediate on CONST_VECTORs.
10440 (aarch64_reg_or_bic_imm): Likewise.
10442 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10444 PR rtl-optimization/82396
10445 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10447 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10449 * graphite-dependences.c (scop_get_reads): Move code to...
10450 (scop_get_must_writes): Move code to...
10451 (scop_get_may_writes): Move code to...
10452 (scop_get_reads_and_writes): ... here.
10453 (scop_get_dependences): Call scop_get_reads_and_writes.
10455 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10457 PR tree-optimization/82434
10458 * fold-const.h (can_native_encode_type_p,
10459 can_native_encode_string_p): Remove.
10460 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10461 don't encode anything, just return what would be otherwise returned.
10462 (native_encode_fixed, native_encode_complex, native_encode_vector):
10464 (native_encode_string): Likewise. Inline by hand
10465 can_native_encode_string_p.
10466 (can_native_encode_type_p): Remove.
10467 (can_native_encode_string_p): Remove.
10468 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10469 STRING_CSTs using can_native_encode_string_p, test all
10470 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10471 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10472 argument from native_encode_expr.
10473 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10474 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10477 2017-10-06 Richard Biener <rguenther@suse.de>
10479 PR tree-optimization/82397
10480 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10481 operand_equal_p but rely on data_ref_compare_tree for detecting
10483 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10484 to match up with dr_group_sort_cmp.
10486 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10489 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10491 * config/s390/s390-builtin-types.def: Regenerate.
10493 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10496 * config/s390/s390-builtin-types.def: Regenerate.
10497 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10498 Change flag from B_VXE to B_VX.
10499 (s390_vec_min_dbl): Remove B_VXE flag.
10501 2017-10-06 Richard Biener <rguenther@suse.de>
10503 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10504 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10505 translate_isl_ast_to_gimple::is_valid_rename,
10506 translate_isl_ast_to_gimple::get_rename,
10507 translate_isl_ast_to_gimple::get_def_bb_for_const,
10508 translate_isl_ast_to_gimple::get_new_name,
10509 translate_isl_ast_to_gimple::collect_all_ssa_names,
10510 translate_isl_ast_to_gimple::copy_loop_phi_args,
10511 translate_isl_ast_to_gimple::collect_all_ssa_names,
10512 translate_isl_ast_to_gimple::copy_loop_phi_args,
10513 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10514 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10515 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10516 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10517 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10518 translate_isl_ast_to_gimple::copy_cond_phi_args,
10519 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10520 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10521 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10522 translate_isl_ast_to_gimple::rename_uses,
10523 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10524 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10525 (set_rename_for_each_def): Likewise.
10526 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10527 here. Handle rewriting SCEV analyzable uses here.
10528 (copy_bb_and_scalar_dependences): Generate code for PHI
10530 (graphite_regenerate_ast_isl): Adjust.
10531 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10532 (add_write, add_read): New functions.
10533 (build_cross_bb_scalars_def): Use it and simplify.
10534 (build_cross_bb_scalars_use): Likewise.
10535 (graphite_find_cross_bb_scalar_vars): Inline into...
10536 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10537 simulating out-of-SSA. Compute liveout and add dependencies.
10538 (build_scops): Force an empty entry block.
10539 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10541 (sese_build_liveouts): Declare.
10542 (sese_trivially_empty_bb_p): Likewise.
10543 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10544 compute liveout and debug_liveout.
10545 (sese_bad_liveouts_use): Remove.
10546 (sese_reset_debug_liveouts_bb): Likewise.
10547 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10548 (sese_build_liveouts): Build liveout and debug_liveout and store
10550 (new_sese_info): Adjust.
10551 (free_sese_info): Likewise.
10552 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10553 do not build liveout here.
10554 (move_sese_in_condition): Adjust region entry.
10555 (scev_analyzable_p): Match up with chrec_apply requirements.
10556 (sese_trivially_empty_bb_p): New.
10557 * tree-into-ssa.c (get_reaching_def): Properly support generating
10558 default-defs for incremental rewrite of anonymous names.
10560 2017-10-06 Richard Biener <rguenther@suse.de>
10562 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10563 precision do not perform modulo reduction.
10565 2017-10-06 Richard Biener <rguenther@suse.de>
10567 PR tree-optimization/82436
10568 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10569 conservatively choose the vectorization factor when checking
10570 whether we can perform the required load permutation.
10571 (vect_transform_slp_perm_load): Assert when we may not fail.
10573 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10575 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10576 message for incompatible -msdata=* and -mcall-* options.
10578 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10580 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10581 rate for post-reload scheduling.
10583 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10585 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10587 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10589 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10590 to improve monte carlo in scimark.
10592 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10594 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10595 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10596 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10597 pentium4_cost, nocona_cost): Set reassociation width to 1.
10598 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10599 width to 2 for fp operations and 1 otherwise.
10600 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10602 (atom_cost): Set reassociation width to 2.
10603 (slm_cost, generic_cost): Set fp reassociation width
10604 to 2 and 1 otherwise.
10605 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10606 (core_cost): Set fp reassociation width to 4 and vector to 2.
10607 (ix86_reassociation_width): Rewrite using cost table; special case
10608 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10609 and TARGET_AVX128_OPTIMAL.
10610 * config/i386/i386.h (processor_costs): Add
10611 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10612 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10613 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10614 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10615 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10616 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10618 2017-10-05 Nathan Sidwell <nathan@acm.org>
10620 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10622 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10624 * config/arm/arm.c (arm_test_fpu_data): New.
10625 (arm_run_selftests): Call arm_test_fpu_data.
10627 2017-10-04 Nathan Sidwell <nathan@acm.org>
10629 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10631 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10633 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10635 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10636 check type for aarch64_simd_valid_immediate.
10637 (aarch64_output_simd_mov_immediate): Update prototype.
10638 (aarch64_simd_valid_immediate): Update prototype.
10639 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10640 support for ORR-immediate.
10641 (and<mode>3): modified pattern to add support for BIC-immediate.
10642 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10643 now checks for valid immediate for BIC and ORR based on new enum
10645 (aarch64_output_simd_mov_immediate): Function now used to output
10646 BIC/ORR imm as well based on new enum argument.
10647 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10649 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10650 (aarch64_reg_or_bic_imm): Likewise.
10652 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10654 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10655 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10657 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10661 PR rtl-optimization/82396
10662 * haifa-sched.c (autopref_multipass_init): Simplify
10664 (autopref_rank_data): Simplify sort order.
10665 * sched-int.h (autopref_multipass_data_): Remove
10666 multi_mem_insn_p, min_offset and max_offset.
10668 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10670 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10672 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10674 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10675 vect_doubleint_cvt.
10677 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10679 * doc/sourcebuild.texi: Document vect_long_mult.
10681 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10683 PR tree-optimization/82413
10684 * fold-const.c (build_range_check): Use widest_int when comparing
10685 the maximum ETYPE value with HIGH.
10687 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10689 PR rtl-optimization/82396
10690 * haifa-sched.c (autopref_multipass_init): Simplify
10692 (autopref_rank_data): Simplify sort order.
10693 * sched-int.h (autopref_multipass_data_): Remove
10694 multi_mem_insn_p, min_offset and max_offset.
10696 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10698 PR tree-optimization/82381
10699 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10700 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10703 PR tree-optimization/82374
10704 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10705 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10706 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10707 current_function_decl to the new decl.
10709 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10711 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10712 helper macro for IEEE float128 hardware built-in functions.
10713 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10715 (TRUNCF128_ODD): Likewise.
10716 (ADDF128_ODD): Likewise.
10717 (SUBF128_ODD): Likewise.
10718 (MULF128_ODD): Likewise.
10719 (DIVF128_ODD): Likewise.
10720 (FMAF128_ODD): Likewise.
10721 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10722 UNSPEC_TRUNC_ROUND_TO_ODD.
10723 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10724 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10725 floating point round to odd instructions.
10726 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10727 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10728 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10729 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10730 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10731 (trunc<mode>sf2_hw): Change the truncate with round to odd
10732 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10733 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10734 to odd hardware instructions.
10735 (sub<mode>3_odd): Likewise.
10736 (mul<mode>3_odd): Likewise.
10737 (div<mode>3_odd): Likewise.
10738 (sqrt<mode>2_odd): Likewise.
10739 (fma<mode>4_odd): Likewise.
10740 (fms<mode>4_odd): Likewise.
10741 (nfma<mode>4_odd): Likewise.
10742 (nfms<mode>4_odd): Likewise.
10743 (trunc<mode>df2_odd): Change the truncate with round to odd
10744 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10746 * doc/extend.texi (PowerPC built-in functions): Update documentation
10747 for existing IEEE float128-bit built-in functions. Add built-in
10748 functions that generate the IEEE 128-bit floating point round to
10751 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10753 PR rtl-optimization/77729
10754 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10755 to (X&(C1&~C2))|C2 transformations.
10757 2017-10-03 Martin Jambor <mjambor@suse.cz>
10759 PR tree-optimization/82363
10760 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10761 mismatch, mark lacc written regardless of racc.
10763 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10765 PR tree-optimization/82381
10766 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10767 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10768 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10769 fallthrough into reassoc_stmt_dominates_stmt_p.
10772 * combine.c (combine_instructions): Don't combine in unreachable
10775 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
10778 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10779 function to not use the have_cpu variable. Do not set cpu_index,
10780 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10781 or the default cpu.
10782 (rs6000_valid_attribute_p): Remove duplicate initializations of
10783 old_optimize and func_optimize.
10784 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10785 (rs6000_activate_target_options): Make global.
10786 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10789 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10791 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10792 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10793 Set *poffset to 0 and *psize and *pmax_size to -1 if
10794 *poffset + *psize overflows in HOST_WIDE_INT.
10796 PR tree-optimization/82387
10797 PR tree-optimization/82388
10798 PR tree-optimization/82389
10799 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10800 instead of live_bytes non-NULL.
10802 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10805 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10808 2017-10-02 Richard Biener <rguenther@suse.de>
10810 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10811 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10812 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10814 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10816 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10817 requested precision matches the type's.
10818 * calls.c (alloc_max_size): Calculate the new candidate size as
10819 a widest_int and use wi::to_widest when comparing it with the
10820 current candidate size.
10821 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10822 zero rather than integer_zero_node.
10823 * match.pd: Check for a no-op conversion before using wi::add
10824 rather than after. Use tree_to_uhwi when summing small shift
10825 counts into an unsigned int.
10827 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10828 Alan Hayward <alan.hayward@arm.com>
10829 David Sherwood <david.sherwood@arm.com>
10832 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10833 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10834 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10835 POINTER_AND_FP_REGS.
10837 2017-10-02 Richard Biener <rguenther@suse.de>
10839 PR tree-optimization/82355
10840 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10841 a mapping for the enclosing loop but avoid generating one for
10842 the loop tree root.
10843 (copy_bb_and_scalar_dependences): Remove premature codegen
10844 error on PHIs in blocks duplicated into multiple places.
10845 * graphite-scop-detection.c
10846 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10847 in the region use it as loop and nest to analyze the DR in.
10848 (try_generate_gimple_bb): Likewise.
10849 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10850 (add_loop_constraints): For blocks in a loop not in the region
10851 create a dimension with a single iteration.
10852 * sese.h (gbb_loop_at_index): Remove assert.
10854 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10856 * omp-expand.c (adjust_context_scope): New function.
10857 (expand_parallel_call): Call adjust_context_scope.
10859 2017-10-01 Jeff Law <law@redhat.com>
10861 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10862 dom_opt_dom_walker class with direct access to private members.
10863 Add comments. Call test_for_singularity.
10864 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10865 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10866 m_dummy_cond anymore.
10867 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10869 (pass_dominator:execute): Build the dummy_cond here and pass it
10870 to the dom_opt_dom_walker ctor.
10871 (test_for_singularity): New function.
10873 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10874 Maya Rashish <coypu@sdf.org>
10876 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10877 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10878 (alpha*-*-netbsd*) Use nbsd_tm_file.
10879 (arm*-*-netbsdelf*) Likewise.
10880 (i[34567]86-*-netbsdelf*) Likewise.
10881 (x86_64-*-netbsd*) Likewise.
10882 (mips*-*-netbsd*) Likewise.
10883 (powerpc-*-netbsd*) Likewise.
10884 (sh*-*-netbsd*) Likewise.
10885 (sparc-*-netbsdelf*) Likewise.
10886 (sparc64-*-netbsd*) Likewise.
10887 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10889 (vax-*-netbsdelf*) Likewise.
10890 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10891 (UINT_FAST8_TYPE) Likewise.
10892 (INT_FAST16_TYPE) Check CHAR_FAST16.
10893 (UINT_FAST16_TYPE) Likewise.
10895 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10898 * config/i386/i386.md
10899 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10900 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10901 *divmodsi4_zext_2): New define_insn_and_split.
10902 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10903 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10904 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10905 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10906 New define_insn_and_split.
10907 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10908 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10909 operands[1] having DImode when mode is SImode.
10911 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10912 always SImode for DIV and MOD in REG_EQUAL notes.
10914 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10916 PR middle-end/82319
10917 * match.pd: Fix handling of NaNs in pattern.
10919 2017-09-29 Jeff Law <law@redhat.com>
10921 * sbitmap.c (bitmap_bit_in_range_p): New function.
10922 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10923 * tree-ssa-dse.c (live_bytes_read): New function.
10924 (dse_classify_store): Ignore reads of dead bytes.
10926 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10927 typos and whitespace errors.
10928 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10929 * config/s390/s390.c (s390_emit_prologue): Likewise.
10931 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10934 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10935 with a symbol for LRA.
10937 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10939 PR rtl-optimization/82338
10940 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10942 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10944 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10945 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10946 (qsort_chk): Declare.
10947 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10948 (qsort_chk): New function.
10950 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10952 PR tree-optimization/82337
10953 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10954 phi definition if the PHI result appears in an abnormal PHI.
10955 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10956 basis appears in an abnormal PHI.
10958 2017-09-29 Richard Biener <rguenther@suse.de>
10960 * graphite-isl-ast-to-gimple.c
10961 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10962 (binary_op_to_tree): Use it.
10963 (get_rename_from_scev): Likewise.
10964 (copy_loop_phi_nodes): Likewise.
10965 (copy_bb_and_scalar_dependences): Likewise.
10966 (translate_pending_phi_nodes): Likewise.
10968 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10971 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10972 for movabsq $(i32 << shift), r64.
10974 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10976 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10977 index when encoding %esp as %rsp to avoid 0x67 prefix.
10979 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10981 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10982 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10984 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10986 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10987 Extensions with more than 16 double VFP registers.
10988 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10989 to_clear_mask and all code related to it. Replace the remaining
10990 entry by a sbitmap and adapt code accordingly.
10992 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10994 * brig-builtins.def: Change pure attributes to const.
10996 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10998 * config.gcc (default_gnu_indirect_function): Default to yes for
10999 sparc*-*-linux* with glibc.
11001 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11003 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11004 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11005 when creating .init_array and .fini_array sections with priority
11008 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11011 * config/aarch64/aarch64.c
11012 (aarch64_builtin_support_vector_misalignment): Always return false
11013 when misalignment is unknown.
11015 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11017 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11018 this function to return false if the definition used by the swap
11019 instruction is artificial, or if the memory address from which the
11020 constant value is loaded is not represented by a base address held
11021 in a register or if the base address register is a frame or stack
11022 pointer. Additionally, return false if the base address of the
11023 loaded constant is a SYMBOL_REF but is not considered to be a
11025 (replace_swapped_load_constant): New function.
11026 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11027 loaded constant vector with a load of a swapped constant vector.
11029 2017-09-27 Carl Love <cel@us.ibm.com>
11031 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11032 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11033 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11036 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11038 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11039 first, always call autopref_rank_data otherwise.
11041 2017-09-27 Richard Biener <rguenther@suse.de>
11043 * graphite-scop-detection.c (find_scop_parameters): Move
11044 loop bound handling ...
11045 (gather_bbs::before_dom_children): ... here, avoiding the need
11046 to build scop_info->loop_nest.
11047 (record_loop_in_sese): Remove.
11048 * sese.h (sese_info_t::loop_nest): Remove.
11049 * sese.c (new_sese_info): Do not allocate loop_nest.
11050 (free_sese_info): Do not free loop_nest.
11052 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11055 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11056 lhs from calls if the lhs has addressable type.
11058 2017-09-27 Richard Biener <rguenther@suse.de>
11060 * graphite.h (scop::max_alias_set): New member.
11061 * graphite-scop-detection.c: Remove references to non-existing
11062 --param in comments.
11063 (build_alias_sets): Record the maximum alias set used for drs.
11064 (build_scops): Support zero as unlimited for
11065 --param graphite-max-arrays-per-scop.
11066 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11067 and inline into ...
11068 (build_poly_sr_1): ... here. Compute alias set based on the
11069 maximum alias set used for drs rather than
11070 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11072 2017-09-27 Richard Biener <rguenther@suse.de>
11074 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11075 --param loop-block-tile-size=0 to disable tiling.
11077 2017-09-27 Richard Biener <rguenther@suse.de>
11079 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11080 (graphite-max-nb-scop-params): Document special value zero.
11081 * domwalk.h (dom_walker::STOP): New symbolical constant.
11082 (dom_walker::dom_walker): Add optional parameter for bb to
11084 (dom_walker::~dom_walker): Declare.
11085 (dom_walker::before_dom_children): Document STOP return value.
11086 (dom_walker::m_user_bb_to_rpo): New member.
11087 (dom_walker::m_bb_to_rpo): Likewise.
11088 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11089 mapping here if not provided by the user.
11090 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11091 provided by the user.
11092 (dom_walker::STOP): Define.
11093 (dom_walker::walk): Do not compute bb to RPO mapping here.
11094 Support STOP return value from before_dom_children to stop
11096 * graphite-optimize-isl.c (optimize_isl): If the schedule
11097 is the same still generate code if -fgraphite-identity
11098 or -floop-parallelize-all are given.
11099 * graphite-scop-detection.c: Include cfganal.h.
11100 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11102 (gather_bbs::before_dom_children): Return STOP for BBs
11104 (build_scops): Compute bb to RPO mapping and pass it to
11105 the domwalk. Treat --param graphite-max-nb-scop-params=0
11106 as not limiting the number of params.
11107 * graphite.c (graphite_initialize): Remove limit on the
11108 number of basic-blocks in a function.
11109 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11110 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11111 default value of 10.
11113 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11115 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11116 Adjust code to eliminate needing to do the shift right 32-bits
11117 operation after XSCVDPSPN.
11119 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11121 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11122 ((X / Y) != 0 -> X >= Y): Likewise.
11124 2017-09-26 Carl Love <cel@us.ibm.com>
11126 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11127 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11128 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11129 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11130 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11131 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11132 definitions and overloading.
11133 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11134 statement for P9V_BUILTIN_XST_LEN_R.
11135 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11136 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11137 define_expand and define_insn for the instructions and builtins.
11138 * doc/extend.texi: Update the built-in documentation file for the new
11139 built-in functions.
11140 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11141 define_insn for the instructions
11143 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11146 * gcc/config/netbsd-protos.h: New file.
11147 * gcc/config/netbsd.c: New file.
11148 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11149 * gcc/config/t-netbsd: New file.
11150 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11151 (tmake_file) Add t-netbsd.
11152 (extra_objs) Add netbsd.o.
11154 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11158 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11160 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11162 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11163 sign extension from a vector register to a GPR by doing a 32-bit
11164 direct move and then an EXTSW.
11165 (extendsi<mode>2 splitter): Likewise.
11166 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11167 right or vector extract after doing XSCVDPSPN. Use
11168 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11170 (movdi_from_sf_zero_ext): Likewise.
11171 (reload_gpr_from_vsxsf): Likewise.
11172 (p8_mfvsrd_4_disf): Delete, no longer used.
11173 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11174 then needing to move the SFmode to a GPR to use the XSCVDPSP
11175 instruction instead of FRSP and XSCVDPSPN.
11176 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11177 it is adjacent to the other XSCVSPDP insns.
11178 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11179 SFmode to be in traditional Altivec registers.
11180 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11181 (vsx_xscvspdpn): Likewise.
11182 (vsx_xscvspdpn_scalar): Likewise.
11184 2017-09-26 Martin Jambor <mjambor@suse.cz>
11186 * tree-sra.c (compare_access_positions): Put integral types first,
11187 stabilize sorting of integral types, remove conditions putting
11188 non-full-precision integers last.
11189 (sort_and_splice_var_accesses): Disable scalarization if a
11190 non-integert would be represented by a non-full-precision integer.
11192 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11194 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11195 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11196 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11197 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11198 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11199 conditionals inside the function instead of around it. Call
11200 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11201 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11203 2017-09-26 Richard Biener <rguenther@suse.de>
11205 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11207 (scop_detection::build_scop_breadth): ... this. Removed.
11208 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11209 (scop_detection::harmful_stmt_in_bb): Likewise.
11210 (scop_detection::graphite_can_represent_stmt): Likewise.
11211 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11212 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11213 (scop_detection::can_represent_loop_1): ... this. Removed.
11214 (scop_detection::harmful_loop_in_region): Simplify after inlining
11215 the above and remove more quadraticness.
11216 (build_scops): Adjust.
11217 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11220 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11223 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11224 REGNO (base) == SP_REG if base is a REG.
11226 PR middle-end/35691
11227 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11228 if it is different SSA_NAME.
11229 (optimize_range_tests_cmp_bitwise): New function.
11230 (optimize_range_tests): Call it.
11232 2017-09-26 Richard Biener <rguenther@suse.de>
11234 PR tree-optimization/82321
11235 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11236 for the def being inside the loop.
11238 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11240 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11242 * config/s390/s390-builtins.def: Fix constraint on op4.
11244 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11246 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11247 independent expanders.
11248 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11249 ("vec_ordered", "vec_unordered"): New expanders.
11251 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11253 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11256 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11258 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11259 vec_unpacks_lo_v16qi.
11260 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11262 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11264 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11265 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11266 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11268 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11270 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11272 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11274 ("*vec_slb<mode>"): New insn pattern.
11275 ("vec_shr_<mode>"): New expander.
11276 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11277 and force the shift count operand to V16QImode.
11278 ("vec_srb<mode>"): Set shift count mode to V16QI.
11280 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11282 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11283 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11284 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11286 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11289 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11291 2017-09-26 Richard Biener <rguenther@suse.de>
11293 PR tree-optimization/82320
11294 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11297 2017-09-25 Jeff Law <law@redhat.com>
11299 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11300 prototype for new argument.
11301 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11302 mostly extracted from rs6000_emit_allocate_stack.
11303 (rs6000_emit_probe_stack_range_stack_clash): New function.
11304 (rs6000_emit_allocate_stack): Call
11305 rs6000_emit_probe_stack_range_stack_clash as needed.
11306 (rs6000_emit_probe_stack_range): Add additional argument
11307 to call to gen_probe_stack_range{si,di}.
11308 (output_probe_stack_range): New.
11309 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11310 (output_probe_stack_range_stack_clash): New.
11311 (rs6000_emit_prologue): Emit notes into dump file as requested.
11312 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11313 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11314 Add additional operand and pass it to output_probe_stack_range.
11316 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11318 PR tree-optimization/82163
11319 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11320 (checking_verify_loop_closed_ssa): New parameter.
11321 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11322 (check_loop_closed_ssa_stmt): Delete.
11323 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11324 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11325 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11328 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11330 * brig-builtins.def: Treat HSAIL barrier builtins as
11331 setjmp/longjump style functions.
11333 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11335 * target.def (constant_alignment): New hook.
11336 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11337 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11338 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11339 * doc/tm.texi: Regenerate.
11340 * targhooks.h (default_constant_alignment): Declare.
11341 (constant_alignment_word_strings): Likewise.
11342 * targhooks.c (default_constant_alignment): New function.
11343 (constant_alignment_word_strings): Likewise.
11344 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11345 instead of CONSTANT_ALIGNMENT.
11346 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11347 (force_const_mem): Likewise.
11348 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11349 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11350 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11351 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11352 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11354 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11355 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11356 constant_alignment_word_strings.
11357 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11358 (CONSTANT_ALIGNMENT): Likewise.
11359 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11360 (arm_constant_alignment): New function.
11361 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11362 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11363 constant_alignment_word_strings.
11364 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11365 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11366 constant_alignment_word_strings.
11367 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11368 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11369 (cris_constant_alignment): New function.
11370 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11371 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11372 (epiphany_constant_alignment): New function.
11373 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11374 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11375 constant_alignment_word_strings.
11376 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11377 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11378 constant_alignment_word_strings.
11379 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11380 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11381 constant_alignment_word_strings.
11382 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11383 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11384 * config/i386/i386.c (ix86_constant_alignment): Make static.
11385 Use the same interface as the target hook.
11386 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11387 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11388 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11389 constant_alignment_word_strings.
11390 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11391 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11392 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11393 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11394 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11395 constant_alignment_word_strings.
11396 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11397 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11398 constant_alignment_word_strings.
11399 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11400 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11401 constant_alignment_word_strings.
11402 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11403 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11405 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11406 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11407 * config/mips/mips.c (mips_constant_alignment): New function.
11408 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11409 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11410 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11411 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11412 (mmix_constant_alignment): Make static. Use the same interface
11413 as the target hook.
11414 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11415 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11416 constant_alignment_word_strings.
11417 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11418 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11419 constant_alignment_word_strings.
11420 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11421 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11422 constant_alignment_word_strings.
11423 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11424 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11425 (rs6000_constant_alignment): New function.
11426 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11427 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11428 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11429 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11430 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11431 (rs6000_constant_alignment): New function.
11432 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11433 * config/s390/s390.c (s390_constant_alignment): New function.
11434 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11435 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11436 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11437 constant_alignment_word_strings.
11438 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11439 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11440 (sparc_constant_alignment): New function.
11441 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11442 * config/spu/spu.c (spu_constant_alignment): New function.
11443 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11444 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11445 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11446 constant_alignment_word_strings.
11447 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11448 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11449 constant_alignment_word_strings.
11450 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11451 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11452 constant_alignment_word_strings.
11453 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11454 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11455 (visium_constant_alignment): New function.
11456 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11457 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11458 (xtensa_constant_alignment): New function.
11459 * system.h (CONSTANT_ALIGNMENT): Poison.
11461 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11463 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11464 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11465 (rs6000_builtin_valid_without_lhs): New helper function.
11466 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11467 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11469 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11471 * target.h (vec_perm_indices): Use unsigned short rather than
11473 (auto_vec_perm_indices): Likewise.
11474 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11475 Use unsigned int rather than unsigned char.
11476 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11478 2017-09-25 Richard Biener <rguenther@suse.de>
11480 * cfgloop.h (sort_sibling_loops): Declare.
11481 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11482 (sort_sibling_loops): New function sorting the sibling loop list
11484 * graphite.c (graphite_transform_loops): Sort sibling loops.
11486 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11488 * target.def (vec_perm_const_ok): Change sel parameter to
11490 * optabs-query.c (can_vec_perm_p): Update accordingly.
11491 * doc/tm.texi: Regenerate.
11492 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11493 auto_vec_perm_indices and remove separate nelt field.
11494 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11495 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11496 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11497 (aarch64_expand_vec_perm_const): Update accordingly.
11498 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11499 to vec_perm_indices.
11500 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11501 auto_vec_perm_indices and remove separate nelt field.
11502 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11503 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11504 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11506 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11507 to vec_perm_indices.
11508 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11509 sel to vec_perm_indices.
11510 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11511 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11512 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11514 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11517 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11520 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11521 on the FUNCTION_DECL function context if it has a DIE that is a
11524 2017-09-25 Richard Biener <rguenther@suse.de>
11526 PR tree-optimization/82285
11527 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11530 2017-09-25 Tom de Vries <tom@codesourcery.com>
11534 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11537 2017-09-25 Richard Biener <rguenther@suse.de>
11539 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11540 ISL errors other than isl_error_quota happen. Dump if the
11541 schedule is the same.
11542 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11543 errors instead of aborting inside ISL.
11545 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11548 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11549 of libgcc_eh for m64.
11550 * config/i386/darwin64.h: Likewise.
11552 2017-09-25 Richard Biener <rguenther@suse.de>
11554 PR middle-end/82144
11555 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11556 attribute for incomplete types nor twice for complete ones.
11558 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11561 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11562 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11565 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11568 * config/i386/i386.opt (mprefer-avx256): Use
11569 ix86_target_flags variable.
11570 * config/i386/i386.c (ix86_target_string): Move
11571 -mprefer-avx256 to flag2_opts.
11573 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11575 PR middle-end/35691
11576 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11577 and x != -1 | y != -1 into (x & y) != -1.
11579 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11581 * config.gcc: Add new case statement to set
11582 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11583 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11584 s390x-*-linux* case statements. Added aarch64 to the list of
11585 supported architectures.
11587 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11589 PR tree-optimization/82289
11590 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11591 STMT_VINFO_RELEVANT_P.
11593 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11598 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11599 (extract_range_from_binary_expr_1): Don't call
11600 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11602 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11603 Alan Hayward <alan.hayward@arm.com>
11604 David Sherwood <david.sherwood@arm.com>
11606 * target.def (preferred_vector_alignment): New hook.
11607 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11609 * doc/tm.texi: Regenerate.
11610 * targhooks.h (default_preferred_vector_alignment): Declare.
11611 * targhooks.c (default_preferred_vector_alignment): New function.
11612 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11614 (DR_TARGET_ALIGNMENT): New macro.
11615 (aligned_access_p): Update commentary.
11616 (vect_known_alignment_in_bytes): New function.
11617 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11619 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11620 Calculate the misalignment based on the target alignment rather than
11622 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11623 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11624 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11625 the target alignment, rather than masking the element misalignment
11626 with the number of elements in a vector. Also use the target
11627 alignment when calculating the maximum number of peels.
11628 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11629 instead of TYPE_ALIGN_UNIT.
11630 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11631 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11632 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11633 (vect_create_data_ref_ptr): Likewise.
11634 (vect_setup_realignment): Realign by ANDing with
11635 -DR_TARGET_MISALIGNMENT.
11636 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11637 the number of peels based on DR_TARGET_ALIGNMENT.
11638 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11639 with the guaranteed alignment boundary when deciding whether
11641 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11642 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11643 (ensure_base_align): Remove stmt_info parameter. Get the
11644 target base alignment from DR_TARGET_ALIGNMENT.
11645 (vectorizable_store): Update call accordingly. Interpret
11646 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11648 (vectorizable_load): Likewise.
11650 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11651 Alan Hayward <alan.hayward@arm.com>
11652 David Sherwood <david.sherwood@arm.com>
11654 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11655 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11656 (vect_enhance_data_refs_alignment): Likewise.
11658 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11660 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11661 error. Only quit immediately if parsing is complete.
11662 (BEGIN): Initialize fatal_err and parse_done.
11663 (begin fpu, end fpu): Check number of arguments.
11664 (begin arch, end arch): Likewise.
11665 (begin cpu, end cpu): Likewise.
11666 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11667 (optalias): Likewise.
11669 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11671 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11672 * config/arm/arm-isa.h: Delete. Move definitions to ...
11673 * arm-cpus.in: ... here. Use new feature and fgroup values.
11674 * config/arm/arm.c (arm_option_override): Use lower case for feature
11676 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11677 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11678 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11680 (print_isa_bits_for): New function.
11681 (gen_isa): New function.
11682 (gen_comm_data): Use print_isa_bits_for.
11683 (define feature): New keyword.
11684 (define fgroup): New keyword.
11685 * config/arm/t-arm (TM_H): Remove.
11686 (GTM_H): Add arm-isa.h.
11687 (arm-isa.h): Add rule to generate file.
11688 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11689 case for feature bit names.
11691 2017-09-22 Richard Biener <rguenther@suse.de>
11693 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11695 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11696 print of no dependency loops ...
11697 * graphite.c (graphite_transform_loops): ... here.
11698 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11700 (same_close_phi_node, remove_duplicate_close_phi,
11701 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11702 (canonicalize_loop_closed_ssa): ... here and simplify.
11703 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11704 (optimize_isl): Use dump_printf_loc to tell when we stopped
11705 optimizing because of an ISL timeout.
11707 2017-09-22 Richard Biener <rguenther@suse.de>
11709 PR tree-optimization/82291
11710 * tree-if-conv.c (predicate_mem_writes): Make sure to
11711 remove writes in blocks predicated with false.
11713 2017-09-22 Richard Biener <rguenther@suse.de>
11715 * sese.c: Include cfganal.h.
11716 (if_region_set_false_region): Remove.
11717 (create_if_region_on_edge): Likewise.
11718 (move_sese_in_condition): Re-implement without destroying
11721 2017-09-22 Richard Biener <rguenther@suse.de>
11723 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11724 Verify both BBs contain loop PHI nodes before dispatching to
11725 copy_loop_phi_args.
11726 (graphite_regenerate_ast_isl): Do not recompute dominators,
11727 do not verify three times. Restructure for clarity.
11728 * graphite-scop-detection.c (same_close_phi_node,
11729 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11730 defined_in_loop_p, canonicalize_loop_closed_ssa,
11731 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11732 checking and SSA rewrite, move to ...
11733 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11734 (graphite_initialize): Do not pass in ctx, do not reset the
11735 SCEV cache, compute only dominators.
11736 (graphite_transform_loops): Allocate ISL ctx after
11737 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11738 Maintain post-dominators only around build_scops.
11739 * sese.c (if_region_set_false_region): Make static. Free
11740 and recompute dominators.
11741 (move_sese_in_condition): Assert we don't get called with
11742 post-dominators computed.
11743 * sese.h (if_region_set_false_region): Remove.
11745 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11747 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11748 mode attribute for TARGET_AVX512VL.
11750 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11752 * config/i386/i386.opt (mprefer-avx256): New option.
11753 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11755 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11756 for TARGET_PREFER_AVX256.
11757 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11759 2017-09-21 Jeff Law <law@redhat.com>
11761 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11762 Fix dump output if the only stack space is for pushed registers.
11764 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11766 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11769 2017-09-21 Martin Sebor <msebor@redhat.com>
11772 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11773 code (in C++) or code that triggers warnings.
11775 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11777 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11779 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11781 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11782 * sched-rgn.c: Ditto.
11783 * sel-sched-ir.c: Ditto.
11785 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11787 * toplev.h (set_random_seed): Adjust return type.
11788 * toplev.c (init_local_tick): Move eager initialization of random_seed
11789 to get_random_seed. Adjust comment.
11790 (init_random_seed): Inline to get_random_seed, delete.
11791 (get_random_seed): Initialize random_seed lazily.
11792 (set_random_seed): Do not return previous value.
11793 (print_switch_value): Do not call get_random_seed.
11795 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11797 * cgraph.c (delete_function_version): New, broken out from...
11798 (cgraph_node::delete_function_version): ...here. Rename to
11799 cgraph_node::delete_function_version_by_decl. Update all uses.
11800 (cgraph_node::remove): Call delete_function_version.
11802 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11805 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11806 VAR_DECLs to which addressable non-volatile parameters are mapped
11807 and for id->retvar after the return value assignment. Clear
11808 id->retval and id->retbnd after inlining.
11810 2017-09-21 Richard Biener <rguenther@suse.de>
11812 PR tree-optimization/82276
11813 PR tree-optimization/82244
11814 * tree-vrp.c (build_assert_expr_for): Set
11815 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11817 (remove_range_assertions): Revert earlier change.
11819 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11822 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11824 2017-09-21 Richard Biener <rguenther@suse.de>
11826 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11827 Restore valid IL after code generation errors.
11828 * graphite.c (graphite_transform_loops): Diagnose code
11829 generation issues as MSG_MISSED_OPTIMIZATION and continue
11830 with processing SCOPs.
11832 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11833 Alan Hayward <alan.hayward@arm.com>
11834 David Sherwood <david.sherwood@arm.com>
11836 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11837 rather than choosing between plus_constant and gen_rtx_<CODE>.
11838 * expr.c (emit_push_insn): Likewise.
11839 (expand_expr_real_2): Likewise.
11841 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11842 Alan Hayward <alan.hayward@arm.com>
11843 David Sherwood <david.sherwood@arm.com>
11845 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11847 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11848 Alan Hayward <alan.hayward@arm.com>
11849 David Sherwood <david.sherwood@arm.com>
11851 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11852 calling tree_to_uhwi.
11854 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11855 Alan Hayward <alan.hayward@arm.com>
11856 David Sherwood <david.sherwood@arm.com>
11858 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11859 INTEGER_CST rather than a negative test for ADDR_EXPR.
11861 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11862 Alan Hayward <alan.hayward@arm.com>
11863 David Sherwood <david.sherwood@arm.com>
11865 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11866 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11868 2017-09-21 Richard Biener <rguenther@suse.de>
11870 PR tree-optimization/71351
11871 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11872 graphite_create_new_loop_guard): Remove, fold remaining parts
11874 (translate_isl_ast_node_for): ... here and simplify.
11876 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11879 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11880 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11881 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11882 alternative always use QI mode, for -Os imov (=R,R) alternative
11883 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11886 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11887 Jeff Law <law@redhat.com>
11889 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11890 (allocate_stack_space): New function, partially extracted from
11891 s390_emit_prologue.
11892 (s390_emit_prologue): Track offset to most recent stack probe.
11893 Code to allocate space moved into allocate_stack_space.
11894 Dump actions when no stack is allocated.
11895 (s390_prologue_plus_offset): New function.
11896 (s390_emit_stack_probe): Likewise.
11898 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11900 * common.opt (Wa, Wl, Wp, g, gz=): Add
11902 (gno-column-info): Remove.
11903 (gcolumn-info): Drop RejectNegative.
11904 (gno-): New prefix.
11905 (gno-record-gcc-switches): Remove.
11906 (grecord-gcc-switches): Drop RejectNegative.
11907 (gno-split-dwarf): Remove.
11908 (gsplit-dwarf): Drop RejectNegative.
11909 (gno-strict-dwarf): Remove.
11910 (gstrict-dwarf): Drop RejectNegative.
11911 * config/darwin.opt (gfull, gused): Add RejectNegative.
11912 * dwarf2out.c (gen_producer_string): Drop
11913 gno-record-gcc-switches handler.
11914 * optc-gen.awk: Add g to prefixes with negative forms.
11915 * opts-common.c (remapping_prefix_p): New.
11916 (find_opt): Check it.
11917 (generate_canonical_option): Test g prefix.
11918 (option_map): Add -gno- mapping.
11919 (add_misspelling_candidates): Check remapping_prefix_p.
11921 2017-09-20 Jeff Law <law@redhat.com>
11923 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11924 thinko in stack clash protection support.
11926 * explow.c (compute_stack_clash_protection_loop_data): Use
11927 CONST_INT_P instead of explicit test. Verify object is a
11928 CONST_INT_P before looking at INTVAL.
11929 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11930 instead of explicit test.
11932 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11935 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11936 address instead of to r1 and r11.
11938 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11940 * config.gcc: Support "knm".
11941 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11942 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11944 * config/i386/i386.c (m_KNM): Define.
11945 (processor_target_table): Add "knm".
11947 (ix86_option_override_internal): Add "knm".
11948 (ix86_issue_rate): Add PROCESSOR_KNM.
11949 (ix86_adjust_cost): Ditto.
11950 (ia32_multipass_dfa_lookahead): Ditto.
11951 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11952 (fold_builtin_cpu): Add M_INTEL_KNM.
11953 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11954 (processor_type): Add PROCESSOR_KNM.
11955 * config/i386/x86-tune.def: Add m_KNM.
11956 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11958 2017-09-20 Richard Biener <rguenther@suse.de>
11960 PR tree-optimization/80213
11961 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11962 are allowed in empty BBs as well.
11963 (canonicalize_loop_closed_ssa): Also look for other complex
11965 (scop_detection::get_sese): Include the loop-closed PHI block
11967 (scop_detection::merge_sese): Remove code adding extra blocks.
11968 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11969 (build_scops): Assert the final returned scop is invalid.
11971 2017-09-20 Richard Biener <rguenther@suse.de>
11973 PR tree-optimization/82264
11974 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11976 (vn_phi_lookup): Likewise.
11977 (vn_phi_insert): Likewise.
11979 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11981 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11982 that fits into uhwi or shwi, add DW_AT_const_value regardless
11983 of early_dwarf without going through RTL, using add_AT_unsigned
11986 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11987 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11988 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11989 (DEBUG_MACRO_SECTION): Likewise.
11990 (DEBUG_LTO_MACRO_SECTION): Likewise.
11991 (DEBUG_STR_DWO_SECTION): Likewise.
11992 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11993 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11994 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11995 (DEBUG_LTO_LINE_STR_SECTION): Define.
11996 (init_sections_and_labels): Initialize debug_line_str_section
11997 variable. Initialize debug_loc_section for -gdwarf-5 to
11998 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12000 2017-09-20 Richard Biener <rguenther@suse.de>
12002 * graphite-sese-to-poly.c (extract_affine): Properly handle
12003 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12005 2017-09-20 Richard Biener <rguenther@suse.de>
12007 PR tree-optimization/81373
12008 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12009 Force SESE live-out defs to be handled even if they are
12012 2017-09-19 Jeff Law <law@redhat.com>
12014 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12015 nothing for stack adjustments with REG_STACK_CHECK.
12016 * sched-deps.c (parse_add_or_inc): Reject insns with
12017 REG_STACK_CHECK from dependency breaking.
12018 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12019 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12020 * reg-notes.def (STACK_CHECK): New note.
12022 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12023 (ix86_expand_prologue): Dump stack clash info as needed.
12024 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12026 * function.c (dump_stack_clash_frame_info): New function.
12027 * function.h (dump_stack_clash_frame_info): Prototype.
12028 (enum stack_clash_probes): New enum.
12030 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12031 flag_stack_clash_protection.
12032 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12033 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12034 (arm_frame_pointer_required): Likewise.
12035 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12036 (ia64_expand_prologue): Likewise.
12037 * config/mips/mips.c (mips_expand_prologue): Likewise.
12038 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12039 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12040 (sparc_flat_expand_prologue): Likewise.
12041 * config/spu/spu.c (spu_expand_prologue): Likewise.
12043 * explow.c: Include "params.h".
12044 (anti_adjust_stack_and_probe_stack_clash): New function.
12045 (get_stack_check_protect): Likewise.
12046 (compute_stack_clash_protection_loop_data): Likewise.
12047 (emit_stack_clash_protection_loop_start): Likewise.
12048 (emit_stack_clash_protection_loop_end): Likewise.
12049 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12050 Use anti_adjust_stack_and_probe_stack_clash.
12051 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12052 (emit_stack_clash_protection_loop_start): Likewise.
12053 (emit_stack_clash_protection_loop_end): Likewise.
12054 * rtl.h (get_stack_check_protect): Prototype.
12055 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12056 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12057 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12059 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12061 * doc/tm.texi: Rebuilt.
12062 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12063 get_stack_check_protect.
12064 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12065 * config/arm/arm.c (arm_expand_prologue): Likewise.
12066 (arm_frame_pointer_required): Likewise.
12067 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12068 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12069 * config/mips/mips.c (mips_expand_prologue): Likewise.
12070 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12071 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12072 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12073 (sparc_flat_expand_prologue): Likewise.
12075 * common.opt (-fstack-clash-protection): New option.
12076 * flag-types.h (enum stack_check_type): Note difference between
12077 -fstack-check= and -fstack-clash-protection.
12078 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12079 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12080 * toplev.c (process_options): Issue warnings/errors for cases
12081 not handled with -fstack-clash-protection.
12082 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12083 (-fstack-check): Note additional problem with -fstack-check=generic.
12084 Note that -fstack-check is primarily for Ada and refer users
12085 to -fstack-clash-protection for stack-clash-protection.
12086 Document new params for stack clash protection.
12088 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12090 * config/i386/i386.c (ix86_split_long_move): Do not handle
12091 address used for LEA in a special way.
12093 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12095 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12097 2017-09-19 Martin Sebor <msebor@redhat.com>
12100 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12101 of incompatible types.
12103 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12105 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12106 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12107 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12108 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12110 2017-09-19 Richard Biener <rguenther@suse.de>
12112 PR tree-optimization/82244
12113 * tree-vrp.c (remove_range_assertions): Do not propagate
12114 a constant to abnormals but replace the assert with a copy.
12116 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12118 PR rtl-optimization/57878
12119 PR rtl-optimization/68988
12120 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12121 avoidance test involving non_reload_pseudos. Move frequency test
12122 below the general fragmentation avoidance test.
12124 2017-09-19 Richard Biener <rguenther@suse.de>
12126 PR tree-optimization/69728
12127 * graphite-sese-to-poly.c (schedule_error): New global.
12128 (add_loop_schedule): Handle empty domain by failing the
12130 (build_original_schedule): Handle schedule_error.
12132 2017-09-19 Richard Biener <rguenther@suse.de>
12134 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12135 Do not iterate to sibling loops but only to siblings of inner
12138 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12141 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12142 that the registers are different.
12144 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12146 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12147 to processor_model and "amdfam17h" to arch_names_table.
12148 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12150 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12153 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12155 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12156 Alan Hayward <alan.hayward@arm.com>
12157 David Sherwood <david.sherwood@arm.com>
12159 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12161 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12163 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12164 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12165 vect_schedule_slp_instance.
12166 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12167 Update call to vect_slp_analyze_node_operations. Simplify return
12169 (vect_slp_analyze_bb_1): Update call accordingly.
12170 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12171 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12172 (vect_schedule_slp): Update call accordingly.
12174 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12175 Alan Hayward <alan.hayward@arm.com>
12176 David Sherwood <david.sherwood@arm.com>
12178 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12179 with types that aren't in fact scalar.
12181 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12183 * tree-vect-slp.c (vect_record_max_nunits): New function,
12185 (vect_build_slp_tree_1): ...here.
12186 (vect_build_slp_tree_2): Call it for phis too.
12188 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12190 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12191 to vect_get_vec_def_for_operand when getting the mask operand.
12193 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12194 Alan Hayward <alan.hayward@arm.com>
12195 David Sherwood <david.sherwood@arm.com>
12197 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12200 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12201 Alan Hayward <alan.hayward@arm.com>
12202 David Sherwood <david.sherwood@arm.com>
12204 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12205 calculation for vector booleans.
12207 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12208 Alan Hayward <alan.hayward@arm.com>
12209 David Sherwood <david.sherwood@arm.com>
12211 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12213 (vect_transform_stmt): ...here.
12214 (vect_analyze_stmt): Use it instead of calling
12215 vectorizable_live_operation directly.
12217 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12219 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12220 non-SIMT targets in acc vector loops.
12222 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12224 * configure.ac: Add arc and check if assembler supports gdwarf2.
12225 * configure: Regenerate.
12227 2017-09-18 Richard Biener <rguenther@suse.de>
12229 PR tree-optimization/82220
12230 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12231 epilogue niters from the min_profitable_iters compute.
12233 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12236 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12237 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12238 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12240 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12243 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12244 switching to a new text section.
12246 2017-09-18 Richard Biener <rguenther@suse.de>
12248 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12250 (build_alias_set): Reject aliases with no access function.
12252 2017-09-18 Richard Biener <rguenther@suse.de>
12254 PR tree-optimization/79622
12255 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12257 (build_cross_bb_scalars_use): Likewise.
12259 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12261 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12263 2017-09-18 Alan Modra <amodra@gmail.com>
12266 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12267 stack_pointer_rtx for count 0. Update comments. Break up
12268 large rtl expression.
12270 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12272 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12273 Increase to 20 bytes.
12274 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12275 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12276 or avx version of the stub.
12278 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12281 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12282 compute the minimum stack alignment. Also update preferred stack
12283 boundary for leaf functions.
12285 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12287 PR tree-optimization/82228
12288 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12291 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12293 * common/config/nds32/nds32-common.c
12294 (nds32_option_optimization_table): Refine formatting.
12295 (nds32_option_optimization_table): Use -fsched-pressure and
12296 -fomit-frame-pointer for specific optimization level.
12298 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12300 * config/nds32/nds32.c: Refine formatting and comments.
12301 * config/nds32/nds32.h: Likewise.
12302 * config/nds32/nds32.md: Likewise.
12303 * config/nds32/nds32-cost.c: Likewise.
12304 * config/nds32/nds32-isr.c: Likewise.
12305 * config/nds32/nds32-md-auxiliary.c: Likewise.
12306 * config/nds32/nds32-multiple.md: Likewise.
12307 * config/nds32/nds32-predicates.c: Likewise.
12309 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12310 Jakub Jelinek <jakub@redhat.com>
12312 Add support for -std=c++2a.
12313 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12315 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12317 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12320 * doc/extend.texi (Common Function Attributes): Add
12321 references to ARM, AArch64, and S/390 specific attributes.
12322 (Function Specific Option Pragmas): Add AArch64 and S/390
12323 to list of back ends that support the target pragma.
12325 2017-09-15 Nathan Sidwell <nathan@acm.org>
12327 * doc/standards.texi: Fix C++17 description. Update URLs for
12330 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12332 * common.opt (Wcast-align=strict): New warning option.
12333 * doc/invoke.texi: Document -Wcast-align=strict.
12335 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12337 * cgraph.h (cgraph_thunk_info): Add comments.
12338 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12339 assert for VIRTUAL_* arguments stricter.
12341 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12343 PR tree-optimization/71026
12344 * match.pd: Move RDIV patterns from fold-const.c
12345 * fold-const.c (distribute_real_division): Removed.
12346 (fold_binary_loc): Remove calls to distribute_real_divison.
12348 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12350 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12351 c++1z and gnu++1z as deprecated. Change other references to
12352 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12353 Change -Wc++1z-compat to -Wc++17-compat.
12354 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12355 * dwarf2out.c (highest_c_language): Handle C++17.
12356 (gen_compile_unit_die): Likewise.
12358 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12360 PR rtl-optimization/82192
12361 * combine.c (make_extraction): Don't look through non-paradoxical
12362 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12365 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12366 Alan Hayward <alan.hayward@arm.com>
12367 David Sherwood <david.sherwood@arm.com>
12369 * target.def (function_arg_offset): New hook.
12370 * targhooks.h (default_function_arg_offset): Declare.
12371 * targhooks.c (default_function_arg_offset): New function.
12372 * function.c (locate_and_pad_parm): Use
12373 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12374 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12375 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12376 * doc/tm.texi: Regenerate.
12377 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12378 * config/spu/spu.c (spu_function_arg_offset): New function.
12379 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12380 * system.h (FUNCTION_ARG_OFFSET): Poison.
12382 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12383 Alan Hayard <alan.hayward@arm.com>
12384 David Sherwood <david.sherwood@arm.com>
12386 * target.def (truly_noop_truncation): New hook.
12387 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12388 than TRULY_NOOP_TRUNCATION.
12389 * hooks.h (hook_bool_uint_uint_true): Declare.
12390 * hooks.c (hook_bool_uint_uint_true): New function.
12391 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12392 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12393 * doc/tm.texi: Regenerate.
12394 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12395 rather than TRULY_NOOP_TRUNCATION in comments.
12396 (simplify_comparison): Likewise.
12397 (record_truncated_value): Likewise.
12398 * expmed.c (extract_bit_field_1): Likewise.
12399 (extract_split_bit_field): Likewise.
12400 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12401 instead of TRULY_NOOP_TRUNCATION.
12402 * function.c (assign_parm_setup_block): Likewise.
12403 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12404 * rtlhooks.c: Include target.h.
12405 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12406 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12407 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12408 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12409 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12410 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12411 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12412 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12413 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12414 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12415 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12416 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12417 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12418 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12419 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12420 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12421 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12422 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12423 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12424 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12425 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12426 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12427 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12428 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12429 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12430 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12431 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12432 rather than TRULY_NOOP_TRUNCATION in comments.
12433 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12434 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12435 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12436 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12437 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12438 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12439 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12440 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12441 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12442 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12443 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12444 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12445 rather than TRULY_NOOP_TRUNCATION in comments.
12446 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12447 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12448 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12449 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12450 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12451 TRULY_NOOP_TRUNCATION condition.
12452 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12453 (TRULY_NOOP_TRUNCATION): Delete.
12454 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12455 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12456 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12457 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12458 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12459 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12460 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12461 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12462 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12463 rather than TRULY_NOOP_TRUNCATION in comments.
12464 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12465 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12466 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12467 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12468 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12469 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12471 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12474 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12475 (*cmp_ior): Likewise.
12476 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12477 (*ior_scc_scc_cmp): Likewise.
12478 (*and_scc_scc): Likewise.
12479 (*and_scc_scc_cmp): Likewise.
12481 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12482 Alan Hayard <alan.hayward@arm.com>
12483 David Sherwood <david.sherwood@arm.com>
12485 * target.def (can_change_mode_class): New hook.
12486 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12487 (hard_regno_nregs): Likewise.
12488 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12489 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12490 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12491 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12492 (LOAD_EXTEND_OP): Update accordingly.
12493 * doc/tm.texi: Regenerate.
12494 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12495 CANNOT_CHANGE_MODE_CLASS.
12496 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12497 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12498 * combine.c (simplify_set): Update accordingly.
12499 * emit-rtl.c (validate_subreg): Likewise.
12500 * recog.c (general_operand): Likewise.
12501 * regcprop.c (mode_change_ok): Likewise.
12502 * reload1.c (choose_reload_regs): Likewise.
12503 (inherit_piecemeal_p): Likewise.
12504 * rtlanal.c (simplify_subreg_regno): Likewise.
12505 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12506 instead of CANNOT_CHANGE_MODE_CLASS.
12507 (reload_cse_simplify_operands): Likewise.
12508 * reload.c (push_reload): Use targetm.can_change_mode_class
12509 instead of CANNOT_CHANGE_MODE_CLASS.
12510 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12511 REG_CANNOT_CHANGE_MODE_P.
12512 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12513 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12514 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12515 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12516 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12517 (arm_can_change_mode_class): New function.
12518 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12519 than CANNOT_CHANGE_MODE_CLASS in comments.
12520 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12521 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12522 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12523 (ix86_can_change_mode_class): ...this new function, inverting the
12524 sense of the return value.
12525 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12526 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12527 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12528 (ia64_can_change_mode_class): New function.
12529 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12530 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12531 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12532 (m32c_can_change_mode_class): ...this new function, inverting the
12533 sense of the return value.
12534 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12535 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12536 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12537 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12538 (mips_can_change_mode_class): ...this new function, inverting the
12539 sense of the return value.
12540 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12541 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12542 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12543 (msp430_can_change_mode_class): New function.
12544 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12545 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12546 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12547 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12548 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12549 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12550 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12551 (pa_cannot_change_mode_class): Replace with...
12552 (pa_can_change_mode_class): ...this new function, inverting the
12553 sense of the return value.
12554 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12555 than CANNOT_CHANGE_MODE_CLASS in comments.
12556 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12557 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12558 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12559 (pdp11_cannot_change_mode_class): Replace with...
12560 (pdp11_can_change_mode_class): ...this new function, inverting the
12561 sense of the return value.
12562 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12563 * config/powerpcspe/powerpcspe-protos.h
12564 (rs6000_cannot_change_mode_class_ptr): Delete.
12565 * config/powerpcspe/powerpcspe.c
12566 (rs6000_cannot_change_mode_class_ptr): Delete.
12567 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12568 (rs6000_option_override_internal): Assign to
12569 targetm.can_change_mode_class instead of
12570 rs6000_cannot_change_mode_class_ptr.
12571 (rs6000_cannot_change_mode_class): Replace with...
12572 (rs6000_can_change_mode_class): ...this new function, inverting the
12573 sense of the return value.
12574 (rs6000_debug_cannot_change_mode_class): Replace with...
12575 (rs6000_debug_can_change_mode_class): ...this new function.
12576 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12577 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12578 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12579 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12580 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12582 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12584 (rs6000_option_override_internal): Assign to
12585 targetm.can_change_mode_class instead of
12586 rs6000_cannot_change_mode_class_ptr.
12587 (rs6000_cannot_change_mode_class): Replace with...
12588 (rs6000_can_change_mode_class): ...this new function, inverting the
12589 sense of the return value.
12590 (rs6000_debug_cannot_change_mode_class): Replace with...
12591 (rs6000_debug_can_change_mode_class): ...this new function.
12592 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12593 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12594 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12595 (s390_can_change_mode_class): ...this new function, inverting the
12596 sense of the return value.
12597 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12598 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12599 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12600 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12601 (sh_cannot_change_mode_class): Replace with...
12602 (sh_can_change_mode_class): ...this new function, inverting the
12603 sense of the return value.
12604 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12605 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12606 (sparc_can_change_mode_class): New function.
12607 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12608 * config/spu/spu.c (spu_can_change_mode_class): New function.
12609 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12610 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12611 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12612 (visium_can_change_mode_class): New function.
12613 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12615 2017-09-15 Richard Biener <rguenther@suse.de>
12617 PR tree-optimization/82217
12618 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12619 but not undefined case.
12621 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12624 * postreload.c (reload_cse_simplify_operands): Skip
12625 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12627 2017-09-15 Richard Biener <rguenther@suse.de>
12629 PR tree-optimization/68823
12630 * graphite-scop-detection.c (build_alias_set): If we have a
12631 possible dependence check whether we can handle them by just
12632 looking at the DRs DR_ACCESS_FNs.
12633 (build_scops): If build_alias_set fails, fail the SCOP.
12635 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12637 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12638 to support float128 built-in functions that require the ISA 3.0
12640 (BU_FLOAT128_3_HW): Likewise.
12641 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12642 built-in functions.
12643 (FMAF128): Likewise.
12645 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12646 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12647 floating point instructions.
12648 (rs6000_invalid_builtin): Likewise.
12649 (rs6000_builtin_mask_names): Likewise.
12650 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12651 (RS6000_BTM_FLOAT128_HW): Likewise.
12652 (RS6000_BTM_COMMON): Likewise.
12653 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12655 * doc/extend.texi (RS/6000 built-in functions): Document the
12656 IEEE 128-bit floating point square root and fused multiply-add
12657 built-in functions.
12659 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12661 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12662 reg (r2) isn't in the set of registers defined in the prologue.
12664 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12665 Alan Hayward <alan.hayward@arm.com>
12666 David Sherwood <david.sherwood@arm.com>
12668 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12669 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12670 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12671 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12672 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12674 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12675 max_vectorization_factor.
12676 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12678 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12679 Alan Hayward <alan.hayward@arm.com>
12680 David Sherwood <david.sherwood@arm.com>
12682 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12683 (vect_worthwhile_without_simd_p): Declare.
12684 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12685 (vectorizable_reduction): Use it.
12686 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12687 (vectorizable_operation): Likewise.
12689 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12690 Alan Hayward <alan.hayward@arm.com>
12691 David Sherwood <david.sherwood@arm.com>
12693 * tree-vectorizer.h (vect_get_num_copies): New function.
12694 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12695 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12696 (vectorizable_induction): Likewise.
12697 (vectorizable_live_operation): Likewise.
12698 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12699 (vectorizable_bswap): Likewise.
12700 (vectorizable_call): Likewise.
12701 (vectorizable_conversion): Likewise.
12702 (vectorizable_assignment): Likewise.
12703 (vectorizable_shift): Likewise.
12704 (vectorizable_operation): Likewise.
12705 (vectorizable_store): Likewise.
12706 (vectorizable_load): Likewise.
12707 (vectorizable_condition): Likewise.
12708 (vectorizable_comparison): Likewise.
12709 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12711 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12712 Alan Hayward <alan.hayward@arm.com>
12713 David Sherwood <david.sherwood@arm.com>
12715 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12716 of vect_init_vector.
12718 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12719 Alan Hayward <alan.hayward@arm.com>
12720 David Sherwood <david.sherwood@arm.com>
12722 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12723 an inline wrapper that provides a location.
12724 (gimple_build_vector): Likewise.
12725 * gimple-fold.c (gimple_build_vector_from_val): New function.
12726 (gimple_build_vector): Likewise.
12727 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12728 functions to build the initial value. Always return a gimple value.
12729 (get_initial_defs_for_reduction): Likewise. Only compute
12731 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12732 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12733 (vectorizable_induction): Use gimple_build_vector rather than
12736 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12737 Alan Hayward <alan.hayward@arm.com>
12738 David Sherwood <david.sherwood@arm.com>
12740 * target.h (vec_perm_indices): New typedef.
12741 (auto_vec_perm_indices): Likewise.
12742 * optabs-query.h: Include target.h
12743 (can_vec_perm_p): Take a vec_perm_indices *.
12744 * optabs-query.c (can_vec_perm_p): Likewise.
12745 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12746 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12747 * tree-vect-generic.c (lower_vec_perm): Likewise.
12748 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12749 (vect_grouped_load_supported): Likewise.
12750 (vect_shift_permute_load_chain): Likewise.
12751 (vect_permute_store_chain): Use auto_vec_perm_indices.
12752 (vect_permute_load_chain): Likewise.
12753 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12754 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12755 Update uses of can_vec_perm_p.
12756 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12757 mode with a number of elements. Take a vec_perm_indices *.
12758 (vect_create_epilog_for_reduction): Update accordingly.
12759 Use auto_vec_perm_indices.
12760 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12761 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12762 (vect_transform_slp_perm_load): Likewise.
12763 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12764 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12765 (vect_gen_perm_mask_checked): Likewise.
12766 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12767 (vect_gen_perm_mask_checked): Likewise.
12768 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12769 (vectorizable_store): Likewise.
12770 (vectorizable_load): Likewise.
12771 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12772 (vectorizable_bswap): Likewise.
12774 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12775 Alan Hayward <alan.hayward@arm.com>
12776 David Sherwood <david.sherwood@arm.com>
12778 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12779 * tree.c (build_vector): Likewise.
12780 (build_vector_from_ctor): Update accordingly.
12781 (build_vector_from_val): Likewise.
12782 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12783 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12784 * tree-vect-generic.c (add_rshift): Likewise.
12785 (expand_vector_divmod): Likewise.
12786 (optimize_vector_constructor): Likewise.
12787 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12788 (vect_transform_slp_perm_load): Likewise.
12789 (vect_schedule_slp_instance): Likewise.
12790 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12791 (vectorizable_call): Likewise.
12792 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12793 * expmed.c (make_tree): Likewise.
12794 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12795 a vector passed to build_vector.
12796 (fold_convert_const): Likewise.
12797 (exact_inverse): Likewise.
12798 (fold_ternary_loc): Likewise.
12799 (fold_relational_const): Likewise.
12800 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12801 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12802 (const_unop): Likewise. Store the reduction accumulator in a
12803 variable rather than an array.
12804 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12805 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12806 the new vector, rather than constructing it after the input arrays.
12807 (native_interpret_vector): Use auto_vec<tree> when building
12808 a vector passed to build_vector. Add elements in order.
12809 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12810 auto_vec<tree> when building a vector passed to build_vector.
12811 (vect_create_epilog_for_reduction): Likewise.
12812 (vectorizable_induction): Likewise.
12813 (get_initial_def_for_reduction): Likewise. Fix indentation of
12815 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12817 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12818 passed to build_vector.
12820 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12821 Alan Hayward <alan.hayward@arm.com>
12822 David Sherwood <david.sherwood@arm.com>
12824 * tree-core.h (tree_base::u): Add an "nelts" field.
12825 (tree_vector): Use VECTOR_CST_NELTS as the length.
12826 * tree.c (tree_size): Likewise.
12827 (make_vector): Initialize VECTOR_CST_NELTS.
12828 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12829 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12830 TYPE_VECTOR_SUBPARTS.
12831 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12832 as the number of units, setting it from VECTOR_CST_NELTS.
12833 (const_vector_from_tree): Likewise.
12834 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12835 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12836 (fold_negate_expr_1): Likewise.
12837 (fold_convert_const): Likewise.
12838 (const_binop): Likewise. Differentiate the number of output and
12840 (const_unop): Likewise.
12841 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12842 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12843 in cases that did the opposite.
12845 2017-09-14 Richard Biener <rguenther@suse.de>
12847 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12850 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12852 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12854 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12857 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12858 if and where to split a bb, except for splitting before debug insn
12859 sequences followed by non-label real insn. Delete debug insns
12860 in between basic blocks.
12862 * combine.c (make_compound_operation_int): Formatting fixes.
12864 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12865 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12866 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12867 * config/sol2.h (LINK_EH_SPEC): Likewise.
12868 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12869 * config/s390/linux.h (LINK_SPEC): Likewise.
12870 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12871 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12872 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12873 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12874 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12875 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12876 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12877 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12879 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12881 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12883 (ENDFILE_LINUX_SPEC): Likewise.
12884 (LINK_EH_SPEC): Likewise.
12885 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12886 (LINK_OS_LINUX_SPEC32): Likewise.
12887 (LINK_OS_LINUX_SPEC64): Likewise.
12888 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12889 (LINK_OS_LINUX_SPEC): Likewise.
12891 2017-09-13 Martin Liska <mliska@suse.cz>
12893 PR middle-end/82154
12894 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12895 CASE_HIGH is NULL_TREE.
12897 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12898 Alan Hayward <alan.hayward@arm.com>
12899 David Sherwood <david.sherwood@arm.com>
12901 * target.def (secondary_memory_needed): New hook.
12902 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12903 instead of SECONDARY_MEMORY_NEEDED.
12904 (secondary_memory_needed_mode): Likewise.
12905 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12906 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12907 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12908 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12909 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12910 * doc/tm.texi: Regenerate.
12911 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12912 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12913 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12914 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12915 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12916 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12917 mode argument first and change the reg_class arguments to reg_class_t.
12918 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12919 Make static. Update the call to inline_secondary_memory_needed.
12920 (ix86_register_move_cost): Update the call to
12921 inline_secondary_memory_needed.
12922 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12923 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12925 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12926 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12928 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12929 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12930 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12931 and match hook interface. Add comment from mips.h.
12932 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12933 * config/mmix/mmix.md (truncdfsf2): Refer to
12934 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12936 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12937 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12938 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12939 (pa_secondary_memory_needed): New function.
12940 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12941 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12942 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12943 (pdp11_secondary_memory_needed): Make static and match hook interface.
12944 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12945 * config/powerpcspe/powerpcspe-protos.h
12946 (rs6000_secondary_memory_needed_ptr): Delete.
12947 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12949 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12950 (rs6000_option_override_internal): Assign to
12951 targetm.secondary_memory_needed rather than
12952 rs6000_secondary_memory_needed_ptr.
12953 (rs6000_secondary_memory_needed): Match hook interface.
12954 (rs6000_debug_secondary_memory_needed): Likewise.
12955 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12956 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12957 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12958 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12959 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12960 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12962 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12963 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12964 (rs6000_option_override_internal): Assign to
12965 targetm.secondary_memory_needed rather than
12966 rs6000_secondary_memory_needed_ptr.
12967 (rs6000_secondary_memory_needed): Match hook interface.
12968 (rs6000_debug_secondary_memory_needed): Likewise.
12969 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12970 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12971 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12972 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12973 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12974 (sparc_secondary_memory_needed): New function.
12975 * lra-constraints.c (check_and_process_move): Refer to
12976 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12978 (curr_insn_transform): Likewise.
12979 (process_alt_operands): Use targetm.secondary_memory_needed
12980 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12981 (check_secondary_memory_needed_p): Likewise.
12982 (choose_split_class): Likewise.
12983 * reload.c: Unconditionally include code that was previously
12984 conditional on SECONDARY_MEMORY_NEEDED.
12985 (push_secondary_reload): Use targetm.secondary_memory_needed
12986 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12987 (push_reload): Likewise.
12988 * reload1.c: Unconditionally include code that was previously
12989 conditional on SECONDARY_MEMORY_NEEDED.
12990 (choose_reload_regs): Use targetm.secondary_memory_needed
12991 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12992 (gen_reload): Likewise.
12993 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12995 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12996 Alan Hayward <alan.hayward@arm.com>
12997 David Sherwood <david.sherwood@arm.com>
12999 * target.def (secondary_memory_needed_mode): New hook:
13000 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13001 * targhooks.h (default_secondary_memory_needed_mode): New function.
13002 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13003 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13004 * doc/tm.texi: Regenerate.
13005 * lra-constraints.c (check_and_process_move): Use
13006 targetm.secondary_memory_needed_mode instead of
13007 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13008 (curr_insn_transform): Likewise.
13009 * reload.c (get_secondary_mem): Likewise.
13010 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13011 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13013 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13014 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13015 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13016 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13017 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13019 * config/powerpcspe/powerpcspe-protos.h
13020 (rs6000_secondary_memory_needed_mode): Delete.
13021 * config/powerpcspe/powerpcspe.c
13022 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13023 (rs6000_secondary_memory_needed_mode): Make static.
13024 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13025 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13027 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13029 (rs6000_secondary_memory_needed_mode): Make static.
13030 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13031 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13032 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13033 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13034 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13036 (sparc_secondary_memory_needed_mode): New function.
13037 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13039 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13041 * config/aarch64/constraints.md (Umq): New constraint.
13042 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13044 (mov<mode>): Update condition.
13046 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13048 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13049 when bitposition is the same.
13051 2017-09-13 Richard Biener <rguenther@suse.de>
13053 * dwarf2out.c (output_die_symbol): Remove.
13054 (output_die): Do not output a DIEs symbol.
13056 2017-09-13 Richard Biener <rguenther@suse.de>
13058 PR middle-end/82128
13059 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13060 default-def to avoid breaking iterator update with the weird
13061 interaction with cgraph_update_edges_for_call_stmt_node.
13063 2017-09-13 Richard Biener <rguenther@suse.de>
13065 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13066 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13067 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13068 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13070 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13072 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13073 Disable pc relative literal load irrespective of
13074 TARGET_FIX_ERR_A53_84341 for default.
13076 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13078 * config/sparc/sparc.c (output_return): Output the source location of
13079 the insn in the delay slot, if any.
13080 (output_sibcall): Likewise.
13082 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13085 * common.opt (-static-pie): New alias.
13086 (shared): Negate static-pie.
13087 (-no-pie): Update help text.
13089 (static-pie): New option.
13090 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13091 -static-pie support.
13092 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13093 (LINK_EH_SPEC): Likewise.
13094 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13095 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13096 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13097 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13098 (init_gcc_specs): Likewise.
13099 (init_spec): Likewise.
13100 (display_help): Update help message for -pie.
13101 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13104 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13106 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13107 (movdi_aarch64): Likewise.
13108 (movti_aarch64): Likewise.
13110 2017-09-12 Simon Wright <simon@pushface.org>
13113 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13114 calculation of the minor version, always output as 0.
13116 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13119 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13120 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13121 on it early, rather than manual conversion late. For
13122 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13123 instead of performing manual conversion.
13125 2017-09-12 Carl Love <cel@us.ibm.com>
13127 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13128 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13130 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13131 VMULOSW): Add definitions.
13132 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13133 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13134 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13135 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13136 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13138 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13140 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13142 (movti_aarch64): Likewise.
13143 (movdf_aarch64): Likewise.
13144 (movtf_aarch64): Likewise.
13145 (load_pairdi): Likewise.
13146 (store_pairdi): Likewise.
13147 (load_pairdf): Likewise.
13148 (store_pairdf): Likewise.
13149 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13150 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13151 (ldr_got_small_<mode>): Likewise.
13152 (ldr_got_small_28k_<mode>): Likewise.
13153 (ldr_got_tiny): Likewise.
13154 * config/aarch64/iterators.md (ldst_sz): New.
13155 (ldpstp_sz): Likewise.
13156 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13158 (thunderx_load): Split load_8 to load_16.
13159 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13161 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13162 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13163 (xgene1_store_pair): Split store_8 to store_16.
13164 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13165 (falkor_st_0_st_sd): Split store_8 to store_16.
13167 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13169 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13170 and store1/2/3/4 to store_4/8/12/16.
13171 * config/aarch64/aarch64.md: Update for rename.
13172 * config/arm/arm.md: Likewise.: Likewise.
13173 * config/arm/arm.c: Likewise.
13174 * config/arm/thumb1.md: Likewise.
13175 * config/arm/thumb2.md: Likewise.
13176 * config/arm/vfp.md: Likewise.
13177 * config/arm/arm-generic.md: Likewise.
13178 * config/arm/arm1020e.md: Likewise.
13179 * config/arm/arm1026ejs.md: Likewise.
13180 * config/arm/arm1136jfs.md: Likewise.
13181 * config/arm/arm926ejs.md: Likewise.
13182 * config/arm/cortex-a15.md: Likewise.
13183 * config/arm/cortex-a17.md: Likewise.
13184 * config/arm/cortex-a5.md: Likewise.
13185 * config/arm/cortex-a53.md: Likewise.
13186 * config/arm/cortex-a57.md: Likewise.
13187 * config/arm/cortex-a7.md: Likewise.
13188 * config/arm/cortex-a8.md: Likewise.
13189 * config/arm/cortex-a9.md: Likewise.
13190 * config/arm/cortex-m4.md: Likewise.
13191 * config/arm/cortex-m7.md: Likewise.
13192 * config/arm/cortex-r4.md: Likewise.
13193 * config/arm/exynos-m1.md: Likewise.
13194 * config/arm/fa526.md: Likewise.
13195 * config/arm/fa606te.md: Likewise.
13196 * config/arm/fa626te.md: Likewise.
13197 * config/arm/fa726te.md: Likewise.
13198 * config/arm/fmp626.md: Likewise.
13199 * config/arm/iwmmxt.md: Likewise.
13200 * config/arm/ldmstm.md: Likewise.
13201 * config/arm/marvell-pj4.md: Likewise.
13202 * config/arm/xgene1.md: Likewise.
13203 * config/aarch64/thunderx.md: Likewise.
13204 * config/aarch64/thunderx2t99.md: Likewise.
13205 * config/aarch64/falkor.md: Likewise.
13207 2017-09-12 Martin Liska <mliska@suse.cz>
13209 * attribs.c (private_lookup_attribute): New function.
13210 * attribs.h (private_lookup_attribute): Declared here.
13211 (lookup_attribute): Called from this place.
13213 2017-09-12 Richard Biener <rguenther@suse.de>
13215 PR tree-optimization/82157
13216 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13217 stmts with side-effects.
13219 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13220 Alan Hayward <alan.hayward@arm.com>
13221 David Sherwood <david.sherwood@arm.com>
13223 * target.def (hard_regno_nregs): New hook.
13224 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13225 * targhooks.h (default_hard_regno_nregs): Declare.
13226 * targhooks.c (default_hard_regno_nregs): New function.
13227 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13228 (TARGET_HARD_REGNO_NREGS): ...this hook.
13229 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13230 (CLASS_MAX_NREGS): Likewise.
13231 * doc/tm.texi: Regenerate.
13232 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13233 instead of HARD_REGNO_NREGS.
13234 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13235 HARD_REGNO_NREGS in the comment.
13236 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13237 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13238 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13239 Return an unsigned int.
13240 (TARGET_HARD_REGNO_NREGS): Redefine.
13241 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13242 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13243 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13244 (arc_hard_regno_nregs): New function.
13245 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13246 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13247 (arm_hard_regno_nregs): New function.
13248 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13249 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13250 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13251 (TARGET_HARD_REGNO_NREGS): Redefine.
13252 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13253 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13254 (HARD_REGNO_NREGS): Delete.
13255 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13256 (cr16_hard_regno_nregs): New function.
13257 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13258 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13259 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13260 (cris_hard_regno_nregs): New function.
13261 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13262 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13263 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13264 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13265 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13266 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13267 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13268 (frv_hard_regno_nregs): Make static. Take and return an
13270 (frv_class_max_nregs): Remove outdated copy of documentation.
13271 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13272 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13273 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13274 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13275 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13276 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13277 (TARGET_HARD_REGNO_NREGS): Redefine.
13278 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13279 (CLASS_MAX_NREGS): Update comment.
13280 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13281 (ia64_hard_regno_nregs): New function.
13282 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13283 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13284 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13285 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13286 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13288 (m32c_hard_regno_nregs): Likewise. Make static.
13289 (TARGET_HARD_REGNO_NREGS): Redefine.
13290 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13291 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13292 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13293 (m68k_hard_regno_nregs): New function.
13294 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13295 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13296 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13297 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13298 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13299 Take and return an unsigned int.
13300 (TARGET_HARD_REGNO_NREGS): Redefine.
13301 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13302 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13303 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13304 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13305 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13306 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13307 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13308 (msp430_hard_regno_nregs): Make static. Take and return an
13310 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13311 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13312 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13313 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13314 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13315 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13316 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13317 (TARGET_HARD_REGNO_NREGS): Redefine.
13318 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13319 (PA_HARD_REGNO_NREGS): ...this.
13320 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13321 (PA_HARD_REGNO_NREGS): ...this.
13322 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13323 (pa_hard_regno_nregs): New function.
13324 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13325 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13326 (pdp11_hard_regno_nregs): New function.
13327 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13328 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13329 (rs6000_hard_regno_nregs_hook): New function.
13330 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13331 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13332 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13333 Take and return an unsigned int. Move earlier in file.
13334 (TARGET_HARD_REGNO_NREGS): Redefine.
13335 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13336 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13337 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13338 (rl78_hard_regno_nregs): Make static. Take and return an
13340 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13341 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13342 (rs6000_hard_regno_nregs_hook): New function.
13343 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13344 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13345 (TARGET_HARD_REGNO_NREGS): Redefine.
13346 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13347 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13348 instead of HARD_REGNO_NREGS.
13349 (s390_hard_regno_nregs): New function.
13350 (s390_hard_regno_mode_ok): Add comment from s390.h.
13351 (TARGET_HARD_REGNO_NREGS): Redefine.
13352 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13353 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13354 (sh_hard_regno_nregs): New function.
13355 (sh_pass_in_reg_p): Use it.
13356 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13357 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13358 (sparc_hard_regno_nregs): New function.
13359 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13360 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13361 (spu_function_arg_advance): Use it, supplying a valid register number.
13362 (TARGET_HARD_REGNO_NREGS): Redefine.
13363 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13364 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13365 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13366 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13367 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13368 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13369 (CLASS_MAX_NREGS): Remove copy of old documentation.
13370 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13371 (visium_hard_regno_nregs): New function.
13372 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13373 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13374 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13375 xtensa_hard_regno_nregs): New function.
13376 * system.h (HARD_REGNO_NREGS): Poison.
13378 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13380 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13381 hard_regno_nregs instead of HARD_REGNO_NREGS.
13382 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13383 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13384 (c6x_expand_epilogue): Likewise.
13385 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13386 (frv_read_iacc_argument): Likewise.
13387 * config/sh/sh.c: Include regs.h.
13388 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13389 (regs_used): Likewise.
13390 (output_stack_adjust): Likewise.
13391 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13392 * expmed.c: Include regs.h.
13393 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13394 * ree.c: Include regs.h.
13395 (combine_reaching_defs): Use hard_regno_nregs instead of
13397 (add_removable_extension): Likewise.
13399 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13401 * regs.h (hard_regno_nregs): Turn into a function.
13402 (end_hard_regno): Update accordingly.
13403 * caller-save.c (setup_save_areas): Likewise.
13404 (save_call_clobbered_regs): Likewise.
13405 (replace_reg_with_saved_mem): Likewise.
13406 (insert_restore): Likewise.
13407 (insert_save): Likewise.
13408 * combine.c (can_change_dest_mode): Likewise.
13409 (move_deaths): Likewise.
13410 (distribute_notes): Likewise.
13411 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13412 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13413 (rs6000_split_multireg_move): Likewise.
13414 (rs6000_register_move_cost): Likewise.
13415 (rs6000_memory_move_cost): Likewise.
13416 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13417 (rs6000_split_multireg_move): Likewise.
13418 (rs6000_register_move_cost): Likewise.
13419 (rs6000_memory_move_cost): Likewise.
13420 * cselib.c (cselib_reset_table): Likewise.
13421 (cselib_lookup_1): Likewise.
13422 * emit-rtl.c (set_mode_and_regno): Likewise.
13423 * function.c (aggregate_value_p): Likewise.
13424 * ira-color.c (setup_profitable_hard_regs): Likewise.
13425 (check_hard_reg_p): Likewise.
13426 (calculate_saved_nregs): Likewise.
13427 (assign_hard_reg): Likewise.
13428 (improve_allocation): Likewise.
13429 (calculate_spill_cost): Likewise.
13430 * ira-emit.c (modify_move_list): Likewise.
13431 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13432 (ira_hard_reg_in_set_p): Likewise.
13433 * ira.c (setup_reg_mode_hard_regset): Likewise.
13434 (clarify_prohibited_class_mode_regs): Likewise.
13435 (check_allocation): Likewise.
13436 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13437 (lra_setup_reg_renumber): Likewise.
13438 (setup_try_hard_regno_pseudos): Likewise.
13439 (spill_for): Likewise.
13440 (assign_hard_regno): Likewise.
13441 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13442 * lra-constraints.c (in_class_p): Likewise.
13443 (lra_constraint_offset): Likewise.
13444 (simplify_operand_subreg): Likewise.
13445 (lra_constraints): Likewise.
13446 (split_reg): Likewise.
13447 (split_if_necessary): Likewise.
13448 (invariant_p): Likewise.
13449 (inherit_in_ebb): Likewise.
13450 * lra-lives.c (process_bb_lives): Likewise.
13451 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13452 (get_hard_regs): Likewise.
13453 (do_remat): Likewise.
13454 * lra-spills.c (assign_spill_hard_regs): Likewise.
13455 * mode-switching.c (create_pre_exit): Likewise.
13456 * postreload.c (reload_combine_recognize_pattern): Likewise.
13457 * recog.c (peep2_find_free_register): Likewise.
13458 * regcprop.c (kill_value_regno): Likewise.
13459 (set_value_regno): Likewise.
13460 (copy_value): Likewise.
13461 (maybe_mode_change): Likewise.
13462 (find_oldest_value_reg): Likewise.
13463 (copyprop_hardreg_forward_1): Likewise.
13464 * regrename.c (check_new_reg_p): Likewise.
13465 (regrename_do_replace): Likewise.
13466 * reload.c (push_reload): Likewise.
13467 (combine_reloads): Likewise.
13468 (find_dummy_reload): Likewise.
13469 (operands_match_p): Likewise.
13470 (find_reloads): Likewise.
13471 (find_equiv_reg): Likewise.
13472 (reload_adjust_reg_for_mode): Likewise.
13473 * reload1.c (count_pseudo): Likewise.
13474 (count_spilled_pseudo): Likewise.
13475 (find_reg): Likewise.
13476 (clear_reload_reg_in_use): Likewise.
13477 (free_for_value_p): Likewise.
13478 (allocate_reload_reg): Likewise.
13479 (choose_reload_regs): Likewise.
13480 (reload_adjust_reg_for_temp): Likewise.
13481 (emit_reload_insns): Likewise.
13482 (delete_output_reload): Likewise.
13483 * rtlanal.c (subreg_get_info): Likewise.
13484 * sched-deps.c (sched_analyze_reg): Likewise.
13485 * sel-sched.c (init_regs_for_mode): Likewise.
13486 (mark_unavailable_hard_regs): Likewise.
13487 (choose_best_reg_1): Likewise.
13488 (verify_target_availability): Likewise.
13489 * valtrack.c (dead_debug_insert_temp): Likewise.
13490 * var-tracking.c (track_loc_p): Likewise.
13491 (emit_note_insn_var_location): Likewise.
13492 * varasm.c (make_decl_rtl): Likewise.
13493 * reginfo.c (choose_hard_reg_mode): Likewise.
13494 (init_reg_modes_target): Refer directly to
13495 this_target_regs->x_hard_regno_nregs.
13497 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13499 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13500 instead of hard_regno_nregs.
13502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13504 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13505 end_hard_regno instead of hard_regno_nregs.
13506 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13507 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13508 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13509 * ira-color.c (improve_allocation): Likewise.
13510 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13511 * lra-lives.c (mark_regno_live): Likewise.
13512 (mark_regno_dead): Likewise.
13513 * lra-remat.c (operand_to_remat): Likewise.
13514 * lra.c (collect_non_operand_hard_regs): Likewise.
13515 * postreload.c (reload_combine_note_store): Likewise.
13516 (move2add_valid_value_p): Likewise.
13517 * reload.c (regno_clobbered_p): Likewise.
13519 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13521 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13523 * config/v850/v850.c (v850_reorg): Likewise.
13524 * reload.c (refers_to_regno_for_reload_p): Likewise.
13525 (find_equiv_reg): Likewise.
13526 * reload1.c (reload_reg_reaches_end_p): Likewise.
13528 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13530 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13532 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13533 * config/arm/arm.c (output_move_neon): Likewise.
13534 (arm_attr_length_move_neon): Likewise.
13535 (neon_split_vcombine): Likewise.
13536 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13537 (c6x_mark_reg_written): Likewise.
13538 (c6x_dwarf_register_span): Likewise.
13539 * config/i386/i386.c (ix86_save_reg): Likewise.
13540 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13541 (rws_access_reg): Likewise.
13542 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13543 * mode-switching.c (create_pre_exit): Likewise.
13544 * ree.c (combine_reaching_defs): Likewise.
13545 (add_removable_extension): Likewise.
13546 * regcprop.c (find_oldest_value_reg): Likewise.
13547 (copyprop_hardreg_forward_1): Likewise.
13548 * reload.c (reload_inner_reg_of_subreg): Likewise.
13549 (push_reload): Likewise.
13550 (combine_reloads): Likewise.
13551 (find_dummy_reload): Likewise.
13552 (reload_adjust_reg_for_mode): Likewise.
13553 * reload1.c (find_reload_regs): Likewise.
13554 (forget_old_reloads_1): Likewise.
13555 (reload_reg_free_for_value_p): Likewise.
13556 (reload_adjust_reg_for_temp): Likewise.
13557 (emit_reload_insns): Likewise.
13558 (delete_output_reload): Likewise.
13559 * sel-sched.c (choose_best_reg_1): Likewise.
13560 (choose_best_pseudo_reg): Likewise.
13562 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13563 Alan Hayward <alan.hayward@arm.com>
13564 David Sherwood <david.sherwood@arm.com>
13566 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13567 * target.def (slow_unaligned_access): New hook.
13568 * targhooks.h (default_slow_unaligned_access): Declare.
13569 * targhooks.c (default_slow_unaligned_access): New function.
13570 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13571 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13572 * doc/tm.texi: Regenerate.
13573 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13574 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13575 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13577 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13578 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13580 (rs6000_slow_unaligned_access): New function.
13581 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13582 (expand_block_compare): Likewise.
13583 (expand_strn_compare): Likewise.
13584 (rs6000_rtx_costs): Likewise.
13585 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13586 (riscv_slow_unaligned_access): Likewise.
13587 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13588 (riscv_slow_unaligned_access_p): ...this and make static.
13589 (riscv_option_override): Update accordingly.
13590 (riscv_slow_unaligned_access): New function.
13591 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13592 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13593 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13594 (rs6000_slow_unaligned_access): New function.
13595 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13596 (rs6000_rtx_costs): Likewise.
13597 * config/rs6000/rs6000-string.c (expand_block_compare)
13598 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13599 of SLOW_UNALIGNED_ACCESS.
13600 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13601 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13602 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13603 of SLOW_UNALIGNED_ACCESS.
13604 * expmed.c (simple_mem_bitfield_p): Likewise.
13605 * expr.c (alignment_for_piecewise_move): Likewise.
13606 (emit_group_load_1): Likewise.
13607 (emit_group_store): Likewise.
13608 (copy_blkmode_from_reg): Likewise.
13609 (emit_push_insn): Likewise.
13610 (expand_assignment): Likewise.
13611 (store_field): Likewise.
13612 (expand_expr_real_1): Likewise.
13613 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13614 * lra-constraints.c (simplify_operand_subreg): Likewise.
13615 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13616 * gimple-ssa-store-merging.c: Likewise in block comment at start
13618 * tree-ssa-strlen.c: Include target.h.
13619 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13620 of SLOW_UNALIGNED_ACCESS.
13621 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13623 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13625 PR rtl-optimization/82185
13626 * expmed.c (emit_store_flag_int): Only test tem if it has been
13629 2017-09-12 Richard Biener <rguenther@suse.de>
13631 PR middle-end/82149
13632 * match.pd ((FTYPE) N CMP CST): Fix typo.
13634 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13636 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13638 (mips_near_type_p): Add 'short_call' attribute as a synonym
13640 * doc/extend.texi (short_call): Document new function attribute.
13642 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13645 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13646 assertion check that in the condition.
13647 (get_atomic_generic_size): Likewise. Before testing if parameter
13648 has pointer type, if it has array type, call for C++
13649 default_conversion to perform array-to-pointer conversion.
13651 2017-09-12 Richard Biener <rguenther@suse.de>
13653 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13654 for operations we cannot scalarize.
13656 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13658 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13659 vectors heap vectors. Clean up comments.
13660 Make visited_bbs a reference.
13661 (profitable_jump_thread_path): Make GC
13662 vectors heap vectors. Clean up comments.
13664 (convert_and_register_jump_thread_path): Make GC vectors heap
13666 (check_subpath_and_update_thread_path): Same. Clean up comments.
13667 Make visited_bbs a reference.
13668 (handle_phi): Abstract common code to to
13669 register_jump_thread_path_if_profitable.
13670 Rename VAR_BB to DEF_BB.
13672 Make GC vectors heap vectors.
13673 Make visited_bbs a reference.
13674 (handle_assignment): Same.
13675 (register_jump_thread_path_if_profitable): New.
13676 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13678 Make GC vectors heap vectors. Clean up comments.
13679 Make visited_bbs a reference.
13680 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13681 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13684 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13687 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13688 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13690 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13692 Revert r251800 and r251799.
13694 2017-09-11 Martin Jambor <mjambor@suse.cz>
13697 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13698 arguments in advance.
13699 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13700 use it to find predecessor edges.
13701 (naive_outof_ssa): Collect vector of predecessors.
13703 2017-09-08 Jason Merrill <jason@redhat.com>
13705 PR c++/70029 - ICE with ref-qualifier and -flto
13706 * langhooks.h (struct lang_hooks_for_types): Add
13707 copy_lang_qualifiers.
13708 * attribs.c (build_type_attribute_qual_variant): Use it.
13709 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13711 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13712 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13714 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13717 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13718 (*mulsi3_sp64): New instruction.
13719 (mulsi3): New expander.
13721 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13723 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13725 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13727 * sancov.c: Include memmodel.h.
13729 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13732 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13735 2017-09-07 Carl Love <cel@us.ibm.com>
13737 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13738 the sldi instruction.
13740 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13742 * sancov.c: Include tm_p.h.
13744 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13747 * output.h (switch_to_other_text_partition): New declaration.
13748 * varasm.c (switch_to_other_text_partition): New function.
13749 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13750 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13751 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13752 to the other text partition before emitting LCL label and switch back
13753 after emitting the word after it.
13755 2017-09-07 Richard Biener <rguenther@suse.de>
13757 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13758 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13759 critical edges here, after loop init.
13760 (pass_data_pre): Remove PROP_no_crit_edges flags.
13761 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13762 for valueization of call args to avoid leaking VN_TOP.
13763 (visit_use): Assert we do not visit default defs.
13764 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13765 Use error_mark_node to more easily detect leaking VN_TOP.
13766 All default-defs are varying, not VN_TOP. Mark them visited.
13767 (run_scc_vn): Make code match comment.
13769 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13771 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13772 OPTION_MASK_FLOAT128_KEYWORD.
13773 (POWERPC_MASKS): Likewise.
13774 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13775 support for the -mfloat128-type option, and make -mfloat128
13776 default on PowerPC Linux systems. Define or undefine
13777 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13778 Define __float128 to be __ieee128 if IEEE 128-bit support is
13779 enabled, or undefine it.
13780 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13781 Delete defining __FLOAT128_TYPE__.
13782 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13783 -mfloat128-type option and make -mfloat128 default on PowerPC
13785 (TARGET_FLOAT128_TYPE): Likewise.
13786 (-mfloat128-type): Likewise.
13787 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13788 Delete the -mfloat128-type option and make -mfloat128 default on
13789 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13790 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13791 128-bit floating point is enabled. Change tests from using
13792 -mfloat128-type to -mfloat128.
13793 (rs6000_mangle_type): Use the correct mangling for the __float128
13794 type even if normal long double is restricted to 64-bits.
13795 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13797 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13798 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13799 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13801 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13802 documentation for -mfloat128.
13804 2017-09-06 Olivier Hainque <hainque@adacore.com>
13806 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13808 2017-09-06 Wish Wu <wishwu007@gmail.com>
13809 Jakub Jelinek <jakub@redhat.com>
13811 * asan.c (initialize_sanitizer_builtins): Add
13812 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13813 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13814 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13815 BT_FN_VOID_UINT64_PTR variables.
13816 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13817 (BT_FN_VOID_UINT16_UINT16): Likewise.
13818 (BT_FN_VOID_UINT32_UINT32): Likewise.
13819 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13820 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13821 (BT_FN_VOID_UINT64_PTR): Likewise.
13822 * common.opt (flag_sanitize_coverage): New variable.
13823 (fsanitize-coverage=trace-pc): Remove.
13824 (fsanitize-coverage=): Add.
13825 * flag-types.h (enum sanitize_coverage_code): New enum.
13826 * fold-const.c (fold_range_test): Disable non-short-circuit
13827 optimization if flag_sanitize_coverage.
13828 (fold_truth_andor): Likewise.
13829 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13830 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13831 (coverage_sanitizer_opts): New array.
13832 (get_closest_sanitizer_option): Add OPTS argument, handle also
13833 OPT_fsanitize_coverage_.
13834 (parse_sanitizer_options): Adjusted to also handle
13835 OPT_fsanitize_coverage_.
13836 (common_handle_option): Add OPT_fsanitize_coverage_.
13837 * sancov.c (instrument_comparison, instrument_switch): New function.
13838 (sancov_pass): Add trace-cmp support.
13839 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13840 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13841 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13842 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13843 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13844 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13845 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13846 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13847 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13848 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13850 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13852 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13853 error. Only quit immediately if parsing is complete.
13854 (BEGIN): Initialize fatal_err and parse_done.
13855 (begin fpu, end fpu): Check number of arguments.
13856 (begin arch, end arch): Likewise.
13857 (begin cpu, end cpu): Likewise.
13858 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13859 (optalias): Likewise.
13861 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13863 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13864 * config/arm/arm-isa.h: Delete. Move definitions to ...
13865 * arm-cpus.in: ... here. Use new feature and fgroup values.
13866 * config/arm/arm.c (arm_option_override): Use lower case for feature
13868 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13869 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13870 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13872 (print_isa_bits_for): New function.
13873 (gen_isa): New function.
13874 (gen_comm_data): Use print_isa_bits_for.
13875 (define feature): New keyword.
13876 (define fgroup): New keyword.
13877 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13878 (arm-isa.h): Add rule to generate file.
13879 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13880 case for feature bit names.
13882 2017-09-06 Richard Biener <rguenther@suse.de>
13884 * tree-ssa-pre.c (NECESSARY): Remove.
13885 (create_expression_by_pieces): Do not touch pass-local flags.
13886 (insert_into_preds_of_block): Likewise.
13887 (do_pre_regular_insertion): Likewise.
13888 (eliminate_insert): Likewise.
13889 (eliminate_dom_walker::before_dom_children): Likewise.
13890 (fini_eliminate): Do not look at inserted_exprs.
13891 (mark_operand_necessary): Remove.
13892 (remove_dead_inserted_code): Replace with simple work-list
13893 algorithm based on inserted_exprs and SSA uses.
13894 (pass_pre::execute): Re-order fini_eliminate and
13895 remove_dead_inserted_code.
13897 2017-09-06 Olivier Hainque <hainque@adacore.com>
13899 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13900 for VxWorks 7. Adjust surrounding comments.
13902 2017-09-06 Richard Biener <rguenther@suse.de>
13904 * gimple-ssa-strength-reduction.c
13905 (find_candidates_dom_walker::before_dom_children): Also allow
13908 2017-09-06 Richard Biener <rguenther@suse.de>
13910 PR tree-optimization/82108
13911 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13912 for gap in the non-permutation SLP case.
13914 2017-09-06 Martin Jambor <mjambor@suse.cz>
13916 PR tree-optimization/82078
13917 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13918 add_access_to_work_queue...
13919 (build_accesses_from_assign): ...here.
13920 (propagate_all_subaccesses): Make sure racc is the group
13921 representative, if there is one.
13923 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13925 PR middle-end/82095
13926 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13929 2017-09-06 Richard Biener <rguenther@suse.de>
13931 * gimple-ssa-strength-reduction.c
13932 (find_candidates_dom_walker::before_doom_children): Use a
13933 type and not a mode check.
13935 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13938 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13940 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13942 2017-09-05 Jeff Law <law@redhat.com>
13944 PR tree-optimization/64910
13945 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13946 cases where we have 3 or more operands.
13948 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13950 PR middle-end/81768
13951 * omp-low.c (lower_omp_for): Recompute tree invariant if
13952 gimple_omp_for_initial/final is ADDR_EXPR.
13954 PR middle-end/81768
13955 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13956 into gimple val before gimplification fo the COND_EXPR.
13958 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13960 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13961 REGION_COPY argument.
13962 (thread_through_all_blocks): Remove unused argument to
13963 duplicate_thread_path.
13965 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13966 Alan Hayward <alan.hayward@arm.com>
13967 David Sherwood <david.sherwood@arm.com>
13969 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13970 Take a scalar_mode rather than a machine_mode.
13971 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13972 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13973 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13974 (aarch64_gen_adjusted_ldpstp): Likewise.
13975 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13977 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13978 Alan Hayward <alan.hayward@arm.com>
13979 David Sherwood <david.sherwood@arm.com>
13981 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13982 Take a scalar_int_mode instead of a machine_mode.
13983 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13984 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13985 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13986 (aarch64_simd_attr_length_rglist): Delete.
13987 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13988 a scalar_int_mode instead of a machine_mode.
13989 (aarch64_add_offset): Likewise.
13990 (aarch64_internal_mov_immediate): Likewise
13991 (aarch64_add_constant_internal): Likewise.
13992 (aarch64_add_constant): Likewise.
13993 (aarch64_movw_imm): Likewise.
13994 (aarch64_rtx_arith_op_extract_p): Likewise.
13995 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13996 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13997 Remove assert that the mode isn't a vector.
13998 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13999 (aarch64_expand_mov_immediate): Update calls after above changes.
14000 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14001 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14002 (aarch64_move_imm): Likewise.
14003 (aarch64_can_const_movi_rtx_p): Likewise.
14004 (aarch64_strip_extend): Likewise.
14005 (aarch64_extr_rtx_p): Likewise.
14006 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14007 a CONST_INT when the mode parameter is VOIDmode.
14008 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14010 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14012 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14013 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14014 (bitwise_type_for_mode): Update accordingly.
14016 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14018 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14019 * stor-layout.c (mode_for_size_tree): Likewise.
14020 (mode_for_array): Update accordingly.
14021 (layout_decl): Likewise.
14022 (compute_record_mode): Likewise. Only set the mode once.
14024 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14026 * target.def (get_mask_mode): Change return type to opt_mode.
14028 * doc/tm.texi: Regenerate.
14029 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14030 * targhooks.c (default_get_mask_mode): Likewise.
14031 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14032 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14033 targetm.get_mask_mode.
14034 * tree.c (build_truth_vector_type): Likewise.
14036 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14038 * machmode.h (mode_for_vector): Return an opt_mode.
14039 * stor-layout.c (mode_for_vector): Likewise.
14040 (mode_for_int_vector): Update accordingly.
14041 (layout_type): Likewise.
14042 * config/i386/i386.c (emit_memmov): Likewise.
14043 (ix86_expand_set_or_movmem): Likewise.
14044 (ix86_expand_vector_init): Likewise.
14045 (ix86_get_mask_mode): Likewise.
14046 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14048 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14049 * expmed.c (extract_bit_field_1): Likewise.
14050 * expr.c (expand_expr_real_2): Likewise.
14051 * optabs-query.c (can_vec_perm_p): Likewise.
14052 (can_vec_mask_load_store_p): Likewise.
14053 * optabs.c (expand_vec_perm): Likewise.
14054 * targhooks.c (default_get_mask_mode): Likewise.
14055 * tree-vect-stmts.c (vectorizable_store): Likewise.
14056 (vectorizable_load): Likewise.
14057 (get_vectype_for_scalar_type_and_size): Likewise.
14059 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14061 * machmode.h (mode_for_int_vector): New function.
14062 * stor-layout.c (mode_for_int_vector): Likewise.
14063 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14064 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14065 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14066 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14067 (s390_expand_vcond): Likewise.
14069 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14071 * machmode.h (opt_machine_mode): New type.
14072 (opt_mode<T>): Allow construction from anything that can be
14074 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14075 (mode_for_size): Return an opt_machine_mode.
14076 * stor-layout.c (mode_for_size): Likewise.
14077 (mode_for_size_tree): Update call accordingly.
14078 (bitwise_mode_for_mode): Likewise.
14079 (make_fract_type): Likewise.
14080 (make_accum_type): Likewise.
14081 * caller-save.c (replace_reg_with_saved_mem): Update call
14083 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14084 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14085 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14086 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14087 * expmed.c (extract_bit_field_1): Likewise.
14088 * reload.c (get_secondary_mem): Likewise.
14089 * varasm.c (assemble_integer): Likewise.
14090 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14093 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14095 * machmode.h (decimal_float_mode_for_size): New function.
14096 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14097 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14098 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14099 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14100 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14101 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14103 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14105 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14106 (get_builtin_sync_mode): Likewise.
14107 (expand_ifn_atomic_compare_exchange): Likewise.
14108 (expand_builtin_atomic_clear): Likewise.
14109 (expand_builtin_atomic_test_and_set): Likewise.
14110 (fold_builtin_atomic_always_lock_free): Likewise.
14111 * calls.c (compute_argument_addresses): Likewise.
14112 (emit_library_call_value_1): Likewise.
14113 (store_one_arg): Likewise.
14114 * combine.c (combine_instructions): Likewise.
14115 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14116 * config/arm/arm.c (arm_function_value): Likewise.
14117 (aapcs_allocate_return_reg): Likewise.
14118 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14119 * config/i386/i386.c (construct_container): Likewise.
14120 (ix86_gimplify_va_arg): Likewise.
14121 (ix86_expand_sse_cmp): Likewise.
14122 (emit_memmov): Likewise.
14123 (emit_memset): Likewise.
14124 (expand_small_movmem_or_setmem): Likewise.
14125 (ix86_expand_pextr): Likewise.
14126 (ix86_expand_pinsr): Likewise.
14127 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14128 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14130 * config/mips/mips.c (mips_function_value_1) Likewise.
14131 (mips_block_move_straight): Likewise.
14132 (mips_expand_ins_as_unaligned_store): Likewise.
14133 * config/powerpcspe/powerpcspe.c
14134 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14135 (rs6000_darwin64_record_arg_flush): Likewise.
14136 * config/rs6000/rs6000.c
14137 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14138 (rs6000_darwin64_record_arg_flush): Likewise.
14139 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14140 (sparc_function_value_1): Likewise.
14141 * config/spu/spu.c (adjust_operand): Likewise.
14142 (spu_emit_branch_or_set): Likewise.
14143 (arith_immediate_p): Likewise.
14144 * emit-rtl.c (gen_lowpart_common): Likewise.
14145 * expr.c (expand_expr_real_1): Likewise.
14146 * function.c (assign_parm_setup_block): Likewise.
14147 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14148 * reload1.c (alter_reg): Likewise.
14149 * stor-layout.c (mode_for_vector): Likewise.
14150 (layout_type): Likewise.
14152 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14154 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14155 (spu_convert_move): Likewise.
14156 * lower-subreg.c (resolve_simple_move): Likewise.
14158 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14161 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14162 define_insn to a define_expand.
14163 (altivec_vsum2sws_direct): New define_insn.
14164 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14166 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14168 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14171 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14176 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14177 realign stack if stack alignment needed is less than incoming
14180 2017-09-05 Marek Polacek <polacek@redhat.com>
14183 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14186 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14188 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14190 2017-09-05 Richard Biener <rguenther@suse.de>
14192 PR tree-optimization/82084
14193 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14195 2017-09-05 Richard Biener <rguenther@suse.de>
14197 PR tree-optimization/82102
14198 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14200 2017-09-05 Martin Liska <mliska@suse.cz>
14202 PR tree-optimization/82032
14203 * tree-cfg.c (generate_range_test): New function.
14204 * tree-cfg.h (generate_range_test): Declared here.
14205 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14206 (cleanup_control_expr_graph): Use it.
14207 * tree-switch-conversion.c (try_switch_expansion): Remove
14209 (emit_case_nodes): Use generate_range_test.
14211 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14214 * config/i386/i386.md (*<btsc><mode>_mask): Add
14215 TARGET_USE_BT to insn constraint.
14216 (*btr<mode>_mask): Ditto.
14218 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14220 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14221 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14223 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14226 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14227 TARGET_NEON and TARGET_IWMMXT.
14228 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14229 TARGET_NEON and TARGET_IWMMXT.
14230 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14232 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14234 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14235 (ix86_rewrite_tls_address): Ditto.
14236 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14237 (ix86_rewrite_tls_address_1): Ditto.
14238 (ix86_rewrite_tls_address): Ditto.
14239 * config/i386/predicates.md (tls_address_pattern): New predicate.
14240 * config/i386/i386.md (TLS address splitter): New splitter.
14242 2017-09-04 Richard Biener <rguenther@suse.de>
14244 PR tree-optimization/82084
14245 * fold-const.h (can_native_encode_string_p): Declare.
14246 * fold-const.c (can_native_encode_string_p): Factor out from ...
14247 (native_encode_string): ... here.
14248 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14249 vectorizing stores from constants we later cannot handle.
14251 2017-09-04 Marek Polacek <polacek@redhat.com>
14254 * doc/invoke.texi: Update -Wtautological-compare documentation.
14256 2017-09-04 Jeff Law <law@redhat.com>
14258 PR tree-optimization/64910
14259 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14260 swap the first and last operand if the last is a constant.
14262 2017-09-04 Marek Polacek <polacek@redhat.com>
14265 * convert.c (do_narrow): When sanitizing signed integer overflows,
14266 bail out for signed types.
14267 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14269 2017-09-04 Richard Biener <rguenther@suse.de>
14271 PR tree-optimization/82060
14272 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14273 Move devirtualization after stmt folding and before EH/AB/noreturn
14274 cleanup to get the stmt refs canonicalized. Use a bool instead
14275 of gimple_modified_p since that doesn't work for NOPs. Schedule
14276 NOPs generated by folding for removal.
14278 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14279 Alan Hayward <alan.hayward@arm.com>
14280 David Sherwood <david.sherwood@arm.com>
14282 * coretypes.h (pad_direction): New enum.
14283 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14284 (FUNCTION_ARG_PADDING): Likewise.
14285 * target.def (function_arg_padding): New hook.
14286 * targhooks.h (default_function_arg_padding): Declare.
14287 * targhooks.c (default_function_arg_padding): New function.
14288 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14289 (TARGET_FUNCTION_ARG_PADDING): ...this.
14290 * doc/tm.texi: Regenerate.
14291 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14292 instead of direction.
14293 (compute_argument_addresses): Likewise.
14294 (load_register_parameters): Likewise.
14295 (emit_library_call_value_1): Likewise.
14296 (store_one_arg): Use targetm.calls.function_arg_padding instead
14297 of FUNCTION_ARG_PADDING.
14298 (must_pass_in_stack_var_size_or_pad): Likewise.
14299 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14300 (emit_group_store): Likewise.
14301 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14302 instead of FUNCTION_ARG_PADDING.
14303 (emit_push_insn): Likewise, and propagate enum change throughout
14305 * function.h (direction): Delete.
14306 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14308 * function.c (assign_parm_find_stack_rtl): Likewise.
14309 (assign_parm_setup_block_p): Likewise.
14310 (assign_parm_setup_block): Likewise.
14311 (gimplify_parameters): Likewise.
14312 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14313 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14315 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14316 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14317 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14318 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14319 (aarch64_function_arg_padding): ...this new function.
14320 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14321 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14322 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14323 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14324 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14325 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14326 (arm_pad_arg_upward): Replace with...
14327 (arm_function_arg_padding): ...this new function.
14328 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14330 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14331 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14332 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14333 (ia64_hpux_function_arg_padding): Replace with...
14334 (ia64_function_arg_padding): ...this new function. Use pad_direction
14335 instead of direction. Check for TARGET_HPUX.
14336 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14337 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14338 (iq2000_function_arg_padding): New function.
14339 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14340 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14341 (mips_function_arg_padding): ...this new function.
14342 (mips_pad_reg_upward): Update accordingly.
14343 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14344 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14345 targetm.calls.function_arg_padding.
14346 (FUNCTION_ARG_PADDING): Delete.
14347 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14348 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14349 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14350 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14351 (nios2_block_reg_padding): Return pad_direction instead of direction.
14352 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14353 instead of direction.
14354 (nios2_function_arg_padding): Likewise. Make static.
14355 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14356 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14357 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14358 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14359 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14360 (pa_function_arg_padding): Make static. Return pad_direction instead
14362 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14363 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14364 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14365 instead of direction. Use targetm.calls.function_arg_padding.
14366 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14367 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14368 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14369 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14370 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14372 (function_arg_padding): Rename to...
14373 (rs6000_function_arg_padding): ...this. Make static. Return
14374 pad_direction instead of direction.
14375 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14376 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14377 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14378 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14379 instead of direction. Use targetm.calls.function_arg_padding.
14380 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14381 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14382 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14383 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14384 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14385 (function_arg_padding): Rename to...
14386 (rs6000_function_arg_padding): ...this. Make static. Return
14387 pad_direction instead of direction.
14388 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14389 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14390 * config/s390/s390.c (s390_function_arg_padding): New function.
14391 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14392 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14393 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14394 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14395 (function_arg_padding): Rename to...
14396 (sparc_function_arg_padding): ...this. Make static. Return
14397 pad_direction instead of direction.
14398 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14399 * config/spu/spu.c (spu_function_arg_padding): New function.
14400 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14401 * system.h (FUNCTION_ARG_PADDING): Poison.
14403 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14404 Alan Hayward <alan.hayward@arm.com>
14405 David Sherwood <david.sherwood@arm.com>
14407 * target.def (modes_tieable_p): New hook.
14408 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14409 (TARGET_MODES_TIEABLE_P): ...this.
14410 * doc/tm.texi.in: Regenerate.
14411 * hooks.h (hook_bool_mode_mode_true): Declare.
14412 * hooks.c (hook_bool_mode_mode_true): New function.
14413 * combine.c (subst): Use targetm.modes_tieable_p instead of
14415 * dse.c (find_shift_sequence): Likewise.
14416 * expmed.c (extract_low_bits): Likewise.
14417 * lower-subreg.c: Include target.h.
14418 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14420 * rtlanal.c (rtx_cost): Likewise.
14421 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14422 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14423 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14424 (TARGET_MODES_TIEABLE_P): Redefine.
14425 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14426 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14427 (TARGET_MODES_TIEABLE_P): Redefine.
14428 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14429 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14430 (arc_modes_tieable_p): New function.
14431 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14432 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14433 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14434 (arm_modes_tieable_p): Make static.
14435 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14436 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14437 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14438 (TARGET_MODES_TIEABLE_P): Redefine.
14439 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14440 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14441 (TARGET_MODES_TIEABLE_P): Redefine.
14442 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14443 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14444 (cr16_modes_tieable_p): New function.
14445 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14446 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14447 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14448 (TRULY_NOOP_TRUNCATION): Update comment.
14449 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14450 (TRULY_NOOP_TRUNCATION): Update comment.
14451 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14452 (frv_modes_tieable_p): New function.
14453 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14454 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14455 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14456 (TARGET_MODES_TIEABLE_P): Redefine.
14457 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14458 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14459 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14460 (TARGET_MODES_TIEABLE_P): Redefine.
14461 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14462 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14463 (ia64_modes_tieable_p): New function.
14464 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14465 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14466 (iq2000_modes_tieable_p): New function.
14467 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14468 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14469 (lm32_modes_tieable_p): New function.
14470 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14471 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14472 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14473 (TARGET_MODES_TIEABLE_P): Redefine.
14474 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14475 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14476 (m32r_modes_tieable_p): New function.
14477 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14478 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14479 (m68k_modes_tieable_p): New function.
14480 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14481 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14482 (mcore_modes_tieable_p): New function.
14483 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14484 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14486 (TARGET_MODES_TIEABLE_P): Redefine.
14487 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14488 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14489 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14490 (TARGET_MODES_TIEABLE_P): Redefine.
14491 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14492 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14493 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14494 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14495 (mn10300_modes_tieable_p): ...this and make static.
14496 (TARGET_MODES_TIEABLE_P): Redefine.
14497 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14498 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14499 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14500 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14501 (msp430_modes_tieable_p): Make static.
14502 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14503 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14504 (TARGET_MODES_TIEABLE_P): Redefine.
14505 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14506 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14507 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14508 (TARGET_MODES_TIEABLE_P): Redefine.
14509 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14510 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14511 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14512 (TARGET_MODES_TIEABLE_P): Redefine.
14513 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14514 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14515 (pdp11_modes_tieable_p): New function.
14516 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14517 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14518 (rs6000_modes_tieable_p): New function.
14519 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14520 * config/powerpcspe/powerpcspe.md: Update comment.
14521 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14522 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14523 (TARGET_MODES_TIEABLE_P): Redefine.
14524 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14525 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14526 (rl78_modes_tieable_p): New function.
14527 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14528 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14529 (rs6000_modes_tieable_p): New function.
14530 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14531 * config/rs6000/rs6000.md: Update comment.
14532 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14533 * config/rx/rx.c (rx_modes_tieable_p): New function.
14534 (TARGET_MODES_TIEABLE_P): Redefine.
14535 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14536 * config/s390/s390.c (s390_modes_tieable_p): New function.
14537 (TARGET_MODES_TIEABLE_P): Redefine.
14538 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14539 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14540 (sh_modes_tieable_p): New function.
14541 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14542 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14543 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14544 (sparc_modes_tieable_p): Make static.
14545 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14546 * config/spu/spu.c (spu_modes_tieable_p): New function.
14547 (TARGET_MODES_TIEABLE_P): Redefine.
14548 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14549 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14550 (TARGET_MODES_TIEABLE_P): Redefine.
14551 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14552 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14553 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14554 * config/v850/v850.c (v850_modes_tieable_p): New function.
14555 (TARGET_MODES_TIEABLE_P): Redefine.
14556 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14557 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14558 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14559 (visium_modes_tieable_p): New function.
14560 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14561 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14562 (xtensa_modes_tieable_p): New function.
14563 * system.h (MODES_TIEABLE_P): Poison.
14565 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14566 Alan Hayward <alan.hayward@arm.com>
14567 David Sherwood <david.sherwood@arm.com>
14569 * target.def (hard_regno_mode_ok): New hook.
14570 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14571 (TARGET_HARD_REGNO_MODE_OK): ...this.
14572 * doc/tm.texi.in: Regenerate.
14573 * hooks.h (hook_bool_uint_mode_true): Declare.
14574 * hooks.c (hook_bool_uint_mode_true): New function.
14575 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14576 HARD_REGNO_MODE_OK.
14577 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14578 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14579 instead of HARD_REGNO_MODE_OK.
14580 * caller-save.c: Include target.h.
14581 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14582 HARD_REGNO_MODE_OK.
14583 * combine.c (can_combine_p): Likewise.
14584 (combinable_i3pat): Likewise.
14585 (can_change_dest_mode): Likewise.
14586 * expr.c (init_expr_target): Likewise.
14587 (convert_move): Likewise.
14588 (convert_modes): Likewise.
14589 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14590 (setup_prohibited_mode_move_regs): Likewise.
14591 * ira.h (target_ira): Likewise.
14592 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14593 * lra-constraints.c (process_alt_operands): Likewise.
14594 (split_reg): Likewise.
14595 * recog.c (peep2_find_free_register): Likewise.
14596 * ree.c (combine_reaching_defs): Likewise.
14597 * regcprop.c (maybe_mode_change): Likewise.
14598 * reginfo.c (init_reg_sets_1): Likewise.
14599 (choose_hard_reg_mode): Likewise.
14600 (simplifiable_subregs): Likewise.
14601 * regrename.c (check_new_reg_p): Likewise.
14602 * reload.c (find_valid_class): Likewise.
14603 (find_valid_class_1): Likewise.
14604 (reload_inner_reg_of_subreg): Likewise.
14605 (push_reload): Likewise.
14606 (combine_reloads): Likewise.
14607 (find_dummy_reload): Likewise.
14608 (find_reloads): Likewise.
14609 * reload1.c (find_reg): Likewise.
14610 (set_reload_reg): Likewise.
14611 (allocate_reload_reg): Likewise.
14612 (choose_reload_regs): Likewise.
14613 (reload_adjust_reg_for_temp): Likewise.
14614 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14615 (simplify_subreg_regno): Likewise.
14616 * sel-sched.c (init_regs_for_mode): Likewise.
14617 * varasm.c (make_decl_rtl): Likewise.
14618 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14619 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14620 HARD_REGNO_MODE_OK.
14621 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14622 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14624 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14625 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14627 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14628 (arc_mode_class): Delete.
14629 (HARD_REGNO_MODE_OK): Delete.
14630 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14631 (arc_hard_regno_mode_ok): Rename old array to...
14632 (arc_hard_regno_mode_ok_modes): ...this.
14633 (arc_conditional_register_usage): Update accordingly.
14634 (arc_mode_class): Make static.
14635 (arc_hard_regno_mode_ok): New function.
14636 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14637 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14638 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14639 (arm_hard_regno_mode_ok): Make static.
14640 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14641 HARD_REGNO_MODE_OK.
14642 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14643 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14644 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14647 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14648 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14649 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14650 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14652 * config/bfin/predicates.md (valid_reg_operand): Use
14653 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14654 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14655 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14656 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14657 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14658 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14659 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14660 (cr16_hard_regno_mode_ok): Make static and return a bool.
14661 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14662 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14663 (cris_hard_regno_mode_ok): New function.
14664 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14665 (epiphany_mode_class): Delete.
14666 (HARD_REGNO_MODE_OK): Delete.
14667 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14668 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14669 (hard_regno_mode_ok): Rename to...
14670 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14671 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14672 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14673 HARD_REGNO_MODE_OK.
14674 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14675 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14676 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14677 (frv_hard_regno_mode_ok): Make static and return a bool.
14678 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14679 HARD_REGNO_MODE_OK.
14680 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14681 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14682 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14683 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14685 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14686 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14687 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14688 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14690 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14691 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14692 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14693 (ia64_hard_regno_mode_ok): New function.
14694 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14695 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14696 (iq2000_hard_regno_mode_ok): New function.
14697 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14698 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14699 (lm32_hard_regno_mode_ok): New function.
14700 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14701 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14702 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14703 instead of HARD_REGNO_MODE_OK.
14704 (m32c_hard_regno_ok): Rename to...
14705 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14706 (m32c_cannot_change_mode_class): Update accordingly.
14707 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14708 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14709 (m32r_mode_class): Delete.
14710 (HARD_REGNO_MODE_OK): Delete.
14711 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14712 (m32r_hard_regno_mode_ok): Rename to...
14713 (m32r_hard_regno_modes): ...this.
14714 (m32r_mode_class): Make static.
14715 (m32r_hard_regno_mode_ok): New function.
14716 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14717 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14718 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14719 (m68k_hard_regno_mode_ok): Make static.
14720 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14721 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14722 (mcore_hard_regno_mode_ok): New function.
14723 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14724 (HARD_REGNO_MODE_OK): Delete.
14725 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14727 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14728 (microblaze_hard_regno_mode_ok): New function.
14729 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14730 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14731 (mips_hard_regno_mode_ok): Delete.
14732 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14733 (mips_hard_regno_mode_ok_p): ...this and make static.
14734 (mips_hard_regno_mode_ok_p): Rename to...
14735 (mips_hard_regno_mode_ok_uncached): ...this.
14736 (mips_hard_regno_mode_ok): New function.
14737 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14738 of HARD_REGNO_MODE_OK.
14739 (mips_option_override): Update after above name changes.
14740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14741 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14742 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14743 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14744 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14745 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14746 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14747 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14748 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14749 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14750 (msp430_hard_regno_mode_ok): Make static and return a bool.
14751 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14752 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14753 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14755 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14756 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14757 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14758 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14759 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14760 (PA_HARD_REGNO_MODE_OK): ...this
14761 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14762 (PA_HARD_REGNO_MODE_OK): ...this.
14763 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14764 (pa_hard_regno_mode_ok): New function.
14765 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14766 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14767 (pdp11_hard_regno_mode_ok): New function.
14768 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14769 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14771 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14773 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14774 (rs6000_hard_regno_mode_ok): Rename to...
14775 (rs6000_hard_regno_mode_ok_uncached): ...this.
14776 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14777 (rs6000_hard_regno_mode_ok): New function.
14778 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14779 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14780 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14781 (riscv_hard_regno_mode_ok): ...this and make static.
14782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14783 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14784 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14785 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14786 (rl78_hard_regno_mode_ok): Make static and return bool.
14787 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14788 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14790 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14791 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14792 (rs6000_hard_regno_mode_ok): Rename to...
14793 (rs6000_hard_regno_mode_ok_uncached): ...this.
14794 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14795 (rs6000_hard_regno_mode_ok): New function.
14796 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14797 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14798 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14799 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14800 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14801 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14802 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14803 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14804 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14805 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14806 (sh_hard_regno_mode_ok): Make static.
14807 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14808 instead of HARD_REGNO_MODE_OK.
14809 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14810 (sparc_mode_class): Delete.
14811 (HARD_REGNO_MODE_OK): Delete.
14812 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14813 (hard_regno_mode_classes): Make static.
14814 (sparc_mode_class): Likewise.
14815 (sparc_hard_regno_mode_ok): New function.
14816 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14817 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14818 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14820 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14821 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14822 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14823 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14824 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14825 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14826 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14827 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14828 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14829 (visium_hard_regno_mode_ok): New function.
14830 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14831 instead of HARD_REGNO_MODE_OK.
14832 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14833 (HARD_REGNO_MODE_OK): Delete.
14834 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14835 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14836 (xtensa_option_override): Update accordingly.
14837 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14838 (xtensa_hard_regno_mode_ok): New function.
14839 * system.h (HARD_REGNO_MODE_OK): 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 * target.def (hard_regno_call_part_clobbered): New hook.
14846 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14847 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14848 * doc/tm.texi: Regenerate.
14849 * hooks.h (hook_bool_uint_mode_false): Declare.
14850 * hooks.c (hook_bool_uint_mode_false): New function.
14851 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14852 * cselib.c (cselib_process_insn): Use
14853 targetm.hard_regno_call_part_clobbered instead of
14854 HARD_REGNO_CALL_PART_CLOBBERED.
14855 * ira-conflicts.c (ira_build_conflicts): Likewise.
14856 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14857 * lra-constraints.c (need_for_call_save_p): Likewise.
14858 * lra-lives.c: Include target.h.
14859 (check_pseudos_live_through_calls): Use
14860 targetm.hard_regno_call_part_clobbered instead of
14861 HARD_REGNO_CALL_PART_CLOBBERED.
14862 * regcprop.c: Include target.h.
14863 (copyprop_hardreg_forward_1): Use
14864 targetm.hard_regno_call_part_clobbered instead of
14865 HARD_REGNO_CALL_PART_CLOBBERED.
14866 * reginfo.c (choose_hard_reg_mode): Likewise.
14867 * regrename.c (check_new_reg_p): Likewise.
14868 * reload.c (find_equiv_reg): Likewise.
14869 * reload1.c (emit_reload_insns): Likewise.
14870 * sched-deps.c (deps_analyze_insn): Likewise.
14871 * sel-sched.c (init_regs_for_mode): Likewise.
14872 (mark_unavailable_hard_regs): Likewise.
14873 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14874 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14875 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14877 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14878 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14879 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14881 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14883 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14884 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14885 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14887 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14888 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14889 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14891 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14892 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14894 * config/powerpcspe/powerpcspe.c
14895 (rs6000_hard_regno_call_part_clobbered): New function.
14896 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14897 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14898 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14900 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14901 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14902 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14904 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14905 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14906 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14908 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14909 Alan Hayward <alan.hayward@arm.com>
14910 David Sherwood <david.sherwood@arm.com>
14912 * rtl.h (subreg_memory_offset): Declare.
14913 * emit-rtl.c (subreg_memory_offset): New function.
14914 * expmed.c (store_bit_field_1): Use it.
14915 * expr.c (undefined_operand_subword_p): Likewise.
14916 * simplify-rtx.c (simplify_subreg): Likewise.
14918 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14920 PR rtl-optimization/57448
14923 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14924 using atomic_load pattern.
14925 (expand_atomic_store): Likewise.
14927 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14930 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14931 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14932 handling. Use replace_call_with_value with NULL instead of
14933 gsi_replace, unlink_stmt_vdef and release_defs.
14935 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14938 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14940 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14943 * rtl.h (emit_library_call_value_1): Declare.
14944 (emit_library_call): Replace declaration with a series of overloads.
14945 Remove the parameter count argument.
14946 (emit_library_call_value): Likewise.
14947 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14948 with an "rtx_mode_t *".
14949 (emit_library_call_value): Delete.
14950 (emit_library_call): Likewise.
14951 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14952 (asan_emit_allocas_unpoison): Likewise.
14953 * builtins.c (expand_builtin_powi): Likewise.
14954 (expand_asan_emit_allocas_unpoison): Likewise.
14955 * cfgexpand.c (expand_main_function): Likewise.
14956 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14957 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14958 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14959 * config/arm/arm.c (arm_trampoline_init): Likewise.
14960 (arm_call_tls_get_addr): Likewise.
14961 (arm_expand_divmod_libfunc): Likewise.
14962 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14963 (smulsi3_highpart): Likewise.
14964 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14965 (c6x_expand_compare): Likewise.
14966 (c6x_expand_movmem): Likewise.
14967 * config/frv/frv.c (frv_trampoline_init): Likewise.
14968 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14969 (ix86_expand_divmod_libfunc): Likewise.
14970 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14971 (ia64_expand_compare): Likewise.
14972 (ia64_profile_hook): Likewise.
14973 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14974 (nonlocal_goto): Likewise.
14975 (restore_stack_nonlocal): Likewise.
14976 * config/m32r/m32r.c (block_move_call): Likewise.
14977 (m32r_trampoline_init): Likewise.
14978 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14979 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14980 (m68k_call_m68k_read_tp): Likewise.
14981 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14982 (microblaze_expand_divide): Likewise.
14983 * config/mips/mips.h (mips_args): Likewise.
14984 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14985 (MIPS_ICACHE_SYNC): Likewise.
14986 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14987 (nios2_trampoline_init): Likewise.
14988 * config/pa/pa.c (hppa_tls_call): Likewise.
14989 (pa_trampoline_init): Likewise.
14990 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14991 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14992 (expand_strn_compare): Likewise.
14993 (rs6000_generate_compare): Likewise.
14994 (rs6000_expand_float128_convert): Likewise.
14995 (output_profile_hook): Likewise.
14996 (rs6000_trampoline_init): Likewise.
14997 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14998 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14999 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15000 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15001 (rs6000_generate_compare): Likewise.
15002 (rs6000_expand_float128_convert): Likewise.
15003 (output_profile_hook): Likewise.
15004 (rs6000_trampoline_init): Likewise.
15005 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15006 * config/sh/sh.c (sh_trampoline_init): Likewise.
15007 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15008 (sparc_emit_float_lib_cmp): Likewise.
15009 (sparc32_initialize_trampoline): Likewise.
15010 (sparc64_initialize_trampoline): Likewise.
15011 (sparc_profile_hook): Likewise.
15012 * config/spu/spu.c (ea_load_store): Likewise.
15013 * config/spu/spu.md (floatunssidf2): Likewise.
15014 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15015 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15016 * config/visium/visium.c (expand_block_move_4): Likewise.
15017 (expand_block_move_2): Likewise.
15018 (expand_block_move_1): Likewise.
15019 (expand_block_set_4): Likewise.
15020 (expand_block_set_2): Likewise.
15021 (expand_block_set_1): Likewise.
15022 (visium_trampoline_init): Likewise.
15023 (visium_profile_hook): Likewise.
15024 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15025 (xtensa_setup_frame_addresses): Likewise.
15026 (xtensa_trampoline_init): Likewise.
15027 * except.c (sjlj_emit_function_enter): Likewise.
15028 (sjlj_emit_function_exit): Likewise.
15029 * explow.c (allocate_dynamic_stack_space): Likewise.
15030 (probe_stack_range): Likewise.
15031 * expr.c (convert_mode_scalar): Likewise.
15032 * optabs.c (expand_binop): Likewise.
15033 (expand_twoval_binop_libfunc): Likewise.
15034 (expand_unop): Likewise.
15035 (prepare_cmp_insn): Likewise.
15036 (prepare_float_lib_cmp): Likewise.
15037 (expand_float): Likewise.
15038 (expand_fix): Likewise.
15039 (expand_fixed_convert): Likewise.
15040 (maybe_emit_sync_lock_test_and_set): Likewise.
15041 (expand_atomic_compare_and_swap): Likewise.
15042 (expand_mem_thread_fence): Likewise.
15043 (expand_atomic_fetch_op): Likewise.
15045 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15047 * doc/generic.texi (OpenACC): Adjust URL.
15048 * doc/invoke.texi (C Dialect Options): Ditto.
15050 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15052 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15053 predicate for operand 1. Add (m,<S>) constraint.
15054 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15055 Prevent memory operand 1 with register operand 2.
15057 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15059 PR rtl-optimization/82024
15060 * combine.c (try_combine): If the combination result is a PARALLEL,
15061 and we only need to retain the SET in there that would be placed
15062 at I2, check that we can place that at I3 instead, before doing so.
15064 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15067 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15069 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15070 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15073 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15074 Jeff Law <law@redhat.com>
15076 * varasm.c (bss_initializer_p): Do not put constants into .bss
15077 (categorize_decl_for_section): Handle bss_initializer_p returning
15078 false when DECL_INITIAL is NULL.
15080 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15083 * config/s390/s390.c (s390_can_inline_p): New function.
15085 2017-09-01 Jeff Law <law@redhat.com>
15087 PR tree-optimization/82052
15088 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15089 Always initialize the returned slot after a hash table miss
15090 when INSERT is true.
15092 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15094 * config/s390/s390.md (mem_signal_fence): Remove.
15095 * doc/md.texi (mem_signal_fence): Remove.
15096 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15098 * target-insns.def (mem_signal_fence): Remove.
15100 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15103 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15106 * asan.c (create_odr_indicator): Strip name encoding from assembler
15107 name before appending it after __odr_asan_.
15109 2017-09-01 Martin Liska <mliska@suse.cz>
15111 PR tree-optimization/82059
15112 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15113 frequency only when an edge is redirected.
15115 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15117 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15118 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15119 (arc_conditional_register_usage): Remove ARC600 lp_count
15121 (arc_file_start): Emit Tag_ARC_CPU_variation.
15122 (arc_can_use_doloop_p): New conditions to use ZOLs.
15123 (hwloop_fail): New function.
15124 (hwloop_optimize): Likewise.
15125 (hwloop_pattern_reg): Likewise.
15126 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15127 (arc_reorg_loops): New function, calls reorg_loops.
15128 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15129 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15131 (arc_loop_hazard): Remove function, functionality moved into
15133 (arc_hazard): Remove arc_loop_hazard call.
15134 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15135 into hwloop_optimize.
15136 (arc_label_align): Remove ZOL handling.
15137 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15138 * config/arc/arc.md (doloop_begin): Remove pattern.
15139 (doloop_begin_i): Likewise.
15140 (doloop_end_i): Likewise.
15141 (doloop_fallback): Likewise.
15142 (doloop_fallback_m): Likewise.
15143 (doloop_end): Reimplement expand.
15144 (arc_lp): New pattern for LP instruction.
15145 (loop_end): New pattern.
15146 (loop_fail): Likewise.
15147 (decrement_and_branch_until_zero): Likewise.
15148 * config/arc/arc.opt (mlpc-width): New option.
15149 * doc/invoke.texi (mlpc-width): Document option.
15151 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15153 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15154 (arc_ccfsm_advance): Fix checking for delay slots.
15155 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15157 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15159 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15161 (movhi_insn): Update store instruction constraint which are saving
15162 6-bit short immediates.
15163 (movsi_insn): Consider also short scaled load operations.
15164 (zero_extendhisi2_i): Use Usd constraint instead of T.
15165 (extendhisi2_i): Add q constraint.
15166 (arc_clzsi2): Add type and length attributes.
15167 (arc_ctzsi2): Likewise.
15168 * config/arc/constraints.md (Usc): Update constraint, the
15169 assembler can parse two relocations for a single instruction.
15171 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15173 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15174 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15176 2017-08-31 Olivier Hainque <hainque@adacore.com>
15178 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15179 match as powerpc-wrs-vxworks*.
15181 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15183 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15184 register constraint for by-element operand.
15185 (aarch64_mls_elt_merge<mode>): Likewise.
15187 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15189 * config/arc/arc.c (arc_can_follow_jump): Check for short
15192 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15194 * config.gcc: Use g.opt for arc.
15195 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15196 functionality moved to ...
15197 (legitimate_scaled_address_p): New function, ...here.
15198 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15199 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15200 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15202 (arc_override_options): Handle G option.
15203 (arc_output_pic_addr_const): Correct function definition.
15204 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15205 (arc_decl_anon_ns_mem_p): Delete.
15206 (arc_in_small_data_p): Overhaul this function to take into
15207 consideration the value given via G option.
15208 (arc_rewrite_small_data_1): Renamed and corrected old
15209 arc_rewrite_small_data function.
15210 (arc_rewrite_small_data): New function.
15211 (small_data_pattern): Don't use pic_offset_table_rtx.
15212 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15213 * config/arc/simdext.md (movmisalignv2hi): Use
15214 prepare_move_operands function.
15216 (movmisalign*): Likewise.
15217 * doc/invoke.texi (ARC options): Document -G option.
15219 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15221 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15223 * config/arc/arc.c (arc_print_operand): Output scalled address for
15224 sdata whenever is possible.
15225 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15226 load/stores are available.
15227 (compact_sda_memory_operand): Check for the alignment required by
15228 code density instructions.
15229 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15231 * config/arc/constraints.md (Usd): Update constraint.
15232 (Us0): New constraint.
15233 (Usc): Update constraint.
15235 2017-08-31 Richard Biener <rguenther@suse.de>
15237 PR middle-end/82054
15238 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15239 function only once.
15241 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15243 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15244 Resize type_signature.
15246 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15247 Alan Hayward <alan.hayward@arm.com>
15248 David Sherwood <david.sherwood@arm.com>
15250 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15251 subregs whose inner modes can be stored in GPRs.
15252 (aarch64_classify_index): Likewise.
15254 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15255 Alan Hayward <alan.hayward@arm.com>
15256 David Sherwood <david.sherwood@arm.com>
15258 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15259 (V_INT_EQUIV): ...this.
15260 (v_cmp_result): Rename to...
15261 (v_int_equiv): ...this.
15262 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15263 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15264 (copysign<mode>3): Likewise.
15265 (aarch64_simd_bsl<mode>_internal): Likewise.
15266 (aarch64_simd_bsl<mode>): Likewise.
15267 (vec_cmp<mode><mode>): Likewise.
15268 (vcond<mode><mode>): Likewise.
15269 (vcond<v_cmp_mixed><mode>): Likewise.
15270 (vcondu<mode><v_cmp_mixed>): Likewise.
15271 (aarch64_cm<optab><mode>): Likewise.
15272 (aarch64_cmtst<mode>): Likewise.
15273 (aarch64_fac<optab><mode>): Likewise.
15274 (vec_perm_const<mode>): Likewise.
15275 (vcond_mask_<mode><v_cmp_result>): Rename to...
15276 (vcond_mask_<mode><v_int_equiv>): ...this.
15277 (vec_cmp<mode><v_cmp_result>): Rename to...
15278 (vec_cmp<mode><v_int_equiv>): ...this.
15280 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15281 Alan Hayward <alan.hayward@arm.com>
15282 David Sherwood <david.sherwood@arm.com>
15284 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15286 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15287 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15288 (UNSPEC_LD4_DREG): New unspecs.
15289 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15290 (aarch64_ld2<mode>_dreg_be): Replace with...
15291 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15293 (aarch64_ld3<mode>_dreg_le)
15294 (aarch64_ld3<mode>_dreg_be): Replace with...
15295 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15297 (aarch64_ld4<mode>_dreg_le)
15298 (aarch64_ld4<mode>_dreg_be): Replace with...
15299 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15302 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15304 PR tree-optimization/81987
15305 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15306 insert an initializer in a location not dominated by the stride
15309 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15311 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15312 on the entire header of the finally block in the fallthru case.
15314 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15316 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15318 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15320 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15321 rs6000_emit_move_from_cr and call renamed function.
15322 (rs6000_emit_prologue): Call renamed functions.
15323 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15324 movesi_from_cr, remove volatile CRs.
15326 2017-08-30 Jon Beniston <jon@beniston.com>
15327 Richard Biener <rguenther@suse.de>
15329 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15330 of VECTOR_MODE_P check.
15331 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15332 element vector types.
15334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15336 * df.h (df_read_modify_subreg_p): Remove in favor of...
15337 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15338 const_rtx instead of an rtx.
15339 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15340 * df-problems.c (df_word_lr_mark_ref): Likewise.
15341 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15342 (mark_pseudo_reg_dead): Likewise.
15343 (mark_ref_dead): Likewise.
15344 * reginfo.c (init_subregs_of_mode): Likewise.
15345 * sched-deps.c (sched_analyze_1): Likewise.
15346 * df-scan.c (df_def_record_1): Likewise.
15347 (df_uses_record): Likewise.
15348 (df_read_modify_subreg_p): Remove in favor of...
15349 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15350 const_rtx instead of an rtx.
15352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15353 Alan Hayward <alan.hayward@arm.com>
15354 David Sherwood <david.sherwood@arm.com>
15356 * rtl.h (partial_subreg_p): New function.
15357 * caller-save.c (save_call_clobbered_regs): Use it.
15358 * calls.c (expand_call): Likewise.
15359 * combine.c (combinable_i3pat): Likewise.
15360 (simplify_set): Likewise.
15361 (make_extraction): Likewise.
15362 (make_compound_operation_int): Likewise.
15363 (gen_lowpart_or_truncate): Likewise.
15364 (force_to_mode): Likewise.
15365 (make_field_assignment): Likewise.
15366 (reg_truncated_to_mode): Likewise.
15367 (record_truncated_value): Likewise.
15368 (move_deaths): Likewise.
15369 * cse.c (record_jump_cond): Likewise.
15370 (cse_insn): Likewise.
15371 * cselib.c (cselib_lookup_1): Likewise.
15372 * expmed.c (extract_bit_field_using_extv): Likewise.
15373 * function.c (assign_parm_setup_reg): Likewise.
15374 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15375 * ira-build.c (create_insn_allocnos): Likewise.
15376 * lra-coalesce.c (merge_pseudos): Likewise.
15377 * lra-constraints.c (match_reload): Likewise.
15378 (simplify_operand_subreg): Likewise.
15379 (curr_insn_transform): Likewise.
15380 * lra-lives.c (process_bb_lives): Likewise.
15381 * lra.c (new_insn_reg): Likewise.
15382 (lra_substitute_pseudo): Likewise.
15383 * regcprop.c (mode_change_ok): Likewise.
15384 (maybe_mode_change): Likewise.
15385 (copyprop_hardreg_forward_1): Likewise.
15386 * reload.c (push_reload): Likewise.
15387 (find_reloads): Likewise.
15388 (find_reloads_subreg_address): Likewise.
15389 * reload1.c (alter_reg): Likewise.
15390 (eliminate_regs_1): Likewise.
15391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15393 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15395 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15396 back to if statements, including unpack.
15398 2017-08-30 Martin Liska <mliska@suse.cz>
15400 PR inline-asm/82001
15401 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15403 (func_checker::compare_asm_inputs_outputs): ... this function.
15404 (func_checker::compare_gimple_asm): Use the function to compare
15405 also ASM constrains.
15406 * ipa-icf-gimple.h: Rename the function.
15408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15409 Alan Hayward <alan.hayward@arm.com>
15410 David Sherwood <david.sherwood@arm.com>
15412 * coretypes.h (complex_mode): New type.
15413 * gdbhooks.py (build_pretty_printer): Handle it.
15414 * machmode.h (complex_mode): New class.
15415 (complex_mode::includes_p): New function.
15416 (is_complex_int_mode): Likewise.
15417 (is_complex_float_mode): Likewise.
15418 * genmodes.c (get_mode_class): Handle complex mode classes.
15419 * function.c (expand_function_end): Use is_complex_int_mode.
15421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15422 Alan Hayward <alan.hayward@arm.com>
15423 David Sherwood <david.sherwood@arm.com>
15425 * coretypes.h (scalar_mode_pod): New typedef.
15426 * gdbhooks.py (build_pretty_printer): Handle it.
15427 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15428 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15429 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15430 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15431 as_a <scalar_mode>.
15433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15434 Alan Hayward <alan.hayward@arm.com>
15435 David Sherwood <david.sherwood@arm.com>
15437 * machmode.h (mode_for_vector): Take a scalar_mode instead
15439 * stor-layout.c (mode_for_vector): Likewise.
15440 * explow.c (promote_mode): Use as_a <scalar_mode>.
15441 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15444 Alan Hayward <alan.hayward@arm.com>
15445 David Sherwood <david.sherwood@arm.com>
15447 * target.def (preferred_simd_mode): Take a scalar_mode
15448 instead of a machine_mode.
15449 * targhooks.h (default_preferred_simd_mode): Likewise.
15450 * targhooks.c (default_preferred_simd_mode): Likewise.
15451 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15452 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15453 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15454 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15455 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15456 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15457 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15458 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15460 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15461 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15462 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15463 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15464 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15465 * doc/tm.texi: Regenerate.
15466 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15470 Alan Hayward <alan.hayward@arm.com>
15471 David Sherwood <david.sherwood@arm.com>
15473 * target.def (scalar_mode_supported_p): Take a scalar_mode
15474 instead of a machine_mode.
15475 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15476 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15477 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15478 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15479 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15480 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15481 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15482 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15483 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15484 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15485 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15486 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15487 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15488 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15490 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15491 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15492 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15493 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15494 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15496 * doc/tm.texi: Regenerate.
15498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15499 Alan Hayward <alan.hayward@arm.com>
15500 David Sherwood <david.sherwood@arm.com>
15502 * coretypes.h (opt_scalar_mode): New typedef.
15503 * gdbhooks.py (build_pretty_printers): Handle it.
15504 * machmode.h (mode_iterator::get_2xwider): Add overload for
15506 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15508 * expr.c (convert_mode_scalar): Likewise.
15509 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15510 * optabs.c (expand_float): Likewise.
15511 (expand_fix): Likewise.
15512 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15515 Alan Hayward <alan.hayward@arm.com>
15516 David Sherwood <david.sherwood@arm.com>
15518 * optabs.c (expand_float): Explicitly check for scalars before
15519 using a branching expansion.
15520 (expand_fix): Likewise.
15522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15523 Alan Hayward <alan.hayward@arm.com>
15524 David Sherwood <david.sherwood@arm.com>
15526 * expr.c (convert_mode): Split scalar handling out into...
15527 (convert_mode_scalar): ...this new function. Treat the modes
15530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15531 Alan Hayward <alan.hayward@arm.com>
15532 David Sherwood <david.sherwood@arm.com>
15534 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15536 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15539 Alan Hayward <alan.hayward@arm.com>
15540 David Sherwood <david.sherwood@arm.com>
15542 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15543 rather than a machine_mode.
15544 (fixed_from_string): Likewise.
15545 (fixed_convert): Likewise.
15546 (fixed_convert_from_int): Likewise.
15547 (fixed_convert_from_real): Likewise.
15548 (real_convert_from_fixed): Likewise.
15549 * fixed-value.c (fixed_from_double_int): Likewise.
15550 (fixed_from_string): Likewise.
15551 (fixed_convert): Likewise.
15552 (fixed_convert_from_int): Likewise.
15553 (fixed_convert_from_real): Likewise.
15554 (real_convert_from_fixed): Likewise.
15555 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15558 Alan Hayward <alan.hayward@arm.com>
15559 David Sherwood <david.sherwood@arm.com>
15561 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15562 of separate mode class checks. Do not allow vector modes here.
15563 (immed_wide_int_const): Use as_a <scalar_mode>.
15564 * explow.c (trunc_int_for_mode): Likewise.
15565 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15566 (wi::shwi): Likewise.
15567 (wi::min_value): Likewise.
15568 (wi::max_value): Likewise.
15569 * dwarf2out.c (loc_descriptor): Likewise.
15570 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15571 for CONST_WIDE_INT.
15573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15574 Alan Hayward <alan.hayward@arm.com>
15575 David Sherwood <david.sherwood@arm.com>
15577 * tree.h (SCALAR_TYPE_MODE): New macro.
15578 * expr.c (expand_expr_addr_expr_1): Use it.
15579 (expand_expr_real_2): Likewise.
15580 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15581 (fold_convert_const_fixed_from_int): Likewise.
15582 (fold_convert_const_fixed_from_real): Likewise.
15583 (native_encode_fixed): Likewise
15584 (native_encode_complex): Likewise
15585 (native_encode_vector): Likewise.
15586 (native_interpret_fixed): Likewise.
15587 (native_interpret_real): Likewise.
15588 (native_interpret_complex): Likewise.
15589 (native_interpret_vector): Likewise.
15590 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15591 (simd_clone_adjust_argument_types): Likewise.
15592 (simd_clone_init_simd_arrays): Likewise.
15593 (simd_clone_adjust): Likewise.
15594 * stor-layout.c (layout_type): Likewise.
15595 * tree.c (build_minus_one_cst): Likewise.
15596 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15597 * tree-inline.c (estimate_move_cost): Likewise.
15598 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15599 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15600 (vectorizable_reduction): Likewise.
15601 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15602 (vect_recog_mixed_size_cond_pattern): Likewise.
15603 (check_bool_pattern): Likewise.
15604 (adjust_bool_pattern): Likewise.
15605 (search_type_for_mask_1): Likewise.
15606 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15607 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15608 (vectorizable_load): Likewise.
15609 (vectorizable_store): Likewise.
15610 * ubsan.c (ubsan_encode_value): Likewise.
15611 * varasm.c (output_constant): Likewise.
15613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15614 Alan Hayward <alan.hayward@arm.com>
15615 David Sherwood <david.sherwood@arm.com>
15617 * coretypes.h (scalar_mode): New class.
15618 * machmode.h (scalar_mode): Likewise.
15619 (scalar_mode::includes_p): New function.
15620 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15621 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15622 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15623 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15624 * expmed.c (store_bit_field_1): Likewise.
15625 (extract_bit_field_1): Likewise.
15626 * expr.c (write_complex_part): Likewise.
15627 (read_complex_part): Likewise.
15628 (emit_move_complex_push): Likewise.
15629 (expand_expr_real_2): Likewise.
15630 * function.c (assign_parm_setup_reg): Likewise.
15631 (assign_parms_unsplit_complex): Likewise.
15632 * optabs.c (expand_binop): Likewise.
15633 * rtlanal.c (subreg_get_info): Likewise.
15634 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15635 * varasm.c (output_constant_pool_2): Likewise.
15637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15638 Alan Hayward <alan.hayward@arm.com>
15639 David Sherwood <david.sherwood@arm.com>
15641 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15643 (expmed_mult_highpart_optab): Likewise.
15644 (expmed_mult_highpart): Likewise.
15646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15647 Alan Hayward <alan.hayward@arm.com>
15648 David Sherwood <david.sherwood@arm.com>
15650 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15651 instead of a machine_mode.
15652 (builtin_memset_read_str): Likewise.
15653 * builtins.c (c_readstr): Likewise.
15654 (builtin_memcpy_read_str): Likewise.
15655 (builtin_strncpy_read_str): Likewise.
15656 (builtin_memset_read_str): Likewise.
15657 (builtin_memset_gen_str): Likewise.
15658 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15659 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15660 instead of a machine_mode.
15661 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15663 (make_extraction): Likewise.
15664 (try_widen_shift_mode): Take and return scalar_int_modes instead
15666 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15667 a scalar_int_mode instead of a machine_mode.
15668 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15669 (avr_addr_space_pointer_mode): Likewise.
15670 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15671 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15672 (msp430_unwind_word_mode): Likewise.
15673 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15674 (spu_addr_space_pointer_mode): Likewise.
15675 (spu_addr_space_address_mode): Likewise.
15676 (spu_libgcc_cmp_return_mode): Likewise.
15677 (spu_libgcc_shift_count_mode): Likewise.
15678 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15679 (rl78_addr_space_pointer_mode): Likewise.
15680 (fl78_unwind_word_mode): Likewise.
15681 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15683 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15684 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15685 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15686 (mips_valid_pointer_mode): Likewise.
15687 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15688 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15689 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15691 (ft32_addr_space_address_mode): Likewise.
15692 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15693 scalar_int_mode instead of a machine_mode.
15694 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15696 (m32c_addr_space_address_mode): Likewise.
15697 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15698 (rs6000_eh_return_filter_mode): Likewise.
15699 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15700 (rs6000_eh_return_filter_mode): Likewise.
15701 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15702 (s390_libgcc_shift_count_mode): Likewise.
15703 (s390_unwind_word_mode): Likewise.
15704 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15706 * target.def (mode_rep_extended): Likewise.
15707 (valid_pointer_mode): Likewise.
15708 (addr_space.valid_pointer_mode): Likewise.
15709 (eh_return_filter_mode): Return a scalar_int_mode rather than
15711 (libgcc_cmp_return_mode): Likewise.
15712 (libgcc_shift_count_mode): Likewise.
15713 (unwind_word_mode): Likewise.
15714 (addr_space.pointer_mode): Likewise.
15715 (addr_space.address_mode): Likewise.
15716 * doc/tm.texi: Regenerate.
15717 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15719 (do_jump): Use scalar_int_mode for local variables.
15720 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15721 rather than a machine_mode.
15722 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15723 (scompare_loc_descriptor_wide): Likewise.
15724 (scompare_loc_descriptor_narrow): Likewise.
15725 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15727 * except.c (sjlj_emit_dispatch_table): Likewise.
15728 (expand_builtin_eh_copy_values): Likewise.
15729 * explow.c (convert_memory_address_addr_space_1): Likewise.
15730 Take a scalar_int_mode rather than a machine_mode.
15731 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15732 than a machine_mode.
15733 (memory_address_addr_space): Use scalar_int_mode for local variables.
15734 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15735 rather than a machine_mode.
15736 * expmed.c (mask_rtx): Likewise.
15737 (init_expmed_one_conv): Likewise.
15738 (expand_mult_highpart_adjust): Likewise.
15739 (extract_high_half): Likewise.
15740 (expmed_mult_highpart_optab): Likewise.
15741 (expmed_mult_highpart): Likewise.
15742 (expand_smod_pow2): Likewise.
15743 (expand_sdiv_pow2): Likewise.
15744 (emit_store_flag_int): Likewise.
15745 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15747 (extract_low_bits): Likewise.
15748 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15750 * expr.c (pieces_addr::adjust): Likewise.
15751 (can_store_by_pieces): Likewise.
15752 (store_by_pieces): Likewise.
15753 (clear_by_pieces_1): Likewise.
15754 (expand_expr_addr_expr_1): Likewise.
15755 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15756 (expand_expr_real_1): Likewise.
15757 (try_casesi): Likewise.
15758 * final.c (shorten_branches): Likewise.
15759 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15760 type of "mode" to machine_mode.
15761 * internal-fn.c (expand_arith_overflow_result_store): Take a
15762 scalar_int_mode rather than a machine_mode.
15763 (expand_mul_overflow): Use scalar_int_mode for local variables.
15764 * loop-doloop.c (doloop_modify): Likewise.
15765 (doloop_optimize): Likewise.
15766 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15767 than a machine_mode.
15768 (expand_doubleword_shift_condmove): Likewise.
15769 (expand_doubleword_shift): Likewise.
15770 (expand_doubleword_clz): Likewise.
15771 (expand_doubleword_popcount): Likewise.
15772 (expand_doubleword_parity): Likewise.
15773 (expand_absneg_bit): Use scalar_int_mode for local variables.
15774 (prepare_float_lib_cmp): Likewise.
15775 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15776 rather than a machine_mode.
15777 (convert_memory_address_addr_space): Likewise.
15778 (get_mode_bounds): Likewise.
15779 (get_address_mode): Return a scalar_int_mode rather than a
15781 * rtlanal.c (get_address_mode): Likewise.
15782 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15783 than a machine_mode.
15784 * targhooks.c (default_mode_rep_extended): Likewise.
15785 (default_valid_pointer_mode): Likewise.
15786 (default_addr_space_valid_pointer_mode): Likewise.
15787 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15788 than a machine_mode.
15789 (default_libgcc_cmp_return_mode): Likewise.
15790 (default_libgcc_shift_count_mode): Likewise.
15791 (default_unwind_word_mode): Likewise.
15792 (default_addr_space_pointer_mode): Likewise.
15793 (default_addr_space_address_mode): Likewise.
15794 * targhooks.h (default_eh_return_filter_mode): Likewise.
15795 (default_libgcc_cmp_return_mode): Likewise.
15796 (default_libgcc_shift_count_mode): Likewise.
15797 (default_unwind_word_mode): Likewise.
15798 (default_addr_space_pointer_mode): Likewise.
15799 (default_addr_space_address_mode): Likewise.
15800 (default_mode_rep_extended): Take a scalar_int_mode rather than
15802 (default_valid_pointer_mode): Likewise.
15803 (default_addr_space_valid_pointer_mode): Likewise.
15804 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15806 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15807 rather than a machine_mode.
15808 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15809 for local variables.
15810 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15811 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15812 than a machine_mode.
15814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15815 Alan Hayward <alan.hayward@arm.com>
15816 David Sherwood <david.sherwood@arm.com>
15818 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15819 the mode argument to scalar_int_mode.
15820 (do_jump_by_parts_zero_rtx): Likewise.
15821 (do_jump_by_parts_equality_rtx): Likewise.
15822 (do_jump_by_parts_greater): Take a mode argument.
15823 (do_jump_by_parts_equality): Likewise.
15824 (do_jump_1): Update calls accordingly.
15826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15827 Alan Hayward <alan.hayward@arm.com>
15828 David Sherwood <david.sherwood@arm.com>
15830 * is-a.h (safe_dyn_cast): New function.
15831 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15832 (jump_table_for_label): Likewise.
15833 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15834 instead of an rtx_insn *.
15835 (shorten_branches): Use dyn_cast instead of LABEL_P and
15836 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15837 rtx_jump_table_data::get_data_mode.
15838 (final_scan_insn): Likewise.
15840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15841 Alan Hayward <alan.hayward@arm.com>
15842 David Sherwood <david.sherwood@arm.com>
15844 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15845 trying to combine a full-register integer set with a subreg
15848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15849 Alan Hayward <alan.hayward@arm.com>
15850 David Sherwood <david.sherwood@arm.com>
15852 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15853 that is always either address_mode or pointer_mode.
15855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15856 Alan Hayward <alan.hayward@arm.com>
15857 David Sherwood <david.sherwood@arm.com>
15859 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15860 when the two are known to be equal.
15862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15863 Alan Hayward <alan.hayward@arm.com>
15864 David Sherwood <david.sherwood@arm.com>
15866 * simplify-rtx.c (simplify_const_unary_operation): Use
15867 is_a <scalar_int_mode> instead of checking for a nonzero
15868 precision. Forcibly convert op_mode to a scalar_int_mode
15869 in that case. More clearly differentiate the operand and
15870 result modes and use the former when deciding what the value
15871 of a count-bits operation should be. Use is_int_mode instead
15872 of checking for a MODE_INT. Remove redundant check for whether
15873 this mode has a zero precision.
15875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15876 Alan Hayward <alan.hayward@arm.com>
15877 David Sherwood <david.sherwood@arm.com>
15879 * optabs.c (widen_leading): Change the type of the mode argument
15880 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15881 (widen_bswap): Likewise.
15882 (expand_parity): Likewise.
15883 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15884 (expand_ffs): Likewise.
15885 (epand_unop): Check for scalar integer modes before calling the
15888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15889 Alan Hayward <alan.hayward@arm.com>
15890 David Sherwood <david.sherwood@arm.com>
15892 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15894 (expand_expr_real_1): Update call accordingly.
15896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15897 Alan Hayward <alan.hayward@arm.com>
15898 David Sherwood <david.sherwood@arm.com>
15900 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15901 value_mode arguments. Use scalar_int_mode internally.
15902 (store_bit_field_1): Rename the new integer mode from imode
15903 to op0_mode and use it instead of GET_MODE (op0). Update calls
15904 to store_split_bit_field, store_bit_field_using_insv and
15905 store_fixed_bit_field.
15906 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15907 Use scalar_int_mode internally. Use a bit count rather than a mode
15908 when calculating the largest bit size for get_best_mode.
15909 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15910 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15911 Remove assertion that OP0 has a scalar integer mode.
15912 (store_split_bit_field): Add op0_mode and value_mode arguments.
15913 Update calls to extract_fixed_bit_field.
15914 (extract_bit_field_using_extv): Add an op0_mode argument.
15915 Use scalar_int_mode internally.
15916 (extract_bit_field_1): Rename the new integer mode from imode to
15917 op0_mode and use it instead of GET_MODE (op0). Update calls to
15918 extract_split_bit_field, extract_bit_field_using_extv and
15919 extract_fixed_bit_field.
15920 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15921 to extract_split_bit_field and extract_fixed_bit_field_1.
15922 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15923 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15924 on the target mode.
15925 (extract_split_bit_field): Add an op0_mode argument. Update call
15926 to extract_fixed_bit_field.
15928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15929 Alan Hayward <alan.hayward@arm.com>
15930 David Sherwood <david.sherwood@arm.com>
15932 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15933 * explow.c (hard_function_value): Likewise.
15934 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15935 convert_to_mode call outside the loop.
15936 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15937 for the mode iterator. Require the mode specified by max_pieces
15939 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15941 (copy_blkmode_to_reg): Likewise.
15942 (set_storage_via_setmem): Likewise.
15943 * optabs.c (prepare_cmp_insn): Likewise.
15944 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15945 * stor-layout.c (finish_bitfield_representative): Likewise.
15947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15948 Alan Hayward <alan.hayward@arm.com>
15949 David Sherwood <david.sherwood@arm.com>
15951 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15952 * expr.c (convert_move): Use them.
15953 (convert_modes): Likewise.
15954 (store_expr_with_bounds): Likewise.
15956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15957 Alan Hayward <alan.hayward@arm.com>
15958 David Sherwood <david.sherwood@arm.com>
15960 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15961 parameter for the mode of "x". Remove the "known_x", "known_mode"
15962 and "known_ret" arguments. Change the type of the mode argument
15963 to scalar_int_mode.
15964 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15965 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15966 (reg_num_sign_bit_copies_for_combine): Likewise.
15967 * rtlanal.c (nonzero_bits1): Likewise.
15968 (num_sign_bit_copies1): Likewise.
15969 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15970 (reg_num_sign_bit_copies_general): Likewise.
15971 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15972 (reg_nonzero_bits_general): Likewise.
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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15979 than in subroutines. Return 1 for non-integer modes.
15980 (cached_num_sign_bit_copies): Change the type of the mode parameter
15981 to scalar_int_mode.
15982 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15983 classes. Handle CONST_INT_P first and then check whether X also
15984 has a scalar integer mode. Check the same thing for inner registers
15985 of a SUBREG and for values that are being extended or truncated.
15987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15988 Alan Hayward <alan.hayward@arm.com>
15989 David Sherwood <david.sherwood@arm.com>
15991 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15992 in subroutines. Return the mode mask for non-integer modes.
15993 (cached_nonzero_bits): Change the type of the mode parameter
15994 to scalar_int_mode.
15995 (nonzero_bits1): Likewise. Remove early exit for other mode
15996 classes. Handle CONST_INT_P first and then check whether X
15997 also has a scalar integer mode.
15999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16000 Alan Hayward <alan.hayward@arm.com>
16001 David Sherwood <david.sherwood@arm.com>
16003 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16004 Return a scalar_int_mode and assert that the size is greater than
16006 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16007 (op_by_pieces_d::op_by_pieces_d): Likewise.
16008 (op_by_pieces_d::run): Likewise.
16009 (can_store_by_pieces): Likewise.
16011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16012 Alan Hayward <alan.hayward@arm.com>
16013 David Sherwood <david.sherwood@arm.com>
16015 * combine.c (extract_left_shift): Add a mode argument and update
16017 (make_compound_operation_int): Change the type of the mode parameter
16018 to scalar_int_mode and update the call to extract_left_shift.
16020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16021 Alan Hayward <alan.hayward@arm.com>
16022 David Sherwood <david.sherwood@arm.com>
16024 * combine.c (simplify_and_const_int): Change the type of the mode
16025 parameter to scalar_int_mode.
16026 (simplify_and_const_int_1): Likewise. Update recursive call.
16028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16029 Alan Hayward <alan.hayward@arm.com>
16030 David Sherwood <david.sherwood@arm.com>
16032 * combine.c (simplify_compare_const): Check that the mode is a
16033 scalar_int_mode (rather than VOIDmode) before testing its
16035 (simplify_comparison): Move COMPARISON_P handling out of the
16036 loop and restrict the latter part of the loop to scalar_int_modes.
16037 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16038 and when considering SUBREG_REGs. Use is_int_mode instead of
16039 checking GET_MODE_CLASS against MODE_INT.
16041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16042 Alan Hayward <alan.hayward@arm.com>
16043 David Sherwood <david.sherwood@arm.com>
16045 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16046 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16047 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
16054 (force_to_mode): ...here. Keep xmode up-to-date and use it
16055 instead of GET_MODE (x).
16057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16058 Alan Hayward <alan.hayward@arm.com>
16059 David Sherwood <david.sherwood@arm.com>
16061 * optabs-query.h (extraction_insn::struct_mode): Change type to
16062 opt_scalar_int_mode and update comment.
16063 (extraction_insn::field_mode): Change type to scalar_int_mode.
16064 (extraction_insn::pos_mode): Likewise.
16065 * combine.c (make_extraction): Update accordingly.
16066 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16067 (get_optab_extraction_insn): Likewise.
16068 * recog.c (simplify_while_replacing): Likewise.
16069 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16070 parameter to opt_scalar_int_mode.
16072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16073 Alan Hayward <alan.hayward@arm.com>
16074 David Sherwood <david.sherwood@arm.com>
16076 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16077 to a scalar_int_mode instead of a machine_mode.
16078 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16079 (get_best_mode): Return a boolean and use a pointer argument to store
16080 the selected mode. Replace the limit mode parameter with a bit limit.
16081 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16082 for the values returned by bit_field_mode_iterator::next_mode.
16083 (store_bit_field): Update call to get_best_mode.
16084 (store_fixed_bit_field): Likewise.
16085 (extract_fixed_bit_field): Likewise.
16086 * expr.c (optimize_bitfield_assignment_op): Likewise.
16087 * fold-const.c (optimize_bit_field_compare): Likewise.
16088 (fold_truth_andor_1): Likewise.
16089 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16090 Update for new type of m_mode.
16091 (get_best_mode): As above.
16093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16094 Alan Hayward <alan.hayward@arm.com>
16095 David Sherwood <david.sherwood@arm.com>
16097 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16098 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16099 (store_bit_field): Check is_a <scalar_int_mode> before calling
16100 strict_volatile_bitfield_p.
16101 (extract_bit_field): Likewise.
16103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16104 Alan Hayward <alan.hayward@arm.com>
16105 David Sherwood <david.sherwood@arm.com>
16107 * target.def (cstore_mode): Return a scalar_int_mode.
16108 * doc/tm.texi: Regenerate.
16109 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16110 * targhooks.h (default_cstore_mode): Likewise.
16111 * targhooks.c (default_cstore_mode): Likewise, using a forced
16113 * expmed.c (emit_cstore): Expect the target of the cstore to be
16116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16117 Alan Hayward <alan.hayward@arm.com>
16118 David Sherwood <david.sherwood@arm.com>
16120 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16122 (niter_desc): Likewise mode.
16123 (iv_analyze): Add a mode parameter.
16125 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16126 and change its type to scalar_int_mode.
16127 * loop-iv.c: Update commentary at head of file.
16128 (iv_constant): Pass the mode paraeter before the rtx it describes
16129 and change its type to scalar_int_mode. Remove VOIDmode handling.
16130 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16131 (iv_extend): Likewise.
16132 (shorten_into_mode): Likewise.
16133 (iv_add): Use scalar_int_mode.
16134 (iv_mult): Likewise.
16135 (iv_shift): Likewise.
16136 (canonicalize_iv_subregs): Likewise.
16137 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16138 it describes and change its mode to scalar_int_mode. Also change
16139 the type of the returned inner_mode to scalar_int_mode.
16140 (get_biv_step): Likewise, turning outer_mode from a pointer
16141 into a direct parameter. Update call to get_biv_step_1.
16142 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16143 iv_constant and get_biv_step.
16144 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16145 and change its type to scalar_int_mode. Don't initialise iv->mode
16146 to VOIDmode and remove later checks for its still being VOIDmode.
16147 Update calls to iv_analyze_op and iv_analyze_expr. Check
16148 is_a <scalar_int_mode> when changing the mode under consideration.
16149 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16150 Update call to iv_analyze_expr.
16151 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16152 inner register is not also a scalar_int_mode. Update call to
16154 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16155 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16156 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16157 separate mode class checks. Update calls to iv_analyze. Remove
16158 fix-up of VOIDmodes after iv_analyze_biv.
16159 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16160 don't have a scalar_int_mode. Update call to biv_p.
16162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16163 Alan Hayward <alan.hayward@arm.com>
16164 David Sherwood <david.sherwood@arm.com>
16166 * cfgexpand.c (convert_debug_memory_address): Use
16167 as_a <scalar_int_mode>.
16168 * combine.c (expand_compound_operation): Likewise.
16169 (make_extraction): Likewise.
16170 (change_zero_ext): Likewise.
16171 (simplify_comparison): Likewise.
16172 * cse.c (cse_insn): Likewise.
16173 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16174 (mem_loc_descriptor): Likewise.
16175 (loc_descriptor): Likewise.
16176 * expmed.c (init_expmed_one_mode): Likewise.
16177 (synth_mult): Likewise.
16178 (emit_store_flag_1): Likewise.
16179 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16180 of a comparison with size.
16181 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16182 (reduce_to_bit_field_precision): Likewise.
16183 * function.c (expand_function_end): Likewise.
16184 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16185 * loop-doloop.c (doloop_modify): Likewise.
16186 * optabs.c (expand_binop): Likewise.
16187 (expand_unop): Likewise.
16188 (expand_copysign_absneg): Likewise.
16189 (prepare_cmp_insn): Likewise.
16190 (maybe_legitimize_operand): Likewise.
16191 * recog.c (const_scalar_int_operand): Likewise.
16192 * rtlanal.c (get_address_mode): Likewise.
16193 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16194 (simplify_cond_clz_ctz): Likewise.
16195 * tree-nested.c (get_nl_goto_field): Likewise.
16196 * tree.c (build_vector_type_for_mode): Likewise.
16197 * var-tracking.c (use_narrower_mode): Likewise.
16199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16200 Alan Hayward <alan.hayward@arm.com>
16201 David Sherwood <david.sherwood@arm.com>
16203 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16204 * builtins.c (expand_builtin_signbit): Use it.
16205 * cfgexpand.c (expand_debug_expr): Likewise.
16206 * dojump.c (do_jump): Likewise.
16207 (do_compare_and_jump): Likewise.
16208 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16209 * expmed.c (make_tree): Likewise.
16210 * expr.c (expand_expr_real_2): Likewise.
16211 (expand_expr_real_1): Likewise.
16212 (try_casesi): Likewise.
16213 * fold-const-call.c (fold_const_call_ss): Likewise.
16214 * fold-const.c (unextend): Likewise.
16215 (extract_muldiv_1): Likewise.
16216 (fold_single_bit_test): Likewise.
16217 (native_encode_int): Likewise.
16218 (native_encode_string): Likewise.
16219 (native_interpret_int): Likewise.
16220 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16221 * internal-fn.c (expand_addsub_overflow): Likewise.
16222 (expand_neg_overflow): Likewise.
16223 (expand_mul_overflow): Likewise.
16224 (expand_arith_overflow): Likewise.
16225 * match.pd: Likewise.
16226 * stor-layout.c (layout_type): Likewise.
16227 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16228 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16229 * tree-ssanames.c (get_range_info): Likewise.
16230 * tree-switch-conversion.c (array_value_type) Likewise.
16231 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16232 (vect_recog_divmod_pattern): Likewise.
16233 (vect_recog_mixed_size_cond_pattern): Likewise.
16234 * tree-vrp.c (extract_range_basic): Likewise.
16235 (simplify_float_conversion_using_ranges): Likewise.
16236 * tree.c (int_fits_type_p): Likewise.
16237 * ubsan.c (instrument_bool_enum_load): Likewise.
16238 * varasm.c (mergeable_string_section): Likewise.
16239 (narrowing_initializer_constant_valid_p): Likewise.
16240 (output_constant): Likewise.
16242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16243 Alan Hayward <alan.hayward@arm.com>
16244 David Sherwood <david.sherwood@arm.com>
16246 * machmode.h (NARROWEST_INT_MODE): New macro.
16247 * expr.c (alignment_for_piecewise_move): Use it instead of
16248 GET_CLASS_NARROWEST_MODE (MODE_INT).
16249 (push_block): Likewise.
16250 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16252 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16255 Alan Hayward <alan.hayward@arm.com>
16256 David Sherwood <david.sherwood@arm.com>
16258 * postreload.c (move2add_valid_value_p): Change the type of the
16259 mode parameter to scalar_int_mode.
16260 (move2add_use_add2_insn): Add a mode parameter and use it instead
16262 (move2add_use_add3_insn): Likewise.
16263 (reload_cse_move2add): Update accordingly.
16265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16266 Alan Hayward <alan.hayward@arm.com>
16267 David Sherwood <david.sherwood@arm.com>
16269 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16271 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16272 * optabs.c (expand_unop): Likewise.
16274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16275 Alan Hayward <alan.hayward@arm.com>
16276 David Sherwood <david.sherwood@arm.com>
16278 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16279 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16280 (popcount_loc_descriptor): Likewise.
16281 (bswap_loc_descriptor): Likewise.
16282 (rotate_loc_descriptor): Likewise.
16283 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16284 calling the functions above.
16286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16287 Alan Hayward <alan.hayward@arm.com>
16288 David Sherwood <david.sherwood@arm.com>
16290 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16292 (try_combine): Likewise.
16293 (simplify_if_then_else): Likewise.
16294 * cse.c (cse_insn): Likewise.
16295 * dwarf2out.c (mem_loc_descriptor): Likewise.
16296 * emit-rtl.c (gen_lowpart_common): Likewise.
16297 * simplify-rtx.c (simplify_truncation): Likewise.
16298 (simplify_binary_operation_1): Likewise.
16299 (simplify_const_relational_operation): Likewise.
16300 (simplify_ternary_operation): Likewise.
16301 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16304 Alan Hayward <alan.hayward@arm.com>
16305 David Sherwood <david.sherwood@arm.com>
16307 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16308 * reload.c (push_reload): Likewise.
16309 (find_reloads): Likewise.
16311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16312 Alan Hayward <alan.hayward@arm.com>
16313 David Sherwood <david.sherwood@arm.com>
16315 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16316 (make_compound_operation_int): Likewise.
16317 (change_zero_ext): Likewise.
16318 * expr.c (convert_move): Likewise.
16319 (convert_modes): Likewise.
16320 * fwprop.c (forward_propagate_subreg): Likewise.
16321 * loop-iv.c (get_biv_step_1): Likewise.
16322 * optabs.c (widen_operand): Likewise.
16323 * postreload.c (move2add_valid_value_p): Likewise.
16324 * recog.c (simplify_while_replacing): Likewise.
16325 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16326 (simplify_binary_operation_1): Likewise. Remove redundant
16327 mode equality check.
16329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16330 Alan Hayward <alan.hayward@arm.com>
16331 David Sherwood <david.sherwood@arm.com>
16333 * combine.c (combine_simplify_rtx): Add checks for
16334 is_a <scalar_int_mode>.
16335 (simplify_if_then_else): Likewise.
16336 (make_field_assignment): Likewise.
16337 (simplify_comparison): Likewise.
16338 * ifcvt.c (noce_try_bitop): Likewise.
16339 * loop-invariant.c (canonicalize_address_mult): Likewise.
16340 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16343 Alan Hayward <alan.hayward@arm.com>
16344 David Sherwood <david.sherwood@arm.com>
16346 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16347 is_a <scalar_int_mode> instead of != BLKmode.
16349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16350 Alan Hayward <alan.hayward@arm.com>
16351 David Sherwood <david.sherwood@arm.com>
16353 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16354 instead of != VOIDmode.
16355 * combine.c (if_then_else_cond): Likewise.
16356 (change_zero_ext): Likewise.
16357 * dwarf2out.c (mem_loc_descriptor): Likewise.
16358 (loc_descriptor): Likewise.
16359 * rtlanal.c (canonicalize_condition): Likewise.
16360 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16363 Alan Hayward <alan.hayward@arm.com>
16364 David Sherwood <david.sherwood@arm.com>
16366 * simplify-rtx.c (simplify_binary_operation_1): Use
16367 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16370 Alan Hayward <alan.hayward@arm.com>
16371 David Sherwood <david.sherwood@arm.com>
16373 * wide-int.h (int_traits<unsigned char>) New class.
16374 (int_traits<unsigned short>) Likewise.
16375 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16376 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16378 * combine.c (set_nonzero_bits_and_sign_copies): Use
16379 is_a <scalar_int_mode>.
16380 (find_split_point): Likewise.
16381 (combine_simplify_rtx): Likewise.
16382 (simplify_logical): Likewise.
16383 (expand_compound_operation): Likewise.
16384 (expand_field_assignment): Likewise.
16385 (make_compound_operation): Likewise.
16386 (extended_count): Likewise.
16387 (change_zero_ext): Likewise.
16388 (simplify_comparison): Likewise.
16389 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16390 (ucompare_loc_descriptor): Likewise.
16391 (minmax_loc_descriptor): Likewise.
16392 (mem_loc_descriptor): Likewise.
16393 (loc_descriptor): Likewise.
16394 * expmed.c (init_expmed_one_mode): Likewise.
16395 * lra-constraints.c (lra_constraint_offset): Likewise.
16396 * optabs.c (prepare_libcall_arg): Likewise.
16397 * postreload.c (move2add_note_store): Likewise.
16398 * reload.c (operands_match_p): Likewise.
16399 * rtl.h (load_extend_op): Likewise.
16400 * rtlhooks.c (gen_lowpart_general): Likewise.
16401 * simplify-rtx.c (simplify_truncation): Likewise.
16402 (simplify_unary_operation_1): Likewise.
16403 (simplify_binary_operation_1): Likewise.
16404 (simplify_const_binary_operation): Likewise.
16405 (simplify_const_relational_operation): Likewise.
16406 (simplify_subreg): Likewise.
16407 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16408 * var-tracking.c (adjust_mems): Likewise.
16409 (prepare_call_arguments): Likewise.
16411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16412 Alan Hayward <alan.hayward@arm.com>
16413 David Sherwood <david.sherwood@arm.com>
16415 * machmode.h (is_int_mode): New fuction.
16416 * combine.c (find_split_point): Use it.
16417 (combine_simplify_rtx): Likewise.
16418 (simplify_if_then_else): Likewise.
16419 (simplify_set): Likewise.
16420 (simplify_shift_const_1): Likewise.
16421 (simplify_comparison): Likewise.
16422 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16423 * cse.c (notreg_cost): Likewise.
16424 (cse_insn): Likewise.
16425 * cselib.c (cselib_lookup_1): Likewise.
16426 * dojump.c (do_jump_1): Likewise.
16427 (do_compare_rtx_and_jump): Likewise.
16428 * dse.c (get_call_args): Likewise.
16429 * dwarf2out.c (rtl_for_decl_init): Likewise.
16430 (native_encode_initializer): Likewise.
16431 * expmed.c (emit_store_flag_1): Likewise.
16432 (emit_store_flag): Likewise.
16433 * expr.c (convert_modes): Likewise.
16434 (store_field): Likewise.
16435 (expand_expr_real_1): Likewise.
16436 * fold-const.c (fold_read_from_constant_string): Likewise.
16437 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16438 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16439 * optabs.c (expand_binop): Likewise.
16440 (expand_unop): Likewise.
16441 (expand_abs_nojump): Likewise.
16442 (expand_one_cmpl_abs_nojump): Likewise.
16443 * simplify-rtx.c (mode_signbit_p): Likewise.
16444 (val_signbit_p): Likewise.
16445 (val_signbit_known_set_p): Likewise.
16446 (val_signbit_known_clear_p): Likewise.
16447 (simplify_relational_operation_1): Likewise.
16448 * tree.c (vector_type_mode): Likewise.
16450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16451 Alan Hayward <alan.hayward@arm.com>
16452 David Sherwood <david.sherwood@arm.com>
16454 * machmode.h (smallest_mode_for_size): Fix formatting.
16455 (smallest_int_mode_for_size): New function.
16456 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16457 instead of smallest_mode_for_size.
16458 * combine.c (make_extraction): Likewise.
16459 * config/arc/arc.c (arc_expand_movmem): Likewise.
16460 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16461 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16462 * config/s390/s390.c (s390_expand_insv): Likewise.
16463 * config/sparc/sparc.c (assign_int_registers): Likewise.
16464 * config/spu/spu.c (spu_function_value): Likewise.
16465 (spu_function_arg): Likewise.
16466 * coverage.c (get_gcov_type): Likewise.
16467 (get_gcov_unsigned_t): Likewise.
16468 * dse.c (find_shift_sequence): Likewise.
16469 * expmed.c (store_bit_field_1): Likewise.
16470 * expr.c (convert_move): Likewise.
16471 (store_field): Likewise.
16472 * internal-fn.c (expand_arith_overflow): Likewise.
16473 * optabs-query.c (get_best_extraction_insn): Likewise.
16474 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16475 * stor-layout.c (layout_type): Likewise.
16476 (initialize_sizetypes): Likewise.
16477 * targhooks.c (default_get_mask_mode): Likewise.
16478 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16481 Alan Hayward <alan.hayward@arm.com>
16482 David Sherwood <david.sherwood@arm.com>
16484 * machmode.h (opt_mode::else_blk): New function.
16485 (int_mode_for_mode): Declare.
16486 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16487 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16489 * cfgexpand.c (expand_debug_expr): Likewise.
16490 * combine.c (gen_lowpart_or_truncate): Likewise.
16491 (gen_lowpart_for_combine): Likewise.
16492 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16493 * config/avr/avr.c (avr_to_int_mode): Likewise.
16494 (avr_out_plus_1): Likewise.
16495 (avr_out_plus): Likewise.
16496 (avr_out_round): Likewise.
16497 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16498 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16499 (s390_expand_vcond): Likewise.
16500 * config/spu/spu.c (spu_split_immediate): Likewise.
16501 (spu_expand_mov): Likewise.
16502 * dse.c (get_stored_val): Likewise.
16503 * expmed.c (store_bit_field_1): Likewise.
16504 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16506 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16507 (extract_low_bits): Likewise.
16508 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16509 handling rather than repeating the check.
16510 (emit_group_store): Likewise.
16511 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16512 * optabs.c (expand_absneg_bit): Likewise.
16513 (expand_copysign_absneg): Likewise.
16514 (expand_copysign_bit): Likewise.
16515 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16516 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16517 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16518 * var-tracking.c (prepare_call_arguments): Likewise.
16519 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16520 int_mode_for_mode instead of mode_for_size.
16521 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16524 Alan Hayward <alan.hayward@arm.com>
16525 David Sherwood <david.sherwood@arm.com>
16527 * machmode.h (int_mode_for_size): New function.
16528 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16529 instead of mode_for_size.
16530 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16532 * combine.c (expand_field_assignment): Use int_mode_for_size
16533 instead of mode_for_size.
16534 (make_extraction): Likewise.
16535 (simplify_shift_const_1): Likewise.
16536 (simplify_comparison): Likewise.
16537 * dojump.c (do_jump): Likewise.
16538 * dwarf2out.c (mem_loc_descriptor): Likewise.
16539 * emit-rtl.c (init_derived_machine_modes): Likewise.
16540 * expmed.c (flip_storage_order): Likewise.
16541 (convert_extracted_bit_field): Likewise.
16542 * expr.c (copy_blkmode_from_reg): Likewise.
16543 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16544 * internal-fn.c (expand_mul_overflow): Likewise.
16545 * lower-subreg.c (simple_move): Likewise.
16546 * optabs-libfuncs.c (init_optabs): Likewise.
16547 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16548 * tree.c (vector_type_mode): Likewise.
16549 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16550 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16551 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16552 * tree-vect-stmts.c (vectorizable_load): Likewise.
16553 (vectorizable_store): Likewise.
16555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16556 Alan Hayward <alan.hayward@arm.com>
16557 David Sherwood <david.sherwood@arm.com>
16559 * coretypes.h (pod_mode): New type.
16560 (scalar_int_mode_pod): New typedef.
16561 * machmode.h (pod_mode): New class.
16562 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16563 * genmodes.c (emit_mode_int_n): Update accordingly.
16564 * lower-subreg.h (target_lower_subreg): Change type to
16565 scalar_int_mode_pod.
16566 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16567 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16574 machine_mode to scalar_int_mode.
16575 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16576 (rs6000_option_override_internal): Remove cast to int.
16577 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16578 machine_mode to scalar_int_mode.
16579 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16580 (rs6000_option_override_internal): Remove cast to int.
16581 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16582 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16584 * config/s390/s390.c (s390_expand_builtin): Likewise.
16585 * coretypes.h (scalar_int_mode): New type.
16586 (opt_scalar_int_mode): New typedef.
16587 * machmode.h (scalar_int_mode): New class.
16588 (scalar_int_mode::includes_p): New function.
16589 (byte_mode): Change type to scalar_int_mode.
16590 (word_mode): Likewise.
16591 (ptr_mode): Likewise.
16592 * emit-rtl.c (byte_mode): Likewise.
16593 (word_mode): Likewise.
16594 (ptr_mode): Likewise.
16595 (init_derived_machine_modes): Update accordingly.
16596 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16597 and MODE_PARTIAL_INT.
16598 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16599 opt_scalar_int_mode.
16601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16602 Alan Hayward <alan.hayward@arm.com>
16603 David Sherwood <david.sherwood@arm.com>
16605 * target.def (libgcc_floating_mode_supported_p): Take a
16607 * doc/tm.texi: Regenerate.
16608 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16610 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16611 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16615 Alan Hayward <alan.hayward@arm.com>
16616 David Sherwood <david.sherwood@arm.com>
16618 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16619 * doc/tm.texi: Regenerate.
16620 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16621 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16622 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16623 * targhooks.h (default_floatn_mode): Likewise.
16624 * targhooks.c (default_floatn_mode): Likewise.
16625 * tree.c (build_common_tree_nodes): Update accordingly.
16627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16628 Alan Hayward <alan.hayward@arm.com>
16629 David Sherwood <david.sherwood@arm.com>
16631 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16632 (mode_iterator::iterate_p): Likewise.
16633 (mode_iterator::get_wider): Likewise.
16634 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16637 Alan Hayward <alan.hayward@arm.com>
16638 David Sherwood <david.sherwood@arm.com>
16640 * coretypes.h (opt_scalar_float_mode): New typedef.
16641 * machmode.h (float_mode_for_size): New function.
16642 * emit-rtl.c (double_mode): Delete.
16643 (init_emit_once): Use float_mode_for_size.
16644 * stor-layout.c (layout_type): Likewise.
16645 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16648 Alan Hayward <alan.hayward@arm.com>
16649 David Sherwood <david.sherwood@arm.com>
16651 * output.h (assemble_real): Take a scalar_float_mode.
16652 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16653 * config/arm/arm.md (consttable_4): Likewise.
16654 (consttable_8): Likewise.
16655 (consttable_16): Likewise.
16656 * config/mips/mips.md (consttable_float): Likewise.
16657 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16658 * varasm.c (assemble_real): Take a scalar_float_mode.
16659 (output_constant_pool_2): Update accordingly.
16660 (output_constant): Likewise.
16662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16663 Alan Hayward <alan.hayward@arm.com>
16664 David Sherwood <david.sherwood@arm.com>
16666 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16667 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16668 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16669 (native_encode_real): Likewise.
16670 (native_interpret_real): Likewise.
16671 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16672 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16675 Alan Hayward <alan.hayward@arm.com>
16676 David Sherwood <david.sherwood@arm.com>
16678 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16679 <scalar_float_mode>. Simplify.
16680 (gen_extend_conv_libfunc): Likewise.
16682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16683 Alan Hayward <alan.hayward@arm.com>
16684 David Sherwood <david.sherwood@arm.com>
16686 * coretypes.h (scalar_float_mode): New type.
16687 * machmode.h (mode_traits::from_int): Use machine_mode if
16688 USE_ENUM_MODES is defined.
16689 (is_a): New function.
16691 (dyn_cast): Likewise.
16692 (scalar_float_mode): New class.
16693 (scalar_float_mode::includes_p): New function.
16694 (is_float_mode): Likewise.
16695 * gdbhooks.py (MachineModePrinter): New class.
16696 (build_pretty_printer): Use it for scalar_float_mode.
16697 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16698 (format_helper::format_helper): Turn into a template.
16699 * genmodes.c (get_mode_class): New function.
16700 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16701 or machine_mode if none.
16702 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16703 as_a <scalar_float_mode>.
16704 * dwarf2out.c (mem_loc_descriptor): Likewise.
16705 (insert_float): Likewise.
16706 (add_const_value_attribute): Likewise.
16707 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16708 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16709 (expand_unop): Update accordingly.
16710 (expand_abs_nojump): Likewise.
16711 (expand_copysign_absneg): Take a scalar_float_mode.
16712 (expand_copysign_bit): Likewise.
16713 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
16720 * machmode.h (opt_mode): Likewise.
16721 (opt_mode::else_void): New function.
16722 (opt_mode::require): Likewise.
16723 (opt_mode::exists): Likewise.
16724 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16725 (GET_MODE_2XWIDER_MODE): Likewise.
16726 (mode_iterator::get_wider): Update accordingly.
16727 (mode_iterator::get_2xwider): Likewise.
16728 (mode_iterator::get_known_wider): Likewise, turning into a template.
16729 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16730 forcing a wider mode to exist.
16731 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16732 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16733 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16734 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16735 * lower-subreg.c (init_lower_subreg): Likewise.
16736 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16737 on the final iteration.
16738 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16739 a wider mode exists before asking for a move pattern.
16740 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16741 forcing a wider mode to exist.
16742 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16743 returning false if no such mode exists.
16744 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16745 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16746 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16747 Avoid checking for a MODE_INT if we already know the mode is not a
16749 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16750 forcing a wider mode to exist.
16751 (expmed_mult_highpart_optab): Likewise.
16752 (expmed_mult_highpart): Likewise.
16753 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16755 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16756 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16757 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16758 * internal-fn.c (expand_mul_overflow): Update use of
16759 GET_MODE_2XWIDER_MODE.
16760 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16761 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16762 GET_MODE_WIDER_MODE.
16763 (convert_plusminus_to_widen): Likewise.
16764 * tree-switch-conversion.c (array_value_type): Likewise.
16765 * var-tracking.c (emit_note_insn_var_location): Likewise.
16766 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16767 Return false inside rather than outside the loop if no wider mode
16769 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16770 and GET_MODE_2XWIDER_MODE
16771 (can_compare_p): Use else_void.
16772 * gdbhooks.py (OptMachineModePrinter): New class.
16773 (build_pretty_printer): Use it for opt_mode.
16775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16776 Alan Hayward <alan.hayward@arm.com>
16777 David Sherwood <david.sherwood@arm.com>
16779 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16780 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16783 Alan Hayward <alan.hayward@arm.com>
16784 David Sherwood <david.sherwood@arm.com>
16786 * machmode.h (mode_traits): New structure.
16787 (get_narrowest_mode): New function.
16788 (mode_iterator::start): Likewise.
16789 (mode_iterator::iterate_p): Likewise.
16790 (mode_iterator::get_wider): Likewise.
16791 (mode_iterator::get_known_wider): Likewise.
16792 (mode_iterator::get_2xwider): Likewise.
16793 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16794 (FOR_EACH_MODE): Likewise.
16795 (FOR_EACH_MODE_FROM): Likewise.
16796 (FOR_EACH_MODE_UNTIL): Likewise.
16797 (FOR_EACH_WIDER_MODE): Likewise.
16798 (FOR_EACH_2XWIDER_MODE): Likewise.
16799 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16800 * combine.c (simplify_comparison): Likewise
16801 * config/i386/i386.c (type_natural_mode): Likewise.
16802 * cse.c (cse_insn): Likewise.
16803 * dse.c (find_shift_sequence): Likewise.
16804 * emit-rtl.c (init_derived_machine_modes): Likewise.
16805 (init_emit_once): Likewise.
16806 * explow.c (hard_function_value): Likewise.
16807 * expmed.c (extract_fixed_bit_field_1): Likewise.
16808 (extract_bit_field_1): Likewise.
16809 (expand_divmod): Likewise.
16810 (emit_store_flag_1): Likewise.
16811 * expr.c (init_expr_target): Likewise.
16812 (convert_move): Likewise.
16813 (alignment_for_piecewise_move): Likewise.
16814 (widest_int_mode_for_size): Likewise.
16815 (emit_block_move_via_movmem): Likewise.
16816 (copy_blkmode_to_reg): Likewise.
16817 (set_storage_via_setmem): Likewise.
16818 (compress_float_constant): Likewise.
16819 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16820 * optabs-query.c (get_best_extraction_insn): Likewise.
16821 * optabs.c (expand_binop): Likewise.
16822 (expand_twoval_unop): Likewise.
16823 (expand_twoval_binop): Likewise.
16824 (widen_leading): Likewise.
16825 (widen_bswap): Likewise.
16826 (expand_parity): Likewise.
16827 (expand_unop): Likewise.
16828 (prepare_cmp_insn): Likewise.
16829 (prepare_float_lib_cmp): Likewise.
16830 (expand_float): Likewise.
16831 (expand_fix): Likewise.
16832 (expand_sfix_optab): Likewise.
16833 * postreload.c (move2add_use_add2_insn): Likewise.
16834 * reg-stack.c (reg_to_stack): Likewise.
16835 * reginfo.c (choose_hard_reg_mode): Likewise.
16836 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16837 * stor-layout.c (mode_for_size): Likewise.
16838 (smallest_mode_for_size): Likewise.
16839 (mode_for_vector): Likewise.
16840 (finish_bitfield_representative): Likewise.
16841 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16842 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16843 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16844 * var-tracking.c (prepare_call_arguments): Likewise.
16846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16847 Alan Hayward <alan.hayward@arm.com>
16848 David Sherwood <david.sherwood@arm.com>
16850 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16851 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16852 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16853 * machmode.h (mode_size): Move earlier in file.
16854 (mode_precision): Likewise.
16855 (mode_inner): Likewise.
16856 (mode_nunits): Likewise.
16857 (mode_unit_size): Likewise.
16858 (unit_unit_precision): Likewise.
16859 (mode_wider): Likewise.
16860 (mode_2xwider): Likewise.
16861 (machine_mode): New class.
16862 (mode_to_bytes): New function.
16863 (mode_to_bits): Likewise.
16864 (mode_to_precision): Likewise.
16865 (mode_to_inner): Likewise.
16866 (mode_to_unit_size): Likewise.
16867 (mode_to_unit_precision): Likewise.
16868 (mode_to_nunits): Likewise.
16869 (GET_MODE_SIZE): Use mode_to_bytes.
16870 (GET_MODE_BITSIZE): Use mode_to_bits.
16871 (GET_MODE_PRECISION): Use mode_to_precision.
16872 (GET_MODE_INNER): Use mode_to_inner.
16873 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16874 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16875 (GET_MODE_NUNITS): Use mode_to_nunits.
16876 * system.h (ALWAYS_INLINE): New macro.
16877 * config/powerpcspe/powerpcspe-c.c
16878 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16879 int for arg1_mode and arg2_mode.
16881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16882 Alan Hayward <alan.hayward@arm.com>
16883 David Sherwood <david.sherwood@arm.com>
16885 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16886 Prefix mode names with E_ in case statements.
16887 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16888 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16889 (aarch64_split_simd_move): Likewise.
16890 (aarch64_gen_storewb_pair): Likewise.
16891 (aarch64_gen_loadwb_pair): Likewise.
16892 (aarch64_gen_store_pair): Likewise.
16893 (aarch64_gen_load_pair): Likewise.
16894 (aarch64_get_condition_code_1): Likewise.
16895 (aarch64_constant_pool_reload_icode): Likewise.
16896 (get_rsqrte_type): Likewise.
16897 (get_rsqrts_type): Likewise.
16898 (get_recpe_type): Likewise.
16899 (get_recps_type): Likewise.
16900 (aarch64_gimplify_va_arg_expr): Likewise.
16901 (aarch64_simd_container_mode): Likewise.
16902 (aarch64_emit_load_exclusive): Likewise.
16903 (aarch64_emit_store_exclusive): Likewise.
16904 (aarch64_expand_compare_and_swap): Likewise.
16905 (aarch64_gen_atomic_cas): Likewise.
16906 (aarch64_emit_bic): Likewise.
16907 (aarch64_emit_atomic_swap): Likewise.
16908 (aarch64_emit_atomic_load_op): Likewise.
16909 (aarch64_evpc_trn): Likewise.
16910 (aarch64_evpc_uzp): Likewise.
16911 (aarch64_evpc_zip): Likewise.
16912 (aarch64_evpc_ext): Likewise.
16913 (aarch64_evpc_rev): Likewise.
16914 (aarch64_evpc_dup): Likewise.
16915 (aarch64_gen_ccmp_first): Likewise.
16916 (aarch64_gen_ccmp_next): Likewise.
16917 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16918 (alpha_emit_xfloating_libcall): Likewise.
16919 (emit_insxl): Likewise.
16920 (alpha_arg_type): Likewise.
16921 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16922 (arc_preferred_simd_mode): Likewise.
16923 (arc_secondary_reload): Likewise.
16924 (get_arc_condition_code): Likewise.
16925 (arc_print_operand): Likewise.
16926 (arc_legitimate_constant_p): Likewise.
16927 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16928 * config/arc/arc.md (casesi_load): Likewise.
16929 (casesi_compact_jump): Likewise.
16930 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16931 (cc_use_register): Likewise.
16932 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16933 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16934 (arm_init_iwmmxt_builtins): Likewise.
16935 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16936 (neon_expand_vector_init): Likewise.
16937 (arm_attr_length_move_neon): Likewise.
16938 (maybe_get_arm_condition_code): Likewise.
16939 (arm_emit_vector_const): Likewise.
16940 (arm_preferred_simd_mode): Likewise.
16941 (arm_output_iwmmxt_tinsr): Likewise.
16942 (thumb1_output_casesi): Likewise.
16943 (thumb2_output_casesi): Likewise.
16944 (arm_emit_load_exclusive): Likewise.
16945 (arm_emit_store_exclusive): Likewise.
16946 (arm_expand_compare_and_swap): Likewise.
16947 (arm_evpc_neon_vuzp): Likewise.
16948 (arm_evpc_neon_vzip): Likewise.
16949 (arm_evpc_neon_vrev): Likewise.
16950 (arm_evpc_neon_vtrn): Likewise.
16951 (arm_evpc_neon_vext): Likewise.
16952 (arm_validize_comparison): Likewise.
16953 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16954 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16955 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16956 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16957 (c6x_preferred_simd_mode): Likewise.
16958 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16959 (epiphany_rtx_costs): Likewise.
16960 * config/epiphany/predicates.md (proper_comparison_operator):
16962 * config/frv/frv.c (condexec_memory_operand): Likewise.
16963 (frv_emit_move): Likewise.
16964 (output_move_single): Likewise.
16965 (output_condmove_single): Likewise.
16966 (frv_hard_regno_mode_ok): Likewise.
16967 (frv_matching_accg_mode): Likewise.
16968 * config/h8300/h8300.c (split_adds_subs): Likewise.
16969 (h8300_rtx_costs): Likewise.
16970 (h8300_print_operand): Likewise.
16971 (compute_mov_length): Likewise.
16972 (output_logical_op): Likewise.
16973 (compute_logical_op_length): Likewise.
16974 (compute_logical_op_cc): Likewise.
16975 (h8300_shift_needs_scratch_p): Likewise.
16976 (output_a_shift): Likewise.
16977 (compute_a_shift_length): Likewise.
16978 (compute_a_shift_cc): Likewise.
16979 (expand_a_rotate): Likewise.
16980 (output_a_rotate): Likewise.
16981 * config/i386/i386.c (classify_argument): Likewise.
16982 (function_arg_advance_32): Likewise.
16983 (function_arg_32): Likewise.
16984 (function_arg_64): Likewise.
16985 (function_value_64): Likewise.
16986 (ix86_gimplify_va_arg): Likewise.
16987 (ix86_legitimate_constant_p): Likewise.
16988 (put_condition_code): Likewise.
16989 (split_double_mode): Likewise.
16990 (ix86_avx256_split_vector_move_misalign): Likewise.
16991 (ix86_expand_vector_logical_operator): Likewise.
16992 (ix86_split_idivmod): Likewise.
16993 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16994 (ix86_build_const_vector): Likewise.
16995 (ix86_build_signbit_mask): Likewise.
16996 (ix86_match_ccmode): Likewise.
16997 (ix86_cc_modes_compatible): Likewise.
16998 (ix86_expand_branch): Likewise.
16999 (ix86_expand_sse_cmp): Likewise.
17000 (ix86_expand_sse_movcc): Likewise.
17001 (ix86_expand_int_sse_cmp): Likewise.
17002 (ix86_expand_vec_perm_vpermi2): Likewise.
17003 (ix86_expand_vec_perm): Likewise.
17004 (ix86_expand_sse_unpack): Likewise.
17005 (ix86_expand_int_addcc): Likewise.
17006 (ix86_split_to_parts): Likewise.
17007 (ix86_vectorize_builtin_gather): Likewise.
17008 (ix86_vectorize_builtin_scatter): Likewise.
17009 (avx_vpermilp_parallel): Likewise.
17010 (inline_memory_move_cost): Likewise.
17011 (ix86_tieable_integer_mode_p): Likewise.
17012 (x86_maybe_negate_const_int): Likewise.
17013 (ix86_expand_vector_init_duplicate): Likewise.
17014 (ix86_expand_vector_init_one_nonzero): Likewise.
17015 (ix86_expand_vector_init_one_var): Likewise.
17016 (ix86_expand_vector_init_concat): Likewise.
17017 (ix86_expand_vector_init_interleave): Likewise.
17018 (ix86_expand_vector_init_general): Likewise.
17019 (ix86_expand_vector_set): Likewise.
17020 (ix86_expand_vector_extract): Likewise.
17021 (emit_reduc_half): Likewise.
17022 (ix86_emit_i387_round): Likewise.
17023 (ix86_mangle_type): Likewise.
17024 (ix86_expand_round_sse4): Likewise.
17025 (expand_vec_perm_blend): Likewise.
17026 (canonicalize_vector_int_perm): Likewise.
17027 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17028 (expand_vec_perm_1): Likewise.
17029 (expand_vec_perm_interleave3): Likewise.
17030 (expand_vec_perm_even_odd_pack): Likewise.
17031 (expand_vec_perm_even_odd_1): Likewise.
17032 (expand_vec_perm_broadcast_1): Likewise.
17033 (ix86_vectorize_vec_perm_const_ok): Likewise.
17034 (ix86_expand_vecop_qihi): Likewise.
17035 (ix86_expand_mul_widen_hilo): Likewise.
17036 (ix86_expand_sse2_abs): Likewise.
17037 (ix86_expand_pextr): Likewise.
17038 (ix86_expand_pinsr): Likewise.
17039 (ix86_preferred_simd_mode): Likewise.
17040 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17041 * config/i386/sse.md (*andnot<mode>3): Likewise.
17042 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17043 (*<code><mode>3): Likewise.
17044 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17045 (ia64_expand_atomic_op): Likewise.
17046 (ia64_arg_type): Likewise.
17047 (ia64_mode_to_int): Likewise.
17048 (ia64_scalar_mode_supported_p): Likewise.
17049 (ia64_vector_mode_supported_p): Likewise.
17050 (expand_vec_perm_broadcast): Likewise.
17051 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17052 (iq2000_function_arg_advance): Likewise.
17053 (iq2000_function_arg): Likewise.
17054 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17055 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17056 (m68k_libcall_value): Likewise.
17057 (m68k_function_value): Likewise.
17058 (sched_attr_op_type): Likewise.
17059 * config/mcore/mcore.c (mcore_output_move): Likewise.
17060 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17062 (microblaze_function_arg): Likewise.
17063 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17064 (mips_print_operand): Likewise.
17065 (mips_mode_ok_for_mov_fmt_p): Likewise.
17066 (mips_vector_mode_supported_p): Likewise.
17067 (mips_preferred_simd_mode): Likewise.
17068 (mips_expand_vpc_loongson_even_odd): Likewise.
17069 (mips_expand_vec_unpack): Likewise.
17070 (mips_expand_vi_broadcast): Likewise.
17071 (mips_expand_vector_init): Likewise.
17072 (mips_expand_vec_reduc): Likewise.
17073 (mips_expand_msa_cmp): Likewise.
17074 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17075 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17076 (cc_flags_for_mode): Likewise.
17077 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17078 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17079 (nds32_output_casesi_pc_relative): Likewise.
17080 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17081 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17082 (nvptx_gen_unpack): Likewise.
17083 (nvptx_gen_pack): Likewise.
17084 (nvptx_gen_shuffle): Likewise.
17085 (nvptx_gen_wcast): Likewise.
17086 (nvptx_preferred_simd_mode): Likewise.
17087 * config/pa/pa.c (pa_secondary_reload): Likewise.
17088 * config/pa/predicates.md (base14_operand): Likewise.
17089 * config/powerpcspe/powerpcspe-c.c
17090 (altivec_resolve_overloaded_builtin): Likewise.
17091 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17093 (rs6000_preferred_simd_mode): Likewise.
17094 (output_vec_const_move): Likewise.
17095 (rs6000_expand_vector_extract): Likewise.
17096 (rs6000_split_vec_extract_var): Likewise.
17097 (reg_offset_addressing_ok_p): Likewise.
17098 (rs6000_legitimate_offset_address_p): Likewise.
17099 (rs6000_legitimize_address): Likewise.
17100 (rs6000_emit_set_const): Likewise.
17101 (rs6000_const_vec): Likewise.
17102 (rs6000_emit_move): Likewise.
17103 (spe_build_register_parallel): Likewise.
17104 (rs6000_darwin64_record_arg_recurse): Likewise.
17105 (swap_selector_for_mode): Likewise.
17106 (spe_init_builtins): Likewise.
17107 (paired_init_builtins): Likewise.
17108 (altivec_init_builtins): Likewise.
17109 (do_load_for_compare): Likewise.
17110 (rs6000_generate_compare): Likewise.
17111 (rs6000_expand_float128_convert): Likewise.
17112 (emit_load_locked): Likewise.
17113 (emit_store_conditional): Likewise.
17114 (rs6000_output_function_epilogue): Likewise.
17115 (rs6000_handle_altivec_attribute): Likewise.
17116 (rs6000_function_value): Likewise.
17117 (emit_fusion_gpr_load): Likewise.
17118 (emit_fusion_p9_load): Likewise.
17119 (emit_fusion_p9_store): Likewise.
17120 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17121 (fusion_gpr_mem_load): Likewise.
17122 (fusion_addis_mem_combo_load): Likewise.
17123 (fusion_addis_mem_combo_store): Likewise.
17124 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17125 (fusion_gpr_mem_load): Likewise.
17126 (fusion_addis_mem_combo_load): Likewise.
17127 (fusion_addis_mem_combo_store): Likewise.
17128 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17130 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17131 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17132 (rs6000_preferred_simd_mode): Likewise.
17133 (output_vec_const_move): Likewise.
17134 (rs6000_expand_vector_extract): Likewise.
17135 (rs6000_split_vec_extract_var): Likewise.
17136 (reg_offset_addressing_ok_p): Likewise.
17137 (rs6000_legitimate_offset_address_p): Likewise.
17138 (rs6000_legitimize_address): Likewise.
17139 (rs6000_emit_set_const): Likewise.
17140 (rs6000_const_vec): Likewise.
17141 (rs6000_emit_move): Likewise.
17142 (rs6000_darwin64_record_arg_recurse): Likewise.
17143 (swap_selector_for_mode): Likewise.
17144 (paired_init_builtins): Likewise.
17145 (altivec_init_builtins): Likewise.
17146 (rs6000_expand_float128_convert): Likewise.
17147 (emit_load_locked): Likewise.
17148 (emit_store_conditional): Likewise.
17149 (rs6000_output_function_epilogue): Likewise.
17150 (rs6000_handle_altivec_attribute): Likewise.
17151 (rs6000_function_value): Likewise.
17152 (emit_fusion_gpr_load): Likewise.
17153 (emit_fusion_p9_load): Likewise.
17154 (emit_fusion_p9_store): Likewise.
17155 * config/rx/rx.c (rx_gen_move_template): Likewise.
17156 (flags_from_mode): Likewise.
17157 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17158 (s390_slb_comparison): Likewise.
17159 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17160 (s390_vector_mode_supported_p): Likewise.
17161 (s390_cc_modes_compatible): Likewise.
17162 (s390_match_ccmode_set): Likewise.
17163 (s390_canonicalize_comparison): Likewise.
17164 (s390_emit_compare_and_swap): Likewise.
17165 (s390_branch_condition_mask): Likewise.
17166 (s390_rtx_costs): Likewise.
17167 (s390_secondary_reload): Likewise.
17168 (__SECONDARY_RELOAD_CASE): Likewise.
17169 (s390_expand_cs): Likewise.
17170 (s390_preferred_simd_mode): Likewise.
17171 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17172 * config/sh/sh.c (sh_print_operand): Likewise.
17173 (dump_table): Likewise.
17174 (sh_secondary_reload): Likewise.
17175 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17176 * config/sh/sh.md (casesi_worker_1): Likewise.
17177 (casesi_worker_2): Likewise.
17178 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17179 (fcc_comparison_operator): Likewise.
17180 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17181 (emit_soft_tfmode_cvt): Likewise.
17182 (sparc_preferred_simd_mode): Likewise.
17183 (output_cbranch): Likewise.
17184 (sparc_print_operand): Likewise.
17185 (sparc_expand_vec_perm_bmask): Likewise.
17186 (vector_init_bshuffle): Likewise.
17187 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17188 (spu_vector_mode_supported_p): Likewise.
17189 (spu_expand_insv): Likewise.
17190 (spu_emit_branch_or_set): Likewise.
17191 (spu_handle_vector_attribute): Likewise.
17192 (spu_builtin_splats): Likewise.
17193 (spu_builtin_extract): Likewise.
17194 (spu_builtin_promote): Likewise.
17195 (spu_expand_sign_extend): Likewise.
17196 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17197 (tilegx_simd_int): Likewise.
17198 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17199 (tilepro_simd_int): Likewise.
17200 * config/v850/v850.c (const_double_split): Likewise.
17201 (v850_print_operand): Likewise.
17202 (ep_memory_offset): Likewise.
17203 * config/vax/vax.c (vax_rtx_costs): Likewise.
17204 (vax_output_int_move): Likewise.
17205 (vax_output_int_add): Likewise.
17206 (vax_output_int_subtract): Likewise.
17207 * config/visium/predicates.md (visium_branch_operator): Likewise.
17208 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17209 (visium_print_operand_address): Likewise.
17210 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17211 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17212 (xtensa_expand_conditional_branch): Likewise.
17213 (xtensa_copy_incoming_a7): Likewise.
17214 (xtensa_output_literal): Likewise.
17215 * dfp.c (decimal_real_maxval): Likewise.
17216 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17219 Alan Hayward <alan.hayward@arm.com>
17220 David Sherwood <david.sherwood@arm.com>
17222 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17223 (mode_nunits_inline): Likewise.
17224 (mode_inner_inline): Likewise.
17225 (mode_unit_size_inline): Likewise.
17226 (mode_unit_precision_inline): Likewise.
17227 (emit_insn_modes_h): Likewise. Also emit a #define of the
17229 (emit_mode_wider): Add an E_ prefix to mode names.
17230 (emit_mode_complex): Likewise.
17231 (emit_mode_inner): Likewise.
17232 (emit_mode_adjustments): Likewise.
17233 (emit_mode_int_n): Likewise.
17234 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17235 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17236 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17237 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17238 (CRC32_BUILTIN, ENTRY): Likewise.
17239 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17240 (aarch64_pop_regs): Likewise.
17241 (aarch64_process_components): Likewise.
17242 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17243 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17244 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17245 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17246 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17247 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17249 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17250 (pcmpistr): Likewise.
17251 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17252 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17253 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17255 * config/rl78/rl78.c (mduc_regs): Likewise.
17256 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17257 (htm_expand_builtin): Likewise.
17258 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17259 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17260 * config/xtensa/xtensa.c (print_operand): Likewise.
17261 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17262 (NUM_MODE_VECTOR_INT): Likewise.
17263 * genoutput.c (null_operand): Likewise.
17264 (output_operand_data): Likewise.
17265 * genrecog.c (print_parameter_value): Likewise.
17266 * lra.c (debug_operand_data): Likewise.
17268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17270 * dwarf2out.c (scompare_loc_descriptor_wide)
17271 (scompare_loc_descriptor_narrow): New functions, split out from...
17272 (scompare_loc_descriptor): ...here.
17273 * expmed.c (emit_store_flag_int): New function, split out from...
17274 (emit_store_flag): ...here.
17276 2017-08-30 Richard Biener <rguenther@suse.de>
17278 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17279 (dwarf2out_early_finish): Move setting of AT_pubnames from
17280 early debug output to early finish.
17282 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17284 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17285 and -mdata-region to the assembler.
17287 2017-08-30 Richard Biener <rguenther@suse.de>
17289 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17291 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17292 (dwarf2out_early_global_decl): Only generate a DIE for the
17293 abstract origin if it doesn't already exist or is a declaration DIE.
17294 (resolve_addr): Do not add the linkage name twice when
17295 generating a stub DIE for the DW_TAG_GNU_call_site target.
17297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17299 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17300 Use machine_mode rather than int for arg1_mode.
17302 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17305 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17306 that the second argument of the built-in functions to unpack
17307 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17308 switch statement instead a lot of if statements.
17309 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17310 Allow 64-bit values to be in Altivec registers as well as
17311 traditional floating point registers.
17312 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17314 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17316 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17317 MAX_REGS_PER_ADDRESS == 1.
17319 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17321 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17322 * config/i386/i386.c (ix86_option_override_internal): Simplify
17323 setting of opts->x_flag_entry.
17325 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17326 Jakub Jelinek <jakub@redhat.com>
17327 Richard Biener <rguenther@suse.de>
17329 PR tree-optimization/81503
17330 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17331 folded constant fits in the target type; reorder tests for clarity.
17333 2017-08-29 Martin Liska <mliska@suse.cz>
17335 * passes.def: Include pass_lower_switch.
17336 * stmt.c (dump_case_nodes): Remove and move to
17337 tree-switch-conversion.
17338 (case_values_threshold): Likewise.
17339 (expand_switch_as_decision_tree_p): Likewise.
17340 (emit_case_decision_tree): Likewise.
17341 (expand_case): Likewise.
17342 (balance_case_nodes): Likewise.
17343 (node_has_low_bound): Likewise.
17344 (node_has_high_bound): Likewise.
17345 (node_is_bounded): Likewise.
17346 (emit_case_nodes): Likewise.
17347 (struct simple_case_node): New struct.
17348 (add_case_node): Remove.
17349 (emit_case_dispatch_table): Use vector instead of case_list.
17350 (reset_out_edges_aux): Remove.
17351 (compute_cases_per_edge): Likewise.
17352 (expand_case): Build list of simple_case_node.
17353 (expand_sjlj_dispatch_table): Use it.
17354 * tree-switch-conversion.c (struct case_node): Moved from
17355 stmt.c and adjusted.
17356 (emit_case_nodes): Likewise.
17357 (node_has_low_bound): Likewise.
17358 (node_has_high_bound): Likewise.
17359 (node_is_bounded): Likewise.
17360 (case_values_threshold): Likewise.
17361 (reset_out_edges_aux): Likewise.
17362 (compute_cases_per_edge): Likewise.
17363 (add_case_node): Likewise.
17364 (dump_case_nodes): Likewise.
17365 (balance_case_nodes): Likewise.
17366 (expand_switch_as_decision_tree_p): Likewise.
17367 (emit_jump): Likewise.
17368 (emit_case_decision_tree): Likewise.
17369 (try_switch_expansion): Likewise.
17370 (do_jump_if_equal): Likewise.
17371 (emit_cmp_and_jump_insns): Likewise.
17372 (fix_phi_operands_for_edge): New function.
17373 (record_phi_operand_mapping): Likewise.
17374 (class pass_lower_switch): New pass.
17375 (pass_lower_switch::execute): New function.
17376 (make_pass_lower_switch): Likewise.
17377 (conditional_probability):
17378 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17379 * tree-pass.h: Add make_pass_lower_switch.
17381 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17384 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17387 2017-08-29 Richard Biener <rguenther@suse.de>
17389 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17390 we do not add a DW_AT_inline attribute twice.
17391 (gen_subprogram_die): Remove code setting DW_AT_inline on
17392 DECL_ABSTRACT_P nodes.
17394 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17396 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17397 calls to internal functions.
17398 (gimplify_modify_expr): Likewise.
17399 * tree-call-cdce.c (use_internal_fn): Likewise.
17400 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17401 (convert_to_divmod): Set the nothrow flag.
17402 * tree-if-conv.c (predicate_mem_writes): Likewise.
17403 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17404 (vectorizable_call): Likewise.
17405 (vectorizable_store): Likewise.
17406 (vectorizable_load): Likewise.
17407 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17408 (vect_recog_mask_conversion_pattern): Likewise.
17410 2017-08-29 Martin Liska <mliska@suse.cz>
17413 * gcc.c (driver_handle_option): Add new argument.
17414 * opts-common.c (handle_option): Pass
17415 target_option_override_hook.
17416 * opts-global.c (lang_handle_option): Add new option.
17417 (set_default_handlers): Add new argument.
17418 (decode_options): Likewise.
17419 * opts.c (target_handle_option): Likewise.
17420 (common_handle_option): Call target_option_override_hook.
17421 * opts.h (struct cl_option_handler_func): Add hook for
17422 target option override.
17423 (struct cl_option_handlers): Likewise.
17424 (set_default_handlers): Add new argument.
17425 (decode_options): Likewise.
17426 (common_handle_option): Likewise.
17427 (target_handle_option): Likewise.
17428 * toplev.c (toplev::main): Pass targetm.target_option.override
17431 2017-08-29 Richard Biener <rguenther@suse.de>
17432 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17434 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17435 life to the active subtree.
17437 2017-08-28 Jeff Law <law@redhat.com>
17439 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17440 derive_equivalences.
17441 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17442 Code moved into....
17443 (edge_info::derive_equivalences): New private member function
17445 * tree-ssa-dom.c (class edge_info): Changed from a struct
17446 to a class. Add ctor/dtor, methods and data members.
17447 (edge_info::edge_info): Renamed from allocate_edge_info.
17448 Initialize additional members.
17449 (edge_info::~edge_info): New.
17450 (free_dom_edge_info): Delete the edge info.
17451 (record_edge_info): Use new class & associated member functions.
17452 Tighten forms for testing for edge equivalences.
17453 (record_temporary_equivalences): Iterate over the simple
17454 equivalences rather than assuming there's only one per edge.
17455 (cprop_into_successor_phis): Iterate over the simple
17456 equivalences rather than assuming there's only one per edge.
17457 (optimize_stmt): Use operand_equal_p rather than pointer
17458 equality for mini-DSE code.
17460 2017-08-28 Nathan Sidwell <nathan@acm.org>
17462 * gcc.c (execute): Fold SIGPIPE handling into switch
17463 statement. Adjust internal error message.
17465 2017-08-28 Richard Biener <rguenther@suse.de>
17468 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17469 Do nothing for removed DIEs.
17471 2017-08-28 Richard Biener <rguenther@suse.de>
17473 PR tree-optimization/81977
17474 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17477 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17480 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17481 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17482 using targetm.gen_mem_thread_fence.
17484 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17487 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17488 predicate to register_operand. Reorder operands.
17489 (*btr<mode>): Ditto.
17490 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17491 (*btr<mode>_mask): Ditto.
17493 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17495 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17496 * config/rs6000/xmmintrin.h: New file.
17497 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17499 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17502 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17503 parameter and_insn and return it.
17504 (recombine_lvx_pattern): Insert a copy to ensure availability of
17505 the base register of the copied masking operation at the point of
17506 the instruction replacement.
17507 (recombine_stvx_pattern): Likewise.
17509 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17511 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17512 undocumented switches.
17513 (-mpower9-dform-vector): Likewise.
17514 (-mpower9-dform): Likewise.
17515 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17516 comments to delete references to -mpower9-dform* switches.
17517 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17518 Delete reference to -mpower9-dform* switches, test for
17519 -mpower9-vector instead.
17520 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17521 (OTHER_P9_VECTOR_MASKS): Likewise.
17522 (POWERPC_MASKS): Likewise.
17523 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17524 tests against -mpower9-dform* to -mpower9-vector. Delete code
17525 that checked for -mpower9-dform* consistancy with other options.
17526 Add test for -mpower9-misc to enable other power9 switches.
17527 (rs6000_init_hard_regno_mode_ok): Likewise.
17528 (rs6000_option_override_internal): Likewise.
17529 (rs6000_emit_prologue): Likewise.
17530 (rs6000_emit_epilogue): Likewise.
17531 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17532 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17533 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17535 (emit_fusion_p9_store): Likewise.
17536 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17537 resetting these macros if the assembler does not support ISA 3.0
17539 (TARGET_P9_DFORM_VECTOR): Likewise.
17540 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17541 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17543 2017-08-25 Alan Modra <amodra@gmail.com>
17545 PR rtl-optimization/81747
17546 * cse.c (cse_extended_basic_block): Don't attempt to record
17547 equivalences for degenerate conditional branches that branch
17548 to their fall-through.
17550 2017-08-24 Martin Sebor <msebor@redhat.com>
17552 PR middle-end/81908
17553 * gimple-fold.c (size_must_be_zero_p): New function.
17554 (gimple_fold_builtin_memory_op): Call it.
17556 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17558 * config/rs6000/mm_malloc.h: New file.
17560 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17562 PR tree-optimization/81913
17563 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17564 analysis when either IVs in condition can wrap.
17566 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17568 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17569 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17571 2017-08-24 Richard Biener <rguenther@suse.de>
17574 * targhooks.c (default_target_can_inline_p): Properly
17575 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17576 is present and always compare.
17577 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17578 infer -mfpmath=sse from TARGET_SSE_P.
17579 (ix86_can_inline_p): Properly use target_option_default_node when
17580 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17582 2017-08-24 Richard Biener <rguenther@suse.de>
17585 * dwarf2out.c (output_die): Handle flag_generate_offload like
17587 (output_comp_unit): Likewise.
17588 (gen_array_type_die): Likewise.
17589 (dwarf2out_early_finish): Likewise.
17590 (note_variable_value_in_expr): Likewise.
17591 (dwarf2out_finish): Likewise. Adjust assert.
17592 * cgraphunit.c (symbol_table::compile): Move setting of
17593 flag_generate_offload earlier ...
17594 (symbol_table::finalize_compilation_unit): ... here, before
17595 early debug finalization.
17597 2017-08-24 Richard Biener <rguenther@suse.de>
17599 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17600 and ipa-fnsummary.h.
17601 (ix86_can_inline_p): When ix86_fpmath flags do not match
17602 check whether the callee uses FP math at all.
17604 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17606 PR middle-end/81931
17607 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17608 instead of TYPE_PRECISION.
17610 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17616 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17618 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17620 (ix86_frame::stack_realign_allocate): New field.
17621 (struct machine_frame_state): Modify comments.
17622 (machine_frame_state::sp_realigned_fp_end): New field.
17623 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17624 layout calculation.
17625 (sp_valid_at): Add assertion to assure no attempt to access invalid
17626 offset of a realigned stack.
17627 (fp_valid_at): Likewise.
17628 (choose_baseaddr): Modify comments.
17629 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17630 ix86_expand_prologue.
17631 (ix86_expand_prologue): Modify stack realignment and allocation.
17632 (ix86_expand_epilogue): Modify comments.
17633 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17634 avx2_runtime, avx512f, and avx512f_runtime.
17636 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17638 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17639 (mstackrealign): Do not init to -1.
17640 * config/i386/i386.c (ix86_option_override_internal):
17641 Check opts_set, not opts when setting default value of
17642 opts->x_ix86_force_align_arg_pointer.
17644 2017-08-23 Richard Biener <rguenther@suse.de>
17646 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17647 lang_hooks.decl_printable_name.
17648 * print-rtl-function.c (print_rtx_function): Likewise.
17649 * tree-pretty-print.c (dump_function_header): Likewise.
17651 2017-08-23 Richard Biener <rguenther@suse.de>
17654 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17655 -g0 at compile-time.
17657 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17659 PR middle-end/19706
17660 * doc/sourcebuild.texi (Other hardware attributes):
17663 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17665 PR middle-end/19706
17666 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17667 Fix single-use check.
17669 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17671 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17673 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17675 * doc/install.texi: Modify to add more details on running selected
17678 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17680 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17681 is combined with -mabi=ms.
17682 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17685 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17687 PR tree-optimization/81488
17688 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17689 and cached_basis fields.
17690 (MAX_SPREAD): New constant.
17691 (alloc_cand_and_find_basis): Initialize new fields.
17692 (clear_visited): New function.
17693 (create_phi_basis_1): Rename from create_phi_basis, set visited
17694 and cached_basis fields.
17695 (create_phi_basis): New wrapper function.
17696 (phi_add_costs_1): Rename from phi_add_costs, add spread
17697 parameter, set visited field, short-circuit when limits reached.
17698 (phi_add_costs): New wrapper function.
17699 (record_phi_increments_1): Rename from record_phi_increments, set
17701 (record_phi_increments): New wrapper function.
17702 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17703 (phi_incr_cost): New wrapper function.
17704 (all_phi_incrs_profitable_1): Rename from
17705 all_phi_incrs_profitable, set visited field.
17706 (all_phi_incrs_profitable): New wrapper function.
17708 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17709 Alan Hayward <alan.hayward@arm.com>
17710 David Sherwood <david.sherwood@arm.com>
17712 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17713 that takes the outer and inner modes.
17714 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17715 comparison as the canonical test for a paradoxical subreg.
17716 * combine.c (simplify_set): Use paradoxical_subreg_p.
17717 (make_extraction): Likewise.
17718 (force_to_mode): Likewise.
17719 (rtx_equal_for_field_assignment_p): Likewise.
17720 (gen_lowpart_for_combine): Likewise.
17721 (simplify_comparison): Likewise.
17722 * cse.c (equiv_constant): Likewise.
17723 * expmed.c (store_bit_field_1): Likewise.
17724 * final.c (alter_subreg): Likewise.
17725 * fwprop.c (propagate_rtx): Likewise.
17726 (forward_propagate_subreg): Likewise.
17727 * ira-conflicts.c (ira_build_conflicts): Likewise.
17728 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17729 * lra-constraints.c (curr_insn_transform): Likewise.
17730 (split_reg): Likewise.
17731 * lra-eliminations.c (move_plus_up): Likewise.
17732 (lra_eliminate_regs_1): Likewise.
17733 * recog.c (general_operand): Likewise.
17734 * ree.c (combine_reaching_defs): Likewise.
17735 * reload.c (push_reload): Likewise.
17736 (find_reloads): Likewise.
17737 * reload1.c (elimination_effects): Likewise.
17738 (compute_reload_subreg_offset): Likewise.
17739 (choose_reload_regs): Likewise.
17740 * rtlanal.c (subreg_lsb_1): Likewise.
17741 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17742 (simplify_subreg): Likewise.
17743 * var-tracking.c (track_loc_p): Likewise.
17744 * emit-rtl.c (byte_lowpart_offset): Likewise.
17745 (paradoxical_subreg_p): Delete out-of-line definition.
17747 2017-08-22 Jeff Law <law@redhat.com>
17749 PR tree-optimization/81741
17750 PR tree-optimization/71947
17751 * tree-ssa-dom.c: Include tree-inline.h.
17752 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17753 equivalences if one is more expensive to compute than the other.
17754 * tree-ssa-scopedtables.h (class const_or_copies): Make
17755 record_const_or_copy_raw method private.
17756 (class avail_exprs_stack): New method simplify_binary_operation.
17757 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17758 avail_exprs_stack::simplify_binary_operation as needed.
17759 (avail_exprs_stack::simplify_binary_operation): New function.
17761 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17763 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17764 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17765 (DOT_SYMBOLS): Likewise.
17766 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17767 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17768 (RS6000_ABI_NAME): Likewise.
17769 (TARGET_CMODEL): Likewise.
17770 (TOC_SECTION_ASM_OP): Likewise.
17771 (SET_CMODEL): New macro.
17772 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17774 2017-08-22 Richard Biener <rguenther@suse.de>
17776 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17777 to allow for free-lang-data replacements similar to verify_type_variant.
17779 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17780 Alan Hayward <alan.hayward@arm.com>
17781 David Sherwood <david.sherwood@arm.com>
17783 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17784 VOIDmode for the LABEL_REF.
17786 2017-08-22 Richard Biener <rguenther@suse.de>
17788 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17790 2017-08-22 Richard Biener <rguenther@suse.de>
17792 * common.opt (feliminate-dwarf2-dups): Ignore.
17793 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17794 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17795 same_die_p_wrap, compute_section_prefix,
17796 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17797 (comdat_symbol_id, comdat_symbol_number): Likewise.
17798 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17800 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17801 (output_die): Mark unreachable path unreachable.
17802 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17803 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17804 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17805 (dwarf2out_early_finish): Likewise.
17807 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17809 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17811 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17814 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17815 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17816 (avr_attribute_table) <io, io_low, address>: Initialize
17817 .decl_required with true.
17819 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17821 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17822 undocumented debugging options.
17823 (-mvsx-scalar-double): Likewise.
17824 (-mallow-df-permute): Likewise.
17825 (-mvectorize-builtins): Likewise.
17826 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17827 (rs6000_builtin_vectorized_function): Likewise.
17828 (rs6000_builtin_md_vectorized_function): Likewise.
17829 (rs6000_opt_vars): Likewise.
17831 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17834 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17835 (*btrq_imm): Rename from *btrq.
17836 (*btcq_imm): Rename from *btcq.
17837 (btsc): New code attribute.
17838 (*<btsc><mode>): New insn pattern.
17839 (*btr<mode>): Ditto.
17840 (*<btsc><mode>_mask): New insn_and_split pattern.
17841 (*btr<mode>_mask): Ditto.
17843 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17844 Alan Hayward <alan.hayward@arm.com>
17845 David Sherwood <david.sherwood@arm.com>
17847 * function.c (pad_below): Simplify padding calculation.
17849 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17850 Alan Hayward <alan.hayward@arm.com>
17851 David Sherwood <david.sherwood@arm.com>
17853 * target.def (function_prologue): Remove frame size argument.
17854 (function_epilogue): Likewise.
17855 * doc/tm.texi: Regenerate.
17856 * final.c (final_start_function): Update call to function_prologue.
17857 (final_end_function): Update call to function_epilogue.
17858 (default_function_pro_epilogue): Remove frame size argument.
17859 * output.h (default_function_pro_epilogue): Likewise.
17860 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17861 (arm_output_function_prologue): Likewise.
17862 * config/frv/frv.c (frv_function_prologue): Likewise.
17863 (frv_function_epilogue): Likewise.
17864 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17865 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17866 (ia64_output_function_epilogue): Likewise.
17867 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17868 (m32r_output_function_epilogue): Likewise.
17869 * config/microblaze/microblaze.c (microblaze_function_prologue)
17870 (microblaze_function_epilogue): Likewise.
17871 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17872 (mips_output_function_epilogue): Likewise.
17873 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17874 (mmix_target_asm_function_epilogue): Likewise.
17875 * config/msp430/msp430.c (msp430_start_function): Likewise.
17876 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17877 (nds32_asm_function_epilogue): Likewise.
17878 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17879 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17880 (pa_output_function_epilogue): Likewise.
17881 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17882 (rs6000_output_function_epilogue): Likewise.
17883 * config/rl78/rl78.c (rl78_start_function): Likewise.
17884 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17885 (rs6000_output_function_epilogue): Likewise.
17886 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17887 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17888 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17889 (sparc_asm_function_epilogue): Likewise.
17891 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17893 * tree.h (type_has_mode_precision_p): New function.
17894 * convert.c (convert_to_integer_1): Use it.
17895 * expr.c (expand_expr_real_2): Likewise.
17896 (expand_expr_real_1): Likewise.
17897 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17898 * match.pd: Likewise.
17899 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17900 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17901 * tree-tailcall.c (process_assignment): Likewise.
17902 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17903 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17904 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17905 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17906 (vectorizable_assignment): Likewise.
17907 (vectorizable_shift): Likewise.
17908 (vectorizable_operation): Likewise.
17909 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17911 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17913 * match.pd: Add pow (C, x) simplification.
17915 2017-08-21 Richard Biener <rguenther@suse.de>
17917 PR tree-optimization/81900
17918 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17919 for blocks with abnormal predecessors.
17920 (compute_antic): Do not set visited flag prematurely.
17922 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17925 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17927 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17929 * stor-layout.h (vector_type_mode): Move to...
17930 * tree.h (vector_type_mode): ...here.
17931 * stor-layout.c (vector_type_mode): Move to...
17932 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17934 2017-08-21 Richard Biener <rguenther@suse.de>
17936 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17937 register_external_die hooks.
17938 (debug_false_tree_charstarstar_uhwistar): Declare.
17939 (debug_nothing_tree_charstar_uhwi): Likewise.
17940 * debug.c (do_nothing_debug_hooks): Adjust.
17941 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17942 (debug_nothing_tree_charstar_uhwi): Likewise.
17943 * dbxout.c (dbx_debug_hooks): Adjust.
17944 (xcoff_debug_hooks): Likewise.
17945 * sdbout.c (sdb_debug_hooks): Likewise.
17946 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17947 * dwarf2out.c (macinfo_label_base): New global.
17948 (dwarf2out_register_external_die): New function for the
17949 register_external_die hook.
17950 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17951 (dwarf2_debug_hooks): Use them.
17952 (dwarf2_lineno_debug_hooks): Adjust.
17953 (struct die_struct): Add with_offset flag.
17954 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17955 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17956 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17957 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17958 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17959 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17960 defining section names for the early LTO debug variants.
17961 (reset_indirect_string): New helper.
17962 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17963 (print_dw_val): Add support for offsetted symbol references.
17964 (get_ultimate_context): Split out from is_cxx.
17965 (is_cxx): Use get_ultimate_context.
17966 (is_fortran): Add decl overload.
17967 (compute_comp_unit_symbol): Split out worker from
17968 compute_section_prefix.
17969 (compute_section_prefix): Call compute_comp_unit_symbol and
17970 set comdat_type_p here.
17971 (output_die): Skip DIE symbol output for the LTO added one.
17972 Handle DIE symbol references with offset.
17973 (output_comp_unit): Guard section name mangling properly.
17974 For LTO debug sections emit a symbol at the section beginning
17975 which we use to refer to its DIEs.
17976 (add_abstract_origin_attribute): For DIEs registered via
17977 dwarf2out_register_external_die directly refer to the early
17978 DIE rather than indirectly through the shadow one we created.
17979 Remove obsolete call to dwarf2out_abstract_function for
17980 non-function/block origins.
17981 (gen_array_type_die): When generating early LTO debug do
17982 not emit DW_AT_string_length.
17983 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17984 late when in LTO. As suggested place a gcc_unreachable for
17985 the DECL_ABSTRACT_P case.
17986 (gen_subprogram_die): Avoid another specification DIE
17987 for early built declarations/definitions for the late LTO case.
17988 (gen_variable_die): Add type references for late duplicated VLA dies
17990 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17991 we have the abstract instance already.
17992 (process_scope_var): Adjust decl DIE contexts in LTO which
17993 first puts them in limbo.
17994 (gen_decl_die): Do not generate type DIEs late apart from
17995 types for VLAs or for decls we do not yet have a DIE. Do not
17996 call dwarf2out_abstract_function late.
17997 (dwarf2out_early_global_decl): Make sure to create DIEs
17998 for abstract instances of a decl first.
17999 (dwarf2out_late_global_decl): Adjust comment.
18000 (output_macinfo_op): With multiple macro sections use
18001 macinfo_label_base to distinguish labels.
18002 (output_macinfo): Likewise. Update macinfo_label_base.
18003 Pass in the line info label.
18004 (note_variable_value_in_expr): When generating LTO resolve
18005 all variable values here by generating DIEs as needed.
18006 (init_sections_and_labels): Add early LTO debug flag parameter
18007 and generate different sections and names if set. Add generation
18008 counter for the labels so we can have multiple of them.
18009 (reset_dies): Helper to allow DIEs to be output multiple times.
18010 (dwarf2out_finish): When outputting DIEs to the fat part of an
18011 LTO object first reset DIEs.
18012 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18013 (modified_type_die): Check for decl_ultimate_origin being self
18015 (gen_type_die_with_usage): Likewise.
18016 (gen_typedef_die): Allow decl_ultimate_origin being self.
18017 (set_decl_abstract_flags): Remove.
18018 (set_block_abstract_flags): Likewise.
18019 (dwarf2out_abstract_function): Treat the early generated DIEs
18020 as the abstract copy and only add DW_AT_inline and
18021 DW_AT_artificial here and call set_decl_origin_self.
18022 If the DIE has an abstract origin don't do anything.
18023 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18024 if we have none yet (Go fails to build one, PR78628).
18025 (variably_modified_type_p): Prevent endless recursion for Ada
18026 cyclic pointer types.
18027 * lto-streamer-in.c: Include debug.h.
18028 (dref_queue): New global.
18029 (lto_read_tree_1): Stream in DIE references.
18030 (lto_input_tree): Register DIE references.
18031 (input_function): Stream DECL_DEBUG_ARGS.
18032 * lto-streamer-out.c: Include debug.h.
18033 (lto_write_tree_1): Output DIE references.
18034 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18035 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18036 (output_function): Stream DECL_DEBUG_ARGS.
18037 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18038 Stream DECL_ABSTRACT_ORIGIN.
18039 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18040 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18041 DECL_CONTEXT for file-scope decls.
18042 * lto-streamer.h (struct dref_entry): Declare.
18043 (dref_queue): Likewise.
18044 * cfgexpand.c (pass_expand::execute): Do not call the
18045 outlining_inline_function hook here.
18046 * lto-wrapper.c (debug_obj): New global.
18047 (tool_cleanup): Unlink it if required.
18048 (debug_objcopy): New function.
18049 (run_gcc): Handle early debug sections in the IL files by
18050 extracting them to separate files, partially linkin them and
18051 feeding the result back as result to the linker.
18052 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18053 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18054 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18055 sections into a separate segment.
18056 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18058 (darwin_asm_dwarf_section): Likewise.
18059 (darwin_asm_output_dwarf_offset): Likewise.
18060 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18062 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18063 Alan Hayward <alan.hayward@arm.com>
18064 David Sherwood <david.sherwood@arm.com>
18066 * read-md.h (md_reader::record_potential_iterator_use): Replace
18067 pointer argument with an rtx and an index.
18068 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18069 (apply_mode_iterator): Likewise.
18070 (apply_code_iterator): Likewise.
18071 (apply_int_iterator): Likewise.
18072 (apply_subst_iterator): Likewise.
18073 (record_iterator_use): Likewise.
18074 (record_attribute_use): Likewise.
18075 (md_reader::record_potential_iterator_use): Likewise. Update calls
18076 to record_iterator_use and apply_iterator.
18077 (iterator_use): Replace ptr with x and index.
18078 (attribute_use): Likewise.
18079 (apply_attribute_uses): Update calls to apply_iterator.
18080 (apply_iterators): Likewise. Update initialization of iterator_use.
18081 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18082 and record_potential_iterator_use.
18083 (rtx_reader::read_rtx_operand): Likewise.
18085 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18086 Alan Hayward <alan.hayward@arm.com>
18087 David Sherwood <david.sherwood@arm.com>
18089 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18092 2017-08-21 Richard Biener <rguenther@suse.de>
18094 PR middle-end/81884
18095 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18096 at struct end conservatively when comparing common bases.
18098 2017-08-21 Richard Biener <rguenther@suse.de>
18100 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18101 (mem_ref_in_stmt): Remove.
18102 (determine_max_movement): Use ref index to get at the reference.
18103 (invariantness_dom_walker::before_dom_children): Deal with
18104 lim data already initialized.
18105 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18107 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18109 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18110 (TARGET_ISA_ROUND): Ditto.
18111 (TARGET_ROUND): Ditto.
18112 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18113 * config/i386/i386.md: Ditto.
18114 * config/i386/sse.md: Ditto.
18115 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18116 with OPTION_MASK_ISA_SSE4_1.
18118 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18121 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18122 __builtin_ia32_lzcnt_u16.
18124 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18127 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18128 (rs6000_set_current_function): Rewrite function to use it.
18130 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18133 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18134 and TYPE_WARN_IF_NOT_ALIGN.
18135 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18136 (handle_warn_if_not_align): New.
18137 (place_union_field): Call handle_warn_if_not_align.
18138 (place_field): Call handle_warn_if_not_align.
18139 Copy TYPE_WARN_IF_NOT_ALIGN.
18140 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18141 (layout_type): Likewise.
18142 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18144 (tree_decl_common): Add warn_if_not_align.
18145 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18146 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18147 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18148 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18149 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18150 * doc/extend.texi: Document warn_if_not_aligned attribute.
18151 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18153 2017-08-17 Martin Liska <mliska@suse.cz>
18156 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18157 (get_data_dependence): Use it as pointer type.
18158 (distribute_loop): Likewise.
18160 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18162 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18163 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18164 (p8_vmrgow_<mode>_direct): New define_insn.
18165 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18166 handle endianness for vmrgew and vmrgow permute patterns.
18168 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18170 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18171 * config/rs6000/rs6000-cpus.def: Remove comment.
18172 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18173 (POWERPC_MASKS): Likewise.
18174 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18175 use of TARGET_VSX_TIMODE.
18176 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18177 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18178 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18179 (rs6000_option_override_internal): Remove dead code.
18180 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18181 (rs6000_legitimize_reload_address): Likewise.
18182 (rs6000_legitimate_address_p): Likewise.
18183 (rs6000_opt_masks): Delete "vsx-timode".
18184 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18185 from function comment.
18186 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18187 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18188 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18190 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18191 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18192 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18193 (VSX_TI): Likewise.
18195 (define_peephole2): Likewise.
18197 2017-08-17 Martin Sebor <msebor@redhat.com>
18200 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18201 past the end of an array.
18202 (test_pp_format): Add test cases.
18204 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18206 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18207 missing ECF_NOTHROW flags.
18209 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18212 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18213 operands residing in integer registers.
18214 (*vsx_le_perm_load_<mode>): Likewise.
18215 (*vsx_le_perm_store_<mode>): Likewise.
18216 (define_peephole2): Add peepholes to optimize the above.
18218 2017-08-17 Marek Polacek <polacek@redhat.com>
18220 PR middle-end/81814
18221 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18222 to mimic what shorten_compare did. Change the return type to bool.
18223 (fold_cond_expr_with_comparison): Update call to
18224 operand_equal_for_comparison_p.
18225 (fold_ternary_loc): Likewise.
18227 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18229 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18231 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18233 2017-08-17 Richard Biener <rguenther@suse.de>
18235 * tree-ssa-structalias.c (solve_graph): When propagating
18236 to successors update the graphs succ edges and avoid duplicate work.
18238 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18241 * config/i386/i386.c (ix86_option_override_internal): Save target
18242 specific options after ix86_stack_protector_guard_reg was changed.
18244 2017-08-17 Richard Biener <rguenther@suse.de>
18246 PR tree-optimization/81827
18247 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18248 (new_var_info): Initialize it conservatively.
18249 (get_call_vi): Mark register vars.
18250 (new_scalar_tmp_constraint_exp): Likewise.
18251 (handle_rhs_call): Likewise.
18252 (handle_const_call): Likewise.
18253 (create_function_info_for): Likewise.
18254 (solve_constraints): Sort varinfos to separate register from
18255 non-register vars to pack points-to solution bitmaps during
18258 2017-08-17 Marek Polacek <polacek@redhat.com>
18260 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18262 2017-08-17 Richard Biener <rguenther@suse.de>
18264 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18265 to true when overflow is undefined and we saturated the result.
18267 2017-08-17 Alan Modra <amodra@gmail.com>
18270 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18271 Don't use store multiple if only one reg needs saving.
18272 (interesting_frame_related_regno): New function.
18273 (rs6000_frame_related): Don't emit frame info for regs that
18275 (rs6000_emit_epilogue): Likewise.
18277 2017-08-16 Nathan Sidwell <nathan@acm.org>
18279 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18280 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18281 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18282 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18283 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18284 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18285 (verify_type): Adjust for TYPE_BINFO move.
18286 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18287 process TYPE_BINFO directly.
18288 (hash_tree): Likewise.
18289 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18291 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18294 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18296 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18298 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18301 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18302 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18303 Add ix86_binary_operator_ok to insn constraint.
18304 (*iordi_1_bts): Ditto.
18305 (*xordi_1_btc): Ditto.
18306 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18307 Update corresponding peephole2 pattern.
18311 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18313 PR tree-optimization/81832
18314 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18315 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18317 2017-08-16 Marek Polacek <polacek@redhat.com>
18320 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18321 regarding size_zero_node.
18323 2017-08-16 Martin Liska <mliska@suse.cz>
18326 * config.gcc: Respect previously set extra_objs in case
18329 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18331 PR tree-optimization/81835
18332 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18333 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18334 not depend on the phi.
18336 2017-08-16 Alan Modra <amodra@gmail.com>
18338 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18341 2017-08-16 Alan Modra <amodra@gmail.com>
18343 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18344 (save_reg_p): ..into this. Update all callers.
18345 (first_reg_to_save): Simplify.
18347 2017-08-16 Alan Modra <amodra@gmail.com>
18349 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18352 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18356 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18357 constant count if that count is less than 32.
18359 2017-08-15 Nathan Sidwell <nathan@acm.org>
18361 * gcc.c (execute): Emit friendlier message if inferior is killed
18362 by an external cause.
18364 2017-08-15 Richard Biener <rguenther@suse.de>
18366 PR tree-optimization/81790
18367 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18368 CONSTRUCTORs from simplifying and VN.
18370 2017-08-14 Martin Sebor <msebor@redhat.com>
18372 * builtin-attrs.def: Add comments.
18374 2017-08-14 Martin Sebor <msebor@redhat.com>
18377 * doc/extend.texi (attribute nonstring): Document new attribute.
18379 2017-08-14 Martin Sebor <msebor@redhat.com>
18382 * tree-diagnostic.c (default_tree_printer): Handle %G.
18383 * gimple-pretty-print.h (percent_G_format): Declare new function.
18384 * gimple-pretty-print.c (percent_G_format): Define.
18385 * tree-pretty-print.c (percent_K_format): Add argument.
18387 2017-08-14 Martin Sebor <msebor@redhat.com>
18389 PR translation/79998
18390 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18391 Remove a stray space.
18393 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18396 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18397 (*iordi_1_bts): Ditto.
18398 (*xordi_1_btc): Ditto.
18400 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18403 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18404 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18406 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18407 quoted strings, and make more translator-friendly.
18408 (darwin_rs6000_override_options): Likewise.
18409 (rs6000_option_override_internal): Likewise.
18410 (rs6000_return_in_memory): Fix overlong line.
18411 (init_cmulative_args): Use quoted strings, and make more
18412 translator-friendly.
18413 (rs6000_pass_by_reference): Fix overlong line.
18414 (def_builtin): Use quoted strings.
18415 (altivec_expand_predicate_builtin): Use quoted strings, and make
18416 more translator-friendly.
18417 (htm_expand_builtin): Use quoted strings.
18418 (cpu_expand_builtin): Use quoted strings, and make more
18419 translator-friendly.
18420 (altivec_expand_builtin): Likewise.
18421 (paired_expand_predicate_builtin): Likewise.
18422 (rs6000_invalid_builtin): Likewise.
18423 (builtin_function_type): Use quoted strings.
18424 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18425 more translator-friendly.
18426 (rs6000_trampoline_init): Likewise.
18427 (rs6000_handle_altivec_attribute): Likewise.
18428 (rs6000_inner_target_options): Use quoted strings.
18429 (rs6000_disable_incompatible_switches): Likewise.
18430 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18431 strings, and make more translator-friendly.
18432 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18434 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18436 PR tree-optimization/81799
18437 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18438 cond_expr to simple gimple operand.
18440 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18442 PR middle-end/46932
18443 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18445 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18449 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18450 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18451 TARGET_GASISR_PROLOGUES.
18452 * config/avr/avr.c (avr_option_override): Same.
18453 (avr_pass_pre_proep::execute): Same.
18455 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18458 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18459 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18461 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18463 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18464 define_insn_and_split. Split to a memory load from 0 in
18465 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18466 using PTR mode iterator.
18467 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18468 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18469 (*add_tp_<mode>): Redefine as define_insn_and_split.
18470 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18471 address space. Merge with *add_tp_x32 using PTR mode iterator.
18472 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18473 Split to an add with a memory load from 0 in
18474 DEFAULT_TLS_SEG_REG address space.
18476 2017-08-12 Andrew Pinski <apinski@cavium.com>
18478 * config/aarch64/aarch64-option-extensions.def (rdma):
18479 Fix feature string to what Linux prints out in /proc/cpuinfo.
18481 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18484 * dwarf2out.c (modified_type_die): For C typedef types that have
18485 an ultimate origin, process the ultimate origin instead of the
18487 (gen_typedef_die): Assert that input DECLs have no ultimate
18489 (gen_type_die_with_usage): For typedef variants that have an
18490 ultimate origin, just call gen_decl_die on the original DECL.
18491 (process_scope_var): Avoid creating DIEs for local typedefs and
18492 concrete static variables.
18494 2017-08-12 Alan Modra <amodra@gmail.com>
18498 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18499 match gnu-user.h startfile.
18500 (ENDFILE_LINUX_SPEC): Similarly.
18502 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18505 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18507 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18509 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18510 * config/aarch64/aarch64.md (mov<mode>): Change.
18511 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18512 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18513 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18515 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18517 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18518 for storage order barriers.
18520 2017-08-11 Martin Liska <mliska@suse.cz>
18523 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18524 variables of void type.
18526 2017-08-11 Martin Liska <mliska@suse.cz>
18528 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18529 TARGET_SUPPORTS_ALIASES.
18530 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18531 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18532 (optimize_weakref): Likewise.
18533 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18534 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18535 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18537 2017-08-11 Martin Liska <mliska@suse.cz>
18540 * config/i386/i386.c (make_resolver_func): Do complete
18541 refactoring of the function.
18543 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18546 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18547 * config/i386/i386.c (ix86_stack_protect_guard): Use
18548 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18549 * doc/invoke.texi (x86 Options): Document
18550 -mstack-protector-guard-symbol= option.
18552 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18554 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18555 * config/i386/i386.c (ix86_split_stack_guard): New function.
18556 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18557 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18558 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18559 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18560 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18561 (split_stack_space_check): Call ix86_split_stack_guard.
18563 2017-08-10 Martin Sebor <msebor@redhat.com>
18565 * print-tree.c (print_node): Print location using the established
18567 Replace spaces with colons.
18568 (debug_raw, debug): Ditto.
18570 2017-08-10 Martin Sebor <msebor@redhat.com>
18573 * pretty-print.c (pp_format): Correct the handling of %s precision.
18575 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18578 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18580 (ix86_finalize_stack_frame_flags): This. Also clear
18581 frame_pointer_needed if -fno-omit-frame-pointer is used without
18583 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18584 with ix86_finalize_stack_frame_flags.
18585 (ix86_expand_epilogue): Likewise.
18586 (ix86_expand_split_stack_prologue): Likewise.
18587 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18589 2017-08-10 Martin Liska <mliska@suse.cz>
18592 * c-attribs.c (handle_target_attribute):
18593 Report warning for an empty string argument of target attribute.
18595 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18598 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18600 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18601 or DECL_NONLOCAL labels.
18602 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18603 or DECL_NONLOCAL labels here.
18605 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18607 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18608 to indicate when early gimple folding has been disabled.
18609 (rs6000_gimple_fold_builtin): Add debug content.
18610 (rs6000_invalid_builtin): Fix whitespace.
18611 (rs6000_expand_builtin): Fix whitespace.
18612 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18614 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18617 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18618 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18620 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18622 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18623 (qdf24xx): Likewise.
18624 * config/aarch64/aarch64.md: Include falkor.md.
18625 * config/aarch64/falkor.md: New.
18627 2017-08-09 Marek Polacek <polacek@redhat.com>
18630 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18631 * diagnostic.c (emit_diagnostic): Add a comment.
18632 (emit_diagnostic_valist): New function.
18634 2017-08-09 Marek Polacek <polacek@redhat.com>
18637 * input.c (make_location): New overload.
18638 * input.h (make_location): Declare.
18640 2017-08-08 Alan Modra <amodra@gmail.com>
18641 H.J. Lu <hongjiu.lu@intel.com>
18644 * gcc.c (NO_PIE_SPEC): Delete.
18645 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18647 (LINK_PIE_SPEC): ..to here.
18648 (LINK_COMMAND_SPEC): Support -no-pie.
18649 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18650 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18652 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18653 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18654 (ENDFILE_CRTEND_SPEC): Similarly.
18656 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18659 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18660 (mstack-protector-guard-offset=): Ditto.
18661 * config/i386/i386.c (ix86_option_override): Handle
18662 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18664 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18665 ix86_stack_protect_guard_offset variables.
18666 (TARGET_STACK_PROTECT_GUARD): Always define.
18667 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18668 and -mstack-protector-guard-offset= options.
18670 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18672 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18673 boundary case for the last candidate.
18675 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18677 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18678 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18680 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18682 PR middle-end/19706
18683 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18684 * config/aarch64/aarch64-builtins.c
18685 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18686 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18687 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18689 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18690 Andrew Pinski <pinskia@gmail.com>
18692 PR middle-end/19706
18693 * internal-fn.def (XORSIGN): New.
18694 * optabs.def (xorsign_optab): New.
18695 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18696 (convert_expand_mult_copysign): New.
18697 (pass_optimize_widening_mul::execute): Call
18698 convert_expand_mult_copysign.
18700 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18702 PR tree-optimization/81354
18703 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18704 Insert on edges rather than explicitly creating landing pads.
18705 (analyze_candidates_and_replace): Commit edge inserts.
18707 2017-08-08 Richard Biener <rguenther@suse.de>
18709 PR middle-end/81719
18710 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18711 (expand_simple_operations): Also look through ADDR_EXPRs with
18712 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18714 2017-08-08 Richard Biener <rguenther@suse.de>
18716 PR tree-optimization/81723
18717 * tree-vect-slp.c (struct bst_traits): New hash traits.
18718 (bst_fail): New global.
18719 (vect_build_slp_tree_2): New worker, split out from ...
18720 (vect_build_slp_tree): ... this now wrapping it with using
18721 bst_fail set to cache SLP tree build fails. Properly handle
18723 (vect_analyze_slp_instance): Allocate and free bst_fail.
18725 2017-08-08 Martin Liska <mliska@suse.cz>
18728 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18729 LABEL_DECLs that can be from a different function.
18731 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18733 PR tree-optimization/81744
18734 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18735 loop's number of iterations.
18737 2017-08-08 Martin Liska <mliska@suse.cz>
18739 * asan.c: Include header files.
18740 * attribs.c (build_decl_attribute_variant): New function moved
18742 (build_type_attribute_qual_variant): Likewise.
18743 (cmp_attrib_identifiers): Likewise.
18744 (simple_cst_list_equal): Likewise.
18745 (omp_declare_simd_clauses_equal): Likewise.
18746 (attribute_value_equal): Likewise.
18747 (comp_type_attributes): Likewise.
18748 (build_type_attribute_variant): Likewise.
18749 (lookup_ident_attribute): Likewise.
18750 (remove_attribute): Likewise.
18751 (merge_attributes): Likewise.
18752 (merge_type_attributes): Likewise.
18753 (merge_decl_attributes): Likewise.
18754 (merge_dllimport_decl_attributes): Likewise.
18755 (handle_dll_attribute): Likewise.
18756 (attribute_list_equal): Likewise.
18757 (attribute_list_contained): Likewise.
18758 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18759 (lookup_attribute_by_prefix): Likewise.
18760 * bb-reorder.c: Include header files.
18761 * builtins.c: Likewise.
18762 * calls.c: Likewise.
18763 * cfgexpand.c: Likewise.
18764 * cgraph.c: Likewise.
18765 * cgraphunit.c: Likewise.
18766 * convert.c: Likewise.
18767 * dwarf2out.c: Likewise.
18768 * final.c: Likewise.
18769 * fold-const.c: Likewise.
18770 * function.c: Likewise.
18771 * gimple-expr.c: Likewise.
18772 * gimple-fold.c: Likewise.
18773 * gimple-pretty-print.c: Likewise.
18774 * gimple.c: Likewise.
18775 * gimplify.c: Likewise.
18776 * hsa-common.c: Likewise.
18777 * hsa-gen.c: Likewise.
18778 * internal-fn.c: Likewise.
18779 * ipa-chkp.c: Likewise.
18780 * ipa-cp.c: Likewise.
18781 * ipa-devirt.c: Likewise.
18782 * ipa-fnsummary.c: Likewise.
18783 * ipa-inline.c: Likewise.
18784 * ipa-visibility.c: Likewise.
18786 * lto-cgraph.c: Likewise.
18787 * omp-expand.c: Likewise.
18788 * omp-general.c: Likewise.
18789 * omp-low.c: Likewise.
18790 * omp-offload.c: Likewise.
18791 * omp-simd-clone.c: Likewise.
18792 * opts-global.c: Likewise.
18793 * passes.c: Likewise.
18794 * predict.c: Likewise.
18795 * sancov.c: Likewise.
18796 * sanopt.c: Likewise.
18797 * symtab.c: Likewise.
18798 * toplev.c: Likewise.
18799 * trans-mem.c: Likewise.
18800 * tree-chkp.c: Likewise.
18801 * tree-eh.c: Likewise.
18802 * tree-into-ssa.c: Likewise.
18803 * tree-object-size.c: Likewise.
18804 * tree-parloops.c: Likewise.
18805 * tree-profile.c: Likewise.
18806 * tree-ssa-ccp.c: Likewise.
18807 * tree-ssa-live.c: Likewise.
18808 * tree-ssa-loop.c: Likewise.
18809 * tree-ssa-sccvn.c: Likewise.
18810 * tree-ssa-structalias.c: Likewise.
18811 * tree-ssa.c: Likewise.
18812 * tree-streamer-in.c: Likewise.
18813 * tree-vectorizer.c: Likewise.
18814 * tree-vrp.c: Likewise.
18815 * tsan.c: Likewise.
18816 * ubsan.c: Likewise.
18817 * varasm.c: Likewise.
18818 * varpool.c: Likewise.
18819 * tree.c: Remove functions moved to attribs.[ch].
18820 * tree.h: Likewise.
18821 * config/aarch64/aarch64.c: Add attrs.h header file.
18822 * config/alpha/alpha.c: Likewise.
18823 * config/arc/arc.c: Likewise.
18824 * config/arm/arm.c: Likewise.
18825 * config/avr/avr.c: Likewise.
18826 * config/bfin/bfin.c: Likewise.
18827 * config/c6x/c6x.c: Likewise.
18828 * config/cr16/cr16.c: Likewise.
18829 * config/cris/cris.c: Likewise.
18830 * config/darwin.c: Likewise.
18831 * config/epiphany/epiphany.c: Likewise.
18832 * config/fr30/fr30.c: Likewise.
18833 * config/frv/frv.c: Likewise.
18834 * config/ft32/ft32.c: Likewise.
18835 * config/h8300/h8300.c: Likewise.
18836 * config/i386/winnt.c: Likewise.
18837 * config/ia64/ia64.c: Likewise.
18838 * config/iq2000/iq2000.c: Likewise.
18839 * config/lm32/lm32.c: Likewise.
18840 * config/m32c/m32c.c: Likewise.
18841 * config/m32r/m32r.c: Likewise.
18842 * config/m68k/m68k.c: Likewise.
18843 * config/mcore/mcore.c: Likewise.
18844 * config/microblaze/microblaze.c: Likewise.
18845 * config/mips/mips.c: Likewise.
18846 * config/mmix/mmix.c: Likewise.
18847 * config/mn10300/mn10300.c: Likewise.
18848 * config/moxie/moxie.c: Likewise.
18849 * config/msp430/msp430.c: Likewise.
18850 * config/nds32/nds32-isr.c: Likewise.
18851 * config/nds32/nds32.c: Likewise.
18852 * config/nios2/nios2.c: Likewise.
18853 * config/nvptx/nvptx.c: Likewise.
18854 * config/pa/pa.c: Likewise.
18855 * config/pdp11/pdp11.c: Likewise.
18856 * config/powerpcspe/powerpcspe.c: Likewise.
18857 * config/riscv/riscv.c: Likewise.
18858 * config/rl78/rl78.c: Likewise.
18859 * config/rx/rx.c: Likewise.
18860 * config/s390/s390.c: Likewise.
18861 * config/sh/sh.c: Likewise.
18862 * config/sol2.c: Likewise.
18863 * config/sparc/sparc.c: Likewise.
18864 * config/spu/spu.c: Likewise.
18865 * config/stormy16/stormy16.c: Likewise.
18866 * config/tilegx/tilegx.c: Likewise.
18867 * config/tilepro/tilepro.c: Likewise.
18868 * config/v850/v850.c: Likewise.
18869 * config/vax/vax.c: Likewise.
18870 * config/visium/visium.c: Likewise.
18871 * config/xtensa/xtensa.c: Likewise.
18873 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18876 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18877 constraints since the -mupper-regs-* switches have been
18879 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18880 into a vector from a double word element that was extracted from
18881 another vector, and eliminate extra XXPERMDI instructions.
18882 (vsx_concat_<mode>_2): Likewise.
18883 (vsx_concat_<mode>_3): Likewise.
18884 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18885 concat to allow optimizing inserts from previous extracts.
18887 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18889 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18890 memory reference to a SSP offset in TLS address space.
18891 (ix86_print_operand) <case '@'>: Remove.
18892 (ix86_print_operand_punct_valid_p): Remove '@' code.
18893 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18894 UNSPEC_SP_TLS_TEST.
18895 (stack_tls_protect_set_<mode>): Remove.
18896 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18897 (stack_tls_protect_test_<mode>): Remove.
18898 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18900 2017-08-07 Olivier Hainque <hainque@adacore.com>
18903 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18905 2017-08-07 Douglas Rupp <rupp@adacore.com>
18907 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18908 variable was referenced as multidir in command.
18910 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18913 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18916 2017-08-07 Martin Liska <mliska@suse.cz>
18918 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18919 * config/rl78/rl78.c: Add include of attribs.h.
18920 * config/sh/sh.c: Likewise.
18921 * config/v850/v850.c: Likewise.
18923 2017-08-07 Tom de Vries <tom@codesourcery.com>
18925 PR middle-end/78266
18926 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18928 2017-08-07 Martin Liska <mliska@suse.cz>
18930 * config/mips/mips.c: Include attribs.h.
18932 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18935 * doc/invoke.texi: Document change in behvaior for -Ofast for
18938 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18940 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18942 (aarch64_pop_regs): Likewise.
18943 (aarch64_gen_load_pair): Likewise.
18944 (aarch64_save_callee_saves): Likewise.
18945 (aarch64_restore_callee_saves): Likewise.
18947 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18949 * config/i386/i386.c: Revert the last change.
18951 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18954 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18956 (ix86_finalize_stack_frame_flags): This. Also clear
18957 frame_pointer_needed if -fno-omit-frame-pointer is used without
18959 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18960 with ix86_finalize_stack_frame_flags.
18961 (ix86_expand_epilogue): Likewise.
18962 (ix86_expand_split_stack_prologue): Likewise.
18964 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18967 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18968 to P_AES for Westmere.
18970 2017-08-07 Jonathan Yong <10walls@gmail.com>
18972 * config/i386/mingw.opt (fset-stack-executable): Removed.
18973 * config/i386/cygming.opt (fset-stack-executable): Moved
18975 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18977 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18979 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18981 2017-08-07 Marek Polacek <polacek@redhat.com>
18983 PR middle-end/81737
18984 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18986 2017-08-07 Martin Liska <mliska@suse.cz>
18988 * attribs.h (canonicalize_attr_name): New function.
18989 (cmp_attribs): Move from c-format.c and adjusted.
18990 (is_attribute_p): Moved from tree.h.
18991 * tree-inline.c: Add new includes.
18992 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18993 (private_is_attribute_p): Remove.
18994 (private_lookup_attribute): Likewise.
18995 (private_lookup_attribute_by_prefix): Simplify.
18996 (remove_attribute): Use is_attribute_p.
18997 * tree.h: Remove removed declarations.
18999 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19001 PR middle-end/81698
19002 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19003 instead of computing it in the function. Formatting fix.
19004 (expand_case): Don't rely on default_edge being the first edge,
19005 clear it if removing it, pass default_edge to
19006 emit_case_dispatch_table.
19007 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19010 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19012 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19013 insn in the function, emit NOP after the insn.
19015 2017-08-06 Tom de Vries <tom@codesourcery.com>
19017 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19020 2017-08-06 Tom de Vries <tom@codesourcery.com>
19022 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19025 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19027 PR tree-optimization/57371
19028 * match.pd: New pattern.
19030 2017-08-04 Marek Polacek <polacek@redhat.com>
19032 PR middle-end/81695
19033 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19034 perform the computation in offset_int.
19036 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19038 PR tree-optimization/81136
19039 * tree-vectorizer.h: Include tree-hash-traits.h.
19040 (vec_base_alignments): New typedef.
19041 (vec_info): Add a base_alignments field.
19042 (vect_record_base_alignments): Declare.
19043 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19045 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19046 (create_data_ref): Add an is_conditional_in_stmt argument.
19047 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19048 the is_conditional_in_stmt field.
19049 (data_ref_loc): Add an is_conditional_in_stmt field.
19050 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19051 (find_data_references_in_stmt): Update call to create_data_ref.
19052 (graphite_find_data_references_in_stmt): Likewise.
19053 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19054 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19055 (vect_record_base_alignment): New function.
19056 (vect_record_base_alignments): Likewise.
19057 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19058 for nested statements even if we fail to compute a misalignment.
19059 Use pooled base alignments for unconditional references.
19060 (vect_find_same_alignment_drs): Compare base addresses instead
19062 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19063 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19065 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19067 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19068 Add an explicit name for the enum. Use auto_vec for slp_instances
19069 and grouped_stores.
19070 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19072 (_bb_vec_info): Add a constructor and destructor.
19073 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19074 (destroy_loop_vec_info): Delete.
19075 (vect_destroy_datarefs): Likewise.
19076 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19077 (vec_info::vec_info): New function.
19078 (vec_info::~vec_info): Likewise.
19079 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19080 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19081 destroy_loop_vec_info.
19082 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19083 (_loop_vec_info::_loop_vec_info): ...this.
19084 (destroy_loop_vec_info): Replace with...
19085 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19086 the stmt_vec_infos. Leave handling of vec_info information to its
19087 destructor. Remove explicit vector releases.
19088 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19089 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19090 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19091 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19092 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19093 (destroy_bb_vec_info): Replace with...
19094 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19095 information to its destructor.
19096 (vect_slp_analyze_bb_1): Use new and delete instead of
19097 new_bb_vec_info and destroy_bb_vec_info.
19098 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19101 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19103 * tree-data-ref.h (subscript): Add access_fn field.
19104 (data_dependence_relation): Add could_be_independent_p.
19105 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19106 (same_access_functions): Move to tree-data-ref.c.
19107 * tree-data-ref.c (ref_contains_union_access_p): New function.
19108 (access_fn_component_p): Likewise.
19109 (access_fn_components_comparable_p): Likewise.
19110 (dr_analyze_indices): Add a reference to access_fn_component_p.
19111 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19113 (constant_access_functions): Likewise.
19114 (add_other_self_distances): Likewise.
19115 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19116 (initialize_data_dependence_relation): Use XCNEW and remove
19117 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19118 of access functions that have the same type. Allow the
19119 subsequence to end with different bases in some circumstances.
19120 Record the chosen access functions in SUB_ACCESS_FN.
19121 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19122 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19123 (subscript_dependence_tester_1): Likewise dra and drb.
19124 (build_classic_dist_vector): Update calls accordingly.
19125 (subscript_dependence_tester): Likewise.
19126 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19127 DDR_COULD_BE_INDEPENDENT_P.
19128 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19129 comp_alias_ddrs instead of may_alias_ddrs.
19130 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19132 (vect_analyze_data_ref_dependence): Use it if
19133 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19134 distance vectors if that fails.
19135 (dependence_distance_ge_vf): New function.
19136 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19137 LOOP_VINFO_MAY_ALIAS_DDRS.
19139 2017-08-04 Richard Biener <rguenther@suse.de>
19141 PR middle-end/81705
19142 * fold-const.c (fold_binary_loc): Properly restrict
19143 minus_var0 && minus_var1 case when associating undefined overflow
19146 2017-08-04 Tom de Vries <tom@codesourcery.com>
19148 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19150 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19152 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19153 Don't start diagnostic messages with a capital letter.
19154 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19156 (rs6000_invalid_builtin): Likewise.
19157 (rs6000_trampoline_init): Likewise.
19159 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19162 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19163 after setting changeable df flags.
19165 2017-08-03 Richard Biener <rguenther@suse.de>
19167 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19168 up if the use is in USE - X.
19170 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19172 * toplev.c (dumpfile.h): New include.
19173 (internal_error_reentered): New static function. Use it...
19174 (internal_error_function): ...here to handle reentered internal_error.
19176 2017-08-03 Richard Biener <rguenther@suse.de>
19178 PR middle-end/81148
19179 * fold-const.c (split_tree): Add minus_var and minus_con
19180 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19181 here but always use minus_*.
19182 (associate_trees): Assert we never associate with MINUS_EXPR
19183 and NULL first operand. Do not recurse for PLUS_EXPR operands
19184 when associating as MINUS_EXPR either.
19185 (fold_binary_loc): Track minus_var and minus_con.
19187 2017-08-03 Tom de Vries <tom@codesourcery.com>
19190 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19191 ACCEL_COMPILER, apply finish_options on
19192 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19194 2017-08-03 Tom de Vries <tom@codesourcery.com>
19197 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19198 function_entry_patch_area_size > 0.
19200 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19203 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19204 instead of 10??LU, perform unit multiplication in wide_int,
19205 don't change alloc_object_size_limit if the limit is larger
19208 PR tree-optimization/81655
19209 PR tree-optimization/81588
19210 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19211 the case when ranges[i].low and high are 1 for unsigned type with
19214 PR middle-end/81052
19215 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19216 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19218 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19220 * tree-vrp.h: Add include guard.
19222 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19225 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19226 (ud2): New insn pattern.
19227 * config/i386/i386.c (ix86_expand_epilogue):
19228 For naked functions, generate ud2 instead of trap insn.
19230 2017-08-02 Marek Polacek <polacek@redhat.com>
19233 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19235 2017-08-02 Tom de Vries <tom@codesourcery.com>
19236 Cesar Philippidis <cesar@codesourcery.com>
19238 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19239 Add missing edge probabilities.
19241 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19243 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19244 Correct endianness.
19246 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19248 PR middle-end/79499
19249 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19250 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19251 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19252 prologue_seq sequences - if any.
19254 2017-08-02 Richard Biener <rguenther@suse.de>
19256 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19257 via vectors if supported, integer extracts via punning if supported
19258 or otherwise vector extracts.
19260 2017-08-02 Richard Biener <rguenther@suse.de>
19262 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19264 (bitmap_insert_into_set): ... this.
19266 2017-08-02 Richard Biener <rguenther@suse.de>
19268 PR tree-optimization/81633
19270 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19272 PR tree-optimization/71752
19273 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19275 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19277 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19278 (machine_function::call_ms2sysv_pad_out): Remove field.
19279 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19280 (ix86_compute_frame_layout): Likewise.
19282 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19285 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19286 attribute with interrupt attribute.
19288 2017-08-01 Andrew Pinski <apinski@cavium.com>
19290 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19291 BIT_INSERT_EXPR's operand 1
19292 to see if the types precision matches.
19294 2017-08-01 Martin Liska <mliska@suse.cz>
19296 PR middle-end/70140
19297 * builtins.c (expand_builtin_memcpy_args): Remove.
19298 (expand_builtin_memcpy): Call newly added function
19299 expand_builtin_memory_copy_args.
19300 (expand_builtin_memcpy_with_bounds): Likewise.
19301 (expand_builtin_mempcpy): Remove last argument.
19302 (expand_builtin_mempcpy_with_bounds): Likewise.
19303 (expand_builtin_memory_copy_args): New function created from
19304 expand_builtin_mempcpy_args with small modifications.
19305 (expand_builtin_mempcpy_args): Remove.
19306 (expand_builtin_stpcpy): Remove unused argument.
19307 (expand_builtin): Likewise.
19308 (expand_builtin_with_bounds): Likewise.
19310 2017-08-01 Martin Liska <mliska@suse.cz>
19313 Make mempcpy more optimal (PR middle-end/70140).
19315 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19319 __builtin_vec_cmpne verify both arguments are compatible vectors
19320 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19321 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19322 move computation of aligned to after checking the argument types.
19326 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19327 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19330 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19331 Doug Rupp <rupp@adacore.com>
19332 Olivier Hainque <hainque@adacore.com>
19334 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19335 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19337 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19338 for TARGET_OS_CPP_BUILTIN.
19339 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19340 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19342 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19343 passing required abi options to the assembler for EABI configurations.
19344 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19345 of .text.hot and .text.unlikely sections for kernel modules when
19346 using ARM style exceptions.
19347 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19348 options. Add EXTRA_CC1_SPEC.
19349 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19351 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19353 (ARM_TARGET2_DWARF_FORMAT): Define.
19354 * config/arm/t-vxworks: Adjust multilib control to removal of the
19355 Diab command line options.
19357 2017-08-01 Martin Liska <mliska@suse.cz>
19359 PR gcov-profile/81561
19360 * gcov.c (unblock): Make unblocking safe as we need to preserve
19361 index correspondence of blocks and block_lists.
19363 2017-08-01 Richard Biener <rguenther@suse.de>
19365 PR tree-optimization/81181
19366 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19367 (compute_antic): ... end of iteration here.
19369 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19371 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19372 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19373 (ftree-slp-vectorize): Likewise.
19374 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19375 can no longer be set independent of flag_tree_loop_vectorize.
19376 * omp-general.c (emp_max_vf): Likewise.
19377 * opts.c (enable_fdo_optimizations): Remove references to
19378 flag_tree_vectorize, these are now implicit.
19379 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19380 and leave it for the options machinery.
19382 2017-08-01 Martin Liska <mliska@suse.cz>
19384 PR middle-end/70140
19385 * builtins.c (expand_builtin_memcpy_args): Remove.
19386 (expand_builtin_memcpy): Call newly added function
19387 expand_builtin_memory_copy_args.
19388 (expand_builtin_memcpy_with_bounds): Likewise.
19389 (expand_builtin_mempcpy): Remove last argument.
19390 (expand_builtin_mempcpy_with_bounds): Likewise.
19391 (expand_builtin_memory_copy_args): New function created from
19392 expand_builtin_mempcpy_args with small modifications.
19393 (expand_builtin_mempcpy_args): Remove.
19394 (expand_builtin_stpcpy): Remove unused argument.
19395 (expand_builtin): Likewise.
19396 (expand_builtin_with_bounds): Likewise.
19398 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19401 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19402 print "ds:" only for immediates in generic address space.
19404 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19407 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19408 (ix86_function_ok_for_sibcall): Return false for naked functions.
19410 2017-08-01 Richard Biener <rguenther@suse.de>
19412 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19413 (compute_antic): Seed worklist with exit block predecessors.
19414 * cfganal.c (dfs_find_deadend): For a cycle return the source
19415 of the edge closing it.
19417 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19419 * config/aarch64/aarch64.c
19420 (aarch64_can_const_movi_rtx_p): Move 0 check.
19422 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19424 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19425 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19427 * match.pd: Ditto in address comparison pattern.
19429 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19431 PR tree-optimization/81627
19432 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19433 closed ssa form for store-store chain.
19435 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19437 PR tree-optimization/81620
19438 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19439 for store-store chain.
19441 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19443 PR tree-optimization/81588
19444 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19445 ranges[i].in_p, invert comparison code ccode. For >/>=,
19446 swap rhs1 and rhs2 and comparison code unconditionally,
19447 for </<= don't do that. Don't swap rhs1/rhs2 again if
19448 ranges[i].in_p, instead invert comparison code ccode if
19449 opcode or oe->rank is BIT_IOR_EXPR.
19452 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19453 a direct optab to conversion optab.
19454 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19455 with GET_MODE_INNER as last argument instead of optab_handler.
19456 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19457 vector extraction if possible and optab is available.
19458 * expr.c (store_constructor): Use convert_optab_handler instead
19459 of optab_handler. Use vector initialization from smaller
19460 vectors if possible and optab is available.
19461 * tree-vect-stmts.c (vectorizable_load): Likewise.
19462 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19463 now have two modes.
19464 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19465 of vec_init from half-sized vectors with the same element mode.
19466 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19467 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19468 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19469 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19470 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19471 after mode in gen_vec_extract* calls.
19472 (vec_extract<mode>): Renamed to ...
19473 (vec_extract<mode><ssescalarmodelower>): ... this.
19474 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19475 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19476 element mode after mode in gen_vec_init* calls.
19477 (VEC_INIT_HALF_MODE): New mode iterator.
19478 (vec_init<mode>): Renamed to ...
19479 (vec_init<mode><ssescalarmodelower>): ... this.
19480 (vec_init<mode><ssehalfvecmodelower>): New expander.
19481 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19482 (vec_extractv2sfsf): ... this.
19483 (vec_initv2sf): Renamed to ...
19484 (vec_initv2sfsf): ... this.
19485 (vec_extractv2si): Renamed to ...
19486 (vec_extractv2sisi): ... this.
19487 (vec_initv2si): Renamed to ...
19488 (vec_initv2sisi): ... this.
19489 (vec_extractv4hi): Renamed to ...
19490 (vec_extractv4hihi): ... this.
19491 (vec_initv4hi): Renamed to ...
19492 (vec_initv4hihi): ... this.
19493 (vec_extractv8qi): Renamed to ...
19494 (vec_extractv8qiqi): ... this.
19495 (vec_initv8qi): Renamed to ...
19496 (vec_initv8qiqi): ... this.
19497 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19498 (vec_init<mode>): Renamed to ...
19499 (vec_init<mode><VEC_base_l>): ... this.
19500 (vec_extract<mode>): Renamed to ...
19501 (vec_extract<mode><VEC_base_l>): ... this.
19502 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19503 (vec_initv2sfsf): ... this.
19504 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19505 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19506 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19507 element mode after mode in gen_vec_init* calls.
19508 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19509 (vec_init<mode><Vel>): ... this.
19510 (vec_extract<mode>): Renamed to ...
19511 (vec_extract<mode><Vel>): ... this.
19512 * config/aarch64/iterators.md (Vel): New mode attribute.
19513 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19514 Add element mode after mode in gen_vec_extract* calls.
19515 * config/s390/vector.md (non_vec_l): New mode attribute.
19516 (vec_extract<mode>): Renamed to ...
19517 (vec_extract<mode><non_vec_l>): ... this.
19518 (vec_init<mode>): Renamed to ...
19519 (vec_init<mode><non_vec_l>): ... this.
19520 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19521 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19523 * config/arm/iterators.md (V_elem_l): New mode attribute.
19524 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19525 (vec_extract<mode><V_elem_l>): ... this.
19526 (vec_extractv2di): Renamed to ...
19527 (vec_extractv2didi): ... this.
19528 (vec_init<mode>): Renamed to ...
19529 (vec_init<mode><V_elem_l>): ... this.
19530 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19531 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19532 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19533 Add element mode after gen_vec_extract* calls.
19534 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19535 (vec_init<mode><unitmode>): ... this.
19536 (vec_extract<mode>): Renamed to ...
19537 (vec_extract<mode><unitmode>): ... this.
19538 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19539 (vec_init<mode><unitmode>): ... this.
19540 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19541 (vec_initv2sfsf): ... this.
19542 (vec_extractv2sf): Renamed to ...
19543 (vec_extractv2sfsf): ... this.
19544 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19545 Add element mode after gen_vec_extract* calls.
19546 * config/mips/mips.md (unitmode): New mode iterator.
19547 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19548 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19549 * config/spu/spu.md (inner_l): New mode attribute.
19550 (vec_init<mode>): Renamed to ...
19551 (vec_init<mode><inner_l>): ... this.
19552 (vec_extract<mode>): Renamed to ...
19553 (vec_extract<mode><inner_l>): ... this.
19554 * config/sparc/sparc.md (veltmode): New mode iterator.
19555 (vec_init<VMALL:mode>): Renamed to ...
19556 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19557 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19558 (vec_initv2sisi): ... this.
19559 (vec_initv2sf): Renamed to ...
19560 (vec_initv2sfsf): ... this.
19561 (vec_extractv2sf): Renamed to ...
19562 (vec_extractv2sfsf): ... this.
19563 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19564 (vec_init<mode>): Renamed to ...
19565 (vec_init<mode><VEC_base_l>): ... this.
19566 (vec_extract<mode>): Renamed to ...
19567 (vec_extract<mode><VEC_base_l>): ... this.
19568 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19569 (vec_initv2sfsf): ... this.
19570 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19571 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19572 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19573 gen_vec_init* calls.
19575 2017-08-01 Richard Biener <rguenther@suse.de>
19577 PR tree-optimization/81297
19578 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19579 TREE_OVERFLOW from INTEGER_CSTs.
19581 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19583 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19585 2017-07-31 Carl Love <cel@us.ibm.com>
19587 * config/rs6000/rs6000-c: Add support for built-in functions
19588 vector signed char vec_xl_be (signed long long, signed char *);
19589 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19590 vector signed int vec_xl_be (signed long long, signed int *);
19591 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19592 vector signed long long vec_xl_be (signed long long, signed long long *);
19593 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19594 vector signed short vec_xl_be (signed long long, signed short *);
19595 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19596 vector double vec_xl_be (signed long long, double *);
19597 vector float vec_xl_be (signed long long, float *);
19598 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19599 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19600 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19602 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19603 (altivec_expand_builtin): Add switch statement to call
19604 altivec_expand_xl_be for each builtin.
19605 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19606 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19607 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19608 __builtin_vsx_le_be_v16qi.
19609 * doc/extend.texi: Update the built-in documentation file for the
19610 new built-in functions.
19612 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19615 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19617 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19619 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19621 * config.gcc: Add z14.
19622 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19623 CPU model numbers for z13s and z14.
19624 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19626 * config/s390/s390-opts.h (enum processor_type): Rename
19627 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19628 * config/s390/s390.c (processor_table): Add field for CPU name to
19629 be passed to Binutils.
19630 (s390_asm_output_machine_for_arch): Use the new field in
19631 processor_table for Binutils.
19632 (s390_expand_builtin): Replace arch12 with z14.
19633 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19634 (s390_get_sched_attrmask): Likewise.
19635 (s390_get_unit_mask): Likewise.
19636 * config/s390/s390.opt: Add z14 to processor_type enum.
19638 2017-07-31 Martin Jambor <mjambor@suse.cz>
19641 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19642 regardless of optimization level.
19644 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19645 Martin Liska <mliska@suse.cz>
19647 * predict.def: Remove old comment and adjust probability.
19648 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19649 PREDICT statements.
19651 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19654 * config/i386/i386.c (ix86_function_naked): New function.
19655 (ix86_can_use_return_insn_p): Return false for naked functions.
19656 (ix86_expand_prologue): Skip prologue for naked functions.
19657 (ix86_expand_epilogue): Skip epilogue for naked functions
19658 and emit trap instruction.
19659 (ix86_warn_func_return): New function.
19660 (ix86_attribute_table): Add "naked" attribute specification.
19661 (TARGET_WARN_FUNC_RETURN): Define.
19662 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19664 2017-07-31 Martin Liska <mliska@suse.cz>
19666 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19667 (dump_gimple_bb_header): Always dump BB info.
19668 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19670 2017-07-31 Martin Liska <mliska@suse.cz>
19673 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19674 also with current_function_decl non-null equality.
19676 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19679 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19680 change type to the element type, instead add eltype variable and
19681 use it where we are interested in the element type.
19683 PR tree-optimization/81603
19684 * ipa-polymorphic-call.c
19685 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19686 offset arithmetic in offset_int, bail out if the resulting bit offset
19687 doesn't fit into shwi.
19689 2017-07-31 Martin Liska <mliska@suse.cz>
19691 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19692 (gimplify_save_expr): Fix comment.
19694 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19697 * config/i386/i386.c (ix86_function_arg): Update arguments for
19699 (ix86_compute_frame_layout): Set the initial stack offset to
19700 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19701 INCOMING_FRAME_SP_OFFSET.
19702 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19703 stack before exception handler returns.
19704 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19705 the 'ERROR_CODE' for exception handler.
19707 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19709 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19710 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19711 (ASM_OUTPUT_REG_POP): Ditto.
19712 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19713 instead of asm_fprintf to output pure string.
19715 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19717 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19718 to imported_module_or_decl hook.
19719 (debug_nothing_tree_tree_tree_bool): Remove.
19720 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19721 * debug.c (do_nothing_debug_hooks): Use
19722 debug_nothing_tree_tree_tree_bool_bool instead of
19723 debug_nothing_tree_tree_tree_bool.
19724 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19725 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19726 * sdbout.c (sdb_debug_hooks): Likewise.
19727 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19728 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19730 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19731 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19732 attribute, don't add anything.
19734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19736 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19737 (fold_build2_stat_loc): Likewise.
19738 (fold_build3_stat_loc): Likewise.
19739 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19740 (fold_build1_loc): Remove macro.
19741 (fold_build2_loc): Likewise.
19742 (fold_build3_loc): Likewise.
19744 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19746 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19747 (gimple_build_debug_bind_source_stat): Likewise.
19748 * gimple.h (gimple_build_debug_bind): Remove macro.
19749 (gimple_build_debug_bind_source): Likewise.
19751 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19753 * bitmap.c (bitmap_alloc): Adjust.
19754 (bitmap_gc_alloc): Likewise.
19755 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19757 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19759 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19760 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19761 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19762 (bitmap_gc_alloc_stat): Likewise.
19763 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19767 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19768 * rtl.h (shallow_copy_rtx): Remove macro.
19770 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19772 * emit-rtl.c (gen_raw_REG): Adjust.
19773 * gengenrtl.c (gendef): Likewise.
19774 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19775 * rtl.h (rtx_alloc): Remove macro.
19777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19779 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19780 (build_tree_list_stat): Likewise.
19781 * tree.h (build_tree_list): Remove macro.
19782 (build_tree_list_vec): Likewise.
19784 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19786 * tree.c (make_vector_stat): Remove _stat from name.
19787 (build_vector_stat): Likewise.
19788 * tree.h (make_vector_stat): Remove macro.
19789 (build_vector_stat): Likewise.
19791 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19793 * tree.h (build_var_debug_value): Remove prototype.
19795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19797 * tree.c (tree_cons_stat): Remove _stat from name.
19798 * tree.h (tree_cons): Remove macro.
19800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19802 * tree.c (build_vl_exp_stat): Remove _stat from name.
19803 * tree.h (build_vl_exp): Remove macro.
19805 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19807 * tree.c (build_decl_stat): Remove _stat from name.
19808 * tree.h (build_decl): Remove macro.
19810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19812 * gimple.c (gimple_build_with_ops_stat): Adjust.
19813 (gimple_alloc_stat): Remove _stat from name.
19814 * gimple.h (gimple_alloc): Remove macro.
19816 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19818 * tree.c (make_tree_vec_stat): Remove _stat from name.
19819 (grow_tree_vec_stat): Likewise.
19820 * tree.h (make_tree_vec_stat): Adjust prototype.
19821 (grow_tree_vec_stat): Likewise.
19822 (make_tree_vec): Remove macro.
19823 (grow_tree_vec): Likewise.
19825 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19827 * fold-const.c (fold_build1_stat_loc): Adjust.
19828 (fold_build2_stat_loc): Likewise.
19829 (fold_build3_stat_loc): Likewise.
19830 * tree.c (build0_stat): Remove _stat from name.
19831 (build1_stat): Likewise.
19832 (build2_stat): Likewise.
19833 (build3_stat): Likewise.
19834 (build4_stat): Likewise.
19835 (build5_stat): Likewise.
19836 * tree.h (build1_loc): Remove macro, and rename _stat function
19838 (build2_loc): Likewise.
19839 (build3_loc): Likewise.
19840 (build4_loc): Likewise.
19841 (build5_loc): Likewise.
19843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19845 * tree.c (make_int_cst_stat): Remove _stat from name.
19846 * tree.h (make_int_cst_stat): Adjust prototype.
19847 (make_int_cst): Remove macro.
19849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19851 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19852 * tree.h (make_tree_binfo_stat): Adjust prototype.
19853 (make_tree_binfo): Remove.
19855 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19857 * tree.c (copy_node_stat): Rename to copy_node.
19858 (build_distinct_type_copy): Adjust.
19859 * tree.h (copy_node_stat): Adjust prototype.
19860 (copy_node): Remove macro.
19862 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19864 * tree.c (make_node_stat): rename to make_node.
19865 (build_tree_list_stat): Adjust.
19866 (build0_stat): Likewise.
19867 (build2_stat): Likewise.
19868 (build3_stat): Likewise.
19869 (build4_stat): Likewise.
19870 (build5_stat): Likewise.
19871 (build_decl_stat): Likewise.
19872 * tree.h (make_node_stat): Adjust prototype.
19873 (make_node): remove macro.
19875 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19877 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19878 (PPC_FEATURE2_SCV): Likewise.
19879 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19881 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19883 * config/aarch64/aarch64.c
19884 (aarch64_internal_mov_immediate): Add new special pattern.
19885 * config/aarch64/aarch64.md (*movdi_aarch64):
19886 Add reg/32bit const mov case.
19888 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19889 Richard Sandiford <richard.sandiford@linaro.org>
19891 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19892 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19893 Add integer and movi cases.
19894 (movi-split-hf-df-sf split, fp16): New.
19895 (enabled): Added TARGET_FP_F16INST.
19896 * config/aarch64/iterators.md (GPF_HF): New.
19897 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19899 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19901 * config/aarch64/aarch64.c
19902 (aarch64_simd_container_mode): Add prototype.
19903 (aarch64_expand_mov_immediate): Add HI support.
19904 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19905 (aarch64_can_const_movi_rtx_p): New.
19906 (aarch64_preferred_reload_class):
19907 Remove restrictions of using FP registers for certain SIMD operations.
19908 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19909 (aarch64_valid_floating_const): Add integer move validation.
19910 (aarch64_simd_imm_scalar_p): Remove.
19911 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19912 (aarch64_legitimate_constant_p): Expand list of supported cases.
19913 * config/aarch64/aarch64-protos.h
19914 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19915 (aarch64_reinterpret_float_as_int): New.
19916 (aarch64_simd_imm_scalar_p): Remove.
19917 * config/aarch64/constraints.md (Uvi): New.
19918 (Dd): Split into Ds and new Dd.
19919 * config/aarch64/aarch64.md (*movsi_aarch64):
19921 (*movdi_aarch64): Add SIMD mov case.
19923 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19925 * tree-predcom.c: (struct chain): Handle store-store chain in which
19926 stores for elimination only store loop invariant values.
19927 (execute_pred_commoning_chain): Ditto.
19928 (prepare_initializers_chain_store_elim): Ditto.
19929 (prepare_finalizers): Ditto.
19930 (is_inv_store_elimination_chain): New function.
19931 (initialize_root_vars_store_elim_1): New function.
19933 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19935 * tree-predcom.c: Revise general description of the pass.
19936 (enum chain_type): New enum type for store elimination.
19937 (struct chain): New field supporting store elimination.
19938 (struct component): Ditto.
19939 (dump_chain): Dump store-stores chain.
19940 (release_chain): Release resources.
19941 (split_data_refs_to_components): Compute and create component
19942 contains only stores for elimination.
19943 (get_chain_last_ref_at): New function.
19944 (make_invariant_chain): Initialization.
19945 (make_rooted_chain): Specify chain type in parameter and record it.
19946 (add_looparound_copies): Skip for store-stores chain.
19947 (determine_roots_comp): Compute type of chain and pass it to
19949 (initialize_root_vars_store_elim_2): New function.
19950 (finalize_eliminated_stores): New function.
19951 (remove_stmt): Handle store for elimination.
19952 (execute_pred_commoning_chain): Execute predictive commoning on
19953 store-store chains.
19954 (determine_unroll_factor): Skip unroll for store-stores chain.
19955 (prepare_initializers_chain_store_elim): New function.
19956 (prepare_initializers_chain): Hanlde store-store chain.
19957 (prepare_finalizers_chain, prepare_finalizers): New function.
19958 (tree_predictive_commoning_loop): Return integer value indicating
19959 if loop is unrolled or lcssa form is corrupted.
19960 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19962 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19964 * tree-predcom.c (initialize_root): Delete.
19965 (execute_pred_commoning_chain): Initialize root vars and replace
19966 reference of non-combined chain directly, rather than call above
19969 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19971 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19972 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19974 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19976 * tree-predcom.c (struct chain): New field init_seq.
19977 (release_chain): Release init_seq.
19978 (prepare_initializers_chain): Record intialization stmts in above
19980 (insert_init_seqs): New function.
19981 (tree_predictive_commoning_loop): Call insert_init_seqs.
19983 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19985 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19987 2017-07-28 Richard Biener <rguenther@suse.de>
19989 * match.pd: Remove superfluous :c.
19990 * genmatch.c (simplify::id): Add member.
19991 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19993 (current_id): New global.
19994 (dt_node::parent): Move from ...
19995 (dt_operand::parent): ... here. Add for_id member.
19996 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19997 (decision_tree::find_node): Relax order requirement when
19998 merging DT_TRUE nodes to ones inbetween the current simplify
19999 and the one we try to merge with. Add diagnostic whenever
20000 we need to enforce pattern order by not merging.
20001 (decision_tree::insert): Set current_id.
20002 (decision_tree::print_node): Dump parent node and for_id.
20003 (parser::last_id): Add member.
20004 (parser::push_simplify): Assign unique id.
20005 (parser::parser): Initialize last_id.
20007 2017-07-28 Martin Liska <mliska@suse.cz>
20010 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20011 gimple_build_debug_bind.
20013 2017-07-28 Richard Biener <rguenther@suse.de>
20015 PR tree-optimization/81502
20016 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20018 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20020 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20021 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20022 for BIT_FIELD_REF args.
20023 * fold-const.c (make_bit_field_ref): Likewise.
20024 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20026 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20029 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20030 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20031 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20032 Or it into SANITIZER_UNDEFINED.
20033 * ubsan.c: Include gimple-fold.h and varasm.h.
20034 (ubsan_expand_ptr_ifn): New function.
20035 (instrument_pointer_overflow): New function.
20036 (maybe_instrument_pointer_overflow): New function.
20037 (instrument_object_size): Formatting fix.
20038 (pass_ubsan::execute): Call instrument_pointer_overflow
20039 and maybe_instrument_pointer_overflow.
20040 * internal-fn.c (expand_UBSAN_PTR): New function.
20041 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20042 * sanitizer.def (__ubsan_handle_pointer_overflow,
20043 __ubsan_handle_pointer_overflow_abort): New builtins.
20044 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20045 * internal-fn.def (UBSAN_PTR): New internal function.
20046 * opts.c (sanitizer_opts): Add pointer-overflow.
20047 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20048 * fold-const.c (build_range_check): Compute pointer range check in
20049 integral type if pointer arithmetics would be needed. Formatting
20052 2017-07-28 Martin Liska <mliska@suse.cz>
20055 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20056 parameters that are of a variable-length.
20058 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20060 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20062 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20063 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20064 (CRT_CALL_STATIC_FUNCTION): Likewise.
20065 (ASM_DEFAULT_SPEC): New define.
20066 (ASM_SPEC32): Likewise.
20067 (ASM_SPEC64): Likewise.
20068 (ASM_SPEC_COMMON): Likewise.
20069 (ASM_SPEC): Likewise.
20070 (INVALID_64BIT): Likewise.
20071 (LINK_OS_DEFAULT_SPEC): Likewise.
20072 (LINK_OS_SPEC32): Likewise.
20073 (LINK_OS_SPEC64): Likewise.
20074 (POWERPC_LINUX): Likewise.
20075 (PTRDIFF_TYPE): Likewise.
20076 (RESTORE_FP_PREFIX): Likewise.
20077 (RESTORE_FP_SUFFIX): Likewise.
20078 (SAVE_FP_PREFIX): Likewise.
20079 (SAVE_FP_SUFFIX): Likewise.
20080 (SIZE_TYPE): Likewise.
20081 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20082 (TARGET_64BIT): Likewise.
20083 (TARGET_64BIT): Likewise.
20084 (TARGET_AIX): Likewise.
20085 (WCHAR_TYPE_SIZE): Likewise.
20086 (WCHAR_TYPE): Undefine.
20087 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20088 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20089 (CPP_OS_RTEMS_SPEC): Delete.
20090 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20091 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20093 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20095 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20097 PR tree-optimization/81578
20098 * tree-parloops.c (build_new_reduction): Bail out if
20099 reduction_code isn't one of the standard OpenMP reductions.
20100 Move the details printing after that decision.
20102 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20104 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20105 related to reload_in_progress.
20106 (splat_input_operand): Likewise.
20107 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20109 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20111 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20112 (TARGET_INSTANTIATE_DECLS): Likewise.
20113 (legitimate_indexed_address_p): Delete reload_in_progress code.
20114 (rs6000_debug_legitimate_address_p): Likewise.
20115 (rs6000_eliminate_indexed_memrefs): Likewise.
20116 (rs6000_emit_le_vsx_store): Likewise.
20117 (rs6000_emit_move_si_sf_subreg): Likewise.
20118 (rs6000_emit_move): Likewise.
20119 (register_to_reg_type): Likewise.
20120 (rs6000_pre_atomic_barrier): Likewise.
20121 (rs6000_machopic_legitimize_pic_address): Likewise.
20122 (rs6000_allocate_stack_temp): Likewise.
20123 (rs6000_address_for_fpconvert): Likewise.
20124 (rs6000_address_for_altivec): Likewise.
20125 (rs6000_secondary_memory_needed_rtx): Delete function.
20126 (rs6000_check_sdmode): Likewise.
20127 (rs6000_alloc_sdmode_stack_slot): Likewise.
20128 (rs6000_instantiate_decls): Likewise.
20129 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20130 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20131 Delete reload_in_progress.
20132 (*vec_reload_and_plus_<mptrsize>): Likewise.
20133 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20134 (vsx_div_v2di): Likewise.
20135 (vsx_udiv_v2di): Likewise.
20137 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20139 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20140 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20141 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20142 (rs6000_debug_reg_global): Delete print of LRA status.
20143 (rs6000_option_override_internal): Delete dead LRA related code.
20144 (rs6000_lra_p): Delete function.
20145 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20147 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20149 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20150 * config/riscv/rtems.h: New file.
20152 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20153 Sudakshina Das <sudi.das@arm.com>
20155 * config/aarch64/aarch64.md
20156 (define_split for and<mode>3nr_compare): Move
20157 non aarch64_logical_operand to a register.
20158 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20159 register immediate operand to a register.
20160 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20162 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20164 PR middle-end/81564
20165 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20167 2017-07-27 Richard Biener <rguenther@suse.de>
20169 PR tree-optimization/81573
20170 PR tree-optimization/81494
20171 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20172 multi defuse cycle case.
20174 2017-07-27 Richard Biener <rguenther@suse.de>
20176 PR tree-optimization/81571
20177 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20180 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20182 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20183 earlier and only if MASK_FPU is set. Adjust formatting.
20185 2017-07-27 Martin Liska <mliska@suse.cz>
20187 * opt-functions.awk: Add validation of value of Init.
20188 * optc-gen.awk: Pass new argument.
20190 2017-07-27 Martin Liska <mliska@suse.cz>
20192 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20193 Fix wrong condition.
20195 2017-07-27 Martin Liska <mliska@suse.cz>
20197 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20198 BBs and edges seen by autoFDO.
20200 2017-07-27 Richard Biener <rguenther@suse.de>
20202 PR tree-optimization/81502
20203 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20204 with incompatible but same sized type.
20205 (execute_update_addresses_taken): Likewise.
20207 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20209 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20210 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20212 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20215 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20216 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20217 Change s_operand to memory_operand.
20219 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20221 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20222 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20223 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20224 Emit instructions rather than returning an expression. Handle TFmode
20225 and KFmode by casting to TImode.
20226 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20227 (rs6000_emit_le_vsx_store): Likewise.
20228 * config/rs6000/vsx.md (VSX_TI): New iterator.
20229 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20230 (*vsx_le_undo_permute_<mode>): Likewise.
20231 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20232 emit the split sequence.
20233 (*vsx_le_perm_store_<mode>): Likewise.
20235 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20237 PR tree-optimization/81555
20238 PR tree-optimization/81556
20239 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20240 if true, force CHANGED for the recursive invocation.
20241 (reassociate_bb): Remember original length of ops array, pass
20242 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20244 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20245 attributes for noipa attribute. For naked attribute use
20246 lookup_attribute first before lookup_attribute_spec.
20247 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20249 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20250 for functions with noipa attribute.
20251 (cgraph_externally_visible_p): Return true for functions with noipa
20253 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20254 for functions with noipa attribute.
20255 * doc/extend.texi: Document noipa function attribute.
20256 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20257 also for functions with noipa attribute.
20258 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20260 2017-07-26 Andrew Pinski <apinski@cavium.com>
20262 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20263 vec_unalign_load_cost and vec_unalign_store_cost.
20265 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20267 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20268 -mvsx-small-integer option.
20269 (ISA_3_0_MASKS_IEEE): Likewise.
20270 (OTHER_VSX_VECTOR_MASKS): Likewise.
20271 (POWERPC_MASKS): Likewise.
20272 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20273 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20274 code, only testing for DImode being allowed in non-VSX floating
20276 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20277 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20279 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20280 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20281 TARGET_P8_VECTOR test.
20282 (rs6000_secondary_reload_simple_move): Likewise.
20283 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20284 since TARGET_P9_VECTOR was already tested.
20285 (rs6000_opt_masks): Remove -mvsx-small-integer.
20286 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20287 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20289 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20290 test for TARGET_VEXTRACTUB was used, and that uses
20292 (p9 extract splitter): Likewise.
20293 (vsx_extract_<mode>_di_p9): Likewise.
20294 (vsx_extract_<mode>_store_p9): Likewise.
20295 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20296 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20297 the elimination of TARGET_VSX_SMALL_INTEGER.
20298 (vsx_extract_<mode>_p8): Likewise.
20299 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20300 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20301 (vsx_set_<mode>_p9): Likewise.
20302 (vsx_set_v4sf_p9): Likewise.
20303 (vsx_set_v4sf_p9_zero): Likewise.
20304 (vsx_insert_extract_v4sf_p9): Likewise.
20305 (vsx_insert_extract_v4sf_p9_2): Likewise.
20306 * config/rs6000/rs6000.md (sign extend splitter): Change
20307 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20308 (floatsi<mode>2_lfiwax_mem): Likewise.
20309 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20310 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20311 since a test for TARGET_P9_VECTOR was used.
20312 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20313 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20314 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20315 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20316 TARGET_P8_VECTOR test.
20317 (fix_trunc<mode>si2_stfiwx): Likewise.
20318 (fix_trunc<mode>si2_internal): Likewise.
20319 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20320 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20322 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20323 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20324 TARGET_P8_VECTOR test.
20325 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20326 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20327 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20329 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20330 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20331 since a test for TARGET_P9_VECTOR was used.
20332 (splitter for loading small constants): Likewise.
20334 2017-07-26 Andrew Pinski <apinski@cavium.com>
20336 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20339 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20342 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20343 (fp_valid_at): Likewise.
20345 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20347 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20348 (qdf24xx_addrcost_table): Likewise.
20349 (cortexa57_tunings): Update to use generic_branch_cost.
20350 (cortexa72_tunings): Likewise.
20351 (cortexa73_tunings): Likewise.
20352 (qdf24xx_tunings): Likewise.
20354 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20356 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20357 (thunderx2t99_branch_cost): Likewise.
20358 (cortexa35_tunings): Update to use generic_branch_cost.
20359 (cortexa53_tunings): Likewise.
20360 (cortexa57_tunings): Likewise.
20361 (cortexa72_tunings): Likewise.
20362 (cortexa73_tunings): Likewise.
20363 (thunderx2t99_tunings): Likewise.
20365 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20367 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20368 (sparc_option_override): Honour MASK_FSMULD.
20369 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20370 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20371 * config/sparc/sparc.opt (mfsmuld): New option.
20372 * doc/invoke.texi (mfsmuld): Document option.
20374 2017-07-26 Marek Polacek <polacek@redhat.com>
20376 PR middle-end/70992
20377 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20379 2017-07-26 Richard Biener <rguenther@suse.de>
20381 * gimple-match-head.c (do_valueize): Return OP if valueize
20383 (get_def): New helper to get at the def stmt of a SSA name
20384 if valueize allows.
20385 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20386 do_valueize to get at the def stmt.
20387 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20389 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20391 PR middle-end/46932
20392 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20394 2017-07-26 Martin Liska <mliska@suse.cz>
20397 * function.c (expand_function_start): Make expansion of
20398 nonlocal_goto_save_area after parm_birth_insn.
20400 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20402 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20403 from all CPU target flags enable members.
20405 2017-07-26 Richard Biener <rguenther@suse.de>
20407 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20408 (decision_tree::gen): Make 'type' const.
20409 (write_predicate): Likewise.
20411 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20413 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20414 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20415 (rs6000_option_override_internal): Likewise.
20416 (rs6000_expand_vector_set): Likewise.
20417 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20418 (TARGET_UPPER_REGS_SF): Likewise.
20419 (TARGET_UPPER_REGS_DI): Likewise.
20420 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20421 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20422 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20423 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20424 (Splitters for DI constants in Altivec registers): Likewise.
20425 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20426 (vsx_set_v4sf_p9): Likewise.
20427 (vsx_set_v4sf_p9_zero): Likewise.
20428 (vsx_insert_extract_v4sf_p9): Likewise.
20429 (vsx_insert_extract_v4sf_p9_2): Likewise.
20431 2017-07-25 Carl Love <cel@us.ibm.com>
20433 * doc/extend.texi: Update the built-in documentation file for the
20434 existing built-in functions
20435 vector signed char vec_cnttz (vector signed char);
20436 vector unsigned char vec_cnttz (vector unsigned char);
20437 vector signed short vec_cnttz (vector signed short);
20438 vector unsigned short vec_cnttz (vector unsigned short);
20439 vector signed int vec_cnttz (vector signed int);
20440 vector unsigned int vec_cnttz (vector unsigned int);
20441 vector signed long long vec_cnttz (vector signed long long);
20442 vector unsigned long long vec_cnttz (vector unsigned long long);
20444 2017-07-25 Andrew Pinski <apinski@cavium.com>
20446 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20447 accesses where the use is for the first operand of a BIT_INSERT.
20449 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20452 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20453 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20455 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20457 * config/i386/gstabs.h: Delete.
20458 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20460 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20462 * config/i386/i386.c (ix86_decompose_address): Do not check for
20463 register RTX when looking at index_reg or base_reg.
20464 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20466 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20468 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20469 to update EH info here.
20471 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20473 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20475 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20477 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20479 2017-07-25 Torsten Duwe <duwe@suse.de>
20481 * common.opt: Introduce -fpatchable-function-entry
20482 command line option, and its variables function_entry_patch_area_size
20483 and function_entry_patch_area_start.
20484 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20485 including a two-value parser.
20486 * target.def (print_patchable_function_entry): New target hook.
20487 * targhooks.h (default_print_patchable_function_entry): New function.
20488 * targhooks.c (default_print_patchable_function_entry): Likewise.
20489 * toplev.c (process_options): Switch off IPA-RA if
20490 patchable function entries are being generated.
20491 * varasm.c (assemble_start_function): Look at the
20492 patchable-function-entry command line switch and current
20493 function attributes and maybe generate NOP instructions by
20494 calling the print_patchable_function_entry hook.
20495 * doc/extend.texi: Document patchable_function_entry attribute.
20496 * doc/invoke.texi: Document -fpatchable_function_entry
20497 command line option.
20498 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20500 * doc/tm.texi: Re-generate.
20502 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20505 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20506 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20508 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20510 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20512 2017-07-25 Richard Biener <rguenther@suse.de>
20514 PR tree-optimization/81455
20515 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20516 not walk in cycles when looking for guards.
20518 2017-07-25 Richard Biener <rguenther@suse.de>
20520 PR tree-optimization/81529
20521 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20522 when optimizing backedge uses.
20524 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20526 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20528 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20529 to dl_section_ref. On AIX, append an expression to subtract
20530 the size of the section length to dl_section_ref.
20532 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20534 * configure.ac: If any of the config.* scripts fail, exit 1.
20535 * configure: Regenerate.
20537 2017-07-25 Richard Biener <rguenther@suse.de>
20539 PR middle-end/81546
20540 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20541 of immediate uses, be more verbose on errors.
20543 2017-07-25 Richard Biener <rguenther@suse.de>
20545 PR tree-optimization/81510
20546 * tree-vect-loop.c (vect_is_simple_reduction): When the
20547 reduction stmt is not inside the loop bail out.
20549 2017-07-25 Richard Biener <rguenther@suse.de>
20551 PR tree-optimization/81303
20552 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20553 profitability check against LOOP_VINFO_NITERSM1.
20555 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20557 * domwalk.c (cmp_bb_postorder): Simplify.
20558 (sort_bbs_postorder): New function. Use it...
20559 (dom_walker::walk): ...here to optimize common cases.
20561 2017-07-25 Martin Liska <mliska@suse.cz>
20564 * ipa-visibility.c (function_and_variable_visibility): Make the
20565 redirection just on target that supports aliasing.
20566 Fix GNU coding style.
20568 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20571 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20573 * config/arm/rtems.h: Likewise.
20574 * config/bfin/rtems.h: Likewise.
20575 * config/i386/rtemself.h: Likewise.
20576 * config/lm32/rtems.h: Likewise.
20577 * config/m32c/rtems.h: Likewise.
20578 * config/m68k/rtemself.h: Likewise.
20579 * config/microblaze/rtems.h: Likewise.
20580 * config/mips/rtems.h: Likewise.
20581 * config/moxie/rtems.h: Likewise.
20582 * config/nios2/rtems.h: Likewise.
20583 * config/powerpcspe/rtems.h: Likewise.
20584 * config/rs6000/rtems.h: Likewise.
20585 * config/rtems.h: Likewise.
20586 * config/sh/rtems.h: Likewise.
20587 * config/sh/rtemself.h: Likewise.
20588 * config/sparc/rtemself.h: Likewise.
20590 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20593 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20594 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20595 * tree-ssa-structalias.c (alias_get_name): Same.
20597 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20600 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20601 instructions if no du chain is found.
20603 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20605 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20607 2017-07-25 Richard Biener <rguenther@suse.de>
20609 PR middle-end/81505
20610 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20613 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20615 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20616 upper-regs options.
20617 (ISA_2_7_MASKS_SERVER): Likewise.
20618 (ISA_3_0_MASKS_IEEE): Likewise.
20619 (OTHER_P8_VECTOR_MASKS): Likewise.
20620 (OTHER_VSX_VECTOR_MASKS): Likewise.
20621 (POWERPC_MASKS): Likewise.
20622 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20623 duplicate list of options.
20624 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20625 explicit -mupper-regs options.
20626 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20627 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20628 alias for -mupper-regs-df.
20629 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20630 (rs6000_init_hard_regno_mode_ok): Likewise.
20631 (rs6000_option_override_internal): Likewise.
20632 (rs6000_opt_masks): Likewise.
20633 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20634 options in terms of whether -mvsx or -mpower8-vector was used.
20635 (TARGET_UPPER_REGS_DI): Likewise.
20636 (TARGET_UPPER_REGS_SF): Likewise.
20637 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20638 -mupper-regs-* options.
20640 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20642 * passes.c (emergency_dump_function): Print some empty lines and a
20643 header before the RTL dump.
20645 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20647 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20649 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20652 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20653 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20655 2017-07-24 Carl Love <cel@us.ibm.com>
20657 * config/rs6000/rs6000-c.c: Add support for built-in functions
20658 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20659 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20660 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20661 vec_extract_fp_from_shortl): Add defines for the two builtins.
20662 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20663 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20665 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20666 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20667 * doc/extend.texi: Update the built-in documentation file for the
20668 new built-in function.
20670 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20673 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20675 * doc/generic.texi: Likewise.
20676 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20677 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20679 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20681 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20682 (aarch64_mls_elt_merge<mode>): Likewise.
20684 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20686 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20687 having __cxa_atexit.
20689 2017-07-23 Michael Collison <michael.collison@arm.com>
20691 * config/arm/arm.c (arm_option_override): Deprecate
20692 use of -mstructure-size-boundary.
20693 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20694 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20696 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20699 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20700 Reduce cost estimate for direct moves.
20702 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20705 * config/i386/i386.c (ix86_option_override_internal): Disable
20706 BMI, BMI2 and TBM instructions for -m16.
20708 2017-07-21 Carl Love <cel@us.ibm.com>
20710 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20711 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20712 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20714 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20715 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20716 VMULOSW): New enum "unspec" values.
20717 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20718 altivec_vmulosw): New patterns.
20719 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20720 VMULOSW): Add definitions.
20722 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20724 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20725 (qdf24xx): Likewise.
20726 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20727 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20728 (AARCH64_FL_V8_1): Renumber.
20729 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20730 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20731 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20732 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20733 rdma to feature modifiers list.
20735 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20737 PR middle-end/56727
20738 * ipa-visibility (function_and_variable_visibility): Convert
20739 recursive PLT call to direct call if appropriate.
20741 2017-07-21 Andrew Pinski <apinski@cavium.com>
20743 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20744 operand 1 to see if the types precision matches.
20745 * fold-const.c (operand_equal_p): Likewise.
20747 2017-07-21 Richard Biener <rguenther@suse.de>
20749 PR tree-optimization/81303
20750 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20751 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20752 (vect_peeling_hash_get_lowest_cost): Adjust.
20753 (vect_enhance_data_refs_alignment): Likewise. Use
20754 vect_get_peeling_costs_all_drs to compute the penalty for no
20755 peeling to match up costs.
20757 2017-07-21 Richard Biener <rguenther@suse.de>
20759 PR tree-optimization/81500
20760 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20761 we didn't identify a reduction path.
20763 2017-07-21 Tom de Vries <tom@codesourcery.com>
20764 Cesar Philippidis <cesar@codesourcery.com>
20766 PR gcov-profile/81442
20767 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20770 2017-07-21 Tom de Vries <tom@codesourcery.com>
20773 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20775 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20776 nvptx_override_options_after_change.
20778 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20780 * dwarf2out.c (output_file_names): Avoid double testing for
20781 dwarf_version >= 5.
20783 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20785 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20787 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20789 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20791 (try_crossjump_to_edge): Do not punt on partitioned functions.
20793 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20795 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20796 Put all BBs reachable only via paths crossing cold region to cold
20798 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20800 2016-07-21 Richard Biener <rguenther@suse.de>
20802 PR tree-optimization/81303
20803 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20804 into account prologue and epilogue iterations when raising
20805 min_profitable_iters to sth at least covering one vector iteration.
20807 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20809 * config/arm/arm.c (arm_test_cpu_arch_dat):
20812 2017-07-20 Nathan Sidwell <nathan@acm.org>
20814 Remove TYPE_METHODS.
20815 * tree.h (TYPE_METHODS): Delete.
20816 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20817 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20818 (dbxout_type_methods): Scan TYPE_FIELDS.
20819 (dbxout_type): Don't check TYPE_METHODS here.
20820 * function.c (use_register_for_decl): Always ignore register for
20821 class types when not optimizing.
20822 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20823 * tree.c (free_lang_data_in_type): Stitch out member functions and
20824 templates from TYPE_FIELDS.
20825 (build_distinct_type_copy, verify_type_variant,
20826 verify_type): Member fns are on TYPE_FIELDS.
20827 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20828 * tree-pretty-print.c (dump_generic_node): Likewise.
20830 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20833 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20834 V2TImode and V4TImode.
20835 (ix86_expand_vector_extract): Likewise.
20836 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20837 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20838 (ssescalarmode): Handle V4TImode and V2TImode.
20839 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20840 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20841 (VEXTRACTI128_MODE): New mode iterator.
20842 (splitter for *vec_extractv?ti first element): New.
20843 (VEC_INIT_MODE): New mode iterator.
20844 (vec_init<mode>): Consolidate 3 expanders into one using
20845 VEC_INIT_MODE mode iterator.
20847 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20849 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20850 non_spilled_static_chain_regno_p.
20852 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20854 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20856 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20858 * bb-reorder.c (connect_traces): Allow copying of blocks within
20861 2017-07-20 Richard Biener <rguenther@suse.de>
20863 * gimple.h (gimple_phi_result): Add gphi * overload.
20864 (gimple_phi_result_ptr): Likewise.
20865 (gimple_phi_arg): Likewise. Adjust index assert to only
20866 allow actual argument accesses rather than all slots available
20868 (gimple_phi_arg_def): Add gphi * overload.
20869 * tree-phinodes.c (make_phi_node): Initialize only actual
20871 (resize_phi_node): Clear memory not covered by old node,
20872 do not initialize excess argument slots.
20873 (reserve_phi_args_for_new_edge): Initialize new argument slot
20876 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20878 PR tree-optimization/81388
20880 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20882 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20883 by removing computation of may_be_zero.
20885 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20886 Tom de Vries <tom@codesourcery.com>
20888 PR middle-end/81030
20889 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20890 when gimple level profile disagrees with what RTL expander did.
20892 2017-07-20 Richard Biener <rguenther@suse.de>
20894 PR tree-optimization/61171
20895 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20896 (vect_analyze_stmt): Add slp instance parameter.
20897 (vectorizable_reduction): Likewise.
20898 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20899 (vect_is_simple_reduction): Deal with chains not detected
20900 as SLP reduction chain, specifically not properly associated
20901 chains containing a mix of plus/minus.
20902 (get_reduction_op): Remove.
20903 (get_initial_defs_for_reduction): Simplify, pass in whether
20904 this is a reduction chain, pass in the SLP node for the PHIs.
20905 (vect_create_epilog_for_reduction): Get the SLP instance as
20907 (vectorizable_reduction): Get the SLP instance as arg.
20908 During analysis remember the SLP node with the PHIs in the
20909 instance. Simplify getting at the vectorized reduction PHIs.
20910 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20911 through SLP instance.
20912 (vect_slp_analyze_operations): Likewise.
20913 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20914 (vect_transform_stmt): Likewise.
20916 2017-07-20 Tom de Vries <tom@codesourcery.com>
20918 PR tree-optimization/81489
20919 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20920 read of phi arg location to before loop that modifies phi.
20922 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20924 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20927 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20929 PR middle-end/81331
20930 * except.c (execute): Fix ordering issue.
20932 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20934 PR rtl-optimization/81423
20935 * combine.c (make_compound_operation_int): Don't try to optimize
20936 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20938 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20940 PR rtl-optimization/81423
20941 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20942 with a constant that is -1 in the truncated to mode.
20944 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20946 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20947 (determine_unlikely_bbs): ... here.
20948 * predict.h (propagate_unlikely_bbs_forward): Declare.
20949 * cfgexpand.c (pass_expand::execute): Use it.
20950 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20952 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20953 propagate_unlikely_bbs_forward.
20955 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20957 PR middle-end/81331
20958 * except.c (maybe_add_nop_after_section_switch): New function.
20961 2017-07-19 Tom de Vries <tom@codesourcery.com>
20963 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20965 2017-07-19 Tom de Vries <tom@codesourcery.com>
20967 * gimple.h (gimple_phi_arg): Make assert more strict.
20969 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20971 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20972 * config/rs6000/mmintrin.h: New file.
20973 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20975 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20977 PR tree-optimization/81346
20978 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20980 2017-07-19 Tom de Vries <tom@codesourcery.com>
20982 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20984 2017-07-19 Tom de Vries <tom@codesourcery.com>
20986 * config/nvptx/nvptx-modes.def: Add V2DImode.
20987 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20988 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20989 (nvptx_output_mov_insn): Handle lack of mov.b128.
20990 (nvptx_print_operand): Handle 'H' and 'L' codes.
20991 (nvptx_vector_mode_supported): Allow V2DImode.
20992 (nvptx_preferred_simd_mode): New function.
20993 (nvptx_data_alignment): New function.
20994 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20995 nvptx_preferred_simd_mode.
20996 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20998 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21000 2017-07-19 Tom de Vries <tom@codesourcery.com>
21002 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21003 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21004 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21005 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21006 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21007 (mov<VECIM>_insn): New define_insn.
21008 (define_expand "mov<VECIM>): New define_expand.
21010 2017-07-19 Tom de Vries <tom@codesourcery.com>
21012 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21014 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21016 PR tree-optimization/81346
21017 * fold-const.h (fold_div_compare, range_check_type): Declare.
21018 * fold-const.c (range_check_type): New function.
21019 (build_range_check): Use range_check_type.
21020 (fold_div_compare): No longer static, rewritten into
21021 a match.pd helper function.
21022 (fold_comparison): Don't call fold_div_compare here.
21023 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21024 as helper function.
21026 2017-07-19 Nathan Sidwell <nathan@acm.org>
21028 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21029 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21030 * tree.c (find_decls_types_r, verify_type): Use
21031 TYPE_{MIN,MAX}_VALUE_RAW.
21032 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21033 (hash_tree): Likewise.
21034 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21036 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21039 2017-07-18 Tom de Vries <tom@codesourcery.com>
21041 PR middle-end/81464
21042 * omp-expand.c (expand_omp_for_static_chunk): Handle
21043 equal-argument loop exit phi.
21045 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21048 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21049 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21050 operand 2 predicate.
21051 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21052 operand 2 predicate.
21053 (ror,rol -> rorx splitters): Use const_int_operand as
21054 operand 2 predicate.
21056 2017-06-18 Richard Biener <rguenther@suse.de>
21058 PR tree-optimization/81410
21059 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21060 the gap in the ! slp_perm SLP case after each group.
21062 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21064 PR middle-end/81463
21065 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21068 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21070 PR middle-end/81462
21071 * predict.c (set_even_probabilities): Cleanup; do not affect
21072 probabilities that are already known.
21073 (combine_predictions_for_bb): Call even when count is set.
21075 2017-07-18 Nathan Sidwell <nathan@acm.org>
21077 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21080 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21083 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21084 optimization for loop niter analysis.
21086 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21089 * config/avr/avr.c (avr_optimize_casesi): Don't use
21090 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21092 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21094 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21095 body_cost_vec from _vect_peel_extended_info.
21096 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21097 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21100 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21102 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21104 2017-07-18 Richard Biener <rguenther@suse.de>
21106 PR tree-optimization/80620
21107 PR tree-optimization/81403
21108 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21109 info when re-using a VN table entry.
21111 2017-07-18 Richard Biener <rguenther@suse.de>
21113 PR tree-optimization/81418
21114 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21115 vectype_in. Verify that with lane-reducing reduction operations
21116 we have a single def-use cycle.
21118 2017-07-17 Carl Love <cel@us.ibm.com>
21120 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21122 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21123 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21124 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21125 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21126 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21127 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21128 VMULOSW): New enum "unspec" values.
21129 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21130 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21131 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21132 altivec_vmulosw): New patterns.
21133 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21134 VMULOSW): Add definitions.
21136 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21138 * config/alpha/alpha.c: Include predict.h.
21140 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21142 * tree-vrp.c (compare_assert_loc): Fix comparison function
21143 to return predictable results.
21145 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21147 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21149 (subdi3): Likewise.
21150 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21151 * doc/invoke.texi (mexpand-adddi): Update text.
21153 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21155 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21156 that also clobbers the CC register. The old expand code is moved
21158 (*arc_clzsi2): ... here.
21159 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21160 the CC register. The old expand code is moved to ...
21161 (arc_ctzsi2): ... here.
21163 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21165 * config/arc/arc.opt (mindexed-loads): Use initial value
21166 TARGET_INDEXED_LOADS_DEFAULT.
21167 (mauto-modify-reg): Use initial value
21168 TARGET_AUTO_MODIFY_REG_DEFAULT.
21169 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21170 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21171 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21172 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21174 2017-07-17 Martin Liska <mliska@suse.cz>
21177 * opts.c (finish_options): Do not allow -fgnu-tm
21178 w/ -fsanitize={kernel-,}address. Say sorry.
21180 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21183 * tree-loop-distribution.c (classify_partition): Only assert on
21184 numer of iterations.
21185 (merge_dep_scc_partitions): Delete prameter. Update function call.
21186 (distribute_loop): Remove code handling loop with unknown niters.
21187 (pass_loop_distribution::execute): Skip loop with unknown niters.
21189 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21192 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21193 function sort_partitions_by_post_order.
21195 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21197 PR tree-optimization/81374
21198 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21199 the max index of basic blocks, rather than number of basic blocks.
21201 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21203 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21205 (arc_legitimate_pic_operand_p): Likewise.
21206 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21208 (arc_needs_pcl_p): Likewise.
21209 (arc_legitimate_pc_offset_p): Likewise.
21210 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21211 function is also used in constrains.md.
21212 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21213 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21214 PLUS. Only return true/false in known cases, otherwise assert.
21215 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21216 is already called in arc_legitimate_constant_p.
21217 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21219 (LEGITIMATE_PIC_OPERAND_P): Use
21220 arc_raw_symbolic_reference_mentioned_p function.
21221 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21226 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21227 Andrew Burgess <andrew.burgess@embecosm.com>
21229 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21230 (arc_return_address_register): New function.
21231 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21232 (arc_handle_fndecl_attribute): Add naked attribute.
21233 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21234 (TARGET_WARN_FUNC_RETURN): Likewise.
21235 (arc_allocate_stack_slots_for_args): New function.
21236 (arc_warn_func_return): Likewise.
21237 (machine_function): Change type fn_type.
21238 (arc_compute_function_type): Consider new naked function type,
21239 change function return type.
21240 (arc_must_save_register): Adapt to handle new
21241 arc_compute_function_type's return type.
21242 (arc_expand_prologue): Likewise.
21243 (arc_expand_epilogue): Likewise.
21244 (arc_return_address_regs): Delete.
21245 (arc_return_address_register): New function.
21246 (arc_epilogue_uses): Use above function.
21247 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21248 (arc_function_type): Change encoding, add naked type.
21249 (ARC_INTERRUPT_P): Change to handle the new encoding.
21250 (ARC_FAST_INTERRUPT_P): Likewise.
21251 (ARC_NORMAL_P): Define.
21252 (ARC_NAKED_P): Likewise.
21253 (arc_compute_function_type): Delete prototype.
21254 * config/arc/arc.md (in_ret_delay_slot): Use
21255 arc_return_address_register function.
21256 (simple_return): Likewise.
21257 (p_return_i): Likewise.
21259 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21261 PR tree-optimization/81428
21262 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21263 can't be built for those types.
21265 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21267 Remove stuff dead since r239246.
21269 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21270 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21271 (avr_inform_devices): Remove dead stuff.
21273 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21275 * config/arm/arm_neon.h: Fix softp typo.
21277 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21279 PR tree-optimization/81365
21280 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21281 aggregate moves onto bb predecessor edges, make sure there are no
21282 loads that could alias the lhs in between the start of bb and the
21285 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21288 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21289 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21290 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21292 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21294 PR tree-optimization/81396
21295 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21296 (init_symbolic_number): Initialize it to 1.
21297 (perform_symbolic_merge): Add n_ops from both operands into the new
21299 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21300 without base_addr as useless if they need more than one operation.
21301 (bswap_replace): Handle !bswap case for NULL base_addr.
21303 2017-07-17 Tom de Vries <tom@codesourcery.com>
21306 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21309 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21311 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21312 conditional builtin define __FIX_LEON3FT_B2BST.
21314 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21316 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21317 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21320 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21322 PR rtl-optimization/81424
21323 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21324 to remove potential trapping from operands if -fnon-call-exceptions.
21326 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21328 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21329 profile_proability for scalling.
21330 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21332 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21334 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21336 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21338 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21339 fixpoint arithmetics.
21341 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21343 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21344 fixpoint arithmetics.
21346 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21348 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21349 fixpoint arithmetics.
21351 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21353 * profile-count.h (profile_probability::from_reg_br_prob_note,
21354 profile_probability::to_reg_br_prob_note): New functions.
21355 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21356 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21357 * predict.c (probability_reliable_p): Update.
21358 (edge_probability_reliable_p): Update.
21359 (br_prob_note_reliable_p): Update.
21360 (invert_br_probabilities): Update.
21361 (add_reg_br_prob_note): New function.
21362 (combine_predictions_for_insn): Update.
21363 * asan.c (asan_clear_shadow): Update.
21364 * cfgbuild.c (compute_outgoing_frequencies): Update.
21365 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21366 (update_br_prob_note): Update.
21367 (rtl_verify_edges): Update.
21368 (purge_dead_edges): Update.
21369 (fixup_reorder_chain): Update.
21370 * emit-rtl.c (try_split): Update.
21371 * ifcvt.c (cond_exec_process_insns): Update.
21372 (cond_exec_process_if_block): Update.
21373 (dead_or_predicable): Update.
21374 * internal-fn.c (expand_addsub_overflow): Update.
21375 (expand_neg_overflow): Update.
21376 (expand_mul_overflow): Update.
21377 * loop-doloop.c (doloop_modify): Update.
21378 * loop-unroll.c (compare_and_jump_seq): Update.
21379 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21380 * predict.h: Update.
21381 * reorg.c (mostly_true_jump): Update.
21383 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21384 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21385 * config/arc/arc.c: (emit_unlikely_jump): Update.
21386 * config/arm/arm.c: (emit_unlikely_jump): Update.
21387 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21388 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21389 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21390 (ix86_print_operand): Update.
21391 (ix86_split_fp_branch): Update.
21392 (predict_jump): Update.
21393 * config/ia64/ia64.c (ia64_print_operand): Update.
21394 * config/mmix/mmix.c (mmix_print_operand): Update.
21395 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21396 (rs6000_expand_split_stack_prologue): Update.
21397 * config/rs6000/rs6000.c: Update.
21398 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21399 (s390_expand_vec_movstr): Update.
21400 (s390_expand_cs_tdsi): Update.
21401 (s390_expand_split_stack_prologue): Update.
21402 * config/sh/sh.c (sh_print_operand): Update.
21403 (expand_cbranchsi4): Update.
21404 (expand_cbranchdi4): Update.
21405 * config/sparc/sparc.c (output_v9branch): Update.
21406 * config/spu/spu.c (get_branch_target): Update.
21407 (ea_load_store_inline): Update.
21408 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21409 * config/tilepro/tilepro.c: Update.
21411 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21413 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21414 (gimplify_save_expr): Likewise.
21416 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21418 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21420 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21422 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21424 * timevar.def (TV_IPA_FNSUMMARY): Define.
21426 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21428 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21429 to back store errata sensitive sequence from being generated.
21430 (sqrtdf2_fix): Likewise.
21432 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21434 * tree-ssa-threadupdate.c (compute_path_counts,
21435 update_joiner_offpath_counts): Use profile_probability.
21437 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21440 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21442 * config/arm/arm-c.c (arm_cpu_builtins): Define
21443 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21445 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21447 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21448 array entries to represent __ieee128 versions of the
21449 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21450 scalar_extract_sig, and scalar_insert_exp built-in functions.
21451 (altivec_resolve_overloaded_builtin): Add special case handling
21452 for the __builtin_scalar_insert_exp function, as represented by
21453 the P9V_BUILTIN_VEC_VSIEDP constant.
21454 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21455 exponent support for __ieee128 argument.
21456 (VSESQP): Add scalar extract signature support for __ieee128
21458 (VSTDCNQP): Add scalar test negative support for __ieee128
21460 (VSIEQP): Add scalar insert exponent support for __int128 argument
21461 with __ieee128 result.
21462 (VSIEQPF): Add scalar insert exponent support for __ieee128
21463 argument with __ieee128 result.
21464 (VSTDCQP): Add scalar test data class support for __ieee128
21466 (VSTDCNQP): Add overload support for scalar test negative with
21467 __ieee128 argument.
21468 (VSTDCQP): Add overload support for scalar test data class
21469 __ieee128 argument.
21470 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21471 UNSPEC_VSX_SXSIGDP.
21472 (UNSPEC_VSX_SIEXPQP): New constant.
21473 (xsxexpqp): New insn for VSX scalar extract exponent quad
21475 (xsxsigqp): New insn for VSX scalar extract significand quad
21477 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21478 precision with floating point argument.
21479 (xststdcqp): New expand for VSX scalar test data class quad
21481 (xststdcnegqp): New expand for VSX scalar test negative quad
21483 (xststdcqp): New insn to match expansions for VSX scalar test data
21484 class quad precision and VSX scalar test negative quad precision.
21485 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21486 special case operand checking to enforce that second operand of
21487 VSX scalar test data class with quad precision argument is a 7-bit
21489 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21490 prototypes and descriptions of __ieee128 versions of
21491 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21492 scalar_test_data_class, and scalar_test_neg built-in functions.
21494 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21496 PR tree-optimization/81162
21497 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21498 replace a negate with an add.
21500 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21502 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21504 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21506 * config/arm/arm-c.c (arm_cpu_builtins): Define
21507 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21509 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21511 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21512 (armv8-r): Set ARM Cortex-R52 as default CPU.
21513 * config/arm/arm-tables.opt: Regenerate.
21514 * config/arm/arm-tune.md: Regenerate.
21515 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21517 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21518 extension for -mcpu=cortex-r52.
21520 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21522 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21523 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21524 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21525 (fp-armv8): Define it as FP_ARMv8 only.
21526 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21527 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21529 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21530 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21531 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21532 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21533 than TARGET_FPU_ARMV8.
21534 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21535 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21536 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21538 * config/arm/neon.md (neon_vrint): Likewise.
21539 (neon_vcvt): Likewise.
21540 (neon_<fmaxmin_op><mode>): Likewise.
21541 (<fmaxmin><mode>3): Likewise.
21542 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21543 * config/arm/predicates.md (arm_cond_move_operator): Check against
21544 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21546 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21548 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21549 to top of function.
21551 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21553 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21554 loop in comment with memset.
21556 2017-07-14 Martin Liska <mliska@suse.cz>
21558 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21559 * dwarf2out.c (is_java): Remove the function.
21560 (output_pubname): Remove usage of the function.
21561 (lower_bound_default): Remove usage of DW_LANG_Java.
21562 (gen_compile_unit_die): Likewise.
21563 * gcc.c: Remove compiler defaults for .java and .zip files.
21564 * gimple-expr.c (remove_suffix): Change as there's no longer
21565 extension than 4-letter one.
21566 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21567 (gimplify_save_expr): Likewise.
21568 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21569 as it's possible even for other languages than Java.
21570 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21571 * lto-opts.c (lto_write_options): Remove reference to Java.
21572 * opts.c (strip_off_ending): Update file extension handling.
21573 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21574 * tree-eh.c (lower_resx): Likewise.
21575 * tree.c (free_lang_data_in_type): Remove dead code.
21576 (find_decls_types_r): Likewise.
21577 (build_common_builtin_nodes): Remove Java from a comment.
21578 (verify_type): Remove dead code.
21579 * varasm.c (assemble_external): Remove Java from a comment.
21581 2017-07-14 Martin Liska <mliska@suse.cz>
21583 * opts.c (finish_options): Add quotes.
21584 (common_handle_option): Likewise.
21586 2017-07-14 Martin Liska <mliska@suse.cz>
21588 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21589 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21590 Remove N_SO_PASCAL.
21591 * dwarf2out.c (lower_bound_default): Do not handle
21593 (gen_compile_unit_die): Likewise.
21594 * gcc.c: Remove default extension binding for GNU Pascal.
21595 * stmt.c: Remove Pascal language from a comment.
21596 * xcoffout.c: Likewise.
21598 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21601 * diagnostic-show-locus.c (fixit_cmp): New function.
21602 (layout::layout): Sort m_fixit_hints.
21603 (column_range::column_range): Assert that the values are valid.
21604 (struct char_span): New struct.
21605 (correction::overwrite): New method.
21606 (struct source_line): New struct.
21607 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21608 calls in terms of classes source_line and char_span, and
21609 correction::overwrite.
21610 (selftest::test_overlapped_fixit_printing_2): New function.
21611 (selftest::diagnostic_show_locus_c_tests): Call it.
21613 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21615 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21616 early if there is no lhs.
21618 2017-07-13 Martin Liska <mliska@suse.cz>
21620 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21621 (gen_reference_type_die): Likewise.
21622 * stor-layout.c: Remove Pascal-related comment.
21624 2017-07-13 Martin Liska <mliska@suse.cz>
21626 * opts.c (finish_options): Add quotes to error messages.
21627 (parse_sanitizer_options): Likewise.
21629 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21631 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21633 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21635 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21637 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21639 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21641 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21643 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21646 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21647 provides the hardware capability bits, define the macro
21648 __BUILTIN_CPU_SUPPORTS__.
21649 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21650 if GLIBC does not provide the hardware capability bits. Add a
21651 gcc_unreachable call if the built-in cpu function is neither
21652 __builtin_cpu_is nor __builtin_cpu_supports.
21653 (rs6000_get_function_versions_dispatcher): Change the warning
21654 that an old GLIBC is used which does not export the capability
21655 bits to be an error.
21656 * doc/extend.texi (target_clones attribute): Document the
21657 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21658 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21659 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21660 the macros defined by GCC if the newer GLIBC is available.
21662 2017-07-12 Jeff Law <law@redhat.com>
21664 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21665 remaining includes slightly.
21666 * config/riscv/riscv-builtins.c: Include profile-count.h.
21668 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21671 * config/avr/avr.c (avr_set_current_function): In diagnostic
21672 messages: Quote keywords and (parts of) identifiers.
21673 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21676 2017-07-12 Carl Love <cel@us.ibm.com>
21678 * config/rs6000/rs6000-c.c: Add support for built-in functions
21679 vector bool char vec_revb (vector bool char);
21680 vector bool short vec_revb (vector short char);
21681 vector bool int vec_revb (vector bool int);
21682 vector bool long long vec_revb (vector bool long long);
21683 * doc/extend.texi: Update the built-in documentation file for the
21684 new built-in functions.
21686 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21688 * config/s390/s390.md: Remove movcc splitter.
21690 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21692 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21693 load/store on condition.
21695 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21698 * config/avr/avr.c (avr_encode_section_info)
21699 [progmem && !TREE_READONLY]: Error if progmem object needs
21702 2017-07-11 Michael Collison <michael.collison@arm.com>
21704 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21707 2017-07-11 Carl Love <cel@us.ibm.com>
21709 * config/rs6000/rs6000-c.c: Add support for builtins
21710 vector unsigned int vec_parity_lsbb (vector signed int);
21711 vector unsigned int vec_parity_lsbb (vector unsigned int);
21712 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21713 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21714 vector unsigned long long vec_parity_lsbb (vector signed long long);
21715 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21716 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21717 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21718 * doc/extend.texi: Update the built-in documentation file for the
21719 new built-in functions.
21721 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21723 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21724 (layout::m_primary_loc): New field.
21725 (layout::layout): Initialize new field. Move location filtering
21726 logic from here to...
21727 (layout::maybe_add_location_range): ...this new method. Add
21728 support for filtering to just the lines already specified by other
21730 (layout::will_show_line_p): New method.
21731 (gcc_rich_location::add_location_if_nearby): New method.
21732 (selftest::test_add_location_if_nearby): New test function.
21733 (selftest::diagnostic_show_locus_c_tests): Call it.
21734 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21737 2017-07-11 Tom de Vries <tom@codesourcery.com>
21739 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21740 (bb_first_real_insn): New function.
21741 (nvptx_single): Add extra initialization of broadcasted condition
21744 2017-07-11 Nathan Sidwell <nathan@acm.org>
21746 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21748 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21750 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21751 Binutils doc as TEXI will mess them up.
21752 * doc/invoke.texi (AVR Options): Same here.
21754 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21756 * config/sparc/sparc.opt (mfix-ut700): New option.
21757 (mfix-gr712rc): Likewise.
21758 (sparc_fix_b2bst): New variable.
21759 * doc/invoke.texi (SPARC options): Document them.
21760 (ARM options): Fix warnings.
21761 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21762 instructions to prevent sequences that can trigger the store-store
21763 errata for certain LEON3FT processors.
21764 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21765 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21766 * config/sparc/sparc.md (fix_b2bst): New attribute.
21767 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21769 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21772 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21774 (*rsqrtsf2_sse): Ditto.
21776 (div<mode>3): Macroize insn from divdf3 and divsf3
21777 using MODEF mode iterator.
21779 2017-07-10 Martin Sebor <msebor@redhat.com>
21781 PR tree-optimization/80397
21782 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21783 instead of testing for equality to INTEGER_TYPE.
21785 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21787 * config.gcc: Remove uclibc from arc target spec.
21789 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21791 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21793 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21796 * lto-wrapper.c (remove_option): New function.
21797 (merge_and_complain): Merge PIC/PIE options more realistically.
21799 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21801 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21805 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21806 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21807 * config.in: Regenerate.
21808 * configure: Regenerate.
21809 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21810 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21811 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21812 (TARGET_GASISR_PROLOGUES): ...target mask.
21813 * common/config/avr/avr-common.c
21814 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21815 Set -mgas-isr-prologues.
21816 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21817 INSERT_PASS_BEFORE for it.
21818 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21819 * config/avr/avr.c (avr_option_override)
21820 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21821 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21822 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21823 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21824 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21826 (make_avr_pass_pre_proep): New function.
21827 (emit_push_sfr) <treg>: Add argument to function and use it
21828 instead of TMP_REG.
21829 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21830 and set machine->gasisr.yes.
21831 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21832 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21833 __gcc_isr.n_pushed to .L__stack_usage.
21834 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21835 (avr_asm_final_postscan_insn): ...this new static function.
21836 * config/avr/avr.h (machine_function)
21837 <is_no_gccisr, use_L__stack_usage>: New fields.
21838 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21839 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21840 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21841 (gasisr, *gasisr): New expander and insn.
21842 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21843 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21844 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21846 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21848 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21851 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21853 Move jump-tables out of .text again.
21856 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21857 (ASM_OUTPUT_ADDR_VEC): New function.
21858 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21859 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21860 INSN_ADDRESSes as asm comment.
21861 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21862 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21863 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21864 * config/avr/avr.md (*tablejump): Adjust comment.
21865 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21866 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21868 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21869 (avr_output_addr_vec): New proto.
21870 (avr_log_t) <insn_addresses>: New field.
21872 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21875 * config/i386/i386.c (ix86_function_arg_advance): Set
21876 outgoing_args_on_stack to true if there are outgoing arguments
21878 (ix86_function_arg): Likewise.
21879 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21880 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21881 * config/i386/i386.h (machine_function): Add
21882 outgoing_args_on_stack.
21884 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21886 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21887 supporting pthreds.
21888 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21890 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21892 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21893 (REAL_H): Remove $(MACHMODE_H).
21894 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21896 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21897 $(MACHMODE_H) and double-int.h.
21898 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21900 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21903 2017-07-07 Andrew Pinski <apinski@cavium.com>
21905 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21906 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21908 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21910 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21911 Add warning if GCC was not configured to link against a GLIBC that
21912 exports the hardware capability bits.
21913 (make_resolver_func): Make resolver function private and not a
21914 COMDAT function. Create the name with clone_function_name instead
21915 of make_unique_name.
21918 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21919 correct operand in doing the split.
21921 2017-07-07 Carl Love <cel@us.ibm.com>
21923 * config/rs6000/rs6000-c: Add support for built-in function
21924 vector unsigned short vec_pack_to_short_fp32 (vector float,
21926 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21927 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21928 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21929 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21930 (convert_4f32_8i16): Add define_expand.
21931 * doc/extend.texi: Update the built-in documentation file for the
21932 new built-in function.
21934 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21936 * config/sparc/m8.md: New file.
21937 * config/sparc/sparc.md: Include m8.md.
21939 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21941 * config/sparc/sparc.opt: New option -mvis4b.
21942 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21943 (sparc_option_override): Handle VIS4B.
21944 (enum sparc_builtins): Define
21945 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21946 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21947 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21948 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21949 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21950 (check_constant_argument): New function.
21951 (sparc_vis_init_builtins): Define builtins
21952 __builtin_vis_dictunpack{8,16,32},
21953 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21954 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21955 __builtin_vis_fpcmpde{8,16,32}shl and
21956 __builtin_vis_fpcmpur{8,16,32}shl.
21957 (sparc_expand_builtin): Check that the constant operands to
21958 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21959 constant and in range.
21960 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21962 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21963 (SPARC_IMM5_P): Likewise.
21964 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21965 (enabled): Handle vis4b.
21966 (UNSPEC_DICTUNPACK): New unspec.
21967 (UNSPEC_FPCMPSHL): Likewise.
21968 (UNSPEC_FPUCMPSHL): Likewise.
21969 (UNSPEC_FPCMPDESHL): Likewise.
21970 (UNSPEC_FPCMPURSHL): Likewise.
21971 (cpu_feature): New CPU feature `vis4b'.
21972 (dictunpack{8,16,32}): New insns.
21973 (FPCSMODE): New mode iterator.
21974 (fpcscond): New code iterator.
21975 (fpcsucond): Likewise.
21976 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21977 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21978 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21979 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21980 * config/sparc/constraints.md: Define constraints `q' for unsigned
21981 2-bit integer constants and `t' for unsigned 5-bit integer
21983 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21985 (imm5_operand_dictunpack16): Likewise.
21986 (imm5_operand_dictunpack32): Likewise.
21987 (imm2_operand): Likewise.
21988 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21989 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21990 ditunpack* and fpcmp*shl builtins.
21992 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21994 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21995 * config.in: Add HAVE_AS_SPARC6 define.
21996 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21998 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22000 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22001 (CPP_CPU_SPEC): Handle m8.
22002 (ASM_CPU_SPEC): Likewise.
22003 * config/sparc/sparc-opts.h (enum processor_type): Add
22005 * config/sparc/sparc.c (m8_costs): New struct.
22006 (sparc_option_override): Handle TARGET_CPU_m8.
22007 (sparc32_initialize_trampoline): Likewise.
22008 (sparc64_initialize_trampoline): Likewise.
22009 (sparc_issue_rate): Likewise.
22010 (sparc_register_move_cost): Likewise.
22011 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22012 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22013 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22014 (CPP_CPU_SPEC): Handle M8.
22015 (ASM_CPU_SPEC): Likewise.
22016 (AS_M8_FLAG): Define.
22017 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22018 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22019 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22021 * configure: Regenerate.
22022 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22025 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22027 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22029 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22030 ("*movdi_insn_sp32"): Do not set v3pipe.
22031 ("*movsi_insn"): Likewise.
22032 ("*movdi_insn_sp64"): Likewise.
22033 ("*movsf_insn"): Likewise.
22034 ("*movdf_insn_sp32"): Likewise.
22035 ("*movdf_insn_sp64"): Likewise.
22036 ("*zero_extendsidi2_insn_sp64"): Likewise.
22037 ("*sign_extendsidi2_insn"): Likewise.
22038 ("*mov<VM32:mode>_insn"): Likewise.
22039 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22040 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22041 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22042 ("<vlop:code><VL:mode>3"): Likewise.
22043 ("*not_<vlop:code><VL:mode>3"): Likewise.
22044 ("*nand<VL:mode>_vis"): Likewise.
22045 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22046 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22047 ("one_cmpl<VL:mode>2"): Likewise.
22048 ("faligndata<VM64:mode>_vis"): Likewise.
22049 ("alignaddrsi_vis"): Likewise.
22050 ("alignaddrdi_vis"): Likweise.
22051 ("alignaddrlsi_vis"): Likewise.
22052 ("alignaddrldi_vis"): Likewise.
22053 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22054 ("bmaskdi_vis"): Likewise.
22055 ("bmasksi_vis"): Likewise.
22056 ("bshuffle<VM64:mode>_vis"): Likewise.
22057 ("cmask8<P:mode>_vis"): Likewise.
22058 ("cmask16<P:mode>_vis"): Likewise.
22059 ("cmask32<P:mode>_vis"): Likewise.
22060 ("pdistn<P:mode>_vis"): Likewise.
22061 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22063 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22065 * config/sparc/sparc.md ("subtype"): New insn attribute.
22066 ("*wrgsr_sp64"): Set insn subtype.
22067 ("*rdgsr_sp64"): Likewise.
22068 ("alignaddrsi_vis"): Likewise.
22069 ("alignaddrdi_vis"): Likewise.
22070 ("alignaddrlsi_vis"): Likewise.
22071 ("alignaddrldi_vis"): Likewise.
22072 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22073 ("fexpand_vis"): Likewise.
22074 ("fpmerge_vis"): Likewise.
22075 ("faligndata<VM64:mode>_vis"): Likewise.
22076 ("bshuffle<VM64:mode>_vis"): Likewise.
22077 ("cmask8<P:mode>_vis"): Likewise.
22078 ("cmask16<P:mode>_vis"): Likewise.
22079 ("cmask32<P:mode>_vis"): Likewise.
22080 ("fchksm16_vis"): Likewise.
22081 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22082 ("fmean16_vis"): Likewise.
22083 ("fp<plusminus_insn>64_vis"): Likewise.
22084 ("<plusminus_insn>v8qi3"): Likewise.
22085 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22086 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22087 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22088 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22089 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22090 ("*movqi_insn"): Likewise.
22091 ("*movhi_insn"): Likewise.
22092 ("*movsi_insn"): Likewise.
22093 ("movsi_pic_gotdata_op"): Likewise.
22094 ("*movdi_insn_sp32"): Likewise.
22095 ("*movdi_insn_sp64"): Likewise.
22096 ("movdi_pic_gotdata_op"): Likewise.
22097 ("*movsf_insn"): Likewise.
22098 ("*movdf_insn_sp32"): Likewise.
22099 ("*movdf_insn_sp64"): Likewise.
22100 ("*zero_extendhisi2_insn"): Likewise.
22101 ("*zero_extendqihi2_insn"): Likewise.
22102 ("*zero_extendqisi2_insn"): Likewise.
22103 ("*zero_extendqidi2_insn"): Likewise.
22104 ("*zero_extendhidi2_insn"): Likewise.
22105 ("*zero_extendsidi2_insn_sp64"): Likewise.
22106 ("ldfsr"): Likewise.
22107 ("prefetch_64"): Likewise.
22108 ("prefetch_32"): Likewise.
22109 ("tie_ld32"): Likewise.
22110 ("tie_ld64"): Likewise.
22111 ("*tldo_ldub_sp32"): Likewise.
22112 ("*tldo_ldub1_sp32"): Likewise.
22113 ("*tldo_ldub2_sp32"): Likewise.
22114 ("*tldo_ldub_sp64"): Likewise.
22115 ("*tldo_ldub1_sp64"): Likewise.
22116 ("*tldo_ldub2_sp64"): Likewise.
22117 ("*tldo_ldub3_sp64"): Likewise.
22118 ("*tldo_lduh_sp32"): Likewise.
22119 ("*tldo_lduh1_sp32"): Likewise.
22120 ("*tldo_lduh_sp64"): Likewise.
22121 ("*tldo_lduh1_sp64"): Likewise.
22122 ("*tldo_lduh2_sp64"): Likewise.
22123 ("*tldo_lduw_sp32"): Likewise.
22124 ("*tldo_lduw_sp64"): Likewise.
22125 ("*tldo_lduw1_sp64"): Likewise.
22126 ("*tldo_ldx_sp64"): Likewise.
22127 ("*mov<VM32:mode>_insn"): Likewise.
22128 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22129 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22131 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22133 * config/sparc/sparc.md ("type"): New insn type viscmp.
22134 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22136 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22137 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22138 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22139 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22141 ("n7_vis_logical_11cycle"): Likewise.
22142 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22143 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22144 * config/sparc/niagara.md ("niag_vis"): Likewise.
22145 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22146 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22148 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22150 * config/sparc/sparc.md: New instruction type `bmask'.
22151 (bmaskdi_vis): Use the `bmask' type.
22152 (bmasksi_vis): Likewise.
22153 * config/sparc/ultra3.md (us3_array): Likewise.
22154 * config/sparc/niagara7.md (n7_array): Likewise.
22155 * config/sparc/niagara4.md (n4_array): Likewise.
22156 * config/sparc/niagara2.md (niag2_vis): Likewise.
22157 (niag3_vis): Likewise.
22158 * config/sparc/niagara.md (niag_vis): Likewise.
22160 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22162 * ipa-comdats.c: Remove optimize check from gate.
22163 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22164 for functions not optimized.
22165 (ipa_fn_summary_read): Skip optimize check.
22166 (ipa_fn_summary_write): Likewise.
22167 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22169 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22171 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22172 (check_callers): Give up on caller which is not optimized.
22173 (inline_small_functions): Likewise.
22174 (ipa_inline): Do not give up when not optimizing.
22175 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22176 away unoptimizes cdtors.
22177 (whole_program_function_and_variable_visibility): Do
22178 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22179 * ipa.c (process_references): Do not check optimize.
22180 (symbol_table::remove_unreachable_nodes): Update optimize check.
22181 (set_writeonly_bit): Update optimize check.
22182 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22183 (pass_ipa_single_use::gate): Remove.
22185 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22187 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22188 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22189 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22190 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22191 permute_load, permute_store, adjust_extract, adjust_splat,
22192 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22193 replace_swap_with_copy, dump_swap_insn_table,
22194 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22195 recombine_lvx_pattern, recombine_stvx_pattern,
22196 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22197 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22198 to file rs6000-p8swap.c.
22199 * config/rs6000/rs6000-p8swap.c: New file.
22200 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22201 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22202 and rs6000*-*-* targets.
22204 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22206 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22208 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22210 * lto-wrapper.c (merge_and_complain): Do not merge
22211 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22212 fsigned_zeros, ftrapping_math, fwrapv.
22213 (append_compiler_options): Do not track these options.
22214 (append_linker_options): Likewie
22216 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22218 * cgraphunit.c (cgraph_node::finalize_function): When
22219 !flag_toplevel_reorde set no_reorder flag.
22220 (varpool_node::finalize_decl): Likewise.
22221 (symbol_table::compile): Drop no toplevel reorder path.
22223 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22225 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22226 edges; zero probability is not better than uninitialized.
22228 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22230 * asan.h (asan_sanitize_allocas_p): Declare.
22231 * asan.c (asan_sanitize_allocas_p): New function.
22232 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22233 (handle_builtin_alloca): Likewise.
22234 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22235 if !asan_sanitize_allocas_p.
22236 * params.def (asan-instrument-allocas): Add new option.
22237 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22238 * opts.c (common_handle_option): Disable allocas sanitization for
22241 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22243 * asan.c: Include gimple-fold.h.
22244 (get_last_alloca_addr): New function.
22245 (handle_builtin_stackrestore): Likewise.
22246 (handle_builtin_alloca): Likewise.
22247 (asan_emit_allocas_unpoison): Likewise.
22248 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22249 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22250 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22251 (instrument_builtin_call): Pass gimple iterator to
22252 get_mem_refs_of_builtin_call.
22253 (last_alloca_addr): New global.
22254 * asan.h (asan_emit_allocas_unpoison): Declare.
22255 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22256 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22257 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22258 if function calls alloca.
22259 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22260 * gimple-fold.h (replace_call_with_value): Declare.
22261 * internal-fn.c: Include asan.h.
22262 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22263 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22265 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22267 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22268 (C_SELFTEST_FLAGS): New.
22269 (CPP_SELFTEST_FLAGS): New.
22270 (SELFTEST_DEPS): New, from deps of s-selftest.
22271 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22272 (CPP_SELFTEST_DEPS): New.
22273 (selftest): Add dependency on s-selftest-c++.
22274 (s-selftest): Rename to...
22275 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22276 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22277 than SELFTEST_FLAGS.
22278 (selftest-gdb): Rename to...
22279 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22281 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22282 (selftest-valgrind): Rename to...
22283 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22285 (selftest-valgrind): Reintroduce as an alias for
22286 selftest-c-valgrind.
22287 (s-selftest-c++): New.
22288 (selftest-c++-gdb): New.
22289 (selftest-c++-valgrind): New.
22291 2017-07-06 Olivier Hainque <hainque@adacore.com>
22293 * gcc.c (process_command): When deciding if undefined variables
22294 should be ignored when processing specs, accept "gcc -v" as well.
22296 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22298 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22299 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22301 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22303 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22304 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22305 * config/arm/arm-tables.opt: Regenerate.
22306 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22308 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22310 2017-07-06 Carl Love <cel@us.ibm.com>
22312 * ChangeLog: Clean up from mid air collision
22314 2017-07-06 Carl Love <cel@us.ibm.com>
22316 * config/rs6000/rs6000-c.c: Add support for built-in functions
22317 vector signed int vec_subc (vector signed int, vector signed int);
22318 vector signed __int128 vec_subc (vector signed __int128,
22319 vector signed __int128);
22320 vector unsigned __int128 vec_subc (vector unsigned __int128,
22321 vector unsigned __int128);
22322 vector signed int vec_sube (vector signed int, vector signed int,
22323 vector signed int);
22324 vector unsigned int vec_sube (vector unsigned int,
22325 vector unsigned int,
22326 vector unsigned int);
22327 vector signed __int128 vec_sube (vector signed __int128,
22328 vector signed __int128,
22329 vector signed__int128);
22330 vector unsigned __int128 vec_sube (vector unsigned __int128,
22331 vector unsigned __int128,
22332 vector unsigned __int128);
22333 vector signed int vec_subec (vector signed int, vector signed int,
22334 vector signed int);
22335 vector unsigned int vec_subec (vector unsigned int,
22336 vector unsigned int,
22337 vector unsigned int);
22338 vector signed __int128 vec_subec (vector signed __int128,
22339 vector signed __int128,
22340 vector signed__int128);
22341 vector unsigned __int128 vec_subec (vector unsigned __int128,
22342 vector unsigned __int128,
22343 vector unsigned __int128);
22344 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22345 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22346 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22347 BU_ALTIVEC_OVERLOAD_X definitions.
22348 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22349 * doc/extend.texi: Update the built-in documentation file for the new
22350 built-in functions.
22352 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22355 * diagnostic-show-locus.c (layout::layout): Use start and finish
22356 spelling location for the start and finish of each range.
22357 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22358 Add unused aspect param.
22359 * input.c (expand_location_1): Add "aspect" param, and use it
22360 to access the correct part of the location.
22361 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22363 (expand_location_to_spelling_point): Likewise.
22364 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22365 param, and pass it to expand_location_1.
22367 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22369 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22370 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22371 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22372 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22373 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22374 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22375 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22376 _mm_maskz_getmant_ss): New intrinsics.
22377 (__builtin_ia32_getexpss128_mask): Changed to ...
22378 __builtin_ia32_getexpss128_round ... this.
22379 (__builtin_ia32_getexpsd128_mask): Changed to ...
22380 __builtin_ia32_getexpsd128_round ... this.
22381 * config/i386/i386-builtin-types.def
22382 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22383 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22384 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22385 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22386 __builtin_ia32_getmantss_mask_round): New builtins.
22387 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22388 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22389 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22390 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22391 * config/i386/sse.md
22392 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22393 avx512f_sgetexp<mode><mask_scalar_name>
22394 <round_saeonly_scalar_name> ... this.
22395 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22396 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22397 vgetexp<ssescalarmodesuffix>
22398 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22399 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22400 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22401 avx512f_vgetmant<mode><mask_scalar_name>
22402 <round_saeonly_scalar_name> ... this.
22403 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22404 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22405 vgetmant<ssescalarmodesuffix>
22406 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22407 %0<mask_scalar_operand4>, %1, %2
22408 <round_saeonly_scalar_mask_op4>, %3} ... this.
22409 * config/i386/subst.md (mask_scalar_operand4,
22410 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22411 round_saeonly_scalar_nimm_predicate): New subst attributes.
22413 2017-07-06 Julia Koval <julia.koval@intel.com>
22415 * config/i386/i386.c (ix86_erase_embedded_rounding):
22416 Remove code for old rounding pattern.
22418 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22420 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22422 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22424 * doc/sourcebuild.texi (Test Directives, Variants of
22425 dg-require-support): Add documentation for dg-require-stack-check.
22427 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22429 * config/i386/subst.md (mask_scalar, round_scalar,
22430 round_saeonly_scalar): New meta-templates.
22431 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22432 round_scalar_mask_operand3, round_scalar_mask_op3,
22433 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22434 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22435 round_saeonly_scalar_constraint,
22436 round_saeonly_scalar_prefix): New subst attribute.
22437 * config/i386/sse.md
22438 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22439 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22440 <round_scalar_name> ... this.
22441 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22442 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22443 <round_scalar_name> ... this.
22444 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22445 <sse>_vm<code><mode>3<mask_scalar_name>
22446 <round_saeonly_scalar_name> ... this.
22447 (v<plusminus_mnemonic><ssescalarmodesuffix>
22448 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22449 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22450 v<plusminus_mnemonic><ssescalarmodesuffix>
22451 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22452 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22453 (v<multdiv_mnemonic><ssescalarmodesuffix>
22454 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22455 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22456 v<multdiv_mnemonic><ssescalarmodesuffix>
22457 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22458 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22459 (v<maxmin_float><ssescalarmodesuffix>
22460 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22461 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22462 v<maxmin_float><ssescalarmodesuffix>
22463 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22464 %0<mask_scalar_operand3>, %1, %<iptr>2
22465 <round_saeonly_scalar_mask_op3>} ... this.
22467 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22469 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22470 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22472 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22473 Alan Hayward <alan.hayward@arm.com>
22474 David Sherwood <david.sherwood@arm.com>
22476 * combine.c (simplify_if_then_else): Remove "enum" before
22478 * compare-elim.c (can_eliminate_compare): Likewise.
22479 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22481 (aarch64_lookup_simd_builtin_type): Likewise.
22482 (aarch64_simd_builtin_type): Likewise.
22483 (aarch64_init_simd_builtin_types): Likewise.
22484 (aarch64_simd_expand_args): Likewise.
22485 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22487 (aarch64_reverse_mask): Likewise.
22488 (aarch64_simd_emit_reg_reg_move): Likewise.
22489 (aarch64_gen_adjusted_ldpstp): Likewise.
22490 (aarch64_ccmp_mode_to_code): Likewise.
22491 (aarch64_operands_ok_for_ldpstp): Likewise.
22492 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22493 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22495 (aarch64_min_divisions_for_recip_mul): Likewise.
22496 (aarch64_reassociation_width): Likewise.
22497 (aarch64_get_condition_code_1): Likewise.
22498 (aarch64_simd_emit_reg_reg_move): Likewise.
22499 (aarch64_simd_attr_length_rglist): Likewise.
22500 (aarch64_reverse_mask): Likewise.
22501 (aarch64_operands_ok_for_ldpstp): Likewise.
22502 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22503 (aarch64_gen_adjusted_ldpstp): Likewise.
22504 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22506 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22507 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22508 (arm_lookup_simd_builtin_type): Likewise.
22509 (arm_simd_builtin_type): Likewise.
22510 (arm_init_simd_builtin_types): Likewise.
22511 (arm_expand_builtin_args): Likewise.
22512 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22513 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22514 (ft32_setup_incoming_varargs): Likewise.
22515 (ft32_function_arg): Likewise.
22516 (ft32_function_arg_advance): Likewise.
22517 (ft32_pass_by_reference): Likewise.
22518 (ft32_arg_partial_bytes): Likewise.
22519 (ft32_valid_pointer_mode): Likewise.
22520 (ft32_addr_space_pointer_mode): Likewise.
22521 (ft32_addr_space_legitimate_address_p): Likewise.
22522 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22524 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22525 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22526 (iamcu_alignment): Likewise.
22527 (canonicalize_vector_int_perm): Likewise.
22528 (ix86_noce_conversion_profitable_p): Likewise.
22529 (ix86_mpx_bound_mode): Likewise.
22530 (ix86_operands_ok_for_move_multiple): Likewise.
22531 * config/microblaze/microblaze-protos.h
22532 (microblaze_expand_conditional_branch_reg): Likewise.
22533 * config/microblaze/microblaze.c
22534 (microblaze_expand_conditional_branch_reg): Likewise.
22535 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22537 (rs6000_reassociation_width): Likewise.
22538 (rs6000_invalid_binary_op): Likewise.
22539 (fusion_p9_p): Likewise.
22540 (emit_fusion_p9_load): Likewise.
22541 (emit_fusion_p9_store): Likewise.
22542 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22544 (riscv_hard_regno_mode_ok_p): Likewise.
22545 (riscv_address_insns): Likewise.
22546 (riscv_split_symbol): Likewise.
22547 (riscv_legitimize_move): Likewise.
22548 (riscv_function_value): Likewise.
22549 (riscv_hard_regno_nregs): Likewise.
22550 (riscv_expand_builtin): Likewise.
22551 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22552 (riscv_build_integer): Likewise.
22553 (riscv_split_integer): Likewise.
22554 (riscv_legitimate_constant_p): Likewise.
22555 (riscv_cannot_force_const_mem): Likewise.
22556 (riscv_regno_mode_ok_for_base_p): Likewise.
22557 (riscv_valid_base_register_p): Likewise.
22558 (riscv_valid_offset_p): Likewise.
22559 (riscv_valid_lo_sum_p): Likewise.
22560 (riscv_classify_address): Likewise.
22561 (riscv_legitimate_address_p): Likewise.
22562 (riscv_address_insns): Likewise.
22563 (riscv_load_store_insns): Likewise.
22564 (riscv_force_binary): Likewise.
22565 (riscv_split_symbol): Likewise.
22566 (riscv_force_address): Likewise.
22567 (riscv_legitimize_address): Likewise.
22568 (riscv_move_integer): Likewise.
22569 (riscv_legitimize_const_move): Likewise.
22570 (riscv_legitimize_move): Likewise.
22571 (riscv_address_cost): Likewise.
22572 (riscv_subword): Likewise.
22573 (riscv_output_move): Likewise.
22574 (riscv_canonicalize_int_order_test): Likewise.
22575 (riscv_emit_int_order_test): Likewise.
22576 (riscv_function_arg_boundary): Likewise.
22577 (riscv_pass_mode_in_fpr_p): Likewise.
22578 (riscv_pass_fpr_single): Likewise.
22579 (riscv_pass_fpr_pair): Likewise.
22580 (riscv_get_arg_info): Likewise.
22581 (riscv_function_arg): Likewise.
22582 (riscv_function_arg_advance): Likewise.
22583 (riscv_arg_partial_bytes): Likewise.
22584 (riscv_function_value): Likewise.
22585 (riscv_pass_by_reference): Likewise.
22586 (riscv_setup_incoming_varargs): Likewise.
22587 (riscv_print_operand): Likewise.
22588 (riscv_elf_select_rtx_section): Likewise.
22589 (riscv_save_restore_reg): Likewise.
22590 (riscv_for_each_saved_reg): Likewise.
22591 (riscv_register_move_cost): Likewise.
22592 (riscv_hard_regno_mode_ok_p): Likewise.
22593 (riscv_hard_regno_nregs): Likewise.
22594 (riscv_class_max_nregs): Likewise.
22595 (riscv_memory_move_cost): Likewise.
22596 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22597 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22598 (rl78_addr_space_address_mode): Likewise.
22599 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22601 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22602 (rs6000_reassociation_width): Likewise.
22603 (rs6000_invalid_binary_op): Likewise.
22604 (fusion_p9_p): Likewise.
22605 (emit_fusion_p9_load): Likewise.
22606 (emit_fusion_p9_store): Likewise.
22607 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22608 (ok_for_simple_move_operands): Likewise.
22609 (ok_for_simple_move_strict_operands): Likewise.
22610 (ok_for_simple_arith_logic_operands): Likewise.
22611 (visium_legitimize_reload_address): Likewise.
22612 (visium_select_cc_mode): Likewise.
22613 (output_cbranch): Likewise.
22614 (visium_split_double_move): Likewise.
22615 (visium_expand_copysign): Likewise.
22616 (visium_expand_int_cstore): Likewise.
22617 (visium_expand_fp_cstore): Likewise.
22618 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22619 (visium_function_arg): Likewise.
22620 (visium_function_arg_advance): Likewise.
22621 (visium_libcall_value): Likewise.
22622 (visium_setup_incoming_varargs): Likewise.
22623 (visium_legitimate_constant_p): Likewise.
22624 (visium_legitimate_address_p): Likewise.
22625 (visium_legitimize_address): Likewise.
22626 (visium_secondary_reload): Likewise.
22627 (visium_register_move_cost): Likewise.
22628 (visium_memory_move_cost): Likewise.
22629 (prepare_move_operands): Likewise.
22630 (ok_for_simple_move_operands): Likewise.
22631 (ok_for_simple_move_strict_operands): Likewise.
22632 (ok_for_simple_arith_logic_operands): Likewise.
22633 (visium_function_value_1): Likewise.
22634 (rtx_ok_for_offset_p): Likewise.
22635 (visium_legitimize_reload_address): Likewise.
22636 (visium_split_double_move): Likewise.
22637 (visium_expand_copysign): Likewise.
22638 (visium_expand_int_cstore): Likewise.
22639 (visium_expand_fp_cstore): Likewise.
22640 (visium_split_cstore): Likewise.
22641 (visium_select_cc_mode): Likewise.
22642 (visium_split_cbranch): Likewise.
22643 (output_cbranch): Likewise.
22644 (visium_print_operand_address): Likewise.
22645 * expmed.c (flip_storage_order): Likewise.
22646 * expmed.h (emit_cstore): Likewise.
22647 (flip_storage_order): Likewise.
22648 * genrecog.c (validate_pattern): Likewise.
22649 * hsa-gen.c (gen_hsa_addr): Likewise.
22650 * internal-fn.c (expand_arith_overflow): Likewise.
22651 * ira-color.c (allocno_copy_cost_saving): Likewise.
22652 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22653 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22654 (process_invariant_for_inheritance): Likewise.
22655 * lra-eliminations.c (move_plus_up): Likewise.
22656 * omp-low.c (lower_oacc_reductions): Likewise.
22657 * simplify-rtx.c (simplify_subreg): Likewise.
22658 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22659 (TARGET_CHKP_BOUND_MODE): Likewise..
22660 * targhooks.c (default_chkp_bound_mode): Likewise.
22661 (default_setup_incoming_vararg_bounds): Likewise.
22662 * targhooks.h (default_chkp_bound_mode): Likewise.
22663 (default_setup_incoming_vararg_bounds): Likewise.
22664 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22665 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22666 (have_whole_vector_shift): Likewise.
22667 * tree-vect-stmts.c (vectorizable_load): Likewise.
22668 * doc/tm.texi: Regenerate.
22670 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22672 Graceful degrade if Binutils PR21472 is not available.
22675 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22676 .rodata in flash test fails.
22677 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22678 * confgure: Regenerate.
22679 * config.in: Regenerate.
22680 * config/avr/avr.c (avr_asm_named_section)
22681 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22682 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22683 (avr_asm_init_sections): Same.
22685 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22687 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22688 (fma<VH:mode>4_intrinsic): Likewise.
22689 (*fmsub<VCVTF:mode>4): Likewise.
22690 (*fmsub<VH:mode>4_intrinsic): Likewise.
22692 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22695 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22696 Don't depend on "optimize > 0".
22697 (out_movhi_r_mr, out_movqi_mr_r): Same.
22698 (out_movhi_mr_r, out_movqi_r_mr): Same.
22699 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22700 io_address_operand on "optimize > 0".
22702 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22704 * tree-loop-distribution.c: Add general explanantion on the pass.
22705 (generate_loops_for_partition): Mark distributed loop.
22706 (pg_add_dependence_edges): New parameter. Handle alias data
22707 dependence specially and record it in the parameter if asked.
22708 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22709 (init_partition_graph_vertices, add_partition_graph_edge): New.
22710 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22711 (free_partition_graph_vdata, build_partition_graph): New.
22712 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22713 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22714 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22715 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22716 (version_for_distribution_p, finalize_partitions): New.
22717 (distribute_loop): Handle alias data dependence specially. Factor
22718 out loop fusion code as functions and call these functions.
22720 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22722 * tree-loop-distribution.c (classify_partition): New parameter and
22723 better handle reduction statement.
22724 (rdg_build_partitions): Revise comment.
22725 (distribute_loop): Compute statements in all partitions and pass it
22726 to classify_partition.
22728 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22730 * tree-loop-distribution.c (enum partition_type): New.
22731 (struct partition): New field type.
22732 (partition_merge_into): Add parameter. Update partition type.
22733 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22734 (build_rdg_partition_for_vertex): Compute partition type.
22735 (rdg_build_partitions): Dump partition type.
22736 (distribute_loop): Update calls to partition_merge_into.
22738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22740 * tree-loop-distribution.c (struct ddr_hasher): New.
22741 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22743 (classify_partition): Call get_data_dependence.
22744 (pg_add_dependence_edges): Ditto.
22745 (distribute_loop): Release data dependence hash table.
22747 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22749 * tree-loop-distribution.c (ref_base_address): Delete.
22750 (similar_memory_accesses): Rename ...
22751 (share_memory_accesses): ... to this. Check if partitions access
22752 the same memory reference.
22753 (distribute_loop): Call share_memory_accesses.
22755 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22757 * tree-loop-distribution.c (struct partition): New field recording
22758 its data reference.
22759 (partition_alloc, partition_free): Init and release data refs.
22760 (partition_merge_into): Merge data refs.
22761 (build_rdg_partition_for_vertex): Collect data refs for partition.
22762 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22764 (distribute_loop): Remve data refs from vertice data of partition
22767 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22769 * tree-loop-distribution.c (params.h): Include header file.
22770 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22771 (datarefs_vec): New global var.
22772 (create_rdg_vertices): Use datarefs_vec directly.
22773 (free_rdg): Don't free data references.
22774 (build_rdg): Update use. Don't free data references.
22775 (distribute_loop): Compute global variable for data references.
22776 Bail out if there are too many data references.
22778 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22780 * tree-loop-distribution.c (loop_nest): New global var.
22781 (build_rdg): Use loop directly, rather than loop nest.
22782 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22784 (distribute_loop): Compute global variable loop nest. Update use.
22786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22788 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22789 (partition_merge_into): New parameter. Dump reason for fusion.
22790 (distribute_loop): Update use of partition_merge_into.
22792 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22794 * tree-loop-distribution.c (bb_top_order_index): New.
22795 (bb_top_order_index_size, bb_top_order_cmp): New.
22796 (stmts_from_loop): Use topological order.
22797 (pass_loop_distribution::execute): Compute and release topological
22798 order for basic blocks.
22800 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22802 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22807 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22808 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22809 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22810 * internal-fn.def (LOOP_DIST_ALIAS): New.
22811 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22812 (fold_loop_internal_call): ... this.
22813 (vect_loop_dist_alias_call): New function.
22814 (set_uid_loop_bbs): Call fold_loop_internal_call.
22815 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22818 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22821 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22822 Require dead FLAGS_REG at the beginning of a peephole.
22824 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22827 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22828 arguments in the call to __builtin_ia32_sbb_u32.
22829 (_subborrow_u64): Swap _X and _Y arguments in the call to
22830 __builtin_ia32_sbb_u64.
22832 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22835 * tree-vrp.c (compare_assert_loc): Turn into a function template
22836 with stable template parameter. Only test if a->e is NULL,
22837 !a->e == !b->e has been verified already. Use e == NULL or
22838 e != NULL instead of e or ! e tests. If stable is true, don't use
22839 iterative_hash_expr, on the other side allow a or b or both NULL
22840 and sort the NULLs last.
22841 (process_assert_insertions): Sort using compare_assert_loc<false>
22842 instead of compare_assert_loc, later sort using
22843 compare_assert_loc<true> before calling process_assert_insertions_for
22844 in a loop. Use break instead of continue once seen NULL pointer.
22846 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22848 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22849 Cortex-R7 and Cortex-R8 processors.
22851 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22853 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22854 uninitialized while src is not.
22856 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22858 * common/config/arm/arm-common.c: Adjust include path for
22860 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22861 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22862 (arm-cpu-data.h): Likewise.
22863 (arm-cpu-cdata.h): Likewise.
22864 * config/arm/arm-cpu.h: Delete.
22865 * config/arm/arm-cpu-cdata.h: Delete.
22866 * config/arm/arm-cpu-data.h: Delete.
22868 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22870 * config/arm/arm-cpus.in (cortex-a55): New.
22871 (cortex-a75): Likewise.
22872 (cortex-a75.cortex-a55): Likewise.
22873 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22875 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22876 * config/arm/arm-cpu-cdata.h: Regenerate.
22877 * config/arm/arm-cpu-data.h: Regenerate.
22878 * config/arm/arm-cpu.h: Regenerate.
22879 * config/arm/arm-tables.opt: Regenerate.
22880 * config/arm/arm-tune.md: Regenerate.
22882 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22884 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22886 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22888 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22891 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22893 PR tree-optimization/81292
22894 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22895 full_string_p, also call adjust_related_strinfos if the adjustment
22896 is simple, otherwise invalidate related strinfos.
22898 2017-07-04 Martin Liska <mliska@suse.cz>
22901 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22902 newly created variable as DECL_IGNORED_P.
22904 2017-07-04 Martin Liska <mliska@suse.cz>
22907 * ipa-inline.c (inline_small_functions):
22908 Use xstrdup_for_dump.
22910 2017-07-04 Tom de Vries <tom@codesourcery.com>
22912 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22914 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22917 * config/darwin.c (darwin_function_switched_text_sections):
22920 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22922 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22924 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22926 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22928 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22930 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22931 min_profitable_iters, and th as inclusive lower bounds.
22932 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22933 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22934 for min_profitable_iters and min_profitable_estimate.
22935 (vect_transform_loop): Treat th as an inclusive lower bound.
22936 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22938 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22941 * config/darwin.c (darwin_function_switched_text_sections):
22942 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22943 in two pieces, and suppress the use of buf.
22945 2017-07-03 Nathan Sidwell <nathan@acm.org>
22947 * hash-table.h (hash_table_mod1): Fix indentation.
22949 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22951 PR middle-end/81290
22952 * predict.c (force_edge_cold): Be more careful about propagation
22954 * profile-count.h (profile_probability::guessed,
22955 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22957 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22959 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22961 * doc/invoke.texi (rcpc architecture extension): Document it.
22963 2017-07-03 Richard Biener <rguenther@suse.de>
22965 PR tree-optimization/60510
22966 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22967 the scalar reduction PHI and use it.
22968 (vectorizable_reduction): Properly guard the single_defuse_cycle
22969 path for non-SLP reduction chains where we cannot use it.
22970 Rework reduc_def/index and vector type deduction. Rework
22971 vector operand gathering during reduction op code-gen.
22972 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22973 chains dissolve the chain and leave it to non-SLP reduction
22976 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22978 * tree-data-ref.h (dr_alignment): Declare.
22979 * tree-data-ref.c (dr_alignment): New function.
22980 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22981 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22983 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22985 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22987 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22988 and base_misalignment fields.
22989 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22990 * tree-data-ref.c: Include builtins.h.
22991 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22992 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22993 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22994 * tree-vect-data-refs.c: Include tree-cfg.h.
22995 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22996 fields instead of calculating an alignment here.
22997 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22998 innermost_loop_behavior fields.
23000 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23002 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23004 (DR_STEP_ALIGNMENT): New macro.
23005 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23006 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23007 (create_data_ref): Print it.
23008 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23009 to tell whether the step preserves vector (mis)alignment.
23010 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23011 Move the check for an integer step and generalise to all INTEGER_CST.
23012 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23013 Print the outer step alignment.
23015 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23017 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23018 with offset_alignment.
23019 (DR_ALIGNED_TO): Delete.
23020 (DR_OFFSET_ALIGNMENT): New macro.
23021 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23022 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23023 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23024 (create_data_ref): Likewise.
23025 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23026 (vect_analyze_data_refs): Likewise.
23027 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23028 creating dummy innermost behavior.
23030 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23032 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23033 with a "innermost_loop_behavior *" and refeence tree.
23034 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23035 (create_data_ref): Update call accordingly.
23036 * tree-predcom.c (find_looparound_phi): Likewise.
23038 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23040 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23041 fields with dr_wrt_vec_loop.
23042 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23043 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23044 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23045 (vect_dr_behavior): New function.
23046 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23048 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23049 track whether the step preserves the misalignment.
23050 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23051 Use vect_dr_behavior.
23052 (vect_setup_realignment): Update call accordingly.
23053 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23054 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23055 call to vect_create_addr_base_for_vector_ref.
23056 (vect_create_cond_for_align_checks): Likewise.
23057 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23058 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23059 (vect_recog_mask_conversion_pattern): Likewise.
23060 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23061 (new_stmt_vec_info): Remove redundant zeroing.
23063 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23065 * common/config/arm/arm-common.c (arm_be8_option): New function.
23066 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23067 (ISA_ARMv6): Add isa_bit_be8.
23068 * config/arm/arm.h (arm_be8_option): Add prototype.
23069 (BE8_SPEC_FUNCTION): New define.
23070 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23071 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23072 (mlittle-endian): Similarly.
23073 (mbe8, mbe32): New options.
23074 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23075 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23077 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23079 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23081 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23083 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23084 (cleanup_tree_cfg_bb): Use it.
23085 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23087 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23089 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23092 * loop-doloop.c (add_test): Update profile.
23094 2017-07-03 Martin Liska <mliska@suse.cz>
23097 * sanopt.c (rewrite_usage_of_param): New function.
23098 (sanitize_rewrite_addressable_params): Likewise.
23099 (pass_sanopt::execute): Call rewrite_usage_of_param.
23101 2017-07-03 Richard Biener <rguenther@suse.de>
23103 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23104 back to using VIEW_CONVERT_EXPR.
23106 2017-07-03 Martin Liska <mliska@suse.cz>
23109 * doc/extend.texi: Document when a resolver function is
23110 generated for target_clones.
23112 2017-07-03 Martin Liska <mliska@suse.cz>
23114 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23115 and shadow memory of auto variables which are subject of
23116 use-after-scope sanitization.
23117 (asan_expand_mark_ifn): Add do set only when is_poison.
23119 2016-07-03 Richard Biener <rguenther@suse.de>
23121 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23123 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23124 (vectorizable_reduction): Perform reduction PHI creation when
23125 visiting a reduction PHI and adjust and simplify code generation
23126 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23127 (vect_transform_loop): Visit reduction PHIs.
23128 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23129 defs into the SLP tree.
23130 (vect_build_slp_tree): Reduction defs terminate the recursion.
23131 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23133 (vect_get_vec_defs_for_stmt_copy): Export.
23134 (vect_get_vec_defs): Likewise.
23135 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23137 (vect_get_vec_defs_for_stmt_copy): Declare.
23138 (vect_get_vec_defs): Likewise.
23140 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23142 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23143 parameter with a "loop" parameter and use it instead of the
23144 loop containing DR_STMT. Don't check simple_iv when doing
23145 BB analysis. Describe the two analysis modes in the comment.
23147 2017-07-03 Tom de Vries <tom@codesourcery.com>
23149 PR tree-optimization/69468
23150 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23151 (find_same_succ_bb): Handle ignore_edge_flags.
23153 2017-07-03 Tom de Vries <tom@codesourcery.com>
23155 PR tree-optimization/81192
23156 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23158 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23160 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23162 2017-07-03 Tom de Vries <tom@codesourcery.com>
23164 PR tree-optimization/81192
23165 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23166 BB_SAME_SUCC (bb) == NULL.
23168 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23170 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23175 * dumpfile.c: Include profile-count.h
23176 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23178 (insert_cond_bb): Update profile.
23179 * tree-cfg.h (insert_cond_bb): Update prototype.
23180 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23181 * tree-dump.c: Do not include tree-cfg.
23183 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23185 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23187 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23189 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23194 * tree-complex.c (expand_complex_div_wide): update profile.
23196 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23197 Alan Hayward <alan.hayward@arm.com>
23198 David Sherwood <david.sherwood@arm.com>
23200 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23201 (CORETYPES_H): New define.
23202 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23203 (insn-modes-inline.h, s-modes-inline-h): New rules.
23204 (generated_files): Add insn-modes-inline.h.
23205 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23206 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23207 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23208 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23209 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23210 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23211 (build/gencodes.o, build/genconditions.o): Likewise.
23212 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23213 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23214 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23215 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23216 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23217 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23218 * coretypes.h: Include everything up to real.h for generators.
23219 Include insn-modes.h first. Include wide-int-print.h after
23220 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23221 * machmode.h: Don't include insn-modes.h here.
23222 * function-tests.c: Remove includes of signop.h, machmode.h,
23223 double-int.h and wide-int.h.
23225 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23227 * optc-save-gen.awk: Likewise.
23228 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23229 * godump.c: Remove include of wide-int-print.h.
23230 * pretty-print.h: Likewise.
23231 * wide-int-print.cc: Likewise.
23232 * wide-int.cc: Likewise.
23233 * hash-map-tests.c: Remove include of signop.h.
23234 * hash-set-tests.c: Likewise.
23235 * rtl-tests.c: Likewise.
23236 * mkconfig.sh: Remove include of machmode.h.
23237 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23239 (emit_insn_modes_inline_h): ...this new function. Emit the code
23240 into an insn-modes-inline.h header file, adding appropriate
23241 include guards and end comments.
23242 (emit_insn_modes_c_header): Remove include of machmode.h.
23243 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23245 (main): Handle -i flag and call emit_insn_modes_inline_h when
23248 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23250 * tree-ssa-strlen.c (strinfo): Rename the length field to
23251 nonzero_chars. Add a full_string_p field.
23252 (compare_nonzero_chars, zero_length_string_p): New functions.
23253 (get_addr_stridx): Add an offset_out parameter.
23254 Use compare_nonzero_chars.
23255 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23256 (new_strinfo): Update after above changes to strinfo.
23257 (set_endptr_and_length): Set full_string_p.
23258 (get_string_length): Update after above changes to strinfo.
23259 (unshare_strinfo): Update call to new_strinfo.
23260 (maybe_invalidate): Likewise.
23261 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23262 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23263 as a uhwi instead of an shwi. Update after above changes to
23264 strinfo and new_strinfo.
23265 (zero_length_string): Assert that chainsi contains full strings.
23266 Use zero_length_string_p. Update call to new_strinfo.
23267 (adjust_related_strinfos): Update after above changes to strinfo.
23268 Copy full_string_p from origsi.
23269 (adjust_last_stmt): Use zero_length_string_p.
23270 (handle_builtin_strlen): Update after above changes to strinfo and
23271 new_strinfo. Install the lhs as the string length if the previous
23272 entry didn't describe a full string.
23273 (handle_builtin_strchr): Update after above changes to strinfo
23275 (handle_builtin_strcpy): Likewise.
23276 (handle_builtin_strcat): Likewise.
23277 (handle_builtin_malloc): Likewise.
23278 (handle_pointer_plus): Likewise.
23279 (handle_builtin_memcpy): Likewise. Track nonzero characters
23280 that aren't necessarily followed by a nul terminator.
23281 (handle_char_store): Likewise.
23283 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23285 PR tree-optimization/80769
23286 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23287 for malloc and calloc. Document the new invariant that all related
23288 strinfos have delayed lengths or none do.
23289 (verify_related_strinfos): Move earlier in file.
23290 (set_endptr_and_length): New function, split out from...
23291 (get_string_length): ...here. Also set the lengths of related
23293 (zero_length_string): Assert that chainsi has known (rather than
23295 (adjust_related_strinfos): Likewise.
23297 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23299 PR tree-optimization/81136
23300 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23301 assert that two references with the same misalignment have the same
23302 compile-time misalignment if those compile-time misalignments
23305 2017-07-01 Andi Kleen <ak@linux.intel.com>
23307 * print-tree.c (print_node): Print all attributes.
23309 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23311 * cfg.c (scale_bbs_frequencies): New function.
23312 * cfg.h (scale_bbs_frequencies): Declare it.
23313 * cfgloopanal.c (single_likely_exit): Cleanup.
23314 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23316 (scale_loop_profile): Likewise.
23317 (loop_version): Likewise.
23318 (create_empty_loop_on_edge): Update.
23319 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23320 scale_loop_frequencies, scale_loop_profile, loopify,
23321 loop_version): Update prototypes.
23322 * modulo-sched.c (sms_schedule): Update.
23323 * predict.c (unlikely_executed_edge_p): Also check probability.
23324 (probably_never_executed_edge_p): Fix typo.
23325 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23326 * tree-parloops.c (gen_parallel_loop): Update.
23327 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23328 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23329 * tree-ssa-loop-split.c (split_loop): Update.
23330 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23331 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23332 (vect_loop_versioning): Update.
23333 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23335 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23337 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23339 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23341 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23342 to keep profile consistent.
23344 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23346 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23347 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23348 * profile-count.h (max_safe_multiplier): Make unsigned.
23349 (profile_count::guessed_zero): New.
23351 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23353 * bb-reorder.c (fix_up_crossing_landing_pad,
23354 fix_crossing_conditional_branches): Use make_single_succ_edge
23355 to keep profile consistent.
23357 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23359 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23362 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23365 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23366 the right scopes, make sure cond_jump isn't preserved between multiple
23367 iterations. Search for fallthru edge whenever there are 3+ edges and
23368 use find_fallthru_edge for it.
23370 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23372 Patch by Alexander Monakov <amonakov@ispras.ru>
23373 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23374 probabilities consistently.
23376 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23378 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23379 emit_cmp_and_jump_insns.
23381 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23384 * tree-inline.c (expand_call_inline): Combine profile statuses.
23386 2017-06-30 Andrew Pinski <apinski@cavium.com>
23388 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23389 fold_stmt returned true.
23391 2017-06-30 Nathan Sidwell <nathan@acm.org>
23393 * ggc.h (empty_string): Delete.
23394 * cfgexpand.c (expand_asm_stmt): Use plain "".
23395 * optabs.c (expand_asm_memory_barrier): Likewise.
23396 * stringpool.c (empty_string): Delete.
23397 (digit_vector, digit_string): Delete.
23398 (ggc_alloc_string): Use plain "", don't optimize single digit
23399 strings. Use ggc_alloc_atomic.
23401 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23403 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23404 comparison set and one other set, use the cost of the non-comparison
23407 2017-06-30 Nathan Sidwell <nathan@acm.org>
23409 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23412 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23414 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23415 loops. Remove now unneeded calls to gimple_switch_set_label() that
23416 just set removed labels to NULL_TREE.
23418 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23420 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23421 (set_range_info): ...here. Only call set_range_info_raw if domain
23423 (set_nonzero_bits): Call set_range_info_raw.
23424 * tree-ssanames.h (set_range_info_raw): New.
23426 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23429 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23430 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23431 of nonimmediate_operand and <store_mask_constraint> instead of m
23432 for the input operand. For V8FI iterator, always split if input
23433 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23434 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23435 <store_mask_predicate> instead of register_operand and
23436 <store_mask_constraint> instead of v for the input operand. Make
23437 sure both operands aren't MEMs for if not <mask_applied>.
23439 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23441 * lto-wrapper.c (copy_file) Close both file descriptors before
23444 2017-06-30 Martin Liska <mliska@suse.cz>
23447 * multiple_target.c (create_dispatcher_calls): Make ifunc
23448 also for function that don't have calls or are not referenced.
23450 2017-06-30 Richard Biener <rguenther@suse.de>
23452 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23453 analyze the first scalar stmt. Move vector type computation
23454 for the BB case here from ...
23455 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23456 live operation processing in the SLP case properly.
23458 2017-06-30 Richard Biener <rguenther@suse.de>
23460 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23462 2017-06-30 Martin Liska <mliska@suse.cz>
23465 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23466 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23468 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23470 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23473 2017-06-30 Richard Biener <rguenther@suse.de>
23475 PR tree-optimization/81249
23476 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23477 condition reduction result to original scalar type.
23479 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23481 * profile-count.h (enum profile_quality): Fix typos and whitespace
23484 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23486 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23487 type for branch probabilities.
23489 2017-06-29 Julian Brown <julian@codesourcery.com>
23490 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23492 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23493 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23494 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23495 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23497 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23499 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23500 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23501 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23502 CC usage from generic code to here.
23503 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23504 CC usage into the target macros.
23506 2017-06-29 Maya Rashish <coypu@sdf.org>
23508 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23511 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23513 * arm/arm-builtins.c: Include profile-count.h
23514 * except.c (sjlj_emit_function_enter): Use
23515 profile_probability::unlikely.
23517 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23519 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23520 and tocrel_offset be pointer args rather than implicitly using
23522 (legitimate_constant_pool_address_p, rs6000_emit_move,
23523 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23524 tocrel_offset and use in toc_relative_expr_p call.
23525 (print_operand, print_operand_address): Use static tocrel_base_oac
23526 and tocrel_offset_oac.
23527 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23530 2017-06-29 Maya Rashish <coypu@sdf.org>
23532 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23534 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23536 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23537 objects, take into account only the alignment of 'op0' and 'mode1' if
23540 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23542 * ccmp.c (ccmp_tree_comparison_p): New function.
23543 (ccmp_candidate_p): Update to use above function.
23544 (get_compare_parts): New function.
23545 (expand_ccmp_next): Update to use new functions.
23546 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23548 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23549 take mode as argument.
23550 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23551 * expr.c (expand_expr_real_1): Pass mode as argument.
23553 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23555 * combine.c (combine_instructions): Print insns to dump_file, together
23558 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23560 * asan.c (asan_emit_stack_protection): Update.
23561 (create_cond_insert_point): Update.
23562 * auto-profile.c (afdo_propagate_circuit): Update.
23563 * basic-block.h (struct edge_def): Turn probability to
23564 profile_probability.
23565 (EDGE_FREQUENCY): Update.
23566 * bb-reorder.c (find_traces_1_round): Update.
23567 (better_edge_p): Update.
23568 (sanitize_hot_paths): Update.
23569 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23570 (make_single_succ_edge): Update.
23571 (check_bb_profile): Update.
23572 (dump_edge_info): Update.
23573 (update_bb_profile_for_threading): Update.
23574 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23575 probabilitycount to 0.
23576 * cfgbuild.c (compute_outgoing_frequencies): Update.
23577 * cfgcleanup.c (try_forward_edges): Update.
23578 (outgoing_edges_match): Update.
23579 (try_crossjump_to_edge): Update.
23580 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23581 (expand_gimple_tailcall): Update.
23582 (construct_init_block): Use make_single_succ_edge.
23583 (construct_exit_block): Use make_single_succ_edge.
23584 * cfghooks.c (verify_flow_info): Update.
23585 (redirect_edge_succ_nodup): Update.
23586 (split_edge): Update.
23587 (account_profile_record): Update.
23588 * cfgloopanal.c (single_likely_exit): Update.
23589 * cfgloopmanip.c (scale_loop_profile): Update.
23590 (set_zero_probability): Remove.
23591 (duplicate_loop_to_header_edge): Update.
23592 * cfgloopmanip.h (loop_version): Update prototype.
23593 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23594 (force_nonfallthru_and_redirect): Update.
23595 (update_br_prob_note): Update.
23596 (rtl_verify_edges): Update.
23597 (purge_dead_edges): Update.
23598 (rtl_lv_add_condition_to_bb): Update.
23599 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23600 * cgraphunit.c (init_lowered_empty_function): Update.
23601 (cgraph_node::expand_thunk): Update.
23602 * cilk-common.c: Include profile-count.h
23603 * dojump.c (inv): Remove.
23604 (jumpifnot): Update.
23605 (jumpifnot_1): Update.
23606 (do_jump_1): Update.
23608 (do_jump_by_parts_greater_rtx): Update.
23609 (do_compare_rtx_and_jump): Update.
23610 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23611 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23612 * dwarf2cfi.c: Include profile-count.h
23613 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23614 (sjlj_emit_dispatch_table): Likewise.
23615 * explow.c: Include profile-count.h
23616 * expmed.c (emit_store_flag_force): Update.
23617 (do_cmp_and_jump): Update.
23618 * expr.c (compare_by_pieces_d::generate): Update.
23619 (compare_by_pieces_d::finish_mode): Update.
23620 (emit_block_move_via_loop): Update.
23621 (store_expr_with_bounds): Update.
23622 (store_constructor): Update.
23623 (expand_expr_real_2): Update.
23624 (expand_expr_real_1): Update.
23625 * expr.h (try_casesi, try_tablejump): Update prototypes.
23626 * gimple-pretty-print.c (dump_probability): Update.
23627 (dump_profile): New.
23628 (dump_gimple_label): Update.
23629 (dump_gimple_bb_header): Update.
23630 * graph.c (draw_cfg_node_succ_edges): Update.
23631 * hsa-gen.c (convert_switch_statements): Update.
23632 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23633 (find_if_case_1): Update.
23634 (find_if_case_2): Update.
23635 * internal-fn.c (expand_arith_overflow_result_store): Update.
23636 (expand_addsub_overflow): Update.
23637 (expand_neg_overflow): Update.
23638 (expand_mul_overflow): Update.
23639 (expand_vector_ubsan_overflow): Update.
23640 * ipa-cp.c (good_cloning_opportunity_p): Update.
23641 * ipa-split.c (split_function): Use make_single_succ_edge.
23642 * ipa-utils.c (ipa_merge_profiles): Update.
23643 * loop-doloop.c (add_test): Update.
23644 (doloop_modify): Update.
23645 * loop-unroll.c (compare_and_jump_seq): Update.
23646 (unroll_loop_runtime_iterations): Update.
23647 * lra-constraints.c (lra_inheritance): Update.
23648 * lto-streamer-in.c (input_cfg): Update.
23649 * lto-streamer-out.c (output_cfg): Update.
23650 * mcf.c (adjust_cfg_counts): Update.
23651 * modulo-sched.c (sms_schedule): Update.
23652 * omp-expand.c (expand_omp_for_init_counts): Update.
23653 (extract_omp_for_update_vars): Update.
23654 (expand_omp_ordered_sink): Update.
23655 (expand_omp_for_ordered_loops): Update.
23656 (expand_omp_for_generic): Update.
23657 (expand_omp_for_static_nochunk): Update.
23658 (expand_omp_for_static_chunk): Update.
23659 (expand_cilk_for): Update.
23660 (expand_omp_simd): Update.
23661 (expand_omp_taskloop_for_outer): Update.
23662 (expand_omp_taskloop_for_inner): Update.
23663 * omp-simd-clone.c (simd_clone_adjust): Update.
23664 * optabs.c (expand_doubleword_shift): Update.
23665 (expand_abs): Update.
23666 (emit_cmp_and_jump_insn_1): Update.
23667 (expand_compare_and_swap_loop): Update.
23668 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23669 * predict.c (predictable_edge_p): Update.
23670 (edge_probability_reliable_p): Update.
23671 (set_even_probabilities): Update.
23672 (combine_predictions_for_insn): Update.
23673 (combine_predictions_for_bb): Update.
23674 (propagate_freq): Update.
23675 (estimate_bb_frequencies): Update.
23676 (force_edge_cold): Update.
23677 * profile-count.c (profile_count::dump): Add missing space into dump.
23678 (profile_count::debug): Add newline.
23679 (profile_count::differs_from_p): Explicitly convert to unsigned.
23680 (profile_count::stream_in): Update.
23681 (profile_probability::dump): New member function.
23682 (profile_probability::debug): New member function.
23683 (profile_probability::differs_from_p): New member function.
23684 (profile_probability::differs_lot_from_p): New member function.
23685 (profile_probability::stream_in): New member function.
23686 (profile_probability::stream_out): New member function.
23687 * profile-count.h (profile_count_quality): Rename to ...
23688 (profile_quality): ... this one.
23689 (profile_probability): New.
23690 (profile_count): Update.
23691 * profile.c (compute_branch_probabilities): Update.
23692 * recog.c (peep2_attempt): Update.
23693 * sched-ebb.c (schedule_ebbs): Update.
23694 * sched-rgn.c (find_single_block_region): Update.
23695 (compute_dom_prob_ps): Update.
23696 (schedule_region): Update.
23697 * sel-sched-ir.c (compute_succs_info): Update.
23698 * stmt.c (struct case_node): Update.
23699 (do_jump_if_equal): Update.
23700 (get_outgoing_edge_probs): Update.
23701 (conditional_probability): Update.
23702 (emit_case_dispatch_table): Update.
23703 (expand_case): Update.
23704 (expand_sjlj_dispatch_table): Update.
23705 (emit_case_nodes): Update.
23706 * targhooks.c: Update.
23707 * tracer.c (better_p): Update.
23708 (find_best_successor): Update.
23709 * trans-mem.c (expand_transaction): Update.
23710 * tree-call-cdce.c: Update.
23711 * tree-cfg.c (gimple_split_edge): Upate.
23712 (move_sese_region_to_fn): Upate.
23713 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23714 * tree-eh.c (lower_resx): Upate.
23715 (cleanup_empty_eh_move_lp): Upate.
23716 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23717 * tree-inline.c (copy_edges_for_bb): Update.
23718 (copy_cfg_body): Update.
23719 * tree-parloops.c (gen_parallel_loop): Update.
23720 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23721 (gimple_gen_time_profiler): Update.
23722 * tree-ssa-dce.c (remove_dead_stmt): Update.
23723 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23724 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23725 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23726 (unloop_loops): Update.
23727 (try_peel_loop): Update.
23728 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23729 * tree-ssa-loop-split.c (connect_loops): Update.
23730 (split_loop): Update.
23731 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23732 (hoist_guard): Update.
23733 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23734 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23735 (value_replacement): Update.
23736 * tree-ssa-reassoc.c (branch_fixup): Update.
23737 * tree-ssa-tail-merge.c (replace_block_by): Update.
23738 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23739 (create_edge_and_update_destination_phis): Update.
23740 (compute_path_counts): Update.
23741 (recompute_probabilities): Update.
23742 (update_joiner_offpath_counts): Update.
23743 (freqs_to_counts_path): Update.
23744 (duplicate_thread_path): Update.
23745 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23746 (struct switch_conv_info): Update.
23747 (gen_inbound_check): Update.
23748 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23749 (vect_do_peeling): Update.
23750 (vect_loop_versioning): Update.
23751 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23752 (optimize_mask_stores): Update.
23753 * ubsan.c (ubsan_expand_null_ifn): Update.
23754 * value-prof.c (gimple_divmod_fixed_value): Update.
23755 (gimple_divmod_fixed_value_transform): Update.
23756 (gimple_mod_pow2): Update.
23757 (gimple_mod_pow2_value_transform): Update.
23758 (gimple_mod_subtract): Update.
23759 (gimple_mod_subtract_transform): Update.
23760 (gimple_ic): Update.
23761 (gimple_stringop_fixed_value): Update.
23762 (gimple_stringops_transform): Update.
23763 * value-prof.h: Update.
23765 2017-06-29 Carl Love <cel@us.ibm.com>
23767 * config/rs6000/rs6000-c.c: Add support for built-in functions
23768 vector signed int vec_signed (vector float);
23769 vector signed long long vec_signed (vector double);
23770 vector signed int vec_signed2 (vector double, vector double);
23771 vector signed int vec_signede (vector double);
23772 vector signed int vec_signedo (vector double);
23773 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23774 instruction generator.
23775 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23776 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23777 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23779 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23780 vunsignede_v2df): Add define_expands.
23781 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23782 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23783 VEC_UNSIGNEDO): Add definitions.
23784 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23785 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23786 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23787 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23788 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23789 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23790 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23791 vec_unsignede, vec_unsignedo): Add builtin defines.
23792 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23794 * doc/extend.texi: Update the built-in documentation file for the
23795 new built-in functions.
23797 2017-06-29 Richard Biener <rguenther@suse.de>
23799 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23800 reduction chains to LOOP_VINFO_REDUCTIONS.
23801 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23802 SLP reductions after processing reduction chains.
23804 2017-06-29 Nathan Sidwell <nathan@acm.org>
23806 * builtins.c (fold_builtin_FUNCTION): Use
23807 lang_hooks.decl_printable_name.
23809 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23811 PR middle-end/81194
23812 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23813 with only one label.
23814 * stmt.c (expand_case): Assert NCASES is greater than one.
23816 2017-06-29 Richard Biener <rguenther@suse.de>
23818 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23820 (group_case_labels): Likewise.
23821 (find_taken_edge): Push sanity checking on val to workers...
23822 (find_taken_edge_cond_expr): ... here
23823 (find_taken_edge_switch_expr): ... and here, handle cases
23824 with just a default label.
23825 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23826 (group_case_labels): Likewise.
23827 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23828 group_case_labels does anything cleanup the CFG again.
23830 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23832 PR tree-optimization/81196
23833 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23834 exit condition comparing two IVs.
23836 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23838 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23839 profile to the dummy entry at the end of the list of architectures.
23840 * config/arm/arm-cpu-cdata.h: Regenerated.
23842 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23843 Michael Collison <michael.collison@arm.com>
23846 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23848 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23849 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23850 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23851 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23852 of shift when the shift amount is masked with constant equal to
23853 the size of the mode.
23854 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23857 2017-06-29 Martin Liska <mliska@suse.cz>
23859 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23861 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23864 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23867 (edge_growth_cache_entry::edge_growth_cache_entry): New
23869 (reset_edge_growth_cache): Update to use constructor.
23871 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23873 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23874 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23875 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23877 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23879 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23880 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23882 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23884 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23885 (*-linux-uclibc*): Add t-uclibc tmake_file.
23886 * config/t-musl: New.
23887 * config/t-uclibc: New.
23889 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23891 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23893 (gen_comm_data): Emit architectural setting of arch_prof.
23894 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23896 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23897 (armv8-m.base, armv8-m.main): Likewise.
23898 * arm-protos.h (arm_build_target): Add profile field.
23899 (arch_option): Likewise.
23900 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23902 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23903 arm_active_target.profile.
23905 2017-06-28 Richard Biener <rguenther@suse.de>
23907 PR middle-end/81227
23908 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23909 TYPE_OVERFLOW_WRAPS.
23910 * match.pd (negate_expr_p): Likewise.
23911 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23912 fold_build2, not fold_binary.
23914 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23916 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23917 Convert memory address to Pmode.
23918 (aarch64_print_operand): Assert MEM operands are always Pmode.
23920 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23923 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23924 Remove redundant if.
23925 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23926 * config/arm/aarch-common-protos.h
23927 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23928 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23930 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23933 * multiple_target.c (create_dispatcher_calls): Set the default
23934 clone to be static, not public.
23936 2017-06-28 Richard Biener <rguenther@suse.de>
23938 * tree-vect-loop.c (vectorizable_reduction): Move special
23939 cond reduction IV var creation ...
23940 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23941 parameter. Use STMT_VINFO_VECTYPE.
23942 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23945 2017-06-28 Martin Liska <mliska@suse.cz>
23948 * ipa-visibility.c (non_local_p): Handle visibility.
23950 2017-06-28 Martin Liska <mliska@suse.cz>
23953 * common.opt: Add IntegerRange to various options.
23954 * opt-functions.awk (integer_range_info): New function.
23955 * optc-gen.awk: Add integer_range_info to cl_options struct.
23956 * opts-common.c (decode_cmdline_option): Handle
23957 CL_ERR_INT_RANGE_ARG.
23958 (cmdline_handle_error): Likewise.
23959 * opts.c (print_filtered_help): Show valid interval in
23960 when --help is provided.
23961 * opts.h (struct cl_option): Add range_min and range_max fields.
23962 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23964 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23966 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23967 (x * C EQ/NE y * C): New transformation.
23969 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23971 * genmultilib (combination_space): Accept '+' in option names.
23973 2017-06-28 Martin Liska <mliska@suse.cz>
23976 * asan.c (instrument_derefs): Bail out inner references
23977 that are hard register variables.
23979 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23982 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23983 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23985 2017-06-28 Richard Biener <rguenther@suse.de>
23987 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23988 (vect_get_slp_defs): Adjust.
23989 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23991 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23993 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23994 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23995 (vectorizable_reduction): Adjust.
23996 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23998 (vect_get_slp_defs): Likewise.
23999 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24000 (vectorizable_bswap): Adjust.
24001 (vectorizable_call): Likewise.
24002 (vectorizable_conversion): Likewise.
24003 (vectorizable_assignment): Likewise.
24004 (vectorizable_shift): Likewise.
24005 (vectorizable_operation): Likewise.
24006 (vectorizable_store): Likewise.
24007 (vectorizable_condition): Likewise.
24008 (vectorizable_comparison): Likewise.
24010 2017-06-28 Michael Collison <michael.collison@arm.com>
24013 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24015 (arm_gen_movmemqi): Removed unused variable 'i'.
24016 Convert 'for' loop into 'while' loop.
24017 (arm_expand_prologue): Remove last unnecessary set of insn.
24018 (thumb_pop): Remove unused variable 'pushed_words'.
24019 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24021 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24023 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24024 * config/s390/s390-protos.h: Add prototype of
24025 s390_rel_address_ok_p.
24026 * config/s390/s390.c (s390_got_symbol): New function.
24027 (s390_rel_address_ok_p): New function.
24028 (legitimize_pic_address): Use s390_rel_address_ok_p.
24029 (s390_load_got): Use s390_got_symbol.
24030 (s390_option_override): Issue error if
24031 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24032 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24034 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24036 2017-06-27 Andrew Pinski <apinski@cavium.com>
24038 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24039 (X * copysign (1.0, X)): New pattern.
24040 (X * copysign (1.0, -X)): New pattern.
24041 (copysign (-1.0, CST)): New pattern.
24043 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24045 * genmultilib (combination_space): Remove variable.
24046 Validate reuse rules against regular expression for any sequence
24047 of multilib options in any order.
24049 2017-06-27 Michael Collison <michael.collison@arm.com>
24051 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24052 call aarch64_split_simd_combine.
24053 * (aarch64_combine_internal<mode>): Delete pattern.
24054 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24055 Allow register and subreg operands.
24057 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24059 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24060 specific need, just fallback on defaults.
24061 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24063 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24064 Olivier Hainque <hainque@adacore.com>
24066 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24068 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24069 targets. Pick a default if no particular attempt applied.
24070 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24073 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24075 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24076 (x86_64-wrs-vxworks7): Likewise.
24078 2017-06-27 Marek Polacek <polacek@redhat.com>
24081 * ubsan.c (instrument_null): Check get_base_address's result for null.
24083 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24085 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24087 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24089 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24090 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24091 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24092 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24093 New function types.
24094 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24095 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24096 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24097 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24098 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24099 BUILT_IN_FEUPDATEENV): New builtins.
24100 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24101 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24102 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24103 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24105 (builtin_structptr_types): Adjust size.
24106 * tree.c (builtin_structptr_types): Add four entries.
24108 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24109 Olivier Hainque <hainque@adacore.com>
24111 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24112 (TLS_SYM): New local macro, forcing reference to __tls__ on
24113 link command lines for VxWorks 7 RTPs, triggering initialization
24115 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24116 OS features TLS support, true for RTPs on VxWorks 7.
24117 * config/vxworks.c (vxworks_override_options): Setup emutls
24120 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24122 * predict.c (test_prediction_value_range): Use -1U instead of -1
24123 to avoid narrowing conversion warning.
24124 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24125 to avoid narrowing conversion warning.
24126 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24128 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24130 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24132 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24133 64bit configurations.
24134 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24135 (SIZE_TYPE): Likewise.
24136 * config/vxworks.c (vxworks_emutls_var_fields): Use
24137 long_unsigned_type_node instead of unsigned_type_node as the offset
24138 field type, which is "pointer" mode in emutls.c.
24140 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24143 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24145 PR middle-end/81207
24146 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24147 gimple_vuse copying separately from gimple_vdef copying.
24149 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24151 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24154 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24155 Olivier Hainque <hainque@adacore.com>
24157 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24158 for all vxworks7 targets.
24159 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24160 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24161 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24162 variations for VX6/VX7 and 32/64bits later on in ...
24163 (VXWORKS_LIB_SPEC): Leverage new macros.
24164 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24165 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24167 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24169 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24170 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24172 2017-06-26 Carl Love <cel@us.ibm.com>
24174 * config/rs6000/rs6000-c.c: Add support for built-in functions
24175 vector bool char vec_reve (vector bool char);
24176 vector signed char vec_reve (vector signed char);
24177 vector unsigned char vec_reve (vector unsigned char);
24178 vector bool int vec_reve (vector bool int);
24179 vector signed int vec_reve (vector signed int);
24180 vector unsigned int vec_reve (vector unsigned int);
24181 vector bool long long vec_reve (vector bool long long);
24182 vector signed long long vec_reve (vector signed long long);
24183 vector unsigned long long vec_reve (vector unsigned long long);
24184 vector bool short vec_reve (vector bool short);
24185 vector signed short vec_reve (vector signed short);
24186 vector double vec_reve (vector double);
24187 vector float vec_reve (vector float);
24188 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24189 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24190 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24191 (altivec_vreve): New pattern.
24192 * config/rs6000/altivec.h (vec_reve): New define.
24193 * doc/extend.texi (vec_rev): Update the built-in documentation file
24194 for the new built-in functions.
24196 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24198 PR tree-optimization/71815
24199 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24201 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24203 (slsr_process_phi): Likewise.
24204 (replace_uncond_cands_and_profitable_phis): Don't replace a
24205 multiply candidate with a stride of 1 (copy or cast).
24206 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24208 (lowest_cost_path): Likewise.
24209 (total_savings): Likewise.
24211 2017-06-26 Richard Biener <rguenther@suse.de>
24214 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24215 Use def_builtin_pure for all gather builtins.
24217 2017-06-26 Richard Biener <rguenther@suse.de>
24219 PR tree-optimization/81203
24220 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24221 non-dominating BBs.
24223 2017-06-26 Marek Polacek <polacek@redhat.com>
24226 * doc/invoke.texi: Document -Wmultistatement-macros.
24228 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24230 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24231 arm_neon_ok_no_float_abi effective target.
24233 2017-06-26 Richard Biener <rguenther@suse.de>
24235 PR tree-optimization/80928
24236 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24237 (copy_bbs): Set BB_DUPLICATED flag early.
24238 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24240 (execute_on_shrinking_pred): Likewise.
24241 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24242 BB_DUPLICATED blocks.
24243 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24244 iterate over all PHIs considering removal of *gsi.
24246 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24248 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24251 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24253 * config/rs6000/rs6000-string.c: (expand_block_clear,
24254 do_load_for_compare, select_block_compare_mode,
24255 compute_current_alignment, expand_block_compare,
24256 expand_strncmp_align_check, expand_strn_compare,
24257 expand_block_move, rs6000_output_load_multiple)
24258 Move functions related to string/block move/compare
24259 to a separate file.
24260 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24261 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24262 for this function which is now used in two files.
24263 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24264 * config.gcc: Add rs6000-string.o to extra_objs for
24265 targets powerpc*-*-* and rs6000*-*-*.
24267 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24270 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24271 32-bit, since indexed is not valid for DImode.
24272 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24273 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24274 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24275 (define_peephole2 for Altivec d-form store): Likewise.
24278 * multiple_target.c (create_dispatcher_calls): Only create the
24279 dispatcher call if the function is the default clone of a
24280 versioned function.
24282 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24284 PR middle-end/80902
24285 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24286 a call, force the call to not be a tail call.
24288 2017-06-23 Jeff Law <law@redhat.com>
24290 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24293 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24295 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24296 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24297 handling for early expansion of vector shifts (sl,sr,sra,rl).
24298 (builtin_function_type): Add vector shift right instructions
24299 to the unsigned argument list.
24301 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24303 rtl-optimizatoin/79286
24304 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24305 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24306 trap. PIC register plus a const unspec without offset can never trap.
24308 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24310 * tree.h (builtin_structptr_type): New type.
24311 (builtin_structptr_types): Declare new array.
24312 * tree.c (builtin_structptr_types): New array.
24313 (free_lang_data, build_common_tree_nodes): Use it.
24315 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24318 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24321 2017-06-22 Matt Turner <mattst88@gmail.com>
24323 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24324 Lake models to skylake case. Assume skylake for unknown
24325 models with clflushopt.
24327 2017-06-22 Jeff Law <law@redhat.com>
24329 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24330 frame sizes that do not satisfy aarch64_uimm12_shift.
24332 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24334 * profile-count.h (apply_probability,
24335 apply_scale, probability_in): Fix checks for zero.
24337 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24339 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24340 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24342 2016-06-22 Richard Biener <rguenther@suse.de>
24344 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24345 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24346 REDUC_MAX_EXPR support.
24347 (vectorizable_reduction): Likewise.
24348 (vect_create_epilog_for_reduction): Likewise.
24350 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24352 * match.pd (A / (1 << B) -> A >> B): New.
24353 * generic-match-head.c: Include optabs-tree.h.
24354 * gimple-match-head.c: Likewise.
24355 * optabs-tree.h (target_supports_op_p): New.
24356 * optabs-tree.c (target_supports_op_p): New.
24358 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24360 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24361 $gcc_cv_ld --help output.
24362 (gcc_cv_ld_demangle): Likewise.
24363 (gcc_cv_ld_eh_frame_hdr): Likewise.
24364 (gcc_cv_ld_pie): Likewise.
24365 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24366 (gcc_cv_ld_buildid): Likewise.
24367 (gcc_cv_ld_sysroot): Likewise.
24368 (ld_bndplt_support): Likewise.
24369 (ld_pushpopstate_support): Likewise.
24370 * configure: Regenerate.
24371 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24373 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24376 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24377 operands indexes to avoid gap between operands and match_dups.
24379 2017-06-21 Andrew Pinski <apinski@cavium.com>
24381 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24382 Increment Arith_shift and Arith_shift_reg by 1.
24383 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24385 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24386 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24387 (aarch64_strip_extend): Add new argument and test for it.
24388 (aarch64_cheap_mult_shift_p): New function.
24389 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24390 add a cost if it is true.
24391 Update calls to aarch64_strip_extend.
24392 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24394 2017-06-21 Andrew Pinski <apinski@cavium.com>
24396 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24398 (thunderxt88): Likewise.
24399 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24400 (thunderx_prefetch_tune): New variable.
24401 (thunderx2t99_prefetch_tune): Update for the correct values.
24402 (thunderxt88_tunings): New variable.
24403 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24404 generic_prefetch_tune.
24405 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24407 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24409 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24410 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24411 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24412 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24413 (aarch64_atomic_cas<mode>, GPI): Likewise.
24415 2017-06-21 Martin Liska <mliska@suse.cz>
24417 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24418 statements on cold and hot labels.
24419 * predict.c (tree_estimate_probability_bb): Remove the
24420 prediction from this place.
24422 2017-06-21 Martin Liska <mliska@suse.cz>
24424 PR tree-optimization/79489
24425 * gimplify.c (maybe_add_early_return_predict_stmt): New
24427 (gimplify_return_expr): Call the function.
24428 * predict.c (tree_estimate_probability_bb): Remove handling
24430 * predict.def: Update comment about early return predictor.
24431 * gimple-predict.h (is_gimple_predict): New function.
24432 * predict.def: Change default value of early return to 66.
24433 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24435 * passes.def: Put pass_strip_predict_hints to the beginning of
24438 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24440 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24441 FUNCTION_DECL declarations.
24442 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24444 (dwaf2out_decl): Likewise.
24445 * godump.c (go_early_global_decl): Skip call to the real debug hook
24446 for FUNCTION_DECL declarations.
24447 * passes.c (rest_of_decl_compilation): Skip call to the
24448 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24449 -fdump-go-spec is passed.
24451 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24453 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24454 Reorder for compactness.
24455 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24456 (def_builtin_pure, def_builtin_pure2): New functions.
24457 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24459 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24461 * match.pd (nop_convert): New predicate.
24462 ((A +- CST1) +- CST2): Allow some NOP conversions.
24464 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24467 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24468 with ctors/dtors if GOVD_SHARED is set.
24470 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24472 * config/aarch64/aarch64.md (movti_aarch64):
24473 Emit mov rather than orr.
24474 (movtf_aarch64): Likewise.
24475 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24476 Emit mov rather than orr.
24478 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24480 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24481 Swap alternatives, make integer dup more expensive.
24483 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24485 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24486 Return true for non-tls symbols.
24488 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24490 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24491 (cortex-a75): Likewise.
24492 (cortex-a75.cortex-a55): Likewise.
24493 * config/aarch64/aarch64-tune.md: Regenerate.
24494 * doc/invoke.texi (-mtune): Document new values for -mtune.
24496 2017-06-21 Tom de Vries <tom@codesourcery.com>
24498 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24499 stack_size feature.
24500 (Effective-Target Keywords, Other attributes): Suggest using
24501 dg-add-options stack_size feature to get stack limit in stack_size
24502 effective target documentation.
24504 2017-06-21 Julian Brown <julian@codesourcery.com>
24505 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24507 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24508 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24509 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24511 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24512 attribute type list for neon_multiply.
24513 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24514 attribute type list for neon_multiply.
24515 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24516 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24517 attribute type list for neon_multiply.
24518 * config/arm/types.md (crypto_pmull): Add.
24519 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24520 attribute type list.
24522 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24524 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24527 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24529 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24530 to make sure not to dereference a NULL cost_classes_ptr pointer.
24532 2017-06-20 Carl Love <cel@us.ibm.com>
24534 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24535 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24536 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24537 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24538 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24539 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24540 VMULOSW): New enum "unspec" values.
24541 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24542 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24543 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24544 altivec_vmulosw): New patterns.
24545 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24546 VMULOSW): Add definitions.
24548 2017-06-20 Julia Koval <julia.koval@intel.com>
24550 * config/i386/i386.c: Fix rounding expand for new pattern.
24551 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24553 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24555 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24556 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24558 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24560 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24563 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24565 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24566 architecture, then by implementer ID.
24567 * config/aarch64/aarch64-tune.md: Regenerate.
24569 2017-06-20 Richard Biener <rguenther@suse.de>
24571 PR middle-end/81097
24572 * fold-const.c (split_tree): Fold to type before negating.
24574 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24576 * diagnostic-show-locus.c
24577 (selftest::test_fixit_deletion_affecting_newline): New function.
24578 (selftest::diagnostic_show_locus_c_tests): Call it.
24580 2017-06-20 Andreas Schwab <schwab@suse.de>
24583 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24586 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24588 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24589 __ARM_FEATURE_COPROC according to support.
24591 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24593 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24594 Rewritten to avoid overflow for > 32-bit pointers.
24597 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24598 by removing enum keyword.
24599 (ubsan_type_descriptor): Likewise. Formatting fix.
24602 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24603 splitter): Require TARGET_SSE2 in the condition.
24605 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24608 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24609 for doing vector set of SFmode on ISA 3.0.
24610 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24611 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24613 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24614 SFmode value into a V4SF variable that was extracted from another
24615 V4SF variable without converting the element to double precision
24616 and back to single precision vector format.
24617 (vsx_insert_extract_v4sf_p9_2): Likewise.
24619 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24621 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24622 in UWHI to avoid undefined overflow.
24625 * ubsan.h (enum ubsan_encode_value_phase): New.
24626 (ubsan_encode_value): Change second argument to
24627 enum ubsan_encode_value_phase with default value of
24628 UBSAN_ENCODE_VALUE_GENERIC.
24629 * ubsan.c (ubsan_encode_value): Change second argument to
24630 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24631 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24632 create_tmp_var_raw instead of create_tmp_var and use a
24634 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24635 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24636 ubsan_encode_value callers.
24639 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24640 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24641 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24643 2017-06-19 Richard Biener <rguenther@suse.de>
24645 PR middle-end/81118
24646 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24647 estimates if we changed anything.
24649 2017-06-19 Richard Biener <rguenther@suse.de>
24651 PR tree-optimization/80887
24652 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24653 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24654 simplified lookups, then reset mprts_hook.
24655 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24657 (try_to_simplify): Likewise.
24659 2017-06-19 Martin Liska <mliska@suse.cz>
24662 * gimplify.c (gimplify_switch_expr):
24663 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24665 2017-06-19 Martin Liska <mliska@suse.cz>
24667 * doc/install.texi: Document that PGO runs in 4 stages.
24669 2017-06-19 Martin Liska <mliska@suse.cz>
24672 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24673 to dispatcher function name.
24674 * multiple_target.c (replace_function_decl): New function.
24675 (create_dispatcher_calls): Redirect both edges and references.
24677 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24679 * profile-count.c (profile_count::dump): Dump quality.
24680 (profile_count::differs_from_p): Update for unsigned val.
24681 * profile-count.h (profile_count_quality): New enum.
24682 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24684 2017-06-19 Richard Biener <rguenther@suse.de>
24686 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24687 struct function as arg.
24688 (estimate_numbers_of_iterations): Export overload with loop arg.
24689 (free_numbers_of_iterations_estimates_loop): Use an overload of
24690 free_numbers_of_iterations_estimates instead.
24691 * tree-cfg.c (remove_bb): Adjust.
24692 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24693 * tree-parloops.c (gen_parallel_loop): Likewise.
24694 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24696 (tree_unroll_loops_completely): Likewise.
24697 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24698 Use an overload instead and export.
24699 (estimated_loop_iterations): Adjust.
24700 (max_loop_iterations): Likewise.
24701 (likely_max_loop_iterations): Likewise.
24702 (estimate_numbers_of_iterations): Take struct function as arg
24704 (loop_exits_before_overflow): Adjust.
24705 (free_numbers_of_iterations_estimates_loop): Use an overload.
24706 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24707 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24709 2017-06-19 Richard Biener <rguenther@suse.de>
24712 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24713 RANGE_EXPR conservatively.
24715 2017-06-16 Carl Love <cel@us.ibm.com>
24717 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24718 definitions for vec_float, vec_float2, vec_floato,
24719 vec_floate built-ins.
24720 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24721 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24723 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24724 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24725 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24726 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24727 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24728 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24729 vec_floato): Add builtin defines.
24730 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24731 Update the built-in documentation file for the new built-in
24734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24736 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24737 (mthumb): Mark as the negative of -marm.
24739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24741 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24743 (ARM Options, -mtune): Document that this accepts the same
24744 extension options as -mcpu.
24745 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24749 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24750 permitted extensions.
24752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24754 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24755 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24756 (armv8-m.main): Add option +nodsp.
24757 * config/arm/arm-cpu-cdata.h: Regenerated.
24759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24761 * config/arm/t-fuchsia: New file.
24762 * config.gcc (arm*-*-fuchsia*): Use it.
24764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24766 * config/arm/t-symbian: Rewrite for new option infrastructure.
24768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24770 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24771 (MULTILIB_REQUIRED): Likewise.
24773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24775 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24776 (MULTILIB_RESUE): Likewise.
24777 (MULTILIB_MATCHES): Likewise.
24778 (MULTLIB_REQUIRED): Likewise.
24780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24782 * config/arm/t-rtems: Rewrite for new option framework.
24784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24786 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24787 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24788 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24789 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24790 * config/arm/t-multilib: ... here.
24791 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24792 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24793 armv7-a and armv8*-a when A-profile libraries have not been built.
24794 * config/arm/t-rmprofile: Rewrite.
24796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24798 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24799 with a backslash. Remove the backslash after substituting unescaped
24801 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24805 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24806 the driver to pass through the default setting of -mfloat-abi.
24807 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24809 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24810 (all_feat_combs): New rule.
24811 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24813 * config/arm/t-aprofile: Rewrite.
24815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24817 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24818 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24819 fpu is not specified by the user/command-line.
24820 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24821 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24822 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24823 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24824 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24825 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24829 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24830 * config/arm/t-arm-elf: Rewritten.
24832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24834 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24835 have some floating-point instructions.
24836 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24837 (TARGET_MAYBE_HARD_FLOAT): New macro.
24838 * config/arm/arm-builtins.c (arm_init_builtins): Use
24839 TARGET_MAYBE_HARD_FLOAT.
24840 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24844 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24845 (configargs.h): Include it.
24846 (arm_print_hint_for_fpu_option): New function.
24847 (arm_parse_fpu_option): New function.
24848 (candidate_extension): New class.
24849 (arm_canon_for_multilib): New function.
24850 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24851 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24852 (ARCH_CANONICAL_SPECS): New macro.
24853 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24855 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24857 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24858 are set after handling multilib fragments. Set target_cpu_default2
24861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24863 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24865 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24866 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24868 (check_arch): Likewise.
24869 * config/arm/arm.c (arm_configure_build_target): Handle
24870 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24871 options in the default.
24873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24875 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24876 when an option is an alias of another.
24877 * config/arm/parsecpu.awk (optalias): New parser token.
24878 (gen_comm_data): Mark non-alias options as such. Emit entries
24879 for extension aliases.
24880 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24881 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24882 (armv6kz, armv6zk, armv6t2): Likewise.
24883 (armv7): Make vfpv3-d16 an alias.
24884 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24886 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24887 Sort in canonical order.
24888 (armv8-a): Sort in canonical order.
24889 (armv8.1-a, armv8.2-a): Likewise.
24890 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24892 (cortex-a9): Sort in canonical order.
24893 * config/arm/arm.c (selftests.h): Include it.
24894 (arm_test_cpu_arch_data): New function.
24895 (arm_run_self_tests): New function.
24896 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24897 (targetm): Move declaration to the end of the file.
24898 * arm-cpu-cdata.h: Regenerated.
24900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24902 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24903 call to target_mode_check describing the type of option passed.
24904 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24905 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24906 arm_parse_cpu_option_name to match parameters against list of
24908 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24909 arm_arch_core_flags data structure.
24910 * config/arm/arm-cpu_cdata.h: Regenerated.
24912 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24914 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24916 (arm_print_hint_for_cpu_option): Likewise.
24917 (arm_print_hint_for_arch_option): Likewise.
24918 (arm_parse_cpu_option_name): Likewise.
24919 (arm_parse_arch_option_name): Likewise.
24920 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24921 of entries in the all_fpus list.
24922 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24923 (arm_parse_cpu_option_name): Declare.
24924 (arm_parse_arch_option_name): Declare.
24925 (arm_parse_option_features): Declare.
24926 (arm_intialize_isa): Declare.
24927 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24929 (gen_comm_data): ... here. Make definitions non-static.
24930 * config/arm/arm-cpu-data.h: Regenerated.
24931 * config/arm/arm-cpu-cdata.h: Regenerated.
24933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24935 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24936 (cpu_arch_extension): New structure.
24937 (cpu_arch_option, arch_option, cpu_option): New structures.
24938 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24939 architecture types.
24940 (gen_data): Generate new format data tables.
24941 * config/arm/arm.c (cpu_tune): New structure.
24942 (cpu_option, processors): Delete.
24943 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24944 (arm_print_hint_for_cpu_option): ... this and ...
24945 (arm_print_hint_for_arch_option): ... this.
24946 (arm_parse_arch_cpu_name): Delete. Replace with ...
24947 (arm_parse_cpu_option_name): ... this and ...
24948 (arm_parse_arch_option_name): ... this.
24949 (arm_unrecognized_feature): Change type of target parameter to
24951 (arm_parse_arch_cpu_features): Delete. Replace with ...
24952 (arm_parse_option_features): ... this.
24953 (arm_configure_build_target): Rework to use new configuration data
24955 (arm_print_tune_info): Rework for new configuration data tables.
24956 * config/arm/arm-cpu-data.h: Regenerated.
24957 * config/arm/arm-cpu.h: Regenerated.
24959 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24961 * Makefile.in (OBJS): Move sbitmap.o from here ...
24962 (OBJS-libcommon): ... to here.
24964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24966 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24967 (ISA_ALL_CRYPTO): New macro.
24968 (ISA_ALL_SIMD): New macro
24969 (ISA_ALL_FP): New macro.
24970 * config/arm/arm.c (fpu_bitlist): Update initializer.
24971 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24973 (arm9e): Add fpu. Add option for nofp
24974 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24975 (arm926ej-s, arm1026ej-s): Likewise.
24976 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24977 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24978 neon-fp16, neon-vfpv4, nofp and nosimd.
24979 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24980 (cortex-a8): Add fpu. Add option for nofp.
24981 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24982 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24983 (cortex-r4f): Add fpu.
24984 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24985 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24987 (cortex-r8): Likewise.
24988 (cortex-m4): Add fpu. Add option for nofp.
24989 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24990 (cortex-a17.cortex-a7): Likewise.
24991 (cortex-a32): Add fpu. Add options for crypto and nofp.
24992 (cortex-a35, cortex-a53): Likewise.
24993 (cortex-a57): Add fpu. Add option for crypto.
24994 (cortex-a72, cortex-a73): Likewise.
24995 (exynos-m1): Likewise.
24996 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24997 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24998 (cortex-m33): Add fpu. Add option for nofp.
24999 * config/arm/arm-cpu-cdata.h: Regenerated
25000 * config/arm/arm-cpu-data.h: Regenerated.
25002 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25004 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25005 (armv5te, armv5tej): Likewise.
25006 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25007 (armv7): Add options fp and vfpv3-d16.
25008 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25009 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25011 (armv7ve): Likewise.
25012 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25013 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25014 (armv8-a): Add nocrypto option.
25015 (armv8.1-a, armv8.2-a): Likewise.
25016 (armv8-m.main): add options fp, fp.dp and nofp.
25018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25020 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25022 (armv8-a+crc): Delete.
25023 (armv8.1-a): Add options simd, crypto and nofp.
25024 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25025 (armv8.2-a+fp16): Delete.
25026 (armv8-m.main): Add option dsp.
25027 (armv8-m.main+dsp): Delete.
25028 (cortex-a8): Add fpu. Add nofp option.
25029 (cortex-a9): Add fpu. Add nofp and nosimd options.
25030 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25031 link to main cpu and architecture data structures.
25032 (gen_comm_data): Only put isa attributes from the main architecture
25034 (option): New statement for architecture and CPU entries.
25035 * arm.c (struct cpu_option): New structure.
25036 (struct processors): Add entry for options.
25037 (arm_unrecognized_feature): New function.
25038 (arm_parse_arch_cpu_name): Ignore any characters after the first
25040 (arm_parse_arch_cpu_feature): New function.
25041 (arm_configure_build_target): Separate out any CPU and architecture
25042 features and parse separately. Don't error out if -mfpu=auto is
25043 used with only an architecture string.
25044 (arm_print_asm_arch_directives): New function.
25045 (arm_file_start): Call it.
25046 * config/arm/arm-cpu-cdata.h: Regenerated.
25047 * config/arm/arm-cpu-data.h: Likewise.
25048 * config/arm/arm-tables.opt: Likewise.
25050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25052 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25053 assembler when it is not -mfpu=auto.
25055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25057 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25058 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25059 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25060 (ASM_CPU_SPEC): Rewrite.
25061 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25062 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25063 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25064 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25065 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25066 copied string is NUL-terminated. Also strip any characters prefixed
25068 (arm_rewrite_selected_arch): New function.
25069 (arm_rewrite_march): New function.
25071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25073 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25074 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25075 (march, mcpu, mtune): Convert to string-based options.
25076 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25077 (arm_parse_arch_cpu_name): New function.
25078 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25079 identify selected architecture or CPU.
25080 (arm_option_save): New function.
25081 (TARGET_OPTION_SAVE): Redefine.
25082 (arm_option_restore): Restore string options.
25083 (arm_option_print): Print string options.
25085 2017-06-16 Martin Sebor <msebor@redhat.com>
25087 PR tree-optimization/80933
25088 PR tree-optimization/80934
25089 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25090 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25091 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25092 (gimple_fold_builtin): Call them.
25094 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25096 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25097 as unlikely; update profile.
25099 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25101 * predict.c (force_edge_cold): Handle declaring edges impossible
25104 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25106 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25108 (try_unroll_loop_completely): Fix reporting.
25110 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25112 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25114 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25117 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25118 if given a non-constant argument for an intrinsic which requires a
25121 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25123 * profile.c (compare_freqs): New function.
25124 (branch_prob): Sort edge list.
25125 (find_spanning_tree): Assume that the list is priority sorted.
25127 2017-06-16 Richard Biener <rguenther@suse.de>
25129 PR tree-optimization/81090
25130 * passes.def (pass_record_bounds): Remove.
25131 * tree-pass.h (make_pass_record_bounds): Likewise.
25132 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25133 make_pass_record_bounds): Likewise.
25134 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25135 not free niter estimates at the beginning but at the end.
25136 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25138 2017-06-16 Richard Biener <rguenther@suse.de>
25140 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25141 initializer to workaround ICE in host GCC 4.8.
25143 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25145 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25147 (clone_inlined_nodes): Update.
25149 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25151 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25152 prefetch settings, and enable prefetching by default at -O3.
25154 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25156 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25157 Set flag_prefetch_loop_arrays according to tuning data.
25159 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25161 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25162 New tune structure.
25163 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25164 [Unrelated to main purpose of the patch] Place the pointer field last
25165 to enable type checking errors when tune structure are wrongly merged.
25166 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25167 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25168 (thunderx2t99_prefetch_tune): New tune constants.
25169 (tune_params *_tunings): Update all tunings (no functional change).
25170 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25171 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25172 from tunings structures.
25174 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25177 * ubsan.c (instrument_null): Add T argument, use it instead
25178 of computing it based on IS_LHS.
25179 (instrument_object_size): Likewise.
25180 (pass_ubsan::execute): Adjust instrument_null and
25181 instrument_object_size callers to pass gimple_get_lhs or
25182 gimple_assign_rhs1 result to it. Use instrument_null instead of
25183 calling get_base_address and instrument_mem_ref. Handle
25184 aggregate call arguments for object-size sanitization.
25186 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25188 PR tree-optimization/81089
25189 * tree-vrp.c (is_masked_range_test): Validate operands of
25192 2017-06-15 Martin Sebor <msebor@redhat.com>
25195 * dumpfile.c (dump_register): Avoid calling memset to initialize
25196 a class with a default ctor.
25197 * gcc.c (struct compiler): Remove const qualification.
25198 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25199 * hash-table.h: Ditto.
25200 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25202 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25203 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25205 * params.h (struct param_info): Make struct members non-const.
25206 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25207 with default initialization.
25208 * vec.h (vec_copy_construct, vec_default_construct): New helper
25210 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25211 with vec_copy_construct.
25212 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25213 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25214 * doc/invoke.texi (-Wclass-memaccess): Document.
25216 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25218 * emit-rtl.h (is_leaf): Update comment about local
25219 register allocator.
25221 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25224 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25225 for a variable to have a section before checking if the section has a
25227 Set section to.persistent if persistent attribute is set.
25228 Warn if .persistent attribute is used on an automatic variable.
25230 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25232 PR rtl-optimization/80474
25233 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25235 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25237 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25240 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25242 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25243 (casesi): Emit bounds checking as RTL.
25244 (casesi_internal_mips16_<mode>): Remove bounds checking.
25246 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25248 * config/xtensa/xtensa.c (xtensa_option_override): Append
25249 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25250 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25251 xtensa_doloop_hooks): Define unconditionally.
25252 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25254 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25255 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25256 for it in xtensa_option_override.
25257 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25258 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25260 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25262 * doc/cppopts.texi: Document '-' special value to -MF.
25264 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25266 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25267 (cortex_a53_fconst): Likewise.
25268 (cortex_a53_fpmul): Likewise.
25269 (cortex_a53_f_load_64): Likewise.
25270 (cortex_a53_f_load_many): Likewise.
25271 (cortex_a53_advsimd_alu): Likewise.
25272 (cortex_a53_advsimd_alu_q): Likewise.
25273 (cortex_a53_advsimd_mul): Likewise.
25274 (cortex_a53_advsimd_mul_q): Likewise.
25275 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25276 Add missing fmul, r2f_cvt and fconst cases.
25278 2017-06-14 Richard Biener <rguenther@suse.de>
25280 PR middle-end/81088
25281 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25283 (fold_binary_loc): When associating do not treat pre-existing
25284 TREE_OVERFLOW on literal constants as a reason to allow
25285 TREE_OVERFLOW on associated literal constants.
25287 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25289 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25290 (MASK_FEATURES): New macro.
25291 * config/sparc/sparc.c (sparc_option_override): Remove the special
25292 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25294 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25296 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25297 a division of 0 if non-call exceptions are enabled.
25299 2017-06-14 Andrew Pinski <apinski@cavium.com>
25300 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25303 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25304 Improve vector initialization code gen for only variable case.
25306 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25308 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25310 2017-06-14 Richard Biener <rguenther@suse.de>
25312 PR tree-optimization/81083
25313 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25316 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25318 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25319 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25320 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25321 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25322 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25323 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25325 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25327 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25328 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25330 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25332 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25336 * config/rs6000/t-rtems: Don't handle SPE.
25338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25340 * config/rs6000/t-linux: Don't handle SPE.
25342 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25344 * config/rs6000/eabispe.h: Delete file.
25346 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25348 * config/rs6000/t-spe: Delete file.
25350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25352 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25353 (rs6000_legitimate_offset_address_p): Return false for anything in
25354 V2SImode or V2SFmode.
25356 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25358 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25359 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25360 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25362 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25363 (rs6000_legitimate_offset_address_p): Ditto.
25364 (rs6000_emit_move): Ditto.
25365 (rs6000_init_builtins): Remove V4HI_type_node.
25367 2017-06-13 Martin Liska <mliska@suse.cz>
25370 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25371 (gate_asan): Likewise.
25372 * asan.h (asan_no_sanitize_address_p): Remove the function.
25373 (sanitize_flags_p): New function.
25374 * builtins.def: Fix coding style.
25375 * common.opt: Use renamed enum value.
25376 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25377 * doc/extend.texi: Document no_sanitize attribute.
25378 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25379 to SANITIZE_UNDEFINED_NONDEFAULT.
25380 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25381 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25382 Use sanitize_flags_p.
25383 * gimplify.c (gimplify_function_tree): Likewise.
25384 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25385 * opts.c (parse_no_sanitize_attribute): New function.
25386 (common_handle_option): Use renamed enum value.
25387 * opts.h (parse_no_sanitize_attribute): Declare.
25388 * tree.c (sanitize_flags_p): New function.
25389 * tree.h: Declared here.
25390 * tsan.c: Use sanitize_flags_p.
25391 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25392 (instrument_mem_ref): Likewise.
25393 (instrument_bool_enum_load): Likewise.
25394 (do_ubsan_in_current_function): Remove the function.
25395 (pass_ubsan::execute): Use sanitize_flags_p.
25396 * ubsan.h: Remove do_ubsan_in_current_function
25397 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25398 (dump_function_to_file): Use it here.
25400 2017-06-13 Martin Jambor <mjambor@suse.cz>
25402 PR tree-optimization/80803
25403 PR tree-optimization/81063
25404 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25405 (propagate_subaccesses_across_link): Enqueue subtree whenever
25406 necessary instead of relying on the caller.
25408 2017-06-13 Martin Jambor <mjambor@suse.cz>
25410 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25411 that have a first_link.
25412 (sort_and_splice_var_accesses): Do not check first_link before
25414 (subtree_mark_written_and_enqueue): Likewise.
25415 (propagate_all_subaccesses): Likewise and do not stop at first
25416 parent with a first_link.
25418 2017-06-13 Martin Jambor <mjambor@suse.cz>
25420 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25423 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25425 * match.pd: New pattern.
25427 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25429 * tree-vrp.c (is_masked_range_test): New function.
25430 (register_edge_assert_for): Determine ranges for
25433 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25435 PR tree-optimization/67328
25436 * fold-const.c (maskable_range_p): New function.
25437 (build_range_check): Generate bittests if possible.
25439 2017-06-13 Martin Liska <mliska@suse.cz>
25441 * gimple-pretty-print.c (dump_probability): Add new argument.
25442 (dump_edge_probability): Dump both probability and count.
25443 (dump_gimple_label): Likewise.
25444 (dump_gimple_bb_header): Likewise.
25446 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25449 * config/avr/avr-devices.c: Fix indentation.
25450 * config/avr/gen-avr-mmcu-specs.c: Dito.
25452 2017-06-13 Richard Biener <rguenther@suse.de>
25454 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25455 instead get vector type from stmt_info.
25456 (vectorizable_reduction): Adjust. Remove dead code.
25458 2017-06-13 Richard Biener <rguenther@suse.de>
25460 PR middle-end/81065
25461 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25462 case of C * (x * C2 + C3).
25463 (fold_addr_of_array_ref_difference): Properly fold index difference.
25465 2017-06-12 David S. Miller <davem@davemloft.net>
25468 * config/sparc/sparc.md (return expander): Emit frame blockage if
25469 function uses alloca.
25471 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25473 * combine.c (make_field_assignment): Check len rather than the mode
25474 precision when calling force_to_mode.
25476 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25478 Support multilibs and devices that see flash in RAM address range.
25481 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25482 (avr_mcu_t) <flash_pm_offset>: New field.
25483 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25484 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25485 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25486 (AVR_TINY_PM_OFFSET): Remove macro.
25487 * config/avr/avr.opt (-mshort-calls): New option.
25488 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25489 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25490 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25491 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25492 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25493 instead of avr_arch->have_jmp_call.
25494 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25495 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25496 avr_arch->flash_pm_offset to define.
25497 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25498 new field flash_pm_offset. Add entry for avrxmega3.
25499 (avr_texinfo): Add entry for avrxmega3.
25500 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25501 attiny212, attiny214,
25502 attiny412, attiny414, attiny416, attiny417,
25503 attiny814, attiny816, attiny817,
25504 attiny1614, attiny1616, attiny1617,
25505 attiny3214, attiny3216, attiny3217.
25506 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25507 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25508 (avr_print_operand_address) [AVR_TINY]: Same.
25509 (avr_asm_init_sections) <readonly_data_section>: Only patch
25510 callback if avr_arch->flash_pm_offset = 0.
25511 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25512 for rodata if avr_arch->flash_pm_offset != 0.
25513 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25514 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25515 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25516 (m_options): Append opt_rcall.
25517 (m_dirnames): Append dir_rcall.
25518 * config/avr/t-multilib: Regenerate.
25520 * configure.ac [target=avr]: Check whether avrxmega3 default
25521 linker description file works as needed.
25522 * configure: Regenerate.
25523 * doc/avr-mmcu.texi: Regenerate.
25524 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25525 <__AVR_ARCH__>: Document avrxmega3 and 103.
25526 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25527 <__AVR_SHORT_CALLS__>: Document it.
25528 <__AVR_PM_BASE_ADDRESS__>: Document it.
25529 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25530 (AVR Variable Attributes) <progmem>: Document this is
25531 not needed for avrxmega3.
25532 (AVR Named Address Spaces) <__flash>: Dito.
25534 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25536 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25538 2017-06-12 Doug Rupp <rupp@adacore.com>
25540 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25541 Append vxworks-stdint.h to the tm_file list.
25542 * config/vxworks-stdint.h: New file.
25544 2017-06-12 Martin Liska <mliska@suse.cz>
25546 PR tree-optimization/81041
25547 * tree-profile.c (gimple_gen_ic_func_profiler):
25548 Create an extra BB in profile-generate
25549 (gimple_gen_time_profiler): Likewise.
25551 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25553 PR tree-optimization/81003
25554 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25555 (update_range_test): Use it instead of force_gimple_operand_gsi.
25557 2017-06-12 Richard Biener <rguenther@suse.de>
25559 PR tree-optimization/81053
25560 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25561 with backedge value not defined in loop. Simplify def stmt
25564 2017-06-11 Tom de Vries <tom@codesourcery.com>
25567 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25569 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25570 nvptx_cannot_force_const_mem.
25572 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25574 * opts.c (finish_options): Move test for flag_split_stack after
25575 it has been initialized.
25577 2017-06-11 Jason Merrill <jason@redhat.com>
25579 * tree.h (id_equal): New.
25580 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25581 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25582 instead of strcmp of IDENTIFIER_POINTER.
25584 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25586 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25587 (mark_all_inlined_calls_cdtor): Fix formating.
25588 (inline_transform): Rescale profile before inlining.
25590 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25592 * cgraph.h (cgraph_edge::clone): Update prototype.
25593 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25594 (cgraph_node::create_clone): Update.
25595 (cgraph_node::create_version_clone): Update.
25596 * tree-inline.c (copy_bb): Update.
25597 (expand_call_inline): Update.
25599 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25601 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25602 factored out from ...
25603 (rs6000_emit_prologue): ... here.
25605 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25607 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25608 factored out from ...
25609 (rs6000_emit_prologue): ... here.
25611 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25613 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25615 (handle_missing_profiles): Fix computation of tp_first_run.
25616 (counts_to_freqs): Do not touch freqs when count is 0.
25618 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25620 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25623 2017-06-10 Tom de Vries <tom@codesourcery.com>
25625 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25626 attributes): Document signal effective target.
25628 2017-06-10 Tom de Vries <tom@codesourcery.com>
25630 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25631 Document effective target stack_size.
25633 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25635 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25636 to the edit_context if they can be auto-applied.
25638 2017-06-9 Ian Lance Taylor <iant@golang.org>
25640 * opts.c (finish_options): If -fsplit-stack, disable implicit
25641 -forder-blocks-and-partition.
25642 * doc/invoke.texi (Optimize Options): Document that when using
25643 -fsplit-stack -forder-blocks-and-partition is not implicitly
25646 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25648 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25649 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25650 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25651 * builtins.def (abort, trap, unreachable): Declare cold.
25652 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25653 * tree-core.h (ECF_COLD): New.
25654 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25655 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25657 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25659 * predict.c (unlikely_executed_stmt_p): Cleanup.
25661 2017-06-09 Richard Biener <rguenther@suse.de>
25663 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25664 model if the ref is always written to.
25666 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25668 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25670 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25672 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25675 2017-06-09 Tom de Vries <tom@codesourcery.com>
25678 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25679 "target cannot support label values" when encountering LABEL_REF.
25681 2017-06-09 Martin Liska <mliska@suse.cz>
25683 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25684 (gimple_gen_ic_func_profiler): Emit direct comparison
25685 of __gcov_indirect_call_callee with NULL.
25686 (gimple_gen_time_profiler): Change probability from
25687 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25689 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25691 * profile.c (edge_gcov_counts): Turn to pointer.
25692 (compute_branch_probabilities, compute_branch_probabilities): Update.
25693 (branch_prob): Do not clear edge_gcov_count.
25694 * profile.h (edge_gcov_counts): Turn to pointer.
25695 (edge_gcov_count): Update.
25697 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25699 * gimple.h (gimple_check_failed): Mark cold.
25701 2017-06-09 Richard Biener <rguenther@suse.de>
25703 PR tree-optimization/66623
25704 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25705 refactor check_reduction into two parts, properly computing
25706 whether we have to check reduction validity for outer loop
25709 2017-06-09 Richard Biener <rguenther@suse.de>
25711 PR tree-optimization/79483
25712 * graphite-scop-detection.c (order): New global.
25713 (get_order): Compute bb to order mapping that satisfies code
25714 generation constraints.
25715 (cmp_pbbs): New helper.
25716 (build_scops): Start domwalk at entry block, sort generated
25719 2017-06-09 Richard Biener <rguenther@suse.de>
25721 PR middle-end/81007
25722 * ipa-polymorphic-call.c
25723 (ipa_polymorphic_call_context::restrict_to_inner_class):
25724 Skip FIELD_DECLs with error_mark_node type.
25725 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25728 2017-06-09 Martin Liska <mliska@suse.cz>
25730 * predict.c (struct branch_predictor): New struct.
25731 (test_prediction_value_range): New test.
25732 (predict_c_tests): New function.
25733 * selftest-run-tests.c (selftest::run_tests): Run the function.
25734 * selftest.h: Declare new tests.
25736 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25739 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25740 gen_add3_insn did not fail.
25741 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25742 r0, construct that number in a temporary reg and add that reg to r0.
25743 If asked to put the result in r0 as well, fail.
25745 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25748 for early expansion of vec_eqv.
25750 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25752 PR middle-end/81005
25753 * ubsan.c (instrument_null): Avoid pointless code temporary.
25754 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25756 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25760 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25763 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25768 * predict.c (unlikely_executed_edge_p): Move ahead.
25769 (probably_never_executed_edge_p): Use it.
25771 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25773 PR middle-end/79988
25774 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25775 gimple_call_builtin_p call.
25777 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25779 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25780 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25781 rtl_check_failed_type2, rtl_check_failed_code1,
25782 rtl_check_failed_code2, rtl_check_failed_code_mode,
25783 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25784 rtvec_check_failed_bounds, rtl_check_failed_flag,
25785 _fatal_insn_not_found, _fatal_insn): Likewise.
25786 * tree.h (tree_contains_struct_check_failed,
25787 tree_check_failed, tree_not_check_failed,
25788 tree_class_check_failed, tree_range_check_failed,
25789 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25790 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25791 tree_operand_check_failed, omp_clause_check_failed,
25792 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25795 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25797 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25798 flag_branch_probabilities.
25799 * ipa-inline.c (edge_badness): Likewise.
25800 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25801 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25802 * predict.c (maybe_hot_frequency_p): Likewise.
25803 (probably_never_executed): Likewise.
25804 * sched-ebb.c (schedule_ebbs): Likewise.
25805 * sched-rgn.c (find_single_block_region): Likewise.
25806 * tracer.c (tail_duplicate): Likewise.
25808 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25810 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25811 longer requires x_flag_profile_use.
25813 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25815 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25816 instead of flag_reorder_blocks_and_partition.
25817 * dbxout.c (dbxout_function_end): Likewise.
25818 * dwarf2out.c (gen_subprogram_die): Likewise.
25819 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25820 * hw-doloop.c (reorg_loops): Likewise.
25821 * varasm.c (assemble_start_function,
25822 assemble_end_function): Likewise.
25823 (decide_function_section): Do not check for
25824 flag_reorder_blocks_and_partition.
25826 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25828 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25830 (chkp_get_hard_register_fake_addr_expr): Ditto.
25831 (chkp_build_addr_expr): Add check for hard reg case.
25832 (chkp_parse_array_and_component_ref): Ditto.
25833 (chkp_find_bounds_1): Ditto.
25834 (chkp_process_stmt): Don't generate bounds store for
25837 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25839 * predict.c (maybe_hot_bb_p): Do not check profile status.
25840 (maybe_hot_edge_p): Likewise.
25841 (probably_never_executed): Check for zero counts even if profile
25843 (unlikely_executed_edge_p): New function.
25844 (unlikely_executed_stmt_p): New function.
25845 (unlikely_executed_bb_p): New function.
25846 (set_even_probabilities): Use unlikely predicates.
25847 (combine_predictions_for_bb): Likewise.
25848 (predict_paths_for_bb): Likewise.
25849 (predict_paths_leading_to_edge): Likewise.
25850 (determine_unlikely_bbs): New function.
25851 (estimate_bb_frequencies): Use it.
25852 (compute_function_frequency): Use zero counts even if profile is
25854 * profile-count.h: Fix typo.
25856 2017-08-08 Julia Koval <julia.koval@intel.com>
25858 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25859 _mm512_mask_cvtsepi16_storeu_epi8,
25860 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25861 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25862 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25863 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25864 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25865 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25866 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25867 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25868 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25869 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25870 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25871 __builtin_ia32_pmovuswb256mem_mask,
25872 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25873 __builtin_ia32_pmovwb512mem_mask): New builtins.
25875 2017-08-08 Julia Koval <julia.koval@intel.com>
25877 PR target/73350,80862
25878 * config/i386/subst.md (round): Fix round pattern.
25879 * config/i386/i386.c (ix86_erase_embedded_rounding):
25880 Fix erasing rounding for the fixed pattern.
25882 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25884 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25886 2017-06-08 Martin Liska <mliska@suse.cz>
25888 PR gcov-profile/80911
25889 * gcov.c (block_info::block_info): New constructor.
25891 2017-06-07 Carl Love <cel@us.ibm.com>
25893 * config/rs6000/rs6000-c: The return type of the following
25894 built-in functions was implemented as int not long long. Fix sign
25895 of return value for the unsigned version of vec_mulo and vec_mule.
25896 vector unsigned long long vec_bperm (vector unsigned long long,
25897 vector unsigned char)
25898 vector signed long long vec_mule (vector signed int,
25900 vector unsigned long long vec_mule (vector unsigned int,
25901 vector unsigned int)
25902 vector signed long long vec_mulo (vector signed int,
25904 vector unsigned long long vec_mulo (vector unsigned int,
25905 vector unsigned int)
25906 * doc/extend.texi: Fix the documentation for the built-in
25909 2017-06-07 Carl Love <cel@us.ibm.com>
25912 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25915 2017-06-07 Carl Love <cel@us.ibm.com>
25917 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25918 support, Generate doublehv for signed int/float for BE case only.
25920 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25922 * doc/invoke.texi (mcx16): Rewrite.
25924 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25926 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25927 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25928 *mov<mode>_softfloat, and an anonymous splitter): Use
25929 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25931 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25933 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25935 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25936 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25937 (rs6000_debug_reg_global): Adjust.
25938 (rs6000_init_hard_regno_mode_ok): Adjust.
25939 (rs6000_dbx_register_number): Adjust.
25940 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25941 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25942 Remove SPE_ACC and SPEFSCR.
25943 (REG_ALLOC_ORDER): Ditto.
25944 (FRAME_POINTER_REGNUM): Change to 111.
25945 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25946 (REG_CLASS_NAMES): Ditto.
25947 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25948 (REGISTER_NAMES): Ditto.
25949 (ADDITIONAL_REG_NAMES): Ditto.
25950 (rs6000_reg_names): Ditto.
25951 * config/rs6000/rs6000.md: Renumber some register number
25954 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25956 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25958 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25959 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25961 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25962 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25963 Delete the SPE high registers.
25964 (REG_ALLOC_ORDER): Ditto.
25965 (enum reg_class): Remove SPE_HIGH_REGS.
25966 (REG_CLASS_NAMES): Ditto.
25967 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25968 (REGISTER_NAMES): Ditto.
25969 (rs6000_reg_names): Ditto.
25970 * doc/tm.texi.in: Remove SPE as example.
25971 * doc/tm.texi: Regenerate.
25973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25975 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25976 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25977 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25978 * config/rs6000/rs6000.md (type): Remove "brinc".
25980 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25982 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25983 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25984 * config/rs6000/linuxspe.h: Delete file.
25985 * config/rs6000/rs6000.md: Don't include spe.md.
25986 * config/rs6000/spe.h: Delete file.
25987 * config/rs6000/spe.md: Delete file.
25988 * config/rs6000/t-rs6000: Remove spe.md.
25990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25992 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25993 (reg_or_none500mem_operand): Delete.
25994 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25995 instead of reg_or_none500mem_operand.
25997 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25999 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26000 handling of SPE flags.
26001 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26003 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26005 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26007 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26008 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26009 paired_divv2sf3): Similar.
26010 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26011 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26012 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26014 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26015 Rename the paired_* instruction patterns.
26016 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26018 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26019 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26020 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26021 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26022 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26023 PAIRED_VECTOR_MODE.
26024 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26025 (spe_func_has_64bit_regs_p): Delete.
26026 (spe_expand_predicate_builtin): Delete.
26027 (spe_expand_evsel_builtin): Delete.
26028 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26029 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26030 (invalid_e500_subreg): Delete.
26031 (rs6000_legitimize_address): Always force_reg op2 as well, for
26032 paired single memory accesses.
26033 (rs6000_member_type_forces_blk): Delete.
26034 (rs6000_spe_function_arg): Delete.
26035 (rs6000_expand_unop_builtin): Delete SPE handling.
26036 (rs6000_expand_binop_builtin): Ditto.
26037 (spe_expand_stv_builtin): Delete.
26038 (bdesc_2arg_spe): Delete.
26039 (spe_expand_builtin): Delete.
26040 (spe_expand_predicate_builtin): Delete.
26041 (spe_expand_evsel_builtin): Delete.
26042 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26043 (spe_init_builtins): Delete.
26044 (spe_func_has_64bit_regs_p): Delete.
26045 (savres_routine_name): Delete "info" parameter. Adjust callers.
26046 (rs6000_emit_stack_reset): Ditto.
26047 (rs6000_dwarf_register_span): Delete.
26048 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26049 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26050 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26052 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26054 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26055 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26056 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26057 mulv2sf3, divv2sf3): Delete expanders.
26059 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26061 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26065 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26066 * config/rs6000/rs6000.c: Ditto.
26068 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26070 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26071 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26072 comparison_operator.
26074 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26076 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26077 * config/rs6000/rs6000.opt: Ditto.
26078 * config/rs6000/t-rtems: Ditto.
26080 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26082 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26083 TARGET_E500_SINGLE by 0, simplify.
26084 * config/rs6000/rs6000.c: Ditto.
26085 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26086 (spe_build_register_parallel): Delete.
26087 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26088 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26089 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26090 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26091 (E500_CONVERT): Delete.
26092 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26094 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26096 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26097 * config/rs6000/dfp.md: Ditto.
26098 (negdd2, *negdd2_fpr): Merge.
26099 (absdd2, *absdd2_fpr): Merge.
26100 (negtd2, *negtd2_fpr): Merge.
26101 (abstd2, *abstd2_fpr): Merge.
26102 * config/rs6000/e500.h: Delete file.
26103 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26104 TARGET_FPRS by 1 and simplify.
26105 * config/rs6000/rs6000-c.c: Ditto.
26106 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26107 TARGET_DF_SPE by 0.
26108 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26110 * config/rs6000/rs6000.md: Ditto.
26111 (floatdidf2, *floatdidf2_fpr): Merge.
26112 (move_from_CR_gt_bit): Delete.
26113 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26114 (E500_CR_IOR_COMPARE): Delete.
26115 (All patterns that require !TARGET_FPRS): Delete.
26116 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26118 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26120 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26124 * graphds.c (add_edge): Intitialize edge's attached data.
26125 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26126 pointer parameter. Call pointed function on each edge during
26127 graph traversing. Skip traversing the edge when the function
26129 (graphds_dfs, graphds_scc): Ditto.
26130 (for_each_edge): New parameter. Pass the new parameter to callback
26132 * graphds.h (skip_edge_callback): New function pointer type.
26133 (graphds_dfs, graphds_scc): New function pointer parameter.
26134 (graphds_edge_callback, for_each_edge): New parameter.
26136 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26138 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26139 out code checking if runtime alias check is possible to below ...
26140 Call the new function.
26141 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26142 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26144 2017-06-07 Marek Polacek <polacek@redhat.com>
26147 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26148 TYPE_OVERFLOW_WRAPS check.
26150 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26152 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26153 if versioning is required.
26154 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26155 peeling with the check for versioning.
26157 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26159 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26160 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26161 Set true to new parameter if new ssa variable is defined.
26162 (vect_gen_vector_loop_niters): Refactor. Set range information
26163 for the new vector loop bound variable.
26164 (vect_do_peeling): Ditto.
26166 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26168 * tree-affine.c (ssa.h): Include header file.
26169 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26170 has wrapping overflow behavior.
26172 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26174 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26178 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26179 (tree_to_aff_combination): ... here.
26181 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26183 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26184 reg_pressure model function.
26185 (ivopts_global_cost_for_size): Delete.
26186 (determine_set_costs, iv_ca_recount_cost): Call new model function
26187 ivopts_estimate_reg_pressure.
26189 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26191 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26192 expensive than udiv. Remove floating point cases from mod.
26194 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26196 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26197 Increase idiv cost.
26199 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26201 * config/aarch64/aarch64.md
26202 (copysignsf3): Fix mask generation.
26204 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26206 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26208 (class dump_manager): Add register_dumps method.
26209 * dumpfile.c: Include langhooks.h.
26210 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26211 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26212 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26213 (dump_manager::dump_register): Start with 512 entries instead of 32.
26214 (dump_manager::register_dumps): New method.
26215 * toplev.c (general_init): Instead of invoking register_dumps
26216 langhook, invoke register_dumps method on the dump manager.
26217 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26220 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26222 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26223 Say that # requires an associated define_split to exist, and that
26224 the define_split must be suitable for use after register allocation.
26226 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26228 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26229 (compute_outgoing_frequencies): Also initialize zero counts.
26230 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26231 around loops; preserve more of profile when nothing changes.
26233 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26235 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26237 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26238 * config/arm/arm-cpu-cdata.h: Regenerate.
26239 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26240 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26241 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26242 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26243 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26245 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26247 * config/arm/t-rmprofile: Likewise.
26248 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26250 2017-06-06 David S. Miller <davem@davemloft.net>
26253 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26254 blockage if function uses alloca.
26256 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26258 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26259 New "uid" fields to hold pretty-print IDs of group and ref.
26260 Memory references are now identified as <group_id>:<ref_id>
26261 instead of using [random] addresses.
26262 (dump_mem_details): Simplify, no functional change.
26263 (dump_mem_ref): Simplify and make output more concise.
26264 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26265 (find_or_create_group): Initialize group uid.
26266 (record_ref): Initialize ref uid. Improve debug output.
26267 (prune_group_by_reuse, should_issue_prefetch_p,)
26268 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26269 (mark_nontemporal_store, determine_loop_nest_reuse):
26270 Improve debug output.
26272 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26274 * dbgcnt.def (prefetch): New debug counter.
26275 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26276 (schedule_prefetches): Stop issueing prefetches if debug counter
26279 2017-06-06 Tom de Vries <tom@codesourcery.com>
26281 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26282 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26284 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26286 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26287 Use aarch64_reg_or_zero predicate for operand 4.
26288 (aarch64_compare_and_swap<mode> define_insn_and_split):
26289 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26290 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26292 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26294 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26295 (arm_compute_save_core_reg_mask): This.
26296 (thumb1_compute_save_reg_mask): Rename into ...
26297 (thumb1_compute_save_core_reg_mask): This.
26298 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26299 (arm_compute_frame_layout): Likewise.
26301 2017-06-06 Richard Biener <rguenther@suse.de>
26303 PR tree-optimization/80974
26304 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26305 keep or clear leaders SSA info.
26307 2017-06-06 Tom de Vries <tom@codesourcery.com>
26309 * config/nvptx/nvptx.c (split_mode_p): New function.
26310 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26312 2017-06-06 Tom de Vries <tom@codesourcery.com>
26314 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26316 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26319 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26322 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26324 * shrink-wrap.c (handle_simple_exit): Update profile.
26325 (try_shrink_wrapping): Upate profile.
26327 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26329 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26330 (tree_guess_outgoing_edge_probabilities): New.
26331 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26332 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26334 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26336 * ipa-split.c (split_function): Initialize return bb profile.
26338 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26340 * profile.c (compute_branch_probabilities): Also initialize
26341 EXIT_BLOCK profile.
26343 2017-06-06 Richard Biener <rguenther@suse.de>
26345 PR tree-optimization/80928
26346 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26347 (vect_analyze_loop_operations): Properly guard analysis for
26349 (vect_transform_loop): Likewise.
26350 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26351 (vect_model_induction_cost): Do not cost for pure SLP.
26352 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26353 of induction in inner loop vectorization.
26354 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26355 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26356 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26358 (vect_analyze_slp_cost_1): Cost induction.
26359 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26360 (vect_get_slp_vect_defs): Likewise.
26361 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26362 (vect_transform_stmt): Handle SLP reductions.
26363 * tree-vectorizer.h (vectorizable_induction): Adjust.
26365 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26367 * config/rs6000/rs6000.c (make_resolver_func): Update
26368 init_lowered_empty_function call.
26370 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26372 * doc/invoke.texi: Document the -fprofile-abs-path option.
26373 * common.opt (fprofile-abs-path): New option.
26374 * gcov-io.h (gcov_write_filename): Declare.
26375 * gcov-io.c (gcov_write_filename): New function.
26376 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26377 * profile.c (output_location): Likewise.
26379 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26381 * shring-wrap.c: Revert accidental commit.
26383 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26385 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26387 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26389 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26391 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26392 profile in callgraph edge.
26393 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26394 (apply_scale): Likewise.
26395 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26396 Also scale profile when inlining function with zero profile.
26397 (initialize_cfun): Update exit block profile even when it is zero.
26398 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26399 when profile is read.
26401 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26403 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26404 (CLONE_*): New constants to define the processors we can generate
26405 code for with the target_clone attribute.
26406 (rs6000_clone_map): New array to identify which clone processors
26407 the current program is running on.
26408 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26409 target_clone attribute.
26410 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26411 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26412 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26413 (cpu_expand_builtin): Add support for target_clone attribute.
26414 (rs6000_valid_attribute_p): Allow "default" attribute.
26415 (get_decl_name): New debug function to simplify printing the
26416 current function name in debugging statements.
26417 (rs6000_clone_priority): New functions to support the target_clone
26418 attribute, and be able to generate code to switch between ISA 2.05
26419 through ISA 3.0 (power6 through power9).
26420 (rs6000_compare_version_priority): Likewise.
26421 (rs6000_get_function_versions_dispatcher): Likewise.
26422 (make_resolver_func): Likewise.
26423 (add_condition_to_bb): Likewise.
26424 (dispatch_function_versions): Likewise.
26425 (rs6000_generate_version_dispatcher_body): Likewise.
26426 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26427 (fusion_gpr_load_p): Fix a spacing issue.
26428 * doc/extend.texi (Common Function Attributes): Document that the
26429 PowerPC supports the target_clone attribute.
26431 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26433 * config/arm/arm.h: explain F symbol found in description of ARM
26434 register allocation in its legend.
26436 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26438 * config/mips/frame-header-opt.c: Include profile-count.h.
26439 * config/riscv/riscv.c: Include profile-count.h
26441 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26443 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26445 (sm_set_flag_if_changed): Add bbs field.
26446 (execute_sm_if_changed_flag_set): Pass BBS.
26447 (execute_sm): Update.
26449 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26451 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26454 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26456 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26457 (peephole2): New peephole2 to emit the above.
26458 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26460 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26462 * config/aarch64/aarch64.c (define_peephole2 above
26463 *sub_<shift>_<mode>): New peephole.
26465 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26467 * config/i386/i386.c (make_resolver_func): Update.
26468 * Makefile.in: Add profile-count.h and profile-count.o
26469 * auto-profile.c (afdo_indirect_call): Update to new API.
26470 (afdo_set_bb_count): Update.
26471 (afdo_propagate_edge): Update.
26472 (afdo_propagate_circuit): Update.
26473 (afdo_calculate_branch_prob): Update.
26474 (afdo_annotate_cfg): Update.
26475 * basic-block.h: Include profile-count.h
26476 (struct edge_def): Turn count to profile_count.
26477 (struct basic_block_def): Likewie.
26478 (REG_BR_PROB_BASE): Move to profile-count.h
26479 (RDIV): Move to profile-count.h
26480 * bb-reorder.c (max_entry_count): Turn to profile_count.
26481 (find_traces): Update.
26482 (rotate_loop):Update.
26483 (connect_traces):Update.
26484 (sanitize_hot_paths):Update.
26485 * bt-load.c (migrate_btr_defs): Update.
26486 * cfg.c (RDIV): Remove.
26487 (init_flow): Use alloc_block.
26488 (alloc_block): Uninitialize count.
26489 (unchecked_make_edge): Uninitialize count.
26490 (check_bb_profile): Update.
26491 (dump_edge_info): Update.
26492 (dump_bb_info): Update.
26493 (update_bb_profile_for_threading): Update.
26494 (scale_bbs_frequencies_int): Update.
26495 (scale_bbs_frequencies_gcov_type): Update.
26496 (scale_bbs_frequencies_profile_count): New.
26497 * cfg.h (update_bb_profile_for_threading): Update.
26498 (scale_bbs_frequencies_profile_count): Declare.
26499 * cfgbuild.c (compute_outgoing_frequencies): Update.
26500 (find_many_sub_basic_blocks): Update.
26501 * cfgcleanup.c (try_forward_edges): Update.
26502 (try_crossjump_to_edge): Update.
26503 * cfgexpand.c (expand_gimple_tailcall): Update.
26504 (construct_exit_block): Update.
26505 * cfghooks.c (verify_flow_info): Update.
26506 (dump_bb_for_graph): Update.
26507 (split_edge): Update.
26508 (make_forwarder_block): Update.
26509 (duplicate_block): Update.
26510 (account_profile_record): Update.
26511 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26512 (get_estimated_loop_iterations): Update.
26513 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26514 (single_likely_exit): Update.
26515 * cfgloopmanip.c (scale_loop_profile): Update.
26517 (set_zero_probability): Update.
26518 (lv_adjust_loop_entry_edge): Update.
26519 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26520 (purge_dead_edges): Update.
26521 (rtl_account_profile_record): Update.
26522 * cgraph.c (cgraph_node::create): Uninitialize count.
26523 (symbol_table::create_edge): Uninitialize count.
26524 (cgraph_update_edges_for_call_stmt_node): Update.
26525 (cgraph_edge::dump_edge_flags): Update.
26526 (cgraph_node::dump): Update.
26527 (cgraph_edge::maybe_hot_p): Update.
26528 * cgraph.h: Include profile-count.h
26529 (create_clone), create_edge, create_indirect_edge): Update.
26530 (cgraph_node): Turn count to profile_count.
26531 (cgraph_edge0: Likewise.
26532 (make_speculative, clone): Update.
26533 (create_edge): Update.
26534 (init_lowered_empty_function): Update.
26535 * cgraphclones.c (cgraph_edge::clone): Update.
26536 (duplicate_thunk_for_node): Update.
26537 (cgraph_node::create_clone): Update.
26538 * cgraphunit.c (cgraph_node::analyze): Update.
26539 (cgraph_node::expand_thunk): Update.
26540 * final.c (dump_basic_block_info): Update.
26541 * gimple-streamer-in.c (input_bb): Update.
26542 * gimple-streamer-out.c (output_bb): Update.
26543 * graphite.c (print_global_statistics): Update.
26544 (print_graphite_scop_statistics): Update.
26545 * hsa-brig.c: Include basic-block.h.
26546 * hsa-dump.c: Include basic-block.h.
26547 * hsa-gen.c (T sum_slice): Update.
26548 (convert_switch_statements):Update.
26549 * hsa-regalloc.c: Include basic-block.h.
26550 * ipa-chkp.c (chkp_produce_thunks): Update.
26551 * ipa-cp.c (struct caller_statistics): Update.
26552 (init_caller_stats): Update.
26553 (gather_caller_stats): Update.
26554 (ipcp_cloning_candidate_p): Update.
26555 (good_cloning_opportunity_p): Update.
26556 (get_info_about_necessary_edges): Update.
26557 (dump_profile_updates): Update.
26558 (update_profiling_info): Update.
26559 (update_specialized_profile): Update.
26560 (perhaps_add_new_callers): Update.
26561 (decide_about_value): Update.
26562 (ipa_cp_c_finalize): Update.
26563 * ipa-devirt.c (struct odr_type_warn_count): Update.
26564 (struct decl_warn_count): Update.
26565 (struct final_warning_record): Update.
26566 (possible_polymorphic_call_targets): Update.
26567 (ipa_devirt): Update.
26568 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26569 * ipa-icf.c (sem_function::merge): Update.
26570 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26571 * ipa-inline.c (compute_uninlined_call_time): Update.
26572 (compute_inlined_call_time): Update.
26573 (want_inline_small_function_p): Update.
26574 (want_inline_self_recursive_call_p): Update.
26575 (edge_badness): Update.
26576 (lookup_recursive_calls): Update.
26577 (recursive_inlining): Update.
26578 (inline_small_functions): Update.
26579 (dump_overall_stats): Update.
26580 (dump_inline_stats): Update.
26581 * ipa-profile.c (ipa_profile_generate_summary): Update.
26582 (ipa_propagate_frequency): Update.
26583 (ipa_profile): Update.
26584 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26585 * ipa-utils.c (ipa_merge_profiles): Update.
26586 * loop-doloop.c (doloop_modify): Update.
26587 * loop-unroll.c (report_unroll): Update.
26588 (unroll_loop_runtime_iterations): Update.
26589 * lto-cgraph.c (lto_output_edge): Update.
26590 (lto_output_node): Update.
26591 (input_node): Update.
26592 (input_edge): Update.
26593 (merge_profile_summaries): Update.
26594 * lto-streamer-in.c (input_cfg): Update.
26595 * lto-streamer-out.c (output_cfg): Update.
26596 * mcf.c (create_fixup_graph): Update.
26597 (adjust_cfg_counts): Update.
26598 (sum_edge_counts): Update.
26599 * modulo-sched.c (sms_schedule): Update.
26600 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26601 * predict.c (maybe_hot_count_p): Update.
26602 (probably_never_executed): Update.
26603 (dump_prediction): Update.
26604 (combine_predictions_for_bb): Update.
26605 (propagate_freq): Update.
26606 (handle_missing_profiles): Update.
26607 (counts_to_freqs): Update.
26608 (rebuild_frequencies): Update.
26609 (force_edge_cold): Update.
26610 * predict.h: Include profile-count.h
26611 (maybe_hot_count_p, counts_to_freqs): UPdate.
26612 * print-rtl-function.c: Do not include cfg.h
26613 * print-rtl.c: Include basic-block.h
26614 * profile-count.c: New file.
26615 * profile-count.h: New file.
26616 * profile.c (is_edge_inconsistent): Update.
26617 (correct_negative_edge_counts): Update.
26618 (is_inconsistent): Update.
26619 (set_bb_counts): Update.
26620 (read_profile_edge_counts): Update.
26621 (compute_frequency_overlap): Update.
26622 (compute_branch_probabilities): Update; Initialize and deinitialize
26624 (branch_prob): Update.
26625 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26626 (edge_gcov_count): New.
26627 (bb_gcov_count): New.
26628 * shrink-wrap.c (try_shrink_wrapping): Update.
26629 * tracer.c (better_p): Update.
26630 * trans-mem.c (expand_transaction): Update.
26631 (ipa_tm_insert_irr_call): Update.
26632 (ipa_tm_insert_gettmclone_call): Update.
26633 * tree-call-cdce.c: Update.
26634 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26635 (gimple_duplicate_sese_tail): Update.
26636 (gimple_account_profile_record): Update.
26637 (execute_fixup_cfg): Update.
26638 * tree-inline.c (copy_bb): Update.
26639 (copy_edges_for_bb): Update.
26640 (initialize_cfun): Update.
26641 (freqs_to_counts): Update.
26642 (copy_cfg_body): Update.
26643 (expand_call_inline): Update.
26644 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26645 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26646 (try_unroll_loop_completely): Update.
26647 (try_peel_loop): Update.
26648 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26649 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26650 * tree-ssa-loop-split.c (connect_loops): Update.
26651 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26652 * tree-ssa-reassoc.c (branch_fixup): Update.
26653 * tree-ssa-tail-merge.c (replace_block_by): Update.
26654 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26655 (compute_path_counts): Update.
26656 (update_profile): Update.
26657 (recompute_probabilities): Update.
26658 (update_joiner_offpath_counts): Update.
26659 (estimated_freqs_path): Update.
26660 (freqs_to_counts_path): Update.
26661 (clear_counts_path): Update.
26662 (ssa_fix_duplicate_block_edges): Update.
26663 (duplicate_thread_path): Update.
26664 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26665 (struct switch_conv_info): Update.
26666 * tree-tailcall.c (decrease_profile): Update.
26667 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26668 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26669 * value-prof.c (check_counter): Update.
26670 (gimple_divmod_fixed_value): Update.
26671 (gimple_mod_pow2): Update.
26672 (gimple_mod_subtract): Update.
26673 (gimple_ic_transform): Update.
26674 (gimple_stringop_fixed_value): Update.
26675 * value-prof.h (gimple_ic): Update.
26677 2017-06-02 Carl Love <cel@us.ibm.com>
26679 * config/rs6000/rs6000-c: Add support for built-in functions
26680 vector double vec_doublee (vector signed int);
26681 vector double vec_doublee (vector unsigned int);
26682 vector double vec_doublee (vector float);
26683 vector double vec_doubleh (vector signed int);
26684 vector double vec_doubleh (vector unsigned int);
26685 vector double vec_doubleh (vector float);
26686 vector double vec_doublel (vector signed int);
26687 vector double vec_doublel (vector unsigned int);
26688 vector double vec_doublel (vector float);
26689 vector double vec_doubleo (vector signed int);
26690 vector double vec_doubleo (vector unsigned int);
26691 vector double vec_doubleo (vector float);.
26692 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26693 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26695 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26696 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26697 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26699 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26700 vec_doublel, vec_doubleh.
26701 * doc/extend.texi: Update the built-in documentation file for the
26702 new built-in functions.
26704 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26707 * ipa-inline-analysis.c (free_growth_caches): Set
26708 edge_removal_hook_holder to NULL after removing it.
26710 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26712 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26713 comparision with zero.
26715 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26717 for early expansion of vec_min and vec_max builtins.
26718 (builtin_function_type): Add min/max unsigned variants to those
26719 identified as having unsigned arguments.
26721 2017-06-02 Olivier Hainque <hainque@adacore.com>
26723 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26725 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26727 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26728 Use VALL_F16 iterator rather than VALL.
26730 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26732 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26733 Emit CBNZ inside loop when doing a strong exchange and comparing
26734 against zero. Generate the CC flags after the loop.
26736 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26738 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26739 (dl_section_ref): New.
26740 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26741 On AIX, append an expression to subtract the size of the
26742 section length to dl_section_ref.
26744 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26747 for early expansion of vector absolute builtins.
26749 2017-06-02 Richard Biener <rguenther@suse.de>
26751 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26752 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26754 2017-06-02 Richard Biener <rguenther@suse.de>
26756 PR tree-optimization/80948
26757 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26758 stmt order as well.
26760 2017-06-02 Richard Biener <rguenther@suse.de>
26762 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26764 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26765 for inductions as relevant.
26767 2017-06-02 Richard Biener <rguenther@suse.de>
26769 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26770 (vectorizable_induction): ... this. Remove dead code.
26772 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26774 * builtins. (expand_builtin_alloca): Remove second parameter and
26775 infer its value from the first parameter instead.
26776 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26778 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26780 PR rtl-optimization/80903
26781 * loop-doloop.c (add_test): Unshare sequence.
26783 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26785 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26787 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26789 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26791 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26792 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26793 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26794 (xlogue_layout::get_stub_rtx): Make static.
26795 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26796 (xlogue_layout::compute_stub_managed_regs): Rename to...
26797 (xlogue_layout::count_stub_managed_regs): ...this.
26798 (xlogue_layout::is_stub_managed_reg): New function.
26799 (xlogue_layout::m_stub_names): Rename to...
26800 (xlogue_layout::s_stub_names): ...this, make static.
26801 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26802 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26803 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26804 xlogue_layout::s_stub_names): Instantiate statics.
26805 (stub_managed_regs): Remove.
26806 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26807 (disable_call_ms2sysv_xlogues): Rename to...
26808 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26809 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26811 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26812 change after reload_completed.
26813 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26815 (ix86_expand_prologue): Likewise.
26816 (ix86_expand_epilogue): Likewise.
26817 (ix86_expand_split_stack_prologue): Likewise.
26818 (ix86_compute_frame_layout): Remove frame parameter ...
26819 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26820 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26822 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26823 (ix86_frame): Move from here ...
26824 * config/i386/i386.h (ix86_frame): ... to here.
26825 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26826 complete ix86_frame data structure instead. Remove some_ld_name.
26828 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26830 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26831 symbols that hold a DECL_VALUE_EXPR.
26833 2017-06-01 Martin Jambor <mjambor@suse.cz>
26835 PR tree-optimization/80898
26836 * tree-sra.c (process_subtree_disqualification): Removed.
26837 (disqualify_candidate): Do not acll
26838 process_subtree_disqualification.
26839 (subtree_mark_written_and_enqueue): New function.
26840 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26841 RHS has been disqualified and re-queue LHS if necessary. Apart
26842 from that, ignore disqualified RHS.
26844 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26846 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26847 address fetch for z10 or later.
26849 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26851 * config/arc/arc.md (tst_movb): Add guard when splitting.
26853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26855 * config/arc/arc.c (arc_can_eliminate): Test against
26856 arc_frame_pointer_needed.
26858 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26860 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26861 to prevent store reordering.
26862 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26863 (type): Add block type.
26864 (stack_tie): Define special instruction to be used in
26867 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26869 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26870 constraint. It is not valid for the pattern.
26871 (noncommutative_binary_comparison): Likewise.
26873 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26875 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26878 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26880 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26881 be used by the reg-alloc.
26883 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26885 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26886 reg-alloc when having mul64 or mul32x16 instructions.
26887 (mulsidi3): Likewise.
26888 (umulsidi3): Likewise.
26889 (mulsi32x16): New pattern.
26890 (mulsi64): Likewise.
26891 (mulsidi64): Likewise.
26892 (umulsidi64): Likewise.
26893 (MUL32x16_REG): Define.
26894 (mul64_600): Use MUL32x16_REG.
26895 (mac64_600): Likewise.
26896 (umul64_600): Likewise.
26897 (umac64_600): Likewise.
26899 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26901 * config/arc/arc.md (mulsi3_700): Make it commutative.
26903 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26905 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26907 (*sign_extendsidi2_insn): Likewise for movstosw.
26909 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26911 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26912 the type of the input discriminant value. Convert the
26913 discriminant value of signedness vary.
26915 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26917 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26918 Add to -Wall section.
26920 2017-06-01 Richard Biener <rguenther@suse.de>
26922 PR middle-end/66313
26923 * fold-const.c (fold_plusminus_mult_expr): If the factored
26924 factor may be zero use a wrapping type for the inner operation.
26925 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26926 and handle moved defs.
26927 (process_assignment): Properly guard the unary op case. Return a
26928 tri-state indicating that moving the stmt before the call may allow
26929 to continue. Pass through to_move.
26930 (find_tail_calls): Handle moving unrelated defs before
26933 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26936 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26937 splitter result in the canonical way.
26939 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26941 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26942 also for 32bit target. Update insn attributes.
26943 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26945 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26947 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26948 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26949 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26950 (_mm_maskz_min_ss): New intrinsics.
26952 2017-05-31 Martin Liska <mliska@suse.cz>
26954 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26955 Change comment style to one we normally use.
26956 (vectorizable_reduction): Likewise.
26957 (vectorizable_induction): Likewise.
26958 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26959 (vectorizable_call): Likewise.
26960 (vectorizable_simd_clone_call): Likewise.
26961 (vectorizable_conversion): Likewise.
26962 (vectorizable_assignment): Likewise.
26963 (vectorizable_shift): Likewise.
26964 (vectorizable_operation): Likewise.
26965 (vectorizable_store): Likewise.
26966 (vectorizable_load): Likewise.
26967 * tree-vectorizer.h: Likewise.
26969 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26971 * passes.c (emergency_dump_function): New.
26972 * tree-pass.h (emergency_dump_function): Declare.
26973 * plugin.c (plugins_internal_error_function): Remove.
26974 * plugin.h (plugins_internal_error_function): Remove declaration.
26975 * toplev.c (internal_error_function): New static function. Use it...
26976 (general_init): ...here.
26978 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26980 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26981 (arc_rtx_costs): Add costs for new patterns.
26982 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26983 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26985 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26987 * tree-ssa-strlen.c (get_next_strinfo): New function.
26988 (get_stridx_plus_constant): Use it.
26989 (zero_length_string): Likewise.
26990 (adjust_related_strinfos): Likewise.
26991 (adjust_last_stmt): Likewise.
26993 2017-05-31 Richard Biener <rguenther@suse.de>
26996 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26997 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26999 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27001 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27002 loop_vinfo argument and use of dependence distance vectors.
27003 Check instead whether the two references differ only in their
27004 initial value and assume that they have the same alignment if the
27005 difference is a multiple of the vector alignment.
27006 (vect_analyze_data_refs_alignment): Update call accordingly.
27008 2017-05-31 Martin Liska <mliska@suse.cz>
27011 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27013 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27015 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27016 (create_intersect_range_checks): Move from ...
27017 * tree-data-ref.c (create_intersect_range_checks_index)
27018 (create_intersect_range_checks): ... to here.
27019 (create_runtime_alias_checks): New function factored from ...
27020 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27021 here. Call above function.
27022 * tree-data-ref.h (create_runtime_alias_checks): New function.
27024 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27026 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27027 segment length for dr_b and compute it in wide_int.
27029 2017-05-31 Richard Biener <rguenther@suse.de>
27031 PR tree-optimization/80906
27032 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27033 and pass through iv_map.
27034 (copy_bb_and_scalar_dependences): Adjust.
27035 (translate_pending_phi_nodes): Likewise.
27036 (copy_loop_close_phi_args): Handle code-generating IVs instead
27039 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27041 * diagnostic-color.c (color_dict): Add "type-diff".
27042 (parse_gcc_colors): Update comment.
27043 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27044 -fdiagnostics-show-template-tree and -fno-elide-type.
27045 (GCC_COLORS): Add type-diff to example.
27047 (-fdiagnostics-show-template-tree): New.
27048 (-fno-elide-type): New.
27049 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27050 the pp_format_decoder callback. Call any m_format_postprocessor's
27052 (pretty_printer::pretty_printer): Initialize
27053 m_format_postprocessor.
27054 (pretty_printer::~pretty_printer): Delete any
27055 m_format_postprocessor.
27056 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27057 (class format_postprocessor): New class.
27058 (struct pretty_printer::format_decoder): Document the new parameters.
27059 (struct pretty_printer::m_format_postprocessor): New field.
27060 * tree-diagnostic.c (default_tree_printer): Update for new
27061 bool and const char ** params.
27062 * tree-diagnostic.h (default_tree_printer): Likewise.
27064 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27066 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27067 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27068 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27069 rs6000_gen_cell_microcode code.
27070 (rs6000_final_prescan_insn): Delete.
27071 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27072 "warn-cell-microcode" entries.
27073 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27074 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27075 throughout. Change cc_reg_not_micro_cr0_operand to
27076 cc_reg_not_cr0_operand throughout.
27077 (*extendhi<mode>2_noload): Delete.
27078 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27079 (mwarn-cell-microcode): Delete.
27080 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27081 -mgen-cell-microcode and -mwarn-cell-microcode.
27083 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27086 * config/i386/constraints.md (Yd): New constraint.
27088 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27089 and (?Yd, r) alternatives. Update insn attributes.
27090 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27091 and (?*Yd, r) alternatives. Update insn attributes.
27092 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27094 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27096 * gimplify.c (gimplify_modify_expr): Don't create a
27097 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27100 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27102 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27104 2017-05-30 Richard Biener <rguenther@suse.de>
27106 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27107 and reduc_def fields.
27108 (STMT_VINFO_REDUC_TYPE): New define.
27109 (STMT_VINFO_REDUC_DEF): Likewise.
27110 (vect_force_simple_reduction): Adjust prototype.
27111 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27112 (vect_is_simple_reduction): Remove check_reduction argument.
27113 (vect_force_simple_reduction): Adjust and set
27114 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27115 (vectorizable_reduction): Do not re-do reduction analysis
27116 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27117 * tree-parloops.c (gather_scalar_reductions): Adjust.
27119 2017-05-30 Richard Biener <rguenther@suse.de>
27121 PR middle-end/80901
27122 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27125 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27127 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27128 Introduce unknown_misalignment parameter and remove vf.
27129 (vect_peeling_hash_get_lowest_cost):
27130 Pass unknown_misalignment parameter.
27131 (vect_enhance_data_refs_alignment):
27132 Fix unsupportable data ref treatment.
27134 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27136 * tree-vect-data-refs.c (vect_get_data_access_cost):
27137 Workaround for SLP handling.
27138 (vect_enhance_data_refs_alignment):
27139 Compute costs for doing no peeling at all, compare to the best
27140 peeling costs so far and avoid peeling if cheaper.
27142 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27144 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27145 Return peeling info and set costs to zero for unlimited cost
27147 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27148 with unknown misalignment. Compute and costs for unknown
27149 misalignment, compare them to the costs for known misalignment
27150 and choose the cheapest for peeling.
27152 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27154 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27155 (vect_get_peeling_costs_all_drs): Create function.
27156 (vect_peeling_hash_get_lowest_cost):
27157 Use vect_get_peeling_costs_all_drs.
27158 (vect_peeling_supportable): Create function.
27159 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27161 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27163 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27164 DR_HAS_NEGATIVE_STEP.
27165 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27166 (vect_enhance_data_refs_alignment): Use.
27167 (vect_duplicate_ssa_name_ptr_info): Use.
27168 * tree-vectorizer.h (dr_misalignment): Use.
27169 (known_alignment_for_access_p): Use.
27171 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27174 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27176 (has_section_name): New function.
27178 2017-05-30 Martin Liska <mliska@suse.cz>
27181 * auto-profile.c (get_function_decl_from_block): Fix
27184 2017-05-30 Richard Biener <rguenther@suse.de>
27186 PR middle-end/80876
27187 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27189 2017-05-30 Martin Liska <mliska@suse.cz>
27191 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27192 * dumpfile.h (struct dump_file_info): Remove ctors.
27194 2017-05-30 Martin Liska <mliska@suse.cz>
27196 * predict.def: Fix GNU coding style.
27198 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27200 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27201 Mark 'to' argument with ATTRIBUTE_UNUSED.
27203 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27205 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27206 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27207 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27210 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27212 * doc/install.texi (Options specification): Restore entry of
27213 --enable-sjlj-exceptions.
27215 2017-05-27 Michael Eager <eager@eagercon.com>
27218 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27220 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27222 * config/microblaze/microblaze.h
27223 (FIXED_REGISTERS): Update in macro.
27224 (CALL_USED_REGISTERS): Update in macro.
27226 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27228 * doc/install.texi: Add links to macOS binary distributions.
27230 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27234 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27236 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27238 2017-05-26 Martin Liska <mliska@suse.cz>
27240 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27242 2017-05-26 Martin Liska <mliska@suse.cz>
27244 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27245 always leading ';; '.
27246 (dump_bb_info): Likewise.
27247 (brief_dump_cfg): Likewise.
27248 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27249 * dumpfile.c: Remove usage of TDF_VERBOSE.
27250 * dumpfile.h (enum dump_kind): Likewise.
27251 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27252 * print-tree.c (debug_verbose): Remove.
27253 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27254 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27255 * tree-diagnostic.c (default_tree_printer): Replace
27256 TDF_DIAGNOSTIC with TDF_SLIM.
27258 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27260 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27261 in parameter loop, rather than loop_vinfo.
27262 (create_intersect_range_checks): Ditto.
27263 (vect_create_cond_for_alias_checks): Update call to above functions.
27265 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27267 PR tree-optimization/80815
27268 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27269 for merging runtime alias checks. Handle negative DR_STEPs.
27271 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27273 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27275 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27276 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27277 out code pruning runtime alias checks.
27278 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27279 factored out from above.
27280 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27282 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27284 (prune_runtime_alias_test_list): New decalaration.
27286 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27288 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27289 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27290 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27291 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27292 (operator==, comp_dr_with_seg_len_pair): Ditto.
27293 (vect_prune_runtime_alias_test_list): Ditto.
27295 2017-05-26 Martin Liska <mliska@suse.cz>
27298 * params.def: Bound partial-inlining-entry-probability param.
27300 2017-05-26 Marek Polacek <polacek@redhat.com>
27303 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27306 2017-05-26 Richard Biener <rguenther@suse.de>
27308 PR tree-optimization/80842
27309 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27312 2017-05-26 Richard Biener <rguenther@suse.de>
27314 PR tree-optimization/80844
27315 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27317 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27319 * doc/md.texi (Machine Constraints): Update x86 family
27320 machine constraints section to match 'config/i386/constraints.md'.
27322 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27324 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27326 2017-05-25 Nathan Sidwell <nathan@acm.org>
27328 * doc/invoke.texi (--enable-languages): Update documentation.
27330 2017-05-25 Martin Liska <mliska@suse.cz>
27332 * dumpfile.c: Add TDF_FOLDING.
27333 * dumpfile.h (enum dump_kind): Likewise.
27334 * genmatch.c (dt_simplify::gen_1): Use it.
27336 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27338 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27340 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27342 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27343 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27345 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27347 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27348 * match.pd (X == C): Rewrite it here.
27349 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27350 with_certain_nonzero_bits2): New predicates.
27351 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27353 2017-05-24 Nathan Sidwell <nathan@acm.org>
27355 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27358 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27361 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27363 * config/powerpcspe: New port. Files are copied from the rs6000
27364 port, with "rs6000" in filenames replaced by "powerpcspe".
27366 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27368 PR rtl-optimization/80754
27369 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27371 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27373 * config/sparc/sparc.md (length): Return the correct value for -mflat
27374 sibcalls to match output_sibcall.
27376 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27380 * config/rs6000/rs6000.c (struct machine_function): Add new field
27382 (rs6000_get_separate_components): Init that field, use it.
27383 (rs6000_components_for_bb): Use the field.
27385 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27387 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27389 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27391 PR middle-end/80823
27392 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27394 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27397 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27398 address against address_operand predicate.
27399 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27401 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27403 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27404 parameters passed indirectly.
27406 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27408 * config/i386/i386.md (*movdi_internal): Remove SSE4
27409 alternative 18 (?r, *v). Update insn attributes.
27410 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27411 Update insn attributes.
27412 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27413 Update insn attributes.
27414 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27415 alternative 1 (r, v). Remove isa attribute.
27416 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27417 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27418 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27420 2017-05-23 Tom de Vries <tom@codesourcery.com>
27422 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27425 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27427 * cgraphunit.c (symbol_table::process_new_functions): Update.
27428 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27429 (inline_generate_summary): Rename to ...
27430 (ipa_fn_summary_generate): ... this one.
27431 (inline_read_summary): Rename to ...
27432 (ipa_fn_summary_read): ... this one.
27433 (inline_write_summary): Rename to ...
27434 (ipa_fn_summary_write): ... this one.
27435 (inline_free_summary): Rename to ...
27436 (ipa_free_fn_summary): ... this one.
27437 (pass_data_local_fn_summary, pass_local_fn_summary,
27438 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27439 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27440 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27441 make_pass_ipa_fn_summary): New.
27442 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27443 inline_write_summary, inline_free_summary): Remove.
27444 (ipa_free_fn_summary) : New.
27445 * ipa-inline.c (ipa_inline): Update.
27446 (pass_ipa_inline): Do not generate summaries.
27447 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27449 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27450 and add pass_ipa_fn_summary.
27451 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27453 (make_pass_inline_parameters): Remove.
27455 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27457 * omp-low.c (struct omp_context): Remove "default_kind" member.
27460 * omp-offload.c (execute_oacc_device_lower): Remove the
27461 parallelism dimensions function attributes for unparallelized
27462 OpenACC kernels constructs.
27464 2017-05-23 Martin Liska <mliska@suse.cz>
27466 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27468 (cgraph_edge::make_speculative): Likewise.
27469 (cgraph_edge::resolve_speculation): Likewise.
27470 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27471 (cgraph_node::dump): Likewise.
27472 * cgraph.h: Likewise.
27473 * cgraphunit.c (analyze_functions): Likewise.
27474 (symbol_table::compile): Likewise.
27475 * ipa-cp.c (print_all_lattices): Likewise.
27476 (determine_versionability): Likewise.
27477 (initialize_node_lattices): Likewise.
27478 (ipcp_verify_propagated_values): Likewise.
27479 (estimate_local_effects): Likewise.
27480 (update_profiling_info): Likewise.
27481 (create_specialized_node): Likewise.
27482 (perhaps_add_new_callers): Likewise.
27483 (decide_about_value): Likewise.
27484 (decide_whether_version_node): Likewise.
27485 (identify_dead_nodes): Likewise.
27486 (ipcp_store_bits_results): Likewise.
27487 * ipa-devirt.c (dump_targets): Likewise.
27488 (ipa_devirt): Likewise.
27489 * ipa-icf.c (sem_item::dump): Likewise.
27490 (sem_function::equals): Likewise.
27491 (sem_variable::equals): Likewise.
27492 (sem_item_optimizer::read_section): Likewise.
27493 (sem_item_optimizer::execute): Likewise.
27494 (congruence_class::dump): Likewise.
27495 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27496 (dump_inline_summary): Likewise.
27497 (estimate_node_size_and_time): Likewise.
27498 (inline_analyze_function): Likewise.
27499 * ipa-inline-transform.c (inline_call): Likewise.
27500 * ipa-inline.c (report_inline_failed_reason): Likewise.
27501 (want_early_inline_function_p): Likewise.
27502 (edge_badness): Likewise.
27503 (update_edge_key): Likewise.
27504 (inline_small_functions): Likewise.
27505 * ipa-profile.c (ipa_profile): Likewise.
27506 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27507 (ipa_make_edge_direct_to_target): Likewise.
27508 (remove_described_reference): Likewise.
27509 (ipa_impossible_devirt_target): Likewise.
27510 (propagate_controlled_uses): Likewise.
27511 (ipa_print_node_params): Likewise.
27512 (ipcp_transform_function): Likewise.
27513 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27514 (propagate_pure_const): Likewise.
27515 * ipa-reference.c (generate_summary): Likewise.
27516 (read_write_all_from_decl): Likewise.
27517 (propagate): Likewise.
27518 (ipa_reference_read_optimization_summary): Likewise.
27519 * ipa-utils.c (ipa_merge_profiles): Likewise.
27520 * ipa.c (walk_polymorphic_call_targets): Likewise.
27521 (symbol_table::remove_unreachable_nodes): Likewise.
27522 (ipa_single_use): Likewise.
27523 * passes.c (execute_todo): Likewise.
27524 * predict.c (drop_profile): Likewise.
27525 * symtab.c (symtab_node::get_dump_name): New function.
27526 (symtab_node::dump_name): Likewise.
27527 (symtab_node::dump_asm_name): Likewise.
27528 (symtab_node::dump_references): Likewise.
27529 (symtab_node::dump_referring): Likewise.
27530 (symtab_node::dump_base): Likewise.
27531 (symtab_node::debug_symtab): Likewise.
27532 * tree-sra.c (convert_callers_for_node): Likewise.
27533 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27534 * value-prof.c (init_node_map): Likewise.
27536 2017-05-23 Martin Liska <mliska@suse.cz>
27538 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27539 and symtab_node::debug_symtab to symbol_table::debug.
27540 * cgraphunit.c (analyze_functions): Use the renamed function.
27541 (symbol_table::compile): Likewise.
27542 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27543 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27544 * passes.c (execute_todo): Likewise.
27545 * symtab.c (symbol_table::dump): New function.
27546 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27548 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27550 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27551 that nonconst implies exec.
27553 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27555 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27556 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27557 (inline_edge_summary_vec): Turn into ...
27558 (ipa_call_summaries): ... this one.
27559 (redirect_to_unreachable, edge_set_predicate,
27560 evaluate_properties_for_edge, inline_summary_alloc,
27561 reset_ipa_call_summary, reset_inline_summary,
27562 inline_summary_t::duplicate): Update.
27563 (inline_edge_duplication_hook): Turn to ...
27564 (ipa_call_summary_t::duplicate): ... this one.
27565 (inline_edge_removal_hook): Turn to ...
27566 (ipa_call_summary_t::remove): ... this one.
27567 (dump_inline_edge_summary): Turn to ...
27568 (dump_ipa_call_summary): ... this one.
27569 (estimate_function_body_sizes): Update.
27570 (inline_update_callee_summaries): Update.
27571 (remap_edge_change_prob): Update.
27572 (remap_edge_summaries): Update.
27573 (inline_merge_summary): Update.
27574 (do_estimate_edge_time): Update.
27575 (inline_generate_summary): Update.
27576 (inline_read_section): Update.
27577 (inline_read_summary): Update.
27578 (inline_free_summary): Update.
27579 * ipa-inline.c (can_inline_edge_p): Update.
27580 (compute_inlined_call_time): Update.
27581 (want_inline_small_function_p): Update.
27582 (edge_badness): Update.
27583 (early_inliner): Update.
27584 * ipa-inline.h (inline_edge_summary): Turn to ...
27585 (ipa_call_summary): ... this one.
27586 (ipa_call_summary_t): New class.
27587 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27588 (ipa_call_summaries): New.
27589 (inline_edge_summary): Remove.
27590 (estimate_edge_growth): Update.
27591 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27592 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27593 * ipa-split.c (execute_split_functions): Update.
27594 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27596 2017-05-23 Tom de Vries <tom@codesourcery.com>
27598 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27599 attributes): Document rdrand effective target.
27601 2017-05-23 Tom de Vries <tom@codesourcery.com>
27603 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27604 attributes): Sort alphabetically.
27606 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27608 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27610 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27613 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27614 V2DF/V2DI splat into two separate patterns, one that handles
27615 registers, and the other that only handles memory. Drop support
27616 for splatting from a GPR on ISA 2.07 and then splitting the
27617 splat into direct move and splat.
27618 (vsx_splat_<mode>_reg): Likewise.
27619 (vsx_splat_<mode>_mem): Likewise.
27621 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27623 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27625 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27627 PR middle-end/80809
27628 * omp-low.c (finish_taskreg_remap): New function.
27629 (finish_taskreg_scan): If unit size of ctx->record_type
27630 is non-constant, unshare the size expression and replace
27631 decls in it with possible outer var refs.
27633 PR middle-end/80809
27634 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27635 GOVD_SHARED rather than GOVD_PRIVATE with it.
27636 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27637 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27639 PR middle-end/80853
27640 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27641 as last argument to build_outer_var_ref for pointer bases of array
27642 section reductions.
27644 2017-05-19 Martin Sebor <msebor@redhat.com>
27646 * print-tree.c (print_node): Print DECL_READ_P flag.
27648 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27650 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27651 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27652 * cgraph.c: Likewise.
27653 * cgraphunit.c: Likewise.
27654 * gengtype.c: Likewise.
27655 * ipa-cp.c: Likewise.
27656 * ipa-devirt.c: Likewise.
27657 * ipa-icf.c: Likewise.
27658 * ipa-predicate.c: Likewise.
27659 * ipa-profile.c: Likewise.
27660 * ipa-prop.c: Likewise.
27661 * ipa-split.c: Likewise.
27663 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27664 edge_predicate_pool, dump_inline_hints,
27665 inline_summary::account_size_time, redirect_to_unreachable,
27666 edge_set_predicate, set_hint_predicate,
27667 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27668 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27669 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27670 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27671 ipa_call_summary_t::remove, initialize_growth_caches,
27672 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27673 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27674 mark_modified, unmodified_parm_1, unmodified_parm,
27675 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27676 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27677 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27678 will_be_nonconstant_predicate, record_modified_bb_info,
27679 get_minimal_bb, record_modified, param_change_prob,
27680 phi_result_unknown_predicate, predicate_for_phi_result,
27681 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27682 estimate_function_body_sizes, compute_inline_parameters,
27683 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27684 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27685 inline_update_callee_summaries, remap_edge_change_prob,
27686 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27687 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27688 inline_analyze_function, inline_summary_t::insert,
27689 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27690 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27691 inline_free_summary): Move to ipa-fnsummary.h
27692 (predicate_t): Remove.
27693 * ipa-fnsummary.c: New file.
27694 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27695 (enum inline_hints_vals, inline_hints, agg_position_info,
27696 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27697 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27698 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27699 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27700 inline_read_summary, inline_write_summary, inline_free_summary,
27701 inline_analyze_function, initialize_inline_failed,
27702 inline_merge_summary, inline_update_overall_summary,
27703 compute_inline_parameters): Move to ipa-fnsummary.h
27704 * ipa-fnsummary.h: New file.
27705 * ipa-inline-transform.h: Include ipa-inline.h.
27706 * ipa-inline.c: LIkewise.
27708 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27710 * ipa-inline.c (edge_badness): Use inlined_time instead of
27711 inline_summaries->get.
27713 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27715 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27717 2017-05-22 Nathan Sidwell <nathan@acm.org>
27719 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27720 (fdump-lang): Document 'raw' option.
27721 * dumpfile.h (TDI_tu): Delete.
27722 * dumpfile.c (dump_files): Remove translation-unit.
27723 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27725 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27727 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27728 command option from $(AWK) call.
27729 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27730 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27731 [FORMAT]: Remove handling of variable.
27732 * config/avr/t-multilib: Regenerate.
27734 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27736 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27738 (dump_inline_summary): Do not print self_time.
27739 (estimate_function_body_sizes): Do not set self_time.
27740 (compute_inline_parameters): Likewise.
27741 (inline_read_section, inline_write_summary): Do not stream self_time.
27742 * ipa-inline.h (inline_summary): Drop self_time.
27744 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27746 * ipa-inline-analysis.c (account_size_time): Rename to ...
27747 (inline_summary::account_size_time): ... this one.
27748 (reset_ipa_call_summary): Turn to ...
27749 (ipa_call_summary::reset): ... this one.
27750 (reset_inline_summary): Turn to ...
27751 (inline_summary::reset): ... this one.
27752 (inline_summary_t::remove): Update.
27753 (inline_summary_t::duplicate): Update.
27754 (ipa_call_summary_t::remove): Update.
27755 (dump_inline_summary): Update.
27756 (estimate_function_body_sizes): Update.
27757 (compute_inline_parameters): Update.
27758 (estimate_node_size_and_time): Update.
27759 (inline_merge_summary): Update.
27760 (inline_update_overall_summary): Update.
27761 (inline_read_section): Update.
27762 (inline_write_summary): Update.
27763 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27764 add account_size_time and reset member functions.
27765 (ipa_call_summary): Add reset function.
27766 * ipa-predicate.h (predicate::operator &): Constify.
27768 2017-05-22 Richard Biener <rguenther@suse.de>
27770 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27772 2017-05-19 Jason Merrill <jason@redhat.com>
27774 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27776 2017-05-19 Marek Polacek <polacek@redhat.com>
27779 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27780 TYPE_OVERFLOW_WRAPS checks.
27782 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27784 * tree-core.h (enum omp_clause_default_kind): Add
27785 "OMP_CLAUSE_DEFAULT_PRESENT".
27786 * tree-pretty-print.c (dump_omp_clause): Handle it.
27787 * gimplify.c (enum gimplify_omp_var_data): Add
27788 "GOVD_MAP_FORCE_PRESENT".
27789 (gimplify_adjust_omp_clauses_1): Map it to
27790 "GOMP_MAP_FORCE_PRESENT".
27791 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27793 * gimplify.c (oacc_default_clause): Clarify.
27795 2017-05-19 Nathan Sidwell <nathan@acm.org>
27797 LANG_HOOK_REGISTER_DUMPS
27798 * toplev.c (general_init): Call register dump lang hook.
27799 * doc/invoke.texi: Document -fdump-lang option family.
27800 * dumpfile.c (dump_files): Remove class dump here.
27801 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27802 * dumpfile.h (tree_dump_index): Remove TDI_class.
27803 * langhooks-def.h (lhd_register_dumps): Declare.
27804 (LANG_HOOKS_REGISTER_DUMPS): Define.
27805 (LANG_HOOKS_INITIALIZER): Add it.
27806 * langhooks.c (lhd_register_dumps): Define.
27807 * langhooks.h (struct lang_hooks): Add register_dumps.
27809 2017-05-19 Nathan Sidwell <nathan@acm.org>
27811 * context.h (context::set_passes): New.
27812 * context.c (context::context): Do not create pass manager.
27813 * toplev.c (general_init): Create pass manager here.
27815 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27817 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27818 use this splitter if two add or or instructions would also work for
27819 the constant we want to generate.
27821 2017-05-19 Richard Biener <rguenther@suse.de>
27824 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27825 predicate evaluation.
27827 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27829 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27831 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27834 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27836 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27837 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27838 (inline_edge_summary_vec): Turn into ...
27839 (ipa_call_summaries): ... this one.
27840 (redirect_to_unreachable, edge_set_predicate,
27841 evaluate_properties_for_edge, inline_summary_alloc,
27842 reset_ipa_call_summary, reset_inline_summary,
27843 inline_summary_t::duplicate): Update.
27844 (inline_edge_duplication_hook): Turn to ...
27845 (ipa_call_summary_t::duplicate): ... this one.
27846 (inline_edge_removal_hook): Turn to ...
27847 (ipa_call_summary_t::remove): ... this one.
27848 (dump_inline_edge_summary): Turn to ...
27849 (dump_ipa_call_summary): ... this one.
27850 (estimate_function_body_sizes): Update.
27851 (inline_update_callee_summaries): Update.
27852 (remap_edge_change_prob): Update.
27853 (remap_edge_summaries): Update.
27854 (inline_merge_summary): Update.
27855 (do_estimate_edge_time): Update.
27856 (inline_generate_summary): Update.
27857 (inline_read_section): Update.
27858 (inline_read_summary): Update.
27859 (inline_free_summary): Update.
27860 * ipa-inline.c (can_inline_edge_p): Update.
27861 (compute_inlined_call_time): Update.
27862 (want_inline_small_function_p): Update.
27863 (edge_badness): Update.
27864 (early_inliner): Update.
27865 * ipa-inline.h (inline_edge_summary): Turn to ...
27866 (ipa_call_summary): ... this one.
27867 (ipa_call_summary_t): New class.
27868 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27869 (ipa_call_summaries): New.
27870 (inline_edge_summary): Remove.
27871 (estimate_edge_growth): Update.
27872 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27873 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27874 * ipa-split.c (execute_split_functions): Update.
27875 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27877 2017-05-19 Richard Biener <rguenther@suse.de>
27879 PR middle-end/80764
27880 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27882 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27884 * config/rs6000/rs6000.c (struct machine_function): Add field
27885 fpr_is_wrapped_separately.
27886 (rs6000_get_separate_components): Use 64 components. Handle the
27887 new FPR components.
27888 (rs6000_components_for_bb): Handle the FPR components.
27889 (rs6000_emit_prologue_components): Handle the FPR components.
27890 (rs6000_emit_epilogue_components): Handle the FPR components.
27891 (rs6000_set_handled_components): Handle the FPR components.
27892 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27893 that are already separately shrink-wrapped.
27894 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27895 that are already separately shrink-wrapped.
27897 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27900 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27903 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27904 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27905 cases where the register allocator uses a move and an offsettable
27906 memory operation to/from a FPR register on ISA 2.06/2.07.
27907 (define_peephole2 for Altivec d-form store): Likewise.
27909 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27912 * config/i386/mmx.md (*mov<mode>_internal): Enable
27913 alternatives 11, 12, 13 and 14 also for 32bit targets.
27914 Remove alternatives 15, 16, 17 and 18.
27915 * config/i386/sse.md (vec_concatv2di): Change
27916 alternative (!x, *y) to (x, ?!*Yn).
27918 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27920 * dumpfile.h (enum dump_kind): Remove stray comma.
27922 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27924 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27925 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27926 predicate::num_conditions
27927 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27928 (CHANGED): turn into predicate::changed.
27929 (agg_position_info): Move to ipa-predicate.h
27930 (add_condition, predicate::add_clause, predicate::operator &=,
27931 predicate::or_with, predicate::evaluate, predicate::probability,
27932 dump_condition, dump_clause, predicate::dump,
27933 predicate::remap_after_duplication, predicate::remap_after_inlining,
27934 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27935 (evaluate_conditions_for_known_args): Update.
27936 (set_cond_stmt_execution_predicate): Update.
27937 * ipa-inline.h: Include ipa-predicate.h
27938 (condition, inline_param_summary, conditions, agg_position_info,
27939 predicate): Move to ipa-predicate.h
27940 * ipa-predicate.c: New file.
27941 * ipa-predicate.h: New file.
27943 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27945 * final.c (leaf_function_p): Check we are not in a sequence.
27947 2017-05-18 Martin Liska <mliska@suse.cz>
27949 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27950 * dumpfile.c (dump_register): Use new enum dump_kind.
27951 (get_dump_file_name): Likewise.
27952 (dump_enable_all): Likewise.
27953 (dump_switch_p_1): Likewise.
27954 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27955 * dumpfile.h (enum dump_kind): New enum type.
27956 (struct dump_file_info): Create constructor and
27957 format fields and comments.
27958 * passes.c (pass_manager::register_one_dump_file):
27960 * statistics.c (statistics_early_init): Likewise.
27961 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27962 TDF_TREE with TDF_SLIM.
27963 (gather_memory_references_ref): Likewise.
27965 2017-05-18 Martin Liska <mliska@suse.cz>
27967 * vec.h (struct vnull): Use it.
27969 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27971 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27972 (true_predicate, false_predicate, true_predicate_p,
27973 false_predicate_p): Remove.
27974 (single_cond_predicate, not_inlined_predicate): Turn to member function
27976 (add_condition): Update.
27977 (add_clause): Turn to...
27978 (predicate::add_clause): ... this one; update; allow passing NULL
27980 (and_predicates): Turn to ...
27981 (predicate::operator &=): ... this one.
27982 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27983 (or_predicates): Turn to ...
27984 (predicate::or_with): ... this one.
27985 (evaluate_predicate): Turn to ...
27986 (predicate::evaluate): ... this one.
27987 (predicate_probability): Turn to ...
27988 (predicate::probability): ... this one.
27989 (dump_condition): Update.
27990 (dump_predicate): Turn to ...
27991 (predicate::dump): ... this one.
27992 (account_size_time): Update.
27993 (edge_set_predicate): Update.
27994 (set_hint_predicate): UPdate.
27995 (evaluate_conditions_for_known_args): Update.
27996 (evaluate_properties_for_edge): Update.
27997 (remap_predicate_after_duplication): Turn to...
27998 (predicate::remap_after_duplication): ... this one.
27999 (remap_hint_predicate_after_duplication): Update.
28000 (inline_summary_t::duplicate): UPdate.
28001 (dump_inline_edge_summary): Update.
28002 (dump_inline_summary): Update.
28003 (set_cond_stmt_execution_predicate): Update.
28004 (set_switch_stmt_execution_predicate): Update.
28005 (compute_bb_predicates): Update.
28006 (will_be_nonconstant_expr_predicate): Update.
28007 (will_be_nonconstant_predicate): Update.
28008 (phi_result_unknown_predicate): Update.
28009 (predicate_for_phi_result): Update.
28010 (array_index_predicate): Update.
28011 (estimate_function_body_sizes): Update.
28012 (estimate_node_size_and_time): Update.
28013 (estimate_ipcp_clone_size_and_time): Update.
28014 (remap_predicate): Rename to ...
28015 (predicate::remap_after_inlining): ... this one.
28016 (remap_hint_predicate): Update.
28017 (inline_merge_summary): Update.
28018 (inline_update_overall_summary): Update.
28019 (estimate_size_after_inlining): Update.
28020 (read_predicate): Rename to ...
28021 (predicate::stream_in): ... this one.
28022 (read_inline_edge_summary): Update.
28023 (write_predicate): Rename to ...
28024 (predicate::stream_out): ... this one.
28025 (write_inline_edge_summary): Update.
28026 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28027 (clause_t): Turn to uint32_t
28028 (predicate): Turn to class; implement constructor and operators
28030 (size_time_entry): Update.
28031 (inline_summary): Update.
28032 (inline_edge_summary): Update.
28034 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28036 * fold-const.c (fold_binary_loc): Move transformation...
28037 * match.pd (C - X CMP X): ... here.
28039 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28041 * config/sparc/sparc.c (sparc_option_override): Set function
28042 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28043 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28045 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28047 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28049 2017-05-18 Marek Polacek <polacek@redhat.com>
28052 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28053 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28054 gimple_assign_load_p.
28056 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28058 PR middle-end/80692
28059 * real.c (do_compare): Give decimal_do_compare preference over
28060 comparing just the signs.
28062 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28064 * doc/md.texi (Canonicalization of Instructions): Describe the
28065 canonical form of instructions that inherently set a condition
28068 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28070 PR middle-end/80775
28071 * tree-cfg.c: Move deletion of unreachable case statements to after
28072 the merging of consecutive case labels.
28074 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28076 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28077 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28078 restoring of callee-saved registers.
28080 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28082 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28083 * config/visium/visium.c (single_set_and_flags): Likewise.
28084 * config/visium/visium.md (Substitutions): Likewise.
28086 2017-05-17 Martin Liska <mliska@suse.cz>
28088 * cfg.c: Introduce dump_flags_t type and
28089 use it instead of int type.
28091 * cfghooks.c: Likewise.
28092 * cfghooks.h (struct cfg_hooks): Likewise.
28093 * cfgrtl.c: Likewise.
28094 * cfgrtl.h: Likewise.
28095 * cgraph.c (cgraph_node::get_body): Likewise.
28096 * coretypes.h: Likewise.
28097 * domwalk.c: Likewise.
28098 * domwalk.h: Likewise.
28099 * dumpfile.c (struct dump_option_value_info): Likewise.
28100 (dump_enable_all): Likewise.
28101 (dump_switch_p_1): Likewise.
28102 (opt_info_switch_p): Likewise.
28103 * dumpfile.h (enum tree_dump_index): Likewise.
28104 (struct dump_file_info): Likewise.
28105 * genemit.c: Likewise.
28106 * generic-match-head.c: Likewise.
28107 * gengtype.c (open_base_files): Likewise.
28108 * gimple-pretty-print.c: Likewise.
28109 * gimple-pretty-print.h: Likewise.
28110 * graph.c (print_graph_cfg): Likewise.
28111 * graphite-scop-detection.c (dot_all_sese): Likewise.
28112 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28113 * loop-unroll.c (report_unroll): Likewise.
28114 * passes.c (pass_manager::register_one_dump_file): Likewise.
28115 * print-tree.c: Likewise.
28116 * statistics.c: Likewise.
28117 * tree-cfg.c: Likewise.
28118 * tree-cfg.h: Likewise.
28119 * tree-dfa.c: Likewise.
28120 * tree-dfa.h: Likewise.
28121 * tree-dump.c (dump_function): Likewise.
28122 * tree-dump.h (struct dump_info): Likewise.
28123 * tree-pretty-print.c: Likewise.
28124 * tree-pretty-print.h: Likewise.
28125 * tree-ssa-live.c: Likewise.
28126 * tree-ssa-live.h: Likewise.
28127 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28128 * tree-vect-loop.c: Likewise.
28129 * tree-vect-slp.c: Likewise.
28131 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28132 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28134 PR tree-optimization/80457
28135 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28136 of all arguments to a statement as scalar_to_vec operations.
28137 (vectorizable_call): Adjust call to vect_model_simple_cost for
28139 (vectorizable_conversion): Likewise.
28140 (vectorizable_assignment): Likewise.
28141 (vectorizable_shift): Likewise.
28142 (vectorizable_operation): Likewise.
28143 (vectorizable_comparison): Likewise.
28144 (vect_is_simple_cond): Record the def types for operands.
28145 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28146 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28147 for statement argument count.
28149 2017-05-16 Carl Love <cel@us.ibm.com>
28151 * config/rs6000/rs6000-c: Add support for built-in functions
28152 vector unsigned long long vec_bperm (vector unsigned long long,
28153 vector unsigned char)
28154 vector signed long long vec_mule (vector signed int,
28156 vector unsigned long long vec_mule (vector unsigned int,
28157 vector unsigned int)
28158 vector signed long long vec_mulo (vector signed int,
28160 vector unsigned long long vec_mulo (vector unsigned int,
28161 vector unsigned int)
28162 vector signed char vec_sldw (vector signed char,
28163 vector signed char,
28165 vector unsigned char vec_sldw (vector unsigned char,
28166 vector unsigned char,
28168 vector signed short vec_sldw (vector signed short,
28169 vector signed short,
28171 vector unsigned short vec_sldw (vector unsigned short,
28172 vector unsigned short,
28174 vector signed int vec_sldw (vector signed int,
28177 vector unsigned int vec_sldw (vector unsigned int,
28178 vector unsigned int,
28180 vector signed long long vec_sldw (vector signed long long,
28181 vector signed long long,
28183 vector unsigned long long vec_sldw (vector unsigned long long,
28184 vector unsigned long long,
28186 * config/rs6000/rs6000-c: Add support for built-in functions
28187 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28188 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28189 * doc/extend.texi: Update the built-in documentation for the
28190 new built-in functions.
28192 2017-05-16 Marek Polacek <polacek@redhat.com>
28196 * tree.c (save_expr): Don't fold the expression.
28198 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28200 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28201 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28202 and (?*y,m). Update insn attributes.
28204 2017-05-16 Martin Liska <mliska@suse.cz>
28206 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28207 flags argument of print_gimple_stmt, print_gimple_expr,
28208 print_generic_stmt and print_generic_expr.
28209 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28210 * coretypes.h: Likewise.
28211 * except.c (dump_eh_tree): Likewise.
28212 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28213 * gimple-pretty-print.h: Likewise.
28214 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28215 (backprop::push_to_worklist): Likewise.
28216 (backprop::pop_from_worklist): Likewise.
28217 (backprop::process_use): Likewise.
28218 (backprop::intersect_uses): Likewise.
28219 (note_replacement): Likewise.
28220 * gimple-ssa-store-merging.c
28221 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28222 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28223 (pass_store_merging::execute): Likewise.
28224 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28225 (ssa_base_cand_dump_callback): Likewise.
28226 (dump_incr_vec): Likewise.
28227 (replace_refs): Likewise.
28228 (replace_mult_candidate): Likewise.
28229 (create_add_on_incoming_edge): Likewise.
28230 (create_phi_basis): Likewise.
28231 (insert_initializers): Likewise.
28232 (all_phi_incrs_profitable): Likewise.
28233 (introduce_cast_before_cand): Likewise.
28234 (replace_one_candidate): Likewise.
28235 * gimplify.c (gimplify_expr): Likewise.
28236 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28237 (set_rename): Likewise.
28238 (rename_uses): Likewise.
28239 (copy_loop_phi_nodes): Likewise.
28240 (add_close_phis_to_merge_points): Likewise.
28241 (copy_loop_close_phi_args): Likewise.
28242 (copy_cond_phi_args): Likewise.
28243 (graphite_copy_stmts_from_block): Likewise.
28244 (translate_pending_phi_nodes): Likewise.
28245 * graphite-poly.c (print_pdr): Likewise.
28246 (dump_gbb_cases): Likewise.
28247 (dump_gbb_conditions): Likewise.
28248 (print_scop_params): Likewise.
28249 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28250 (build_cross_bb_scalars_use): Likewise.
28251 (gather_bbs::before_dom_children): Likewise.
28252 * hsa-dump.c (dump_hsa_immed): Likewise.
28253 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28254 (get_replacement_map): Likewise.
28255 * ipa-inline-analysis.c (dump_condition): Likewise.
28256 (estimate_function_body_sizes): Likewise.
28257 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28258 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28259 * ipa-prop.c (ipa_dump_param): Likewise.
28260 (ipa_print_node_jump_functions_for_edge): Likewise.
28261 (ipa_modify_call_arguments): Likewise.
28262 (ipa_modify_expr): Likewise.
28263 (ipa_dump_param_adjustments): Likewise.
28264 (ipa_dump_agg_replacement_values): Likewise.
28265 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28266 * ipa-pure-const.c (check_stmt): Likewise.
28267 (pass_nothrow::execute): Likewise.
28268 * ipa-split.c (execute_split_functions): Likewise.
28269 * omp-offload.c (dump_oacc_loop_part): Likewise.
28270 (dump_oacc_loop): Likewise.
28271 * trans-mem.c (tm_log_emit): Likewise.
28272 (tm_memopt_accumulate_memops): Likewise.
28273 (dump_tm_memopt_set): Likewise.
28274 (dump_tm_memopt_transform): Likewise.
28275 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28276 (print_loop): Likewise.
28277 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28278 (chkp_gather_checks_info): Likewise.
28279 (chkp_get_check_result): Likewise.
28280 (chkp_remove_check_if_pass): Likewise.
28281 (chkp_use_outer_bounds_if_possible): Likewise.
28282 (chkp_reduce_bounds_lifetime): Likewise.
28283 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28284 (chkp_mark_completed_bounds): Likewise.
28285 (chkp_register_incomplete_bounds): Likewise.
28286 (chkp_mark_invalid_bounds): Likewise.
28287 (chkp_maybe_copy_and_register_bounds): Likewise.
28288 (chkp_build_returned_bound): Likewise.
28289 (chkp_get_bound_for_parm): Likewise.
28290 (chkp_build_bndldx): Likewise.
28291 (chkp_get_bounds_by_definition): Likewise.
28292 (chkp_generate_extern_var_bounds): Likewise.
28293 (chkp_get_bounds_for_decl_addr): Likewise.
28294 * tree-chrec.c (chrec_apply): Likewise.
28295 * tree-data-ref.c (dump_data_reference): Likewise.
28296 (dump_subscript): Likewise.
28297 (dump_data_dependence_relation): Likewise.
28298 (analyze_overlapping_iterations): Likewise.
28299 * tree-inline.c (expand_call_inline): Likewise.
28300 (tree_function_versioning): Likewise.
28301 * tree-into-ssa.c (dump_defs_stack): Likewise.
28302 (dump_currdefs): Likewise.
28303 (dump_names_replaced_by): Likewise.
28304 (dump_update_ssa): Likewise.
28305 (update_ssa): Likewise.
28306 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28307 * tree-parloops.c (build_new_reduction): Likewise.
28308 (try_create_reduction_list): Likewise.
28309 (ref_conflicts_with_region): Likewise.
28310 (oacc_entry_exit_ok_1): Likewise.
28311 (oacc_entry_exit_single_gang): Likewise.
28312 * tree-pretty-print.h: Likewise.
28313 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28314 (get_scalar_evolution): Likewise.
28315 (add_to_evolution): Likewise.
28316 (get_loop_exit_condition): Likewise.
28317 (analyze_evolution_in_loop): Likewise.
28318 (analyze_initial_condition): Likewise.
28319 (analyze_scalar_evolution): Likewise.
28320 (instantiate_scev): Likewise.
28321 (number_of_latch_executions): Likewise.
28322 (gather_chrec_stats): Likewise.
28323 (final_value_replacement_loop): Likewise.
28324 (scev_const_prop): Likewise.
28325 * tree-sra.c (dump_access): Likewise.
28326 (disqualify_candidate): Likewise.
28327 (create_access): Likewise.
28328 (reject): Likewise.
28329 (maybe_add_sra_candidate): Likewise.
28330 (create_access_replacement): Likewise.
28331 (analyze_access_subtree): Likewise.
28332 (analyze_all_variable_accesses): Likewise.
28333 (sra_modify_assign): Likewise.
28334 (initialize_constant_pool_replacements): Likewise.
28335 (find_param_candidates): Likewise.
28336 (decide_one_param_reduction): Likewise.
28337 (replace_removed_params_ssa_names): Likewise.
28338 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28339 * tree-ssa-copy.c (dump_copy_of): Likewise.
28340 (copy_prop_visit_cond_stmt): Likewise.
28341 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28342 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28343 (record_equivalences_from_stmt): Likewise.
28344 * tree-ssa-dse.c (compute_trims): Likewise.
28345 (delete_dead_call): Likewise.
28346 (delete_dead_assignment): Likewise.
28347 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28348 (forward_propagate_into_cond): Likewise.
28349 (pass_forwprop::execute): Likewise.
28350 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28351 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28353 (move_computations_worker): Likewise.
28354 (execute_sm): Likewise.
28355 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28356 (remove_exits_and_undefined_stmts): Likewise.
28357 (remove_redundant_iv_tests): Likewise.
28358 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28359 (adjust_iv_update_pos): Likewise.
28360 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28361 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28362 (value_replacement): Likewise.
28363 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28364 * tree-ssa-pre.c (print_pre_expr): Likewise.
28365 (get_representative_for): Likewise.
28366 (create_expression_by_pieces): Likewise.
28367 (insert_into_preds_of_block): Likewise.
28368 (eliminate_insert): Likewise.
28369 (eliminate_dom_walker::before_dom_children): Likewise.
28370 (eliminate): Likewise.
28371 (remove_dead_inserted_code): Likewise.
28372 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28373 * tree-ssa-reassoc.c (get_rank): Likewise.
28374 (eliminate_duplicate_pair): Likewise.
28375 (eliminate_plus_minus_pair): Likewise.
28376 (eliminate_not_pairs): Likewise.
28377 (undistribute_ops_list): Likewise.
28378 (eliminate_redundant_comparison): Likewise.
28379 (update_range_test): Likewise.
28380 (optimize_range_tests_var_bound): Likewise.
28381 (optimize_vec_cond_expr): Likewise.
28382 (rewrite_expr_tree): Likewise.
28383 (rewrite_expr_tree_parallel): Likewise.
28384 (linearize_expr): Likewise.
28385 (break_up_subtract): Likewise.
28386 (linearize_expr_tree): Likewise.
28387 (attempt_builtin_powi): Likewise.
28388 (attempt_builtin_copysign): Likewise.
28389 (transform_stmt_to_copy): Likewise.
28390 (transform_stmt_to_multiply): Likewise.
28391 (dump_ops_vector): Likewise.
28392 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28393 (print_scc): Likewise.
28394 (set_ssa_val_to): Likewise.
28395 (visit_reference_op_store): Likewise.
28396 (visit_use): Likewise.
28397 (sccvn_dom_walker::before_dom_children): Likewise.
28398 (run_scc_vn): Likewise.
28399 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28401 (expr_hash_elt::print): Likewise.
28402 (const_and_copies::pop_to_marker): Likewise.
28403 (const_and_copies::record_const_or_copy_raw): Likewise.
28404 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28405 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28406 (dump_predicates): Likewise.
28407 (find_uninit_use): Likewise.
28408 (warn_uninitialized_phi): Likewise.
28409 (pass_late_warn_uninitialized::execute): Likewise.
28410 * tree-ssa.c (verify_vssa): Likewise.
28411 (verify_ssa): Likewise.
28412 (maybe_optimize_var): Likewise.
28413 * tree-vrp.c (dump_value_range): Likewise.
28414 (dump_all_value_ranges): Likewise.
28415 (dump_asserts_for): Likewise.
28416 (register_edge_assert_for_2): Likewise.
28417 (vrp_visit_cond_stmt): Likewise.
28418 (vrp_visit_switch_stmt): Likewise.
28419 (vrp_visit_stmt): Likewise.
28420 (vrp_visit_phi_node): Likewise.
28421 (simplify_cond_using_ranges_1): Likewise.
28422 (fold_predicate_in): Likewise.
28423 (evrp_dom_walker::before_dom_children): Likewise.
28424 (evrp_dom_walker::push_value_range): Likewise.
28425 (evrp_dom_walker::pop_value_range): Likewise.
28426 (execute_early_vrp): Likewise.
28428 2017-05-16 Richard Biener <rguenther@suse.de>
28430 * dwarf2out.c (loc_list_from_tree_1): Do not create
28431 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28433 2017-05-16 Richard Biener <rguenther@suse.de>
28435 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28437 (note_variable_value_in_expr): If we resolved the decl ref
28438 do not push to the stack.
28440 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28442 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28443 operations in fast-math mode.
28444 (vaddq_f16): Likewise.
28445 (vmul_f16): Likewise.
28446 (vmulq_f16): Likewise.
28447 (vsub_f16): Likewise.
28448 (vsubq_f16): Likewise.
28449 * config/arm/neon.md (add<mode>3): New.
28451 (fma:<VH:mode>3): New. Also remove outdated comment.
28454 2017-05-16 Martin Liska <mliska@suse.cz>
28458 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28459 (odr_types_equivalent_p): Likewise.
28461 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28463 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28465 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28468 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28469 non-interunit SSE move alternatives with '?'.
28470 (zero-extendsidi peephole2): New peephole to skip intermediate
28471 general register in SSE zero-extend sequence.
28473 2017-05-15 Jeff Law <law@redhat.com>
28475 * reorg.c (relax_delay_slots): Create a new variable to hold
28476 the temporary target rather than clobbering TARGET_LABEL.
28478 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28479 missing argument to extract_bit_field call.
28480 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28482 2017-05-15 Martin Liska <mliska@suse.cz>
28485 * gcc.c (process_command): Do not allow empty argument of -o option.
28487 2017-05-15 Renlin Li <renlin.li@arm.com>
28489 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28490 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28491 * config/aarch64/constraints.md (Usf): Add long call check.
28492 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28493 (call_value): Likewise.
28494 (sibcall): Likewise.
28495 (sibcall_value): Likewise.
28497 (call_value_insn): New.
28498 (sibcall_insn): Update rtx pattern.
28499 (sibcall_value_insn): Likewise.
28500 (call_internal): Remove.
28501 (call_value_internal): Likewise.
28502 (sibcall_internal): Likewise.
28503 (sibcall_value_internal): Likewise.
28504 (call_reg): Likewise.
28505 (call_symbol): Likewise.
28506 (call_value_reg): Likewise.
28507 (call_value_symbol): Likewise.
28509 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28512 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28514 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28516 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28517 compatible with CCGOCmode and with CCZmode.
28519 2017-05-14 Martin Sebor <msebor@redhat.com>
28521 PR middle-end/77671
28522 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28523 (gimple_fold_builtin_snprintf): Same.
28524 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28525 (gimple_fold_builtin_snprintf): Same.
28526 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28527 of character types.
28528 (is_call_safe): New function.
28529 (try_substitute_return_value): Call it.
28530 (try_simplify_call): New function.
28531 (pass_sprintf_length::handle_gimple_call): Call it.
28533 2017-05-14 Martin Sebor <msebor@redhat.com>
28535 PR middle-end/80669
28536 * builtins.c (expand_builtin_stpncpy): Simplify.
28538 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28540 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28541 * config/i386/i386.h
28542 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28543 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28544 (struct machine_function): Add new members call_ms2sysv,
28545 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28546 (struct machine_frame_state): New fields sp_realigned and
28547 sp_realigned_offset.
28548 * config/i386/i386.c
28549 (enum xlogue_stub): New enum.
28550 (enum xlogue_stub_sets): New enum.
28551 (class xlogue_layout): New class.
28552 (struct ix86_frame): New fields stack_realign_allocate_offset,
28553 stack_realign_offset and outlined_save_offset. Modify comments to
28554 detail stack layout when using out-of-line stubs.
28555 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28556 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28557 -mcall-ms2sysv-xlogues.
28558 (stub_managed_regs): New static variable.
28559 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28560 registers managed by out-of-line stub.
28561 (disable_call_ms2sysv_xlogues): New function.
28562 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28563 m->call_ms2sysv when appropriate and compute frame layout for
28565 (sp_valid_at, fp_valid_at): New inline functions.
28566 (choose_basereg): New function.
28567 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28569 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28570 Use align parameter of choose_baseaddr to generated aligned SSE movs
28572 (pro_epilogue_adjust_stack): Modify to track
28573 machine_frame_state::sp_realigned.
28574 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28575 (ix86_nsaved_sseregs): Likewise.
28576 (ix86_emit_save_regs): Likewise.
28577 (ix86_emit_save_regs_using_mov): Likewise.
28578 (ix86_emit_save_sse_regs_using_mov): Likewise.
28579 (get_scratch_register_on_entry): Likewise.
28580 (gen_frame_set): New function.
28581 (gen_frame_load): Likewise.
28582 (gen_frame_store): Likewise.
28583 (emit_outlined_ms2sysv_save): Likewise.
28584 (emit_outlined_ms2sysv_restore): Likewise.
28585 (ix86_expand_prologue): Modify stack re-alignment code and call
28586 emit_outlined_ms2sysv_save when appropriate.
28587 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28588 parameter rtx_insn *insn, which allows the function to be used to only
28589 generate the notes.
28590 (ix86_expand_epilogue): Modify validity checks of frame and stack
28591 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28592 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28593 * config/i386/predicates.md
28594 (save_multiple): New predicate.
28595 (restore_multiple): Likewise.
28596 * config/i386/sse.md
28597 (save_multiple<mode>): New pattern.
28598 (save_multiple_realign<mode>): Likewise.
28599 (restore_multiple<mode>): Likewise.
28600 (restore_multiple_and_return<mode>): Likewise.
28601 (restore_multiple_leave_return<mode>): Likewise.
28602 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28604 2017-05-14 Julia Koval <julia.koval@intel.com>
28606 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28607 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28608 (__builtin_ia32_xsetbv): New builtins.
28609 * config/i386/i386.c (ix86_expand_special_args_builtin):
28611 (ix86_expand_builtin): Special expand for new intrinsics.
28612 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28613 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28614 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28616 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28618 * cfganal.c (inverted_post_order_compute): Change argument type
28620 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28621 * df-core.c (rest_of_handle_df_initialize): Adjust.
28622 (rest_of_handle_df_finish): Likewise.
28623 (df_analyze_1): Likewise.
28624 (df_analyze): Likewise.
28625 (loop_inverted_post_order_compute): Change argument to be a vec *.
28626 (df_analyze_loop): Adjust.
28627 (df_get_n_blocks): Likewise.
28628 (df_get_postorder): Likewise.
28629 * df.h (struct df_d): Change field to be a vec.
28630 * lcm.c (compute_laterin): Adjust.
28631 (compute_available): Likewise.
28632 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28633 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28634 * tree-ssa-pre.c (compute_antic): Likewise.
28636 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28638 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28639 (depth_first_search::depth_first_search): Change structure init
28640 function to this constructor.
28641 (depth_first_search::add_bb): Rename function to this member.
28642 (depth_first_search::execute): Likewise.
28643 (flow_dfs_compute_reverse_finish): Adjust.
28645 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28647 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28648 (longest_simple_path): Likewise.
28649 * shrink-wrap.c (spread_components): Likewise.
28650 (disqualify_problematic_components): Likewise.
28651 (emit_common_heads_for_components): Likewise.
28652 (emit_common_tails_for_components): Likewise.
28653 (insert_prologue_epilogue_for_components): Likewise.
28655 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28657 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28660 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28662 * df-core.c (df_set_blocks): Start using auto_bitmap.
28663 (df_compact_blocks): Likewise.
28664 * df-problems.c (df_rd_confluence_n): Likewise.
28665 * df-scan.c (df_insn_rescan_all): Likewise.
28666 (df_process_deferred_rescans): Likewise.
28667 (df_update_entry_block_defs): Likewise.
28668 (df_update_exit_block_uses): Likewise.
28669 (df_entry_block_bitmap_verify): Likewise.
28670 (df_exit_block_bitmap_verify): Likewise.
28671 (df_scan_verify): Likewise.
28672 * lra-constraints.c (lra_constraints): Likewise.
28673 (undo_optional_reloads): Likewise.
28674 (lra_undo_inheritance): Likewise.
28675 * lra-remat.c (calculate_gen_cands): Likewise.
28676 (do_remat): Likewise.
28677 * lra-spills.c (assign_spill_hard_regs): Likewise.
28678 (spill_pseudos): Likewise.
28679 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28680 (bitmap_set_subtract_values): Likewise.
28682 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28684 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28685 management with auto_bitmap.
28686 (fix_inter_tick): Likewise.
28687 (fix_recovery_deps): Likewise.
28688 * ira.c (add_store_equivs): Likewise.
28689 (find_moveable_pseudos): Likewise.
28690 (split_live_ranges_for_shrink_wrap): Likewise.
28691 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28692 (rtx_reuse_manager::seen_def_p): Likewise.
28693 (rtx_reuse_manager::set_seen_def): Likewise.
28694 * print-rtl.h (class rtx_reuse_manager): Likewise.
28696 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28698 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28700 (migrate_btr_def): Likewise.
28701 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28702 * df-core.c (loop_post_order_compute): Likewise.
28703 (loop_inverted_post_order_compute): Likewise.
28704 * hsa-common.h: Likewise.
28705 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28706 * init-regs.c (initialize_uninitialized_regs): Likewise.
28707 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28708 (inline_small_functions): Likewise.
28709 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28710 * ira.c (combine_and_move_insns): Likewise.
28711 (build_insn_chain): Likewise.
28712 * loop-invariant.c (find_invariants): Likewise.
28713 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28714 * predict.c (tree_predict_by_opcode): Likewise.
28715 (predict_paths_leading_to): Likewise.
28716 (predict_paths_leading_to_edge): Likewise.
28717 (estimate_loops_at_level): Likewise.
28718 (estimate_loops): Likewise.
28719 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28720 (spread_components): Likewise.
28721 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28722 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28723 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28724 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28725 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28726 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28727 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28728 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28729 (mark_threaded_blocks): Likewise.
28730 (thread_through_all_blocks): Likewise.
28731 * tree-ssa.c (verify_ssa): Likewise.
28732 (execute_update_addresses_taken): Likewise.
28733 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28735 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28737 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28739 (post_order_compute): Likewise.
28740 (inverted_post_order_compute): Likewise.
28741 (pre_and_rev_post_order_compute_fn): Likewise.
28743 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28745 * genrecog.c (int_set::int_set): Explicitly construct our
28746 auto_vec base class.
28747 * vec.h (auto_vec::auto_vec): New constructor.
28749 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28751 * bitmap.h (class auto_bitmap): New constructor taking
28752 bitmap_obstack * argument.
28754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28756 * bitmap.h (class auto_bitmap): Change type of m_bits to
28757 bitmap_head, and adjust ctor / dtor and member operators.
28759 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28761 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28762 when returned register mode doesn't match original mode.
28764 2017-05-12 Jeff Law <law@redhat.com>
28765 Jakub Jelinek <jakub@redhat.com>
28767 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28768 we look for cc setter after the compare-elim changes.
28769 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28770 within the vector to match what compare-elim now expects.
28771 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28772 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28774 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28775 after the compare-elim changes.
28776 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28777 the vector to match what compare-elim now expects.
28778 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28779 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28780 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28781 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28782 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28784 * config/visium/visium.c (single_set_and_flags): Fix where
28785 we look for cc setter after the compare-elim changes.
28786 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28787 with the vector to match what compare-elim now expects.
28788 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28789 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28790 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28791 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28792 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28793 (neg<mode>2_insn_set_overflow): Likewise.
28795 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28797 PR middle-end/79794
28798 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28799 maybe_expand_insn call, set ops[0].target. If still set after call,
28800 set alt_rtl. Add extra arg to recursive calls.
28801 (extract_bit_field): Add alt_rtl argument. Pass to
28803 * expmed.h (extract_bit_field): Fix prototype.
28804 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28805 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28806 to extract_bit_field_calls.
28807 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28808 Pass alt_rtl to extract_bit_field calls.
28809 * calls.c (store_unaligned_arguments_into_psuedos)
28810 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28811 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28813 * optabs.h (struct expand_operand): Add target bitfield.
28815 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28817 * compare-elim.c (try_eliminate_compare): Canonicalize
28818 operation with embedded compare to
28819 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28820 (set (reg) (operation)].
28822 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28824 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28827 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28828 cost of adding a carry flag for ADC instruction.
28829 [case MINUS]: Ignore the cost of subtracting a carry flag
28830 for SBB instruction.
28832 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28834 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28836 * config/rs6000/bmiintrin.h: New file.
28837 * config/rs6000/bmi2intrin.h: New file.
28838 * config/rs6000/x86intrin.h: New file.
28840 2017-05-12 Jeff Law <law@redhat.com>
28842 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28845 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28847 PR middle-end/80707
28848 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28850 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28853 early expansion of vector divide builtins.
28854 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28855 builtins identified as having unsigned arguments.
28857 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28859 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28860 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28861 expansion of vector logical operations (and, andc, or, xor,
28864 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28866 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28867 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28869 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28872 early GIMPLE expansion of vector multiplies.
28874 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28876 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28877 TARGET_HAVE_MOVT conditional.
28878 (movt splitter): Likewise.
28880 2017-05-12 Richard Biener <rguenther@suse.de>
28882 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28883 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28884 Fold all stmts not inplace.
28886 2017-05-12 Richard Biener <rguenther@suse.de>
28888 PR tree-optimization/80713
28889 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28890 inserted_exprs bit for not removed stmts.
28892 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28894 PR middle-end/69921
28895 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28896 parallelized" attribute for parallelized OpenACC kernels.
28897 * omp-offload.c (execute_oacc_device_lower): Use it.
28899 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28900 Set "oacc kernels" attribute.
28901 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28902 parameter. Adjust all users.
28903 (oacc_fn_attrib_kernels_p): Remove function.
28904 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28905 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28906 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28907 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28908 assert "oacc kernels" attribute is set.
28910 2017-05-11 Carl Love <cel@us.ibm.com>
28912 * config/rs6000/rs6000-c: Add support for built-in functions
28913 vector unsigned char vec_popcnt (vector signed char)
28914 vector unsigned char vec_popcnt (vector unsigned char)
28915 vector unsigned short vec_popcnt (vector signed short)
28916 vector unsigned short vec_popcnt (vector unsigned short)
28917 vector unsigned int vec_popcnt (vector signed int)
28918 vector unsigned int vec_popcnt (vector unsigned int)
28919 vector unsigned long long vec_popcnt (vector signed long long)
28920 vector unsigned long long vec_popcnt (vector unsigned long long)
28921 vector signed long long vec_slo (vector signed long long,
28922 vector signed char)
28923 vector signed long long vec_slo (vector signed long long,
28924 vector unsigned char)
28925 vector unsigned long long vec_slo (vector unsigned long long,
28926 vector signed char)
28927 vector unsigned long long vec_slo (vector unsigned long long,
28928 vector unsigned char)
28929 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28930 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28931 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28932 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28933 * doc/extend.texi: Update the built-in documentation file for the
28934 new built-in functions.
28936 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28938 * attribs.h (sorted_attr_string): Move machine independent
28939 functions for target clone support from the i386 port to common
28940 code. Rename ix86_function_versions to common_function_versions.
28941 Rename make_name to make_unique_name.
28942 (common_function_versions): Likewise.
28943 (make_unique_name): Likewise.
28944 (make_dispatcher_decl): Likewise.
28945 (is_function_default_version): Likewise.
28946 * attribs.c (attr_strcmp): Likewise.
28947 (sorted_attr_string): Likewise.
28948 (common_function_versions): Likewise.
28949 (make_unique_name): Likewise.
28950 (make_dispatcher_decl): Likewise.
28951 (is_function_default_version): Likewise.
28952 * config/i386/i386.c (attr_strcmp): Likewise.
28953 (sorted_attr_string): Likewise.
28954 (ix86_function_versions): Likewise.
28955 (make_name): Likewise.
28956 (make_dispatcher_decl): Likewise.
28957 (is_function_default_version): Likewise.
28958 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28960 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28963 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28964 Account for direct move costs for vec_construct of integer
28967 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28970 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28971 (UNSPEC_STX_ATOMIC): Ditto.
28972 (loaddi_via_sse): New insn.
28973 (storedi_via_sse): Ditto.
28974 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28975 Update corresponding peephole2 patterns.
28976 (atomic_storedi_fpu): Ditto.
28978 2017-05-11 Julia Koval <julia.koval@intel.com>
28980 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28981 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28983 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28984 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28985 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28987 2017-05-11 Nathan Sidwell <nathan@acm.org>
28989 * graphite-poly.c: Include dumpfile.h.
28991 * dumpfle.h (dump_function): Declare here ...
28992 * tree-dump.h (dump_function): ... not here.
28993 * dumpfile.c: #include tree-cfg.h.
28994 (dump_function): Move here from ...
28995 * tree-dump.c (dump_function): ... here.
28996 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28997 * graphite-poly.c: Don't include tree-dump.h.
28998 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28999 * print-tree.c: Likewise.
29000 * stor-layout.c: Likewise.
29001 * tree-nested.c: Likewise.
29003 * dumpfile.c (dump_start): Use TDF_FLAGS.
29004 (dump_enable_all): Fix TDF_KIND check thinko.
29006 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29008 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29009 array entries to represent two legal parameterizations of the
29010 overloaded __builtin_cmpb function, as represented by the
29011 P6_OV_BUILTIN_CMPB constant.
29012 (altivec_resolve_overloaded_builtin): Add special case handling
29013 for the __builtin_cmpb function, as represented by the
29014 P6_OV_BUILTIN_CMPB constant.
29015 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29016 (BU_P6_64BIT_2): New macro.
29017 (BU_P6_OVERLOAD_2): New macro
29018 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29019 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29020 (CMPB): Add overload support to represent both 32-bit and 64-bit
29021 compare-bytes function.
29022 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29023 support for TARGET_CMPB.
29024 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29025 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29026 documentation of the __builtin_cmpb overloaded built-in function.
29028 2017-05-11 Richard Biener <rguenther@suse.de>
29030 PR tree-optimization/80705
29031 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29032 bases are not vectorizable.
29034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29036 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29037 when counting register pressure.
29039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29041 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29042 (iv_ca_more_deps): Renamed to ...
29043 (iv_ca_compare_deps): ... this.
29044 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29046 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29048 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29050 (determine_group_iv_costs): ... here.
29051 (find_inv_vars_cb): Record inv var if it's not recorded before.
29053 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29055 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29056 (get_shiftadd_cost): Ditto.
29058 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29060 * tree-ssa-address.c: Include header file.
29061 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29063 (add_to_parts): Refactor.
29064 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29065 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29068 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29070 PR tree-optimization/53090
29071 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29073 (extract_cond_operands): Detect condition with IV on both sides
29074 and return COMP_IV_EXPR_2.
29075 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29076 (rewrite_use_compare): Simplify by removing call to function
29077 extract_cond_operands.
29079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29081 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29082 (extract_cond_operands): Detect condition comparing against non-
29083 invariant bound and return appropriate enum value.
29084 (find_interesting_uses_cond): Update use of extract_cond_operands.
29085 Handle its return value accordingly.
29086 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29090 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29091 nonlinear iv_use computation in loop invariant sensitive way.
29093 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29095 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29096 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29098 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29100 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29101 (dump_cand): Support iv_cand.inv_exprs.
29102 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29104 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29107 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29109 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29111 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29112 as local function. Include necessary header files.
29113 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29117 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29119 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29121 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29122 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29123 RSHIFT_EXPR and BIT_NOT_EXPR.
29125 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29127 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29128 (adjust_setup_cost): New parameter supporting round up adjustment.
29129 (struct address_cost_data): Delete.
29130 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29131 (split_address_cost, ptr_difference_cost): Delete.
29132 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29133 (struct ainc_cost_data): New struct.
29134 (get_address_cost_ainc): New function.
29135 (get_address_cost, get_computation_cost): Reimplement.
29136 (determine_group_iv_cost_address): Record inv_expr for all uses of
29138 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29139 (iv_ca_has_deps): Reimplemented to ...
29140 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29142 (iv_ca_extend): Call iv_ca_more_deps.
29144 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29146 * tree-ssa-address.c (struct mem_address): Move to header file.
29147 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29148 * tree-ssa-address.h (struct mem_address): Move from C file.
29149 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29153 * tree-affine.h (aff_combination_type): New interface.
29154 (aff_combination_zero_p): Remove static.
29155 (aff_combination_const_p): New interface.
29156 (aff_combination_singleton_var_p): New interfaces.
29158 2017-05-11 Richard Biener <rguenther@suse.de>
29160 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29161 Skip unreachable blocks and destinations.
29162 (eliminate): Move stmt removal and fixup ...
29163 (fini_eliminate): ... here. Skip inserted exprs.
29164 (pass_pre::execute): Move fini_pre after fini_eliminate.
29165 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29166 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29167 PRE to get rid of dead code that has invalid SSA form and
29168 split critical edges again.
29170 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29172 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29174 2017-05-11 Richard Biener <rguenther@suse.de>
29176 * passes.c (execute_function_todo): Verify loops if they are
29177 said to be up-to-date.
29178 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29179 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29181 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29184 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29185 handle calling assemble_external ourself.
29188 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29189 modes with zero size. Enhance comment.
29191 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29193 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29194 built-ins for vec_xl and vec_xst with short and char pointer
29197 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29199 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29200 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29201 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29202 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29203 (_mm_maskz_min_round_ss): New intrinsics.
29204 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29205 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29206 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29207 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29208 (__builtin_ia32_minss_mask_round): New builtins.
29209 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29210 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29211 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29213 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29214 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29216 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29219 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29221 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29222 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29223 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29224 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29225 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29226 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29227 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29228 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29229 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29230 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29231 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29232 (__builtin_ia32_mulss_mask_round): New builtins.
29233 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29234 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29235 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29237 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29238 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29240 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29243 2017-05-10 Julia Koval <julia.koval@intel.com>
29245 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29246 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29247 (_mm256_setr_m128i): New intrinsics.
29249 2017-05-10 Julia Koval <julia.koval@intel.com>
29251 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29252 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29253 (_mm_maskz_rcp14_ss): New intrinsics.
29254 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29255 (__builtin_ia32_rcp14ss_mask): New builtins.
29256 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29258 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29260 PR tree-optimization/51513
29261 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29262 (assert_unreachable_fallthru_edge_p): Use it.
29263 (group_case_labels_stmt): Likewise.
29264 * tree-cfg.h: Prototype it.
29265 * stmt.c: Include cfghooks.h and tree-cfg.h.
29266 (emit_case_dispatch_table) <gap_label>: New local variable.
29267 Use it to fill dispatch table gaps.
29268 Test for default_label before updating probabilities.
29269 (expand_case) <default_label>: Remove unneeded initialization.
29270 Test for unreachable default case statement and remove its edge.
29271 Set default_label accordingly.
29272 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29274 2017-05-10 Carl Love <cel@us.ibm.com>
29276 * config/rs6000/rs6000-c: Add support for built-in functions
29277 vector signed char vec_neg (vector signed char)
29278 vector signed short int vec_neg (vector short int)
29279 vector signed int vec_neg (vector signed int)
29280 vector signed long long vec_neg (vector signed long long)
29281 vector float vec_neg (vector float)
29282 vector double vec_neg (vector double)
29283 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29285 * config/rs6000/altivec.h: Add define for vec_neg
29286 * doc/extend.texi: Update the built-in documentation for the
29287 new built-in functions.
29289 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29291 PR tree-optimization/77644
29292 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29294 2017-05-10 Nathan Sidwell <nathan@acm.org>
29296 * dumpfile.h (TDI_lang_all): New.
29297 (TDF_KIND): New. Renumber others
29298 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29300 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29302 (get_dump_file_name): Adjust suffix generation.
29303 (dump_enable_all): Use TDF_KIND.
29304 * doc/invoke.texi (-fdump-lang-all): Document.
29306 * dumpfile.h: Tabify.
29308 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29311 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29312 Move member access before delete.
29314 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29316 * tree-inline.c (expand_call_inline): Split block at stmt
29319 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29322 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29323 are now unused after splitting mov{sf,sd}_hardfloat.
29324 (f32_lr2): Likewise.
29325 (f32_lm): Likewise.
29326 (f32_lm2): Likewise.
29327 (f32_li): Likewise.
29328 (f32_li2): Likewise.
29329 (f32_lv): Likewise.
29330 (f32_sr): Likewise.
29331 (f32_sr2): Likewise.
29332 (f32_sm): Likewise.
29333 (f32_sm2): Likewise.
29334 (f32_si): Likewise.
29335 (f32_si2): Likewise.
29336 (f32_sv): Likewise.
29337 (f32_dm): Likewise.
29338 (f32_vsx): Likewise.
29339 (f32_av): Likewise.
29340 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29341 For movsf, order stores so the VSX stores occur before the GPR
29342 store which encourages the register allocator to use a traditional
29343 FPR instead of a GPR. For movsd, order the stores so that the GPR
29344 store comes before the VSX stores to allow the power6 to work.
29345 This is due to the power6 not having a 32-bit integer store
29346 instruction from a FPR.
29347 (movsf_hardfloat): Likewise.
29348 (movsd_hardfloat): Likewise.
29350 2017-05-09 Martin Sebor <msebor@redhat.com>
29352 PR translation/80280
29353 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29356 PR translation/80280
29357 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29358 data member added in r247778.
29359 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29361 2017-05-09 Nathan Sidwell <nathan@acm.org>
29363 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29365 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29368 2017-05-09 Marek Polacek <polacek@redhat.com>
29370 * doc/invoke.texi: Fix typo.
29372 2017-05-09 Richard Biener <rguenther@suse.de>
29374 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29375 (vrp_val_is_min): Likewise.
29376 (set_value_range_to_value): Likewise.
29377 (set_value_range_to_nonnegative): Likewise.
29378 (gimple_assign_nonzero_p): Likewise.
29379 (gimple_stmt_nonzero_p): Likewise.
29380 (vrp_int_const_binop): Likewise. Remove unreachable case.
29381 (adjust_range_with_scev): Adjust comments.
29382 (compare_range_with_value): Likewise.
29383 (extract_range_from_phi_node): Likewise.
29384 (test_for_singularity): Likewise.
29386 2017-05-09 Richard Biener <rguenther@suse.de>
29388 * tree-vrp.c (get_single_symbol): Add assert that we don't
29389 get overflowed constants as invariant part.
29390 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29391 checks. Use wi::cmp instead of recursing for integer constants.
29392 (compare_values): Just ignore whether we assumed undefined
29393 overflow instead of failing the compare.
29394 (extract_range_for_var_from_comparison_expr): Add comment before the
29395 TREE_NO_WARNING sets.
29396 (test_for_singularity): Likewise.
29397 (extract_range_from_comparison): Do not disable optimization
29398 when we assumed undefined overflow.
29399 (extract_range_basic): Remove init of unused var.
29401 2017-05-09 Richard Biener <rguenther@suse.de>
29403 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29404 (extract_range_from_multiplicative_op_1): Adjust.
29405 (extract_range_from_binary_expr_1): Use int_const_binop.
29407 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29410 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29411 rs6000_store_data_bypass_p in seven define_bypass directives and
29412 in several comments.
29413 * config/rs6000/rs6000-protos.h: Add prototype for
29414 rs6000_store_data_bypass_p function.
29415 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29416 function implements slightly different (rs6000-specific) semantics
29417 than store_data_bypass_p, returning false rather than aborting
29418 with assertion error when arguments do not satisfy the
29419 requirements of store data bypass.
29420 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29421 rs6000_store_data_bypass_p.
29423 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29425 * config/xtensa/xtensa-protos.h
29426 (xtensa_initial_elimination_offset): New declaration.
29427 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29428 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29429 macro definition, add case for FRAME_POINTER_REGNUM when
29430 FRAME_GROWS_DOWNWARD.
29431 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29432 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29433 xtensa_initial_elimination_offset.
29435 2017-05-08 Nathan Sidwell <nathan@acm.org>
29437 * doc/invoke.texi: Alphabetize -fdump options.
29439 2017-05-08 Martin Sebor <msebor@redhat.com>
29441 PR translation/80280
29442 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29444 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29446 * target.def (compute_frame_layout): New optional target hook.
29447 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29448 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29449 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29451 * reload1.c (verify_initial_elim_offsets): Likewise.
29452 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29453 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29454 (arm_get_frame_offsets): Split up into this ...
29455 (arm_compute_frame_layout): ... and this function.
29457 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29459 * config/aarch64/constraints.md (Usa): New constraint.
29460 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29462 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29464 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29465 with_multilib_list after it has been checked.
29467 2017-05-08 Richard Biener <rguenther@suse.de>
29469 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29470 (bitmap_set_subtract_values): Likewise.
29472 2017-05-08 Richard Biener <rguenther@suse.de>
29474 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29475 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29477 (gimple_stmt_nonzero_warnv_p): Rename to ...
29478 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29480 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29481 (extract_range_basic): Adjust, do not disable propagation on
29482 strict overflow sensitive simplification.
29483 (vrp_visit_cond_stmt): Likewise.
29485 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29487 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29488 body size unconditionally.
29490 2017-05-07 Jeff Law <law@redhat.com>
29493 2017-05-06 Jeff Law <law@redhat.com>
29494 PR tree-optimization/78496
29495 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29498 PR tree-optimization/78496
29499 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29500 (simplify_stmt_using_ranges): Call it.
29501 (vrp_dom_walker::before_dom_children): Extract equivalences
29502 from an ASSERT_EXPR with an equality comparison against a
29505 2017-05-06 Jeff Law <law@redhat.com>
29507 PR tree-optimization/78496
29508 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29511 PR tree-optimization/78496
29512 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29513 (simplify_stmt_using_ranges): Call it.
29514 (vrp_dom_walker::before_dom_children): Extract equivalences
29515 from an ASSERT_EXPR with an equality comparison against a
29518 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29520 * lra-constraints.c (lra_copy_reg_equiv): New function.
29521 (split_reg): Use it to copy equivalence information from the
29522 original register to the spill register.
29524 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29526 PR rtl-optimization/75964
29527 * simplify-rtx.c (simplify_const_relational_operation): Remove
29528 invalid handling of comparisons of integer ABS.
29530 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29532 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29533 initialize to zero.
29534 (init_regs): Remove declaration.
29535 (function_arg_advance_32): Initialize error_p as boolean variable.
29537 2017-05-05 Nathan Sidwell <nathan@acm.org>
29539 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29540 lines. Use for (;;).
29542 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29544 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29545 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29546 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29547 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29548 VF=2 that require versioning.
29550 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29552 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29555 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29557 * diagnostic.h (diagnostic_override_option_index): Convert from
29558 macro to inline function.
29560 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29562 * diagnostic.c (last_module_changed_p): New function.
29563 (set_last_module): New function.
29564 (diagnostic_report_current_module): Convert macro usage to
29565 the above functions.
29566 * diagnostic.h (diagnostic_context::last_module): Strengthen
29567 from const line_map * to const line_map_ordinary *.
29568 (diagnostic_last_module_changed): Delete macro.
29569 (diagnostic_set_last_module): Delete macro.
29571 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29573 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29574 with diagnostic_report_diagnostic.
29575 (diagnostic_n_impl_richloc): Likewise.
29576 * diagnostic.h (report_diagnostic): Delete macro.
29577 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29578 with diagnostic_report_diagnostic.
29579 * substring-locations.c (format_warning_va): Likewise.
29581 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29583 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29584 save/restor of format_spec. Move option-printing code to...
29585 (print_option_information): ...this new function, and
29586 reimplement by simply printing to the pretty_printer,
29587 rather than appending to the format string.
29589 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29591 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29592 handling logic into...
29593 (update_effective_level_from_pragmas): ...this new function.
29595 2017-05-04 Andrew Waterman <andrew@sifive.com>
29597 * config/riscv/riscv.opt (mstrict-align): New option.
29598 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29599 (SLOW_UNALIGNED_ACCESS): Define.
29600 (riscv_slow_unaligned_access): Declare.
29601 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29603 (riscv_slow_unaligned_access): New variable.
29604 (rocket_tune_info): Set slow_unaligned_access to true.
29605 (optimize_size_tune_info): Set slow_unaligned_access to false.
29606 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29607 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29608 (riscv_option_override): Set riscv_slow_unaligned_access.
29609 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29611 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29613 * config/riscv/riscv.md: Unify indentation.
29615 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29620 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29622 (extendsi<mode>2): Add support for doing sign extension via
29623 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29624 don't have ISA 3.0 instructions.
29625 (extendsi<mode>2 splitter): Likewise.
29626 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29627 generate the normal insns since SImode can now go in vector
29628 registers. Disallow the special UNSPECs needed for previous
29629 machines to hide SImode being used. Add new insns
29630 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29631 (fix_trunc<mode>si2_stfiwx): Likewise.
29632 (fix_trunc<mode>si2_internal): Likewise.
29633 (fixuns_trunc<mode>si2): Likewise.
29634 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29635 (fctiw<u>z_<mode>_smallint): Likewise.
29636 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29637 of floating point to 32-bit integer from doing a direct move to
29638 the GPR registers to do a store.
29639 (fctiwz_<mode>): Break long line.
29641 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29643 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29644 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29645 (addr_list, addr_offset_valid_p): New.
29646 (split_address_groups): Check offset validity with above function.
29647 (gt-tree-ssa-loop-ivopts.h): Include header file.
29649 2017-05-05 Nathan Sidwell <nathan@acm.org>
29651 * config.gcc (arm*-*-*): Add missing 'fi'.
29653 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29655 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29656 included in -fopt-info does not matter.
29657 * doc/optinfo.texi (-fopt-info): Fix description of default
29658 behavour. Explicitly say order of options included in -fopt-info
29661 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29663 * config.gcc: Allow combinations of aprofile and rmprofile values for
29664 --with-multilib-list.
29665 * config/arm/t-multilib: New file.
29666 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29667 variables. Remove setting of ISA and floating-point ABI in
29668 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29669 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29670 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29671 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29673 * config/arm/t-rmprofile: Likewise except for the matches changes.
29674 * doc/install.texi (--with-multilib-list): Document the combination of
29675 aprofile and rmprofile values and warn about pitfalls in doing that.
29677 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29679 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29680 (movdi_aarch64): Likewise.
29682 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29684 PR tree-optimization/80632
29685 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29687 (build_arrays): Initialize it for virtual phis.
29688 (fix_phi_nodes): Use it for virtual phis.
29690 PR tree-optimization/80558
29691 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29692 [x, y] op z into [x op, y op z] for op & or | if conditions
29695 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29696 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29699 * config/arm/arm.md (use_literal_pool): Remove.
29700 (64-bit immediate split): No longer takes cost into consideration
29701 if arm_disable_literal_pool is enabled.
29702 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29703 used when arm_disable_literal_pool is enabled.
29704 (arm_max_const_double_inline_cost): Remove use of
29705 arm_disable_literal_pool.
29706 (push_minipool_fix): Add assert.
29707 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29708 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29709 (no_literal_pool_sf_immediate): New.
29711 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29713 PR tree-optimization/80613
29714 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29715 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29717 2017-05-05 Richard Biener <rguenther@suse.de>
29719 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29721 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29723 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29724 of this flag from insn conditions due to removal from r247495.
29726 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29728 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29730 (arm_early_store_addr_dep_ptr): Likewise.
29731 * config/arm/aarch-common-protos.h
29732 (arm_early_load_addr_dep_ptr): Add prototype.
29733 (arm_early_store_addr_dep_ptr): Likewise.
29734 * config/arm/cortex-a53.md: Add new bypasses.
29736 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29738 * tree.c (next_type_uid): Change type to unsigned.
29739 (type_hash_canon): Decrement back next_type_uid if
29740 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29741 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29744 2017-05-04 Martin Sebor <msebor@redhat.com>
29746 * builtins.c: Fix a trivial typo in a comment.
29748 PR middle-end/79234
29749 * builtins.c (check_sizes): Adjust to handle reading past the end.
29750 Avoid printing excessive upper bound of ranges. Use %E to print
29751 tree nodes instead of converting them to %wu.
29752 (expand_builtin_memchr): New function.
29753 (compute_dest_size): Rename...
29754 (compute_objsize): ...to this.
29755 (expand_builtin_memcpy): Adjust.
29756 (expand_builtin_mempcpy): Adjust.
29757 (expand_builtin_strcat): Adjust.
29758 (expand_builtin_strcpy): Adjust.
29759 (check_strncat_sizes): Adjust.
29760 (expand_builtin_strncat): Adjust.
29761 (expand_builtin_strncpy): Adjust and simplify.
29762 (expand_builtin_memset): Adjust.
29763 (expand_builtin_bzero): Adjust.
29764 (expand_builtin_memcmp): Adjust.
29765 (expand_builtin): Handle memcmp.
29766 (maybe_emit_chk_warning): Check strncat just once.
29768 2017-05-04 Martin Sebor <msebor@redhat.com>
29770 PR preprocessor/79214
29771 PR middle-end/79222
29772 PR middle-end/79223
29773 * builtins.c (check_sizes): Add inlining context and issue
29774 warnings even when -Wno-system-headers is set.
29775 (check_strncat_sizes): Same.
29776 (expand_builtin_strncat): Same.
29777 (expand_builtin_memmove): New function.
29778 (expand_builtin_stpncpy): Same.
29779 (expand_builtin): Handle memmove and stpncpy.
29781 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29783 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29784 which is not used any more.
29786 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29788 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29790 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29792 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29793 (cortexa53_tunings): Likewise.
29794 (cortexa57_tunings): Likewise.
29795 (cortexa72_tunings): Likewise.
29796 (cortexa73_tunings): Likewise.
29798 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29800 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29801 Set loop alignment to 8.
29803 2017-05-04 Martin Sebor <msebor@redhat.com>
29805 PR translation/80280
29806 * builtins.c (expand_builtin_object_size): Add missing quoting to
29807 %D and like directives.
29808 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29809 (hsa_type_for_tree_type): Same.
29810 (verify_function_arguments): Same.
29811 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29812 * varasm.c (get_section): Same.
29815 2017-05-04 Martin Sebor <msebor@redhat.com>
29817 PR translation/80280
29818 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29820 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29822 * config/aarch64/aarch64.c (generic_addrcost_table):
29823 Change HI/TI mode setting.
29825 2017-05-04 Martin Jambor <mjambor@suse.cz>
29827 PR tree-optimization/80622
29828 * tree-sra.c (comes_initialized_p): New function.
29829 (build_accesses_from_assign): Only set write lazily when
29830 comes_initialized_p is false.
29831 (analyze_access_subtree): Use comes_initialized_p.
29832 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29833 instead of testing for PARM_DECL.
29835 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29837 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29838 constraint on operand 0 to allow more general addressing modes.
29839 Adjust output template.
29840 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29842 * config/aarch64/aarch64-protos.h
29843 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29844 * config/aarch64/constraints.md (Dp): New address constraint.
29845 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29848 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29850 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29851 update use of estimate_ipcp_clone_size_and_time.
29852 (estimate_local_effects): Update use of
29853 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29854 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29855 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29856 Return nonspecialized time.
29858 2017-05-04 Richard Biener <rguenther@suse.de>
29860 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29861 for the last VUSE which def dominates the PHI. Directly call
29863 (get_continuation_for_phi_1): Remove.
29865 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29867 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29868 to explain the use of truncating division. Cap the number of
29869 iterations to the maximum given by nb_iterations_upper_bound,
29872 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29874 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29875 * configure: Regenerate.
29876 * config.in: Regenerate.
29877 * config/i386/driver-mingw32.c: new file.
29878 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29879 * config.host: Link driver-mingw32.o on MinGW host.
29880 * doc/install.texi: Document new --enable-mingw-wildcard configure
29883 2017-05-04 Marek Polacek <polacek@redhat.com>
29885 PR tree-optimization/80612
29886 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29888 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29889 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29891 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29892 (movt splitter): Likewise.
29893 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29894 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29895 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29896 block for Thumb-1 with MOVT.
29897 (thumb2_legitimate_address_p): Move code block ...
29898 (can_avoid_literal_pool_for_label_p): ... into this new function.
29899 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29901 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29902 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29903 "M-profile targets with the MOVT instruction".
29905 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29907 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29908 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29909 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29911 2017-05-04 Martin Liska <mliska@suse.cz>
29913 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29914 variable cond_code.
29916 2017-05-04 Richard Biener <rguenther@suse.de>
29918 * tree.c (array_at_struct_end_p): Handle arrays at struct
29919 end with flexarrays more conservatively. Refactor and treat
29920 arrays of arrays or aggregates more strict. Fix
29921 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29922 * tree.c (array_at_struct_end_p): Adjust prototype.
29923 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29924 * gimple-fold.c (get_range_strlen): Likewise.
29925 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29927 2017-05-04 Richard Biener <rguenther@suse.de>
29929 PR tree-optimization/31130
29930 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29932 (supports_overflow_infinity): Likewise.
29933 (is_negative_overflow_infinity): Likewise.
29934 (is_positive_overflow_infinity): Likewise.
29935 (is_overflow_infinity): Likewise.
29936 (stmt_overflow_infinity): Likewise.
29937 (overflow_infinity_range_p): Likewise.
29938 (usable_range_p): Remove as always returning true.
29939 (make_overflow_infinity): Remove.
29940 (negative_overflow_infinity): Likewise.
29941 (positive_overflow_infinity): Likewise.
29942 (avoid_overflow_infinity): Likewise.
29943 (set_value_range): Adjust accordingly.
29944 (set_value_range_to_nonnegative): Likewise, remove now unused
29945 overflow_infinity arg.
29946 (vrp_operand_equal_p): Adjust.
29947 (update_value_range): Likewise.
29948 (range_int_cst_singleton_p): Likewise.
29949 (operand_less_p): Likewise.
29950 (compare_values_warnv): Likewise.
29951 (extract_range_for_var_from_comparison_expr): Likewise.
29952 (vrp_int_const_binop): Likewise.
29953 (zero_nonzero_bits_from_vr): Likewise.
29954 (extract_range_from_multiplicative_op_1): Likewise.
29955 (extract_range_from_binary_expr_1): Likewise.
29956 (extract_range_from_unary_expr): Likewise.
29957 (extract_range_from_comparison): Likewise.
29958 (extract_range_basic): Likewise.
29959 (adjust_range_with_scev): Likewise.
29960 (compare_ranges): Likewise.
29961 (compare_range_with_value): Likewise.
29962 (dump_value_range): Likewise.
29963 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29965 (simplify_cond_using_ranges): Adjust.
29967 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29969 * brig-builtins.def: Added a builtin for class_f64.
29970 * builtin-types.def: Added a builtin type needed by class_f64.
29972 2017-05-03 Jason Merrill <jason@redhat.com>
29974 * timevar.def: Add TV_CONSTEXPR.
29976 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29978 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29980 2017-05-03 Martin Jambor <mjambor@suse.cz>
29982 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29983 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29984 * ipa-cp.c (ipcp_propagate_stage): Do not call
29985 ipa_update_after_lto_read.
29986 * ipa-inline.c (ipa_inline): Likewise.
29988 2017-05-03 Martin Jambor <mjambor@suse.cz>
29990 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29991 tag. Added a default constructor and a destructor.
29992 (ipa_edge_args_sum_t): New class;
29993 (ipa_edge_args_sum): Declare.
29994 (ipa_edge_args_vector): Remove declaration.
29995 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29996 (ipa_free_edge_args_substructures): Remove declaration.
29997 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29998 (ipa_edge_args_info_available_for_edge_p): Likewise.
29999 * ipa-prop.c (ipa_edge_args_vector): Removed.
30000 (edge_removal_hook_holder): Likewise.
30001 (edge_duplication_hook_holder): Likewise.
30002 (ipa_edge_args_sum): New variable.
30003 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30004 ipa_edge_args_vector.
30005 (ipa_free_edge_args_substructures): Likewise.
30006 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30007 ipa_edge_args_vector.
30008 (ipa_edge_removal_hook): Turned into method
30009 ipa_edge_args_sum_t::remove.
30010 (ipa_edge_duplication_hook): Turned into method
30011 ipa_edge_args_sum_t::duplicate.
30012 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30013 registering edge hooks.
30014 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30015 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30016 ipa_edge_args_sum instead of ipa_edge_args_vector.
30017 * ipa-profile.c (ipa_profile): Likewise.
30019 2017-05-03 Martin Jambor <mjambor@suse.cz>
30021 * symbol-summary.h (function_summary): New method exists.
30022 (function_summary::symtab_removal): Deallocate through release.
30023 (call_summary): New class.
30024 (gt_ggc_mx): New overload.
30025 (gt_pch_nx): Likewise.
30026 (gt_pch_nx): Likewise.
30028 2017-05-03 Jeff Law <law@redhat.com>
30030 PR tree-optimization/78496
30031 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30032 from simplify_cond_using_ranges. Split off code to walk
30033 backwards through casts into ...
30034 (simplify_cond_using_ranges_2): New function.
30035 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30036 (execute_vrp): After identifying jump threads, call
30037 simplify_cond_using_ranges_2.
30039 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30042 * ipa-inline.h (inline_summary): Add ctor.
30043 (create_ggc): Do not use ggc_cleared_alloc.
30045 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30046 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30048 * gcc.c (handle_braces): Support escaping in switch matching
30050 * doc/invoke.texi (Spec Files): Document it.
30051 Remove superfluous @code markup in items.
30053 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30055 * diagnostic-show-locus.c (struct column_range): New struct.
30056 (get_affected_columns): New function.
30057 (get_printed_columns): New function.
30058 (struct correction): New struct.
30059 (correction::ensure_capacity): New function.
30060 (correction::ensure_terminated): New function.
30061 (struct line_corrections): New struct.
30062 (line_corrections::~line_corrections): New dtor.
30063 (line_corrections::add_hint): New function.
30064 (layout::print_trailing_fixits): Reimplement in terms of the new
30066 (selftest::test_overlapped_fixit_printing): New function.
30067 (selftest::diagnostic_show_locus_c_tests): Call it.
30069 2017-05-03 Nathan Sidwell <nathan@acm.org>
30071 Canonicalize canonical type hashing
30072 * tree.h (type_hash_canon_hash): Declare.
30073 * tree.c (type_hash_list, attribute_hash_list): Move into
30074 type_hash_canon_hash.
30075 (build_type_attribute_qual_variant): Break out hash code calc into
30076 type_hash_canon_hash.
30077 (type_hash_canon_hash): New. Generic type hash computation.
30078 (build_range_type_1, build_array_type_1, build_function_type,
30079 build_method_type_directly, build_offset_type, build_complex_type,
30080 make_vector_type): Call it.
30082 2017-05-03 Richard Biener <rguenther@suse.de>
30084 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30085 When all DRs have unknown misaligned do not always peel
30086 when there is a store but apply the same costing model as if
30087 there were only loads.
30089 2017-05-03 Richard Biener <rguenther@suse.de>
30092 PR tree-optimization/80492
30093 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30094 compare_base_decls returning dont-know properly.
30096 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30098 * config/arm/iterators.md (CCSI): New mode iterator.
30099 (arch): New mode attribute.
30100 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30101 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30102 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30103 code iterator for success result mode.
30104 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30105 the corresponding new insn generators.
30107 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30110 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30111 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30113 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30115 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30116 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30117 (DDR_A): Wrap DDR argument in brackets.
30118 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30119 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30120 (DDR_REVERSED_P): Likewise.
30122 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30124 PR tree-optimization/79472
30125 * tree-switch-conversion.c (struct switch_conv_info): Add
30126 contiguous_range and default_case_nonstandard fields.
30127 (collect_switch_conv_info): Compute contiguous_range and
30128 default_case_nonstandard fields, don't clear final_bb if
30129 contiguous_range and only the default case doesn't have the required
30131 (check_all_empty_except_final): Set default_case_nonstandard instead
30132 of failing if contiguous_range and the default case doesn't have empty
30134 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30135 and only the default case doesn't have the required constants. Skip
30137 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30138 if default_case_nonstandard.
30139 (build_constructors): Build constant 1 just once. Assert that default
30140 values aren't inserted in between cases if contiguous_range. Skip
30142 (build_arrays): Skip virtual phis.
30143 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30144 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30145 Handle virtual phis.
30146 (gen_inbound_check): Handle default_case_nonstandard case.
30147 (process_switch): Adjust check_final_bb caller. Call
30148 gather_default_values with the first non-default case instead of
30149 default case if default_case_nonstandard.
30151 2017-05-02 Nathan Sidwell <nathan@acm.org>
30153 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30154 check. Fix formatting.
30156 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30158 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30159 errors when comparing specialized and unspecialized times.
30161 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30163 * diagnostic-show-locus.c
30164 (layout::should_print_annotation_line_p): Make private.
30165 (layout::print_annotation_line): Make private.
30166 (layout::annotation_line_showed_range_p): Make private.
30167 (layout::show_ruler): Make private.
30168 (layout::print_source_line): Make private. Pass in line and
30169 line_width, rather than calling location_get_source_line. Drop
30171 (layout::print_leading_fixits): New method.
30172 (layout::print_any_fixits): Rename to...
30173 (layout::print_trailing_fixits): ...this, and make private.
30174 Don't print newline fixits.
30175 (diagnostic_show_locus): Move logic for printing one row into...
30176 (layout::print_line): ...this new function. Move the
30177 location_get_source_line call and error-handling from
30178 print_source_line to here. Call print_leading_fixits, and rename
30179 print_any_fixits to print_trailing_fixits.
30180 (selftest::test_fixit_insert_containing_newline): Update now that
30181 newlines are partially supported.
30182 (selftest::test_fixit_insert_containing_newline_2): New test.
30183 (selftest::test_fixit_replace_containing_newline): Update comments.
30184 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30185 * edit-context.c (class added_line): New class.
30186 (class edited_line): Describe newline handling in comment.
30187 (edited_line::actually_edited_p): New method.
30188 (edited_line::print_content): Delete redundant decl.
30189 (edited_line::m_predecessors): New field.
30190 (edited_file::print_content): Call edited_line::print_content.
30191 (edited_file::print_diff): Update to support newlines.
30192 (edited_file::print_diff_hunk): Likewise.
30193 (edited_file::print_run_of_changed_lines): New function.
30194 (edited_file::print_diff_line): Convert to...
30195 (print_diff_line): ...this.
30196 (edited_file::get_effective_line_count): New function.
30197 (edited_line::edited_line): Initialize new field m_predecessors.
30198 (edited_line::~edited_line): Clean up m_predecessors.
30199 (edited_line::apply_fixit): Handle newlines.
30200 (edited_line::get_effective_line_count): New function.
30201 (edited_line::print_content): New function.
30202 (edited_line::print_diff_lines): New function.
30203 (selftest::test_applying_fixits_insert_containing_newline): New
30205 (selftest::test_applying_fixits_replace_containing_newline): New
30207 (selftest::insert_line): New function.
30208 (selftest::test_applying_fixits_multiple_lines): Add example of
30210 (selftest::edit_context_c_tests): Call the new tests.
30212 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30214 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30215 parameter cand. Update dump information.
30216 (get_computation_cost): Update uses.
30218 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30220 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30221 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30222 (get_computation_at, rewrite_use_address): Update use of
30223 get_computation_aff.
30225 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30227 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30228 (get_computation): Delete.
30229 (get_computation_cost): Implement like get_computation_cost_at.
30230 Use get_computation_at.
30231 (get_computation_cost_at): Delete.
30232 (rewrite_use_nonlinear_expr): Use get_computation_at.
30233 (rewrite_use_compare, remove_unused_ivs): Ditto.
30235 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30237 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30239 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30241 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30242 (ivopts_global_cost_for_size): Rename parameter and update uses.
30243 (iv_ca_recount_cost): Update uses.
30244 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30245 candidates seperately in n_invs and n_cands.
30246 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30248 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30250 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30251 (find_inv_vars_cb): New.
30252 (find_depends): Renamed to ...
30253 (find_inv_vars): ... this.
30254 (add_candidate_1, force_var_cost): Call find_inv_vars.
30255 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30257 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30259 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30260 inv_vars. Add inv_exprs.
30261 (struct iv_cand): Rename depends_on to inv_vars.
30262 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30263 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30264 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30265 (dump_cand): Dump inv_vars.
30266 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30267 (record_invariant, find_depends, add_candidate_1): Ditto.
30268 (set_group_iv_cost, force_var_cost): Ditto.
30269 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30270 (get_computation_cost_at, get_computation_cost): Ditto.
30271 (determine_group_iv_cost_generic): Ditto.
30272 (determine_group_iv_cost_address): Ditto.
30273 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30274 (determine_group_iv_costs): Ditto.
30275 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30276 (iv_ca_set_remove_invariants): Renamed to ...
30277 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30278 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30279 (iv_ca_set_add_invariants): Renamed to ...
30280 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30281 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30282 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30283 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30284 (create_new_ivs): Remove useless dump.
30286 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30288 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30290 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30291 (iv_ca_set_no_cp, create_new_iv): Ditto.
30293 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30295 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30297 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30299 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30300 function tree_check2.
30302 2017-05-02 Martin Liska <mliska@suse.cz>
30304 * doc/gcov.texi: Add missing preposition.
30305 * gcov.c (function_info::function_info): Properly fill up
30306 all member variables.
30308 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30310 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30312 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30314 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30316 2017-05-02 Martin Liska <mliska@suse.cz>
30319 * lto-streamer-in.c (lto_read_tree_1): Remove
30320 LTO_STREAMER_DEBUG.
30321 * lto-streamer.c (struct tree_hash_entry): Likewise.
30322 (struct tree_entry_hasher): Likewise.
30323 (tree_entry_hasher::hash): Likewise.
30324 (tree_entry_hasher::equal): Likewise.
30325 (lto_streamer_init): Likewise.
30326 (lto_orig_address_map): Likewise.
30327 (lto_orig_address_get): Likewise.
30328 (lto_orig_address_remove): Likewise.
30329 * lto-streamer.h: Likewise.
30330 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30331 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30333 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30335 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30336 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30337 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30338 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30339 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30340 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30341 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30342 (mm_maskz_sub_ss): New intrinsics.
30343 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30344 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30345 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30346 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30347 (__builtin_ia32_subss_mask_round): New builtins.
30348 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30349 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30350 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30352 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30353 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30355 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30358 2017-05-02 Martin Jambor <mjambor@suse.cz>
30360 PR tree-optimization/78687
30361 * tree-sra.c (access): New field parent.
30362 (process_subtree_disqualification): New function.
30363 (disqualify_candidate): Call it.
30364 (build_accesses_from_assign): Reset write flag if creating an
30366 (build_access_subtree): Fill in parent field and also prpagate
30367 down grp_write flag.
30368 (create_artificial_child_access): New parameter set_grp_write, set
30369 grp_write to its value.
30370 (propagate_subaccesses_across_link): Also propagate grp_write flag
30372 (propagate_all_subaccesses): Push the closest parent back to work
30373 queue if add_access_to_work_queue returned true.
30375 2017-05-02 Richard Biener <rguenther@suse.de>
30377 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30378 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30379 -fstrict-overflow documentation.
30380 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30381 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30382 flag_strict_overflow.
30383 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30384 * lto-opts.c (lto_write_options): Do not stream it.
30385 * lto-wrapper.c (merge_and_complain): Do not handle it.
30386 * opts.c (default_options_table): Do not set -fstrict-overflow.
30387 (finish_options): Likewise do not clear it when sanitizing.
30388 * simplify-rtx.c (simplify_const_relational_operation): Do not
30389 test flag_strict_overflow.
30391 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30393 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30394 using enabled attribute.
30395 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30396 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30397 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30398 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30399 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30400 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30401 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30402 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30403 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30404 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30406 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30408 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30410 2017-05-02 Richard Biener <rguenther@suse.de>
30412 PR tree-optimization/80591
30414 2017-04-10 Richard Biener <rguenther@suse.de>
30416 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30419 2017-05-02 Richard Biener <rguenther@suse.de>
30421 PR tree-optimization/80549
30422 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30423 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30424 headers if they do not have a preheader.
30426 2017-05-02 Martin Liska <mliska@suse.cz>
30429 * common.opt: Fix typo.
30430 * doc/invoke.texi: Likewise.
30432 2017-05-01 Jan Beulich <jbeulich@suse.com>
30434 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30435 swapping, add (x,x,m,x,n) alternative.
30437 2017-05-01 Nathan Sidwell <nathan@acm.org>
30439 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30440 unnecessary unadjusted_alignment check.
30442 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30445 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30447 * gimplify.c (gimplify_cilk_detach): New function.
30448 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30449 * tree-core.h: Document EXPR_CILK_SPAWN.
30450 * tree.h (EXPR_CILK_SPAWN): Define.
30452 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30454 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30455 to use new fixit_hint representation, using the "replace" logic.
30456 (get_line_span_for_fixit_hint): Likewise.
30457 (layout::print_any_fixits): Likewise.
30458 (selftest::test_one_liner_many_fixits): Rename to...
30459 (selftest::test_one_liner_many_fixits_1): ...this, and update
30460 comment and expected output to reflect that the multiple fix-it
30461 hints are now consolidated into one insertion.
30462 (selftest::test_one_liner_many_fixits_2): New test.
30463 (selftest::test_diagnostic_show_locus_one_liner): Update for
30465 (selftest::test_fixit_consolidation): Update for fix-it API
30467 * diagnostic.c (print_parseable_fixits): Likewise.
30468 * edit-context.c (edited_line::m_line_events): Convert from
30469 auto_vec <line_event *> to auto_vec <line_event>.
30470 (class line_event): Convert from abstract base class to a concrete
30471 class, taking over the role of replace_event.
30472 (class insert_event): Delete.
30473 (class replace_event): Rename to class line_event. Convert to
30475 (edit_context::add_fixits): Reimplement.
30476 (edit_context::apply_insert): Delete.
30477 (edit_context::apply_replace): Rename to...
30478 (edit_context::apply_fixit): ...this. Convert to half-open range.
30479 (edited_file::apply_insert): Delete.
30480 (edited_file::apply_replace): Rename to...
30481 (edited_file::apply_fixit): ...this.
30482 (edited_line::~edited_line): Drop deletion of events.
30483 (edited_line::apply_insert): Delete.
30484 (edited_line::apply_replace): Rename to...
30485 (edited_line::apply_fixit): ...this. Convert to half-open range.
30486 Update for change to type of m_line_events.
30487 * edit-context.h (edit_context::apply_insert): Delete.
30488 (edit_context::apply_replace): Rename to...
30489 (edit_context::apply_fixit): ...this.
30491 2017-05-01 Martin Sebor <msebor@redhat.com>
30493 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30496 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30499 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30500 __get_cpuid_max returns 0.
30501 (__get_cpuid_count): Ditto.
30503 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30505 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30506 replacement expression is another instance of one of its arguments.
30508 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30511 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30512 check for stack push/pop autoinc.
30513 * config/i386/i386.c (ix86_agi_dependent): Return false
30514 if the only reason why modified_in_p returned true is that
30515 addr is SP based and set_insn is a push or pop.
30517 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30519 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30522 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30525 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30526 (account_size_time): Use two predicates - exec_pred and
30528 (evaluate_conditions_for_known_args): Compute both clause and
30530 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30531 (inline_summary_t::duplicate): Update.
30532 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30534 (compute_inline_parameters): Likewise.
30535 (estimate_edge_size_and_time): Update caluclation of time.
30536 (estimate_node_size_and_time): Compute both time and nonspecialized
30538 (estimate_ipcp_clone_size_and_time): Update.
30539 (inline_merge_summary): Update.
30540 (do_estimate_edge_time): Update.
30541 (do_estimate_edge_size): Update.
30542 (do_estimate_edge_hints): Update.
30543 (inline_read_section, inline_write_summary): Stream both new predicates.
30544 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30546 (compute_inlined_call_time): Cleanup.
30547 (big_speedup_p): Update.
30548 (edge_badness): Update.
30549 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30550 (size_time_entry): Replace predicate by exec_predicate and
30551 nonconst_predicate.
30552 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30553 (estimate_edge_time): Return also nonspec_time.
30554 (reset_edge_growth_cache): Update.
30556 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30558 PR rtl-optimization/80491
30559 * ifcvt.c (noce_process_if_block): When looking for x setter
30560 with missing else_bb, don't check only the insn right before
30561 cond_earliest, but look for the last insn that x is modified in
30562 within the same bb.
30564 PR rtl-optimization/80491
30565 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30567 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30569 PR tree-optimization/80487
30570 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30572 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30574 PR tree-optimization/79697
30575 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30576 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30577 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30579 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30580 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30582 2017-04-28 Martin Sebor <msebor@redhat.com>
30584 PR tree-optimization/80523
30585 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30586 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30588 (maybe_warn, format_directive, parse_directive): Use new functions.
30589 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30591 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30593 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30595 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30597 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30598 target_header_dir): Set correctly.
30599 * configure: Regenerated.
30600 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30601 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30602 instead of SYSTEM_HEADER_DIR.
30604 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30606 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30607 (estimate_local_effects): Likewise.
30608 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30609 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30610 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30611 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30612 do_estimate_edge_time, estimate_edge_time): Likewise.
30613 * ipa-inline-analysis.c (estimate_node_size_and_time,
30614 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30615 (estimate_time_after_inlining): Remove.
30617 2017-04-28 Martin Liska <mliska@suse.cz>
30619 * doc/gcov.texi: Enhance documentation of gcov.
30621 2017-04-28 Martin Liska <mliska@suse.cz>
30623 * doc/gcov.texi: Sort options in alphabetic order.
30624 * doc/gcov-dump.texi: Likewise.
30625 * doc/gcov-tool.texi: Likewise.
30626 * gcov.c (print_usage): Likewise.
30627 * gcov-dump.c (print_usage): Likewise.
30628 * gcov-tool.c (print_merge_usage_message): Likewise.
30629 (print_rewrite_usage_message): Likewise.
30630 (print_overlap_usage_message): Likewise.
30632 2017-04-28 Martin Liska <mliska@suse.cz>
30634 PR gcov-profile/53915
30635 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30637 2017-04-28 Martin Liska <mliska@suse.cz>
30639 PR gcov-profile/79891
30640 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30641 is marked by compiler as living on a line.
30642 (get_cycles_count): Remove usage of the union.
30643 (output_intermediate_file): Likewise.
30644 (find_source): Fix GNU coding style.
30645 (accumulate_line_counts): Remove old non-all block mode.
30646 (output_lines): Remove usage of the union.
30647 * profile.c (output_location): Include all BBs, even if
30648 belonging to a same line (and file) as a previous BB.
30650 2017-04-28 Martin Liska <mliska@suse.cz>
30652 * gcov.c (process_args): Handle new argument 'w'.
30653 (read_graph_file): Assign ID to BBs.
30654 (output_branch_count): Display BB # if verbose flag is set.
30655 (output_lines): Likewise for arcs.
30656 (print_usage): Add '--verbose' option help.
30657 * doc/gcov.texi: Document --verbose (-w) option.
30659 2017-04-28 Martin Liska <mliska@suse.cz>
30661 * gcov.c (struct block_location_info): New struct.
30662 (process_file): Fill up the new structure.
30663 (read_graph_file): Replace usage of encoding by the newly added
30665 (add_line_counts): Likewise.
30666 (accumulate_line_counts): Remove usage of the union.
30667 (function_info::function_info): New function.
30668 (function_info::~function_info): Likewise.
30669 (process_file): Call delete instead of release_function.
30670 (release_function): Release the function.
30671 (release_structures): Call delete instead of release_function.
30672 (solve_flow_graph): Replace usage of num_blocks.
30673 (find_exception_blocks): Likewise.
30674 (output_lines): Fix GNU coding style.
30676 2017-04-28 Martin Liska <mliska@suse.cz>
30679 * coverage.c (coverage_remove_note_file): New function.
30680 * coverage.h: Declare the function.
30681 * toplev.c (finalize): Clean if an error has been seen.
30683 2017-04-28 Martin Liska <mliska@suse.cz>
30685 PR gcov-profile/80031
30686 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30687 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30688 * gcov.c (read_graph_file): Read just number of blocks.
30689 * profile.c (branch_prob): Do not stream 0 flags per a basic
30692 2017-04-28 Martin Liska <mliska@suse.cz>
30694 * gcov-dump.c (tag_*): Add new argument to declarations.
30695 (dump_gcov_file): Likewise.
30696 (tag_blocks): Add and use new argument depth.
30697 (tag_arcs): Likewise.
30698 (tag_lines): Likewise.
30699 (tag_counters): Likewise.
30700 (tag_summary): Likewise.
30701 (dump_working_sets): Use depth to do a proper indentation.
30703 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30706 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30707 * symtab.c (symtab_node::debug_symtab): Move definition here.
30709 2017-04-28 Richard Biener <rguenther@suse.de>
30711 * lto-streamer.h (LTO_major_version): Bump to 7.
30713 2017-04-28 Richard Biener <rguenther@suse.de>
30715 * tree-vrp.c (assert_info): New struct.
30716 (add_assert_info): New helper.
30717 (register_edge_assert_for_2): Refactor to add asserts to a vector
30719 (register_edge_assert_for_1): Likewise.
30720 (register_edge_assert_for): Likewise.
30721 (finish_register_edge_assert_for): New helper actually registering
30722 asserts where live on edge.
30723 (find_conditional_asserts): Adjust.
30724 (find_switch_asserts): Likewise.
30725 (evrp_dom_walker::try_find_new_range): Generalize.
30726 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30728 2017-04-27 Marek Polacek <polacek@redhat.com>
30731 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30732 arg10 and arg11 to itype.
30734 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30736 * doc/extend.texi (Object Size Checking): Improve grammar.
30738 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30741 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30742 that the logic for permitting reciprocal estimates matches that
30745 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30748 * tree.c (type_cache_hasher::equal): Only compare
30749 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30750 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30751 non-aggregate element types.
30752 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30753 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30755 2017-04-27 Richard Biener <rguenther@suse.de>
30757 PR middle-end/80533
30758 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30759 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30760 keeping a reference to a trailing array.
30762 2017-04-27 Richard Biener <rguenther@suse.de>
30764 PR middle-end/80539
30765 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30766 being in loop-closed SSA form conservatively.
30767 (chrec_fold_multiply_poly_poly): Likewise.
30769 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30771 PR middle-end/79665
30772 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30773 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30775 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30778 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30779 (aarch64_function_arg_alignment): Return unsigned int again, but still
30780 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30781 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30782 Don't emit -Wpsabi note.
30783 (aarch64_function_arg_boundary): Likewise.
30784 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30787 2017-04-26 Nathan Sidwell <nathan@acm.org>
30789 * tree.h (crc32_unsigned_n): Declare.
30790 (crc32_unsigned, crc32_unsigned): Make inline.
30791 * tree.c (crc32_unsigned_bits): Replace with ...
30792 (crc32_unsigned_n): ... this.
30793 (crc32_unsigned, crc32_byte): Remove.
30794 (crc32_string): Remove unnecessary braces.
30796 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30798 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30799 * ipa-inline-analysis.c (MAX_TIME): Remove.
30800 (account_size_time): Use sreal for time.
30801 (dump_inline_summary): Update.
30802 (estimate_function_body_sizes): Update.
30803 (estimate_edge_size_and_time): Update.
30804 (estimate_calls_size_and_time): Update.
30805 (estimate_node_size_and_time): Update.
30806 (inline_merge_summary): Update.
30807 (inline_update_overall_summary): Update.
30808 (estimate_time_after_inlining): Update.
30809 (inline_read_section): Update.
30810 (inline_write_summary): Update.
30811 * ipa-inline.c (compute_uninlined_call_time): Update.
30812 (compute_inlined_call_time): Update.
30813 (recursive_inlining): Update.
30814 (inline_small_functions): Update.
30815 (dump_overall_stats): Update.
30816 * ipa-inline.h: Include sreal.h.
30817 (size_time_entry): Turn time to sreal.
30818 (inline_summary): Turn self_time nad time to sreal.
30820 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30822 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30824 (sreal::stream_out, sreal::stream_in): New.
30825 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30827 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30829 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30832 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30835 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30837 (dimode_scalar_chain::compute_convert_gain): Ditto.
30838 (dimode_scalar_chain::make_vector_copies): Ditto.
30839 (dimode_scalar_chain::convert_reg): Ditto.
30840 (dimode_scalar_chain::convert_insn): Ditto.
30841 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30842 (VI248_AVX512BW_1): New mode iterator.
30843 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30844 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30847 2017-04-25 Martin Sebor <msebor@redhat.com>
30849 PR tree-optimization/80497
30850 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30851 constants are representable in HOST_WIDE_INT.
30852 (parse_directive): Ditto.
30854 2017-04-25 Martin Sebor <msebor@redhat.com>
30857 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30858 (new_zero_array): Adjust signature.
30859 (dom_info::dom_init): Used unsigned rather that size_t.
30860 (dom_info::dom_info): Same.
30862 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30863 Jakub Jelinek <jakub@redhat.com>
30866 * config/arm/arm.c: Include gimple.h.
30867 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30868 returns negative, increment ncrn only if it returned positive.
30869 (arm_needs_doubleword_align): Return int instead of bool,
30870 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30871 members, but if there is any such non-FIELD_DECL
30872 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30873 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30874 returns negative, increment nregs only if it returned positive.
30875 (arm_setup_incoming_varargs): Likewise.
30876 (arm_function_arg_boundary): Emit -Wpsabi note if
30877 arm_needs_doubleword_align returns negative, return
30878 DOUBLEWORD_ALIGNMENT only if it returned positive.
30880 2017-04-25 Marek Polacek <polacek@redhat.com>
30883 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30884 first argument to type.
30886 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30889 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30890 type checks to test for compatibility instead of equality.
30892 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30893 Jakub Jelinek <jakub@redhat.com>
30896 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30898 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30899 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30900 the alignment computation, but return their maximum in warn_alignment.
30901 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30902 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30904 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30905 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30908 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30910 * config/arc/simdext.md (dmpyh): Fix typo.
30912 2017-04-25 Richard Biener <rguenther@suse.de>
30914 PR tree-optimization/80492
30915 * alias.c (compare_base_decls): Handle registers with asm
30916 specification conservatively.
30917 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30918 compare_base_decls returning dont-know properly.
30920 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30922 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30923 (legitimate_offset_address_p): New function.
30924 (arc_legitimate_address_p): Use above function.
30926 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30928 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30930 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30932 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30933 ACCH registers whenever they are available.
30935 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30937 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30938 double regs fix when not used.
30940 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30942 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30944 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30945 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30947 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30949 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30950 when emitting short ADD instructions.
30952 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30954 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30956 (cmpsi_cc_c_insn): Likewise.
30957 (cbranchsi4_scratch): Compute proper instruction length using
30958 compact_hreg_operand.
30959 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30961 2017-04-25 Richard Biener <rguenther@suse.de>
30963 PR middle-end/80509
30964 * passes.c (pass_manager::pass_manager): Initialize
30965 m_name_to_pass_map.
30967 2017-04-25 Richard Biener <rguenther@suse.de>
30969 PR tree-optimization/79201
30970 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30972 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30975 * config/s390/vector.md: Split MEM->GPR vector moves for
30976 non-s_operand addresses.
30978 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30981 * config/s390/predicates.md (reload_const_wide_int_operand): New
30983 * config/s390/s390.md ("movti"): Remove d/P alternative.
30984 ("movti_bigconst"): New pattern definition.
30986 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30989 * s390-protos.h (s390_expand_cs_hqi): Removed.
30990 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30991 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30992 modes as well as CCZ1mode and CCZmode.
30993 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30994 signature of s390_emit_compare_and_swap.
30995 (s390_expand_cs_hqi): Likewise, make static.
30996 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30997 compare-and-swap, in some cases.
30998 (s390_expand_cs): Wrapper function.
30999 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31001 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31002 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31003 patterns for small and large integers. Forbid symref memory operands.
31004 Move expander to s390.c. Require cc register.
31005 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31006 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31007 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31008 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31009 symref memory operands. Remove CC mode and call s390_match_ccmode
31011 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31013 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31015 * config/s390/s390.md (define_peephole2): New peephole to help
31016 combining the load-and-test pattern with volatile memory.
31018 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31020 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31021 with CCZmode for TARGET_Z196.
31023 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31025 PR rtl-optimization/80501
31026 * combine.c (make_compound_operation_int): Set subreg_code to SET
31027 even for AND with mask of the sign bit of mode.
31029 PR rtl-optimization/80500
31030 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31031 sum's initial value.
31033 2017-04-25 Julian Brown <julian@codesourcery.com>
31034 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31036 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31038 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31040 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31042 2017-04-25 Julian Brown <julian@codesourcery.com>
31043 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31045 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31046 (thunderx2t99_sha): New Reservation.
31048 2017-04-25 Julian Brown <julian@codesourcery.com>
31049 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31051 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31052 type for 1-element load.
31054 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31056 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31058 2017-04-24 Martin Jambor <mjambor@suse.cz>
31060 PR tree-optimization/80293
31061 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31062 char arrays not totally scalarizable if it is false.
31063 (analyze_all_variable_accesses): Pass correct value in the new
31064 parameter. Add a statistics counter.
31066 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31068 PR middle-end/79931
31069 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31071 2017-04-24 Richard Biener <rguenther@suse.de>
31073 PR tree-optimization/80494
31074 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31075 out for complex types.
31077 2017-04-24 Richard Biener <rguenther@suse.de>
31079 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31080 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31081 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31082 (DFS): Adjust and never fail.
31083 (sccvn_dom_walker::fail): Remove.
31084 (sccvn_dom_walker::before_dom_children): Adjust.
31085 (run_scc_vn): Likewise and never fail.
31086 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31087 (pass_fre::execute): Likewise.
31089 2017-04-24 Richard Biener <rguenther@suse.de>
31091 PR tree-optimization/79725
31092 * tree-ssa-sink.c (statement_sink_location): Return whether
31093 failure reason was zero uses. Move that check later.
31094 (sink_code_in_bb): Deal with zero uses by removing the stmt
31097 2017-04-24 Richard Biener <rguenther@suse.de>
31100 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31101 pointer-based references.
31103 2017-04-24 Richard Biener <rguenther@suse.de>
31106 * pass_manager.h (pass_manager::operator new): Remove.
31107 (pass_manager::operator delete): Likewise.
31108 * passes.c (pass_manager::operator new): Remove.
31109 (pass_manager::operator delete): Likewise.
31110 (pass_manager::pass_manager): Zero individual pass members.
31112 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31115 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31116 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31117 Check "XEXP (src, 1)" operand here.
31118 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31119 Check "XEXP (src, 1)" operand here.
31120 (dimode_scalar_chain::make_vector_copies): Detect count register
31121 of a shift instruction. Zero extend count register from QImode
31122 to DImode to satisfy vector shift pattern count operand predicate.
31123 Substitute vector shift count operand with a DImode copy.
31124 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31127 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31129 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31130 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31131 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31132 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31133 (UNSPEC_NOREX_MEM): Remove definition.
31135 2017-04-21 Richard Biener <rguenther@suse.de>
31137 PR tree-optimization/79547
31138 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31139 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31140 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31141 without any constraints.
31143 2017-04-21 Richard Biener <rguenther@suse.de>
31145 PR tree-optimization/78847
31146 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31148 2017-04-21 Richard Biener <rguenther@suse.de>
31150 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31151 (build_distinct_type_copy): Likewise.
31152 (build_variant_type_copy): Likewise.
31153 * tree.c (build_qualified_type): Pass down mem-stat info.
31154 (build_distinct_type_copy): Likewise.
31155 (build_variant_type_copy): Likewise.
31157 2017-04-21 Richard Biener <rguenther@suse.de>
31159 PR tree-optimization/80237
31160 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31162 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31163 for a simplified result.
31165 2016-04-21 Richard Biener <rguenther@suse.de>
31167 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31168 sth as strict as a simple_iv but a chrec without symbols and an
31169 operand defined in the loop we are peeling (and not some subloop).
31170 (propagate_constants_for_unrolling): Propagate all constants.
31172 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31175 * config/i386/i386.c (print_reg): Remove assert for disalowed
31176 regno values, call output_operand_lossage instead.
31178 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31181 * config/i386/constraints.md (Yc): New register constraint.
31182 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31183 Use Yc constraint for alternative 2 of operand 0. Remove
31184 preferred_for_speed attribute.
31186 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31188 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31189 lastprivate clauses in SIMT case.
31191 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31193 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31195 2017-04-20 Richard Biener <rguenther@suse.de>
31197 PR tree-optimization/57796
31198 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31199 as N scalar stores.
31200 (vect_model_load_cost): Cost gathers as N scalar loads.
31202 2017-04-20 Richard Biener <rguenther@suse.de>
31204 * ggc-page.c (ggc_allocated_p): Rename to ...
31205 (safe_lookup_page_table_entry): ... this and return the lookup
31207 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31209 2017-04-20 Richard Biener <rguenther@suse.de>
31211 PR tree-optimization/80453
31212 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31213 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31214 from the conditions.
31215 (vn_phi_eq): Pass them down.
31216 (vn_phi_lookup): Record them.
31217 (vn_phi_insert): Likewise.
31219 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31221 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31222 uninitialized variable warning to avoid buffer overrun.
31224 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31227 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31228 is suppressed for '{ 0 }' in C.
31230 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31232 * BASE-VER: Set to 8.0.0.
31234 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31236 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31237 priority .init_array and .fini_array section with SECTION_NOTYPE
31240 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31242 PR middle-end/80423
31243 * tree.h (build_array_type): Add typeless_storage default argument.
31244 * tree.c (type_cache_hasher::equal): Also compare
31245 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31246 (build_array_type): Add typeless_storage argument, set
31247 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31249 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31250 (build_array_type): Likewise. Add typeless_storage argument.
31252 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31253 Jakub Jelinek <jakub@redhat.com>
31255 PR tree-optimization/80426
31256 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31257 operation on symbolic operands, also compute the overflow for the
31258 invariant part when the operation degenerates into a negation.
31260 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31263 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31264 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31267 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31269 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31272 * config/avr/avr.c (tree.h): Include it.
31273 (cgraph.h): Include it.
31274 (avr_encode_section_info): Don't warn for uninitialized progmem
31275 variable if it's just an alias.
31277 2017-04-19 Richard Biener <rguenther@suse.de>
31280 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31281 when needed by AutoPGO.
31283 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31286 * doc/lto.texi: Remove an extra 'that'.
31288 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31290 PR rtl-optimization/80429
31291 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31292 are only used in debug insns.
31294 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31295 Vladimir Makarov <vmakarov@redhat.com>
31297 * config/sparc/predicates.md (input_operand): Add comment. Return
31298 true for any memory operand when LRA is in progress.
31299 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31301 2017-04-18 Jeff Law <law@redhat.com>
31304 * mips.md ({return,simple_return}_internal): Do not overwrite
31307 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31309 PR tree-optimization/80443
31310 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31311 instead of adding 1, subtract -1 and similarly instead of subtracting
31314 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31316 PR rtl-optimization/80357
31317 * haifa-sched.c (tmp_bitmap): New variable.
31318 (model_recompute): Handle duplicate use records.
31319 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31320 (free_global_sched_pressure_data): Free it.
31322 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31325 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31326 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31327 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31328 instead of SYSTEM_HEADER_DIR.
31330 2017-04-18 Jeff Law <law@redhat.com>
31332 PR middle-end/80422
31333 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31334 predecessors after walking up the insn chain.
31336 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31339 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31340 sizetype type into debug info.
31342 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31345 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31346 unneeded test for TARGET_UPPER_REGS_SF.
31347 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31349 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31352 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31353 instead of gsi_after_labels.
31355 2017-04-18 Jeff Law <law@redhat.com>
31357 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31361 2017-04-13 Jeff Law <law@redhat.com>
31362 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31363 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31365 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31368 * config/avr/avr.c (intl.h): Include it.
31369 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31371 2017-04-18 Martin Liska <mliska@suse.cz>
31373 PR gcov-profile/78783
31374 * gcov-tool.c (gcov_output_files): Validate that destination
31375 file is either removed by the tool or by a user.
31377 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31378 Guy Benyei <guybe@mellanox.com>
31380 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31381 block, and do not negate it, the stored id is already negative.
31383 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31385 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31387 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31390 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31391 masks of options that should be turned off if the VSX vector
31392 options are turned off.
31393 (OTHER_P8_VECTOR_MASKS): Likewise.
31394 (OTHER_VSX_VECTOR_MASKS): Likewise.
31395 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31396 rs6000_disable_incompatible_switches to validate no type switches
31398 (rs6000_incompatible_switch): New function to disallow turning on
31399 other vector options if -mno-vsx, -mno-power8-vector, or
31400 -mno-power9-vector are specified.
31402 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31404 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31406 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31408 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31409 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31410 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31411 (ARG_POINTER_CFA_OFFSET): Likewise.
31413 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31415 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31416 conditions to take advantage of various optimizations.
31418 2017-04-13 Jeff Law <law@redhat.com>
31420 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31421 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31422 (zero_extendsidi2_dext): Likewise.
31424 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31427 * fold-const.c (fold_ternary_loc): Revert
31428 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31431 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31433 PR rtl-optimization/80343
31434 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31435 new scratch pseudo.
31437 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31440 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31441 to ubsan_encode_value.
31443 2017-04-13 Jeff Law <law@redhat.com>
31445 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31446 appearing in DEBUG_INSNs.
31448 2017-04-13 Martin Liska <mliska@suse.cz>
31450 PR gcov-profile/80413
31451 * gcov-io.c (gcov_write_string): Copy to buffer just when
31452 allocated size is greater than zero.
31454 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31457 * dwarf2out.c (decls_for_scope): Ignore declarations of
31458 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31460 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31463 * ipa-visibility.c (non_local_p): Fix typos.
31464 (localize_node): When localizing symbol in same comdat group,
31465 dissolve the group only when we know external symbols are going
31467 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31469 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31471 PR tree-optimization/79390
31472 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31473 order does not result in usable sequence, retry with reversed operand
31479 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31480 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31481 op0 instead of fold_convert_loc (loc, type, arg0).
31483 2017-04-12 Jeff Law <law@redhat.com>
31485 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31486 has a delay slot in the generated code.
31488 * config/cris/cris.md (cris_preferred_reload_class): Return
31489 GENNONACR_REGS rather than GENERAL_REGS.
31491 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31494 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31495 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31496 signedness of the result type.
31498 2017-04-12 Richard Biener <rguenther@suse.de>
31499 Jeff Law <law@redhat.com>
31501 PR tree-optimization/80359
31502 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31503 trim stores to TARGET_MEM_REFs.
31505 2017-04-12 Richard Biener <rguenther@suse.de>
31507 PR tree-optimization/79390
31508 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31509 threading case even more.
31511 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31514 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31515 for quad_address_p for TImode, instead of just not indexed_address.
31517 2017-04-12 Richard Biener <rguenther@suse.de>
31518 Bernd Edlinger <bernd.edlinger@hotmail.de>
31520 PR middle-end/79671
31521 * alias.c (component_uses_parent_alias_set_from): Handle
31522 TYPE_TYPELESS_STORAGE.
31523 (get_alias_set): Likewise.
31524 * tree-core.h (tree_type_common): Add typeless_storage flag.
31525 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31526 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31527 for types containing members with TYPE_TYPELESS_STORAGE.
31528 (place_field): Likewise.
31529 (layout_type): Likewise for ARRAY_TYPE.
31530 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31531 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31532 TYPE_TYPELESS_STORAGE.
31533 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31535 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31538 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31539 first argument to type.
31541 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31545 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31546 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31547 (rs6000_expand_binop_builtin): Likewise.
31548 (rs6000_expand_ternop_builtin): Likewise; also add missing
31549 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31550 vshasigma built-ins.
31551 * doc/extend.texi: Document that vec_xxpermdi's third argument
31552 must be a constant.
31554 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31556 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31557 Use shift_const cost parameter when calculating gain of STV shifts.
31559 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31561 PR rtl-optimization/70478
31562 * lra-constraints.c (process_alt_operands): Check memory for
31563 disfavoring memory insn operand.
31565 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31567 PR middle-end/80100
31568 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31569 left shift in unsigned HOST_WIDE_INT type.
31571 PR rtl-optimization/80385
31572 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31573 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31576 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31577 if they have any depend clauses.
31579 2017-04-11 Martin Liska <mliska@suse.cz>
31582 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31583 * ipa-split.c (split_function): Create a local comdat symbol
31584 if caller is in a comdat group.
31586 2017-04-11 Martin Liska <mliska@suse.cz>
31589 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31592 2017-04-11 Martin Sebor <msebor@redhat.com>
31594 PR middle-end/80364
31595 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31596 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31598 (directive::set_width, directive::set_precision, format_character):
31600 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31603 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31606 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31607 conflict, set target->arch_name instead of target->cpu_name.
31609 2017-04-11 Richard Biener <rguenther@suse.de>
31611 PR tree-optimization/80374
31612 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31613 build_zero_cst, remove fold_convertible_p check again.
31615 2017-04-11 Martin Liska <mliska@suse.cz>
31618 * ubsan.c (instrument_object_size): Do not instrument register
31621 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31624 * config/i386/i386-builtin-types.def
31625 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31626 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31627 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31628 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31629 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31630 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31631 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31632 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31633 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31634 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31635 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31636 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31637 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31638 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31639 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31640 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31641 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31642 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31643 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31644 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31645 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31646 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31647 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31648 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31649 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31650 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31651 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31652 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31653 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31654 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31655 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31656 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31657 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31658 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31659 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31660 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31661 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31662 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31663 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31664 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31665 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31666 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31667 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31668 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31669 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31670 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31672 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31673 flag to second_arg_count, handle 4 argument function type _COUNT
31674 aliases, handle second_arg_count on second argument rather than last.
31676 2017-04-10 Jeff Law <law@redhat.com>
31678 PR tree-optimization/80374
31679 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31680 record anything if we can not convert integer_zero_node to the
31683 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31686 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31687 Enhance special handling given to the TARGET_P9_MINMAX option in
31688 relation to certain other options.
31690 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31692 PR tree-optimization/80153
31693 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31694 remove POINTER_PLUS_EXPR's base part directly, rather than through
31697 2017-04-10 Richard Biener <rguenther@suse.de>
31698 Bin Cheng <bin.cheng@arm.com>
31700 PR tree-optimization/80153
31701 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31702 the first element of pointer type aff_tree. Build result expr in
31704 (add_elt_to_tree): Convert to type unconditionally. Remove other
31705 fold_convert calls.
31706 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31707 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31709 2017-04-10 Richard Biener <rguenther@suse.de>
31711 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31714 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31716 PR rtl-optimization/70478
31717 * lra-constraints.c (curr_small_class_check): New.
31718 (update_and_check_small_class_inputs): New.
31719 (process_alt_operands): Update curr_small_class_check. Disfavor
31720 alternative insn memory operands. Check available regs for small
31723 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31726 * config/mips/mips.opt (-mvirt): Update description.
31727 * doc/invoke.texi (-mvirt): Likewise.
31729 2017-04-10 Richard Biener <rguenther@suse.de>
31731 PR middle-end/80362
31732 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31733 looking for NEGATE_EXPR in -A / -B to A / B folding.
31735 2017-04-10 Martin Liska <mliska@suse.cz>
31737 PR gcov-profile/80224
31738 * gcov.c (print_usage): Fix usage string.
31739 (get_gcov_intermediate_filename): Remove.
31740 (output_gcov_file): Use both for normal and intermediate format.
31741 (generate_results): Do not initialize special file for
31742 intermediate format.
31744 2017-04-10 Richard Biener <rguenther@suse.de>
31746 PR tree-optimization/80304
31747 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31750 2017-04-10 Nathan Sidwell <nathan@acm.org>
31753 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31754 (rs6000_init_builtins): Use it.
31756 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31758 * config/arm/arm.md (<mrc>): Add mode to SET source.
31759 (<mrrc>): Likewise.
31761 2017-04-10 Richard Biener <rguenther@suse.de>
31763 PR middle-end/80344
31764 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31766 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31769 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31770 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31771 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31772 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31773 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31774 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31775 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31776 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31777 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31778 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31779 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31780 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31781 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31782 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31783 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31784 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31785 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31786 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31787 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31788 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31789 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31790 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31791 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31793 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31795 PR rtl-optimization/70478
31796 * lra-constraints.c: Reverse the last patch.
31798 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31800 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31801 Add comment for WCHAR_T.
31803 2017-04-08 Martin Liska <mliska@suse.cz>
31806 2017-04-07 Martin Liska <mliska@suse.cz>
31809 * ipa-split.c (split_function): Add function part to a same comdat
31812 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31815 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31817 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31819 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31820 (vec_pairing): ...to this.
31821 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31822 (rs6000_sched_init): Adjust for name change.
31823 (struct rs6000_sched_context): Likewise.
31824 (rs6000_init_sched_context): Likewise.
31825 (rs6000_set_sched_context): Likewise.
31827 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31833 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31835 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31836 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31837 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31839 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31841 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31843 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31845 PR rtl-optimization/70703
31846 * ira-color.c (update_conflict_hard_regno_costs): Use
31847 int64_t instead of HOST_WIDE_INT.
31849 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31851 PR rtl-optimization/70478
31852 * lra-constraints.c (process_alt_operands): Disfavor alternative
31853 insn memory operands.
31855 2017-04-07 Jeff Law <law@redhat.com>
31857 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31858 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31860 2017-04-07 Martin Liska <mliska@suse.cz>
31863 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31864 Show error message instead of an ICE.
31866 2017-04-07 Martin Liska <mliska@suse.cz>
31869 * ipa-split.c (split_function): Add function part to a same comdat
31872 2017-04-07 Richard Biener <rguenther@suse.de>
31874 PR middle-end/80341
31875 * tree.c (get_unwidened): Also handle ! for_type case for
31877 * convert.c (do_narrow): Split out from ...
31878 (convert_to_integer_1): ... here. Do not pass final truncation
31879 type to get_unwidened for TRUNC_DIV_EXPR.
31881 2017-04-07 Richard Biener <rguenther@suse.de>
31883 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31885 (aff_combination_const): Adjust.
31886 (aff_combination_scale): Likewise.
31887 (aff_combination_add_elt): Likewise.
31888 (aff_combination_add_cst): Likewise.
31889 (aff_combination_convert): Likewise.
31890 (add_elt_to_tree): Likewise. Remove unused argument.
31891 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31893 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31895 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31897 * config/arm/arm.c (arm_default_short_enums): Use
31898 ARM_DEFAULT_SHORT_ENUMS.
31899 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31901 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31904 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31905 members with redundant out-of-class redeclaration.
31907 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31910 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31911 * config/i386/i386.md (*zero_extendsidi2):
31912 Add (?*x,*x) and (?*v,*v) alternatives.
31914 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31917 * config/i386/i386.c (ix86_expand_builtin)
31918 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31919 mode from insn data. Convert operands to insn operand mode.
31920 Copy operands that don't satisfy insn predicate to a register.
31922 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31924 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31927 2017-04-06 Richard Biener <rguenther@suse.de>
31929 PR tree-optimization/80334
31930 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31931 preserve alignment of accesses.
31933 2017-04-06 Richard Biener <rguenther@suse.de>
31935 PR tree-optimization/80262
31936 * tree-sra.c (build_ref_for_offset): Preserve address-space
31938 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31939 Drop useless address-space information on MEM_REF offsets.
31941 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31943 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31945 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31947 PR rtl-optimization/70703
31948 * ira-color.c (update_conflict_hard_regno_costs): Use
31949 HOST_WIDE_INT instead of long.
31951 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31954 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31955 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31957 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31958 for x86_64 target. Handle -m3dnowa option.
31960 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31962 PR rtl-optimization/70703
31963 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31964 (update_conflict_hard_regno_costs): Use long instead of unsigned
31965 arithmetic for cost calculation.
31967 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31968 Bernd Edlinger <bernd.edlinger@hotmail.de>
31971 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31974 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31977 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31978 ptr_mode with Pmode throughout.
31979 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31980 into probe_stack_range and use DImode.
31982 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31985 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31986 call_eh_return is true.
31988 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31990 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31991 Initialize last_match_fntype_index.
31993 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31996 * tree-nvr.c: Include internal-fn.h.
31997 (pass_return_slot::execute): Ignore internal calls without
32000 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32001 Richard Biener <rguenther@suse.de>
32004 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32005 captures used multiple times, except for the last use.
32006 * generic-match-head.c: Include gimplify.h.
32008 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32010 PR tree-optimization/79390
32011 * target.h (struct noce_if_info): Declare.
32012 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32013 * target.def (noce_conversion_profitable_p): New target hook.
32014 * ifcvt.h (struct noce_if_info): New type, moved from ...
32015 * ifcvt.c (struct noce_if_info): ... here.
32016 (noce_conversion_profitable_p): Renamed to ...
32017 (default_noce_conversion_profitable_p): ... this. No longer
32019 (noce_try_store_flag_constants, noce_try_addcc,
32020 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32021 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32022 instead of noce_conversion_profitable_p.
32023 * config/i386/i386.c: Include ifcvt.h.
32024 (ix86_option_override_internal): Don't override
32025 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32026 (ix86_noce_conversion_profitable_p): New function.
32027 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32028 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32029 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32030 * doc/tm.texi: Regenerated.
32032 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32034 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32037 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32040 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32041 instructions for small multiply cores.
32043 2017-04-04 Jeff Law <law@redhat.com>
32045 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32047 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32048 that are not set by the loop over the elements.
32050 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32053 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32054 int mode, convert_modes it to mode as unsigned, otherwise use
32055 lowpart_subreg to mode rather than SImode.
32056 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32057 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32058 Use DImode instead of SImode for the shift count operand.
32059 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32062 2017-04-04 Richard Biener <rguenther@suse.de>
32064 PR middle-end/80281
32065 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32066 arithmetic done for the negate or the plus. Simplify.
32067 (A - (-B) -> A + B): Likewise.
32068 * fold-const.c (split_tree): Make sure to not negate pointers.
32070 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32072 PR rtl-optimization/60818
32073 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32074 a compare of comparisons with the thing compared if this results
32075 in a different machine mode.
32077 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32079 * alias.c (base_alias_check): Fix typo in comment.
32080 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32081 * cgraphunit.c (symbol_table::compile): Likewise.
32082 * collect2.c (maybe_run_lto_and_relink): Likewise.
32083 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32084 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32085 * config/avr/avr.c (avr_map_op_t): Likewise.
32086 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32087 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32088 * config/epiphany/epiphany.md (movcc): Likewise.
32089 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32090 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32092 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32093 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32094 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32095 * config/sh/sh.c (sh_rtx_costs): Likewise.
32096 * fold-const.c (fold_truth_andor): Likewise.
32097 * genautomata.c (collapse_flag): Likewise.
32098 * gengtype.h (struct type::u::s): Likewise.
32099 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32100 * input.c (FORMAT_AMOUNT): Likewise.
32101 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32102 (known_aggs_to_agg_replacement_list): Likewise.
32103 * ipa-inline-analysis.c: Likewise.
32104 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32105 * ipa-polymorphic-call.c
32106 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32107 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32108 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32110 * modulo-sched.c (apply_reg_moves): Likewise.
32111 * omp-expand.c (build_omp_regions_1): Likewise.
32112 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32113 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32114 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32115 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32116 * value-prof.c: Likewise.
32117 * var-tracking.c (val_reset): Likewise.
32119 2017-04-03 Richard Biener <rguenther@suse.de>
32121 PR tree-optimization/80275
32122 * fold-const.c (split_address_to_core_and_offset): Handle
32125 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32127 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32128 descriptors is at least equal to that of functions.
32130 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32132 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32134 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32137 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32138 (mov<IMOD4:mode>): New expander.
32139 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32141 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32143 PR rtl-optimization/79405
32144 * fwprop.c (propagations_left): New variable.
32145 (forward_propagate_into): Decrement it.
32146 (fwprop_init): Initialize it.
32147 (fw_prop): If the variable has reached zero, stop propagating.
32148 (fwprop_addr): Ditto.
32150 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32153 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32154 a FUNCTION_DECL, pass it as decl instead of origin to
32157 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32159 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32162 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32165 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32166 TARGET_VSX_SMALL_INTEGER.
32168 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32170 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32171 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32173 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32175 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32176 extraction from odd-numbered MSA register.
32178 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32180 PR middle-end/80173
32181 * expmed.c (store_bit_field_1): Don't attempt to create
32182 a word subreg out of hard registers wider than word if they
32183 have HARD_REGNO_NREGS of 1 for their mode.
32185 PR middle-end/80163
32186 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32187 conversions to integer types wider than word and pointer.
32190 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32191 (rtx_equal_for_cselib_p): Pass 0 to it.
32192 * cselib.c (cselib_hasher::equal): Likewise.
32193 (rtx_equal_for_cselib_1): Add depth argument. If depth
32194 is 128, don't look up VALUE locs and punt. Increment
32195 depth in recursive calls when walking VALUE locs.
32197 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32199 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32200 (make_gcov_file_name): Use the canonical path name for generating
32202 (read_line): Fix handling of files with ascii null bytes.
32204 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32206 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32207 to initialise a vector register instead
32208 of using a const_int.
32210 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32212 PR translation/80189
32213 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32214 diagnostic messages.
32216 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32219 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32220 (dfp_diex_<mode>): Update mode of operand 1.
32221 * doc/extend.texi (dxex, dxexq): Document change to return type.
32222 (diex, diexq): Document change to argument type.
32224 2017-03-30 Martin Jambor <mjambor@suse.cz>
32227 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32228 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32229 it reflects the signature changes performed at the callee side.
32230 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32231 to cgraph_build_function_type_skip_args.
32232 (build_function_decl_skip_args): Adjust call to the above function.
32234 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32237 * config/i386/sse.md
32238 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32239 register as dest whenever it is a MEM not rtx_equal_p to the
32240 corresponding dup operand, and when forcing into reg move the
32241 reg into the memory afterwards.
32242 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32243 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32244 for the force_reg mode.
32245 (avx512vl_vextractf128<mode>): Use register as dest either
32246 always when a MEM, or when it is a MEM not rtx_equal_p to the
32247 corresponding dup operand, or even not when it is a CONST_VECTOR
32248 depending on the mode and lo vs. hi.
32249 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32251 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32252 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32253 Likewise. Require that operands[2] is even.
32254 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32255 Remove extraneous parens. Require that operands[2] is a multiple
32257 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32258 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32259 disallow memory then.
32261 2017-03-30 Richard Biener <rguenther@suse.de>
32263 PR tree-optimization/77498
32264 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32265 to non-constants over backedges.
32267 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32269 PR rtl-optimization/80233
32270 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32271 as last_combined_insn. Do not test for BARRIER_P separately.
32273 2017-03-29 Andreas Schwab <schwab@suse.de>
32276 * calls.c (prepare_call_address): Convert funexp to Pmode before
32277 copying to temp reg.
32279 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32281 PR tree-optimization/80158
32282 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32283 Handle possible future case of more than one alternate
32285 (replace_rhs_if_not_dup): Likewise.
32286 (replace_one_candidate): Likewise.
32288 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32290 PR rtl-optimization/80193
32291 * ira.c (ira): Do not check allocation for LRA.
32293 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32295 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32296 (nvptx_output_simt_exit): Declare.
32297 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32298 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32299 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32300 (nvptx_declare_function_name): ...here. Emit declaration of local
32301 memory space buffer for omp_simt_enter insn.
32302 (nvptx_output_unisimt_switch): New.
32303 (nvptx_output_softstack_switch): New.
32304 (nvptx_output_simt_enter): New.
32305 (nvptx_output_simt_exit): New.
32306 * config/nvptx/nvptx.h (struct machine_function): New fields
32307 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32308 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32309 (UNSPECV_SIMT_EXIT): Ditto.
32310 (omp_simt_enter_insn): New insn.
32311 (omp_simt_enter): New expansion.
32312 (omp_simt_exit): New insn.
32313 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32315 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32316 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32317 (expand_GOMP_SIMT_EXIT): New.
32318 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32319 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32320 (GOMP_SIMT_EXIT): Ditto.
32321 * target-insns.def (omp_simt_enter): New insn.
32322 (omp_simt_exit): Ditto.
32323 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32325 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32326 (lower_rec_input_clauses): Likewise.
32327 (lower_lastprivate_clauses): Handle SIMT privatization.
32329 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32330 (ompdevlow_adjust_simt_enter): New.
32331 (find_simtpriv_var_op): New.
32332 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32333 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32335 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32336 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32337 (copy_decl_for_dup_finish): Ditto.
32339 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32341 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32344 * config/i386/i386.c (ix86_option_override_internal): Always
32345 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32347 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32349 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32351 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32353 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32354 mark new edge's irreducible flag accordign to it.
32355 (vect_do_peeling): Check loop preheader edge's irreducible flag
32356 and pass it to function slpeel_add_loop_guard.
32358 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32360 PR tree-optimization/80218
32361 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32362 Update block frequencies and counts.
32364 2017-03-28 Richard Biener <rguenther@suse.de>
32366 PR tree-optimization/78644
32367 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32368 of a simplification result we may not use it at all.
32370 2017-03-28 Richard Biener <rguenther@suse.de>
32373 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32374 without arguments, generate default definition of a SSA name.
32376 2017-03-28 Richard Biener <rguenther@suse.de>
32378 PR middle-end/80222
32379 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32380 TYPE_REF_CAN_ALIAS_ALL references.
32381 * fold-const.c (fold_indirect_ref_1): Likewise.
32383 2017-03-28 Martin Liska <mliska@suse.cz>
32386 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32387 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32389 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32390 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32392 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32393 (EXTRA_SPECS): Define.
32394 (SUBTARGET_EXTRA_SPECS): Likewise.
32395 (SUBTARGET_CPP_SPEC): Likewise.
32396 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32397 SUBTARGET_EXTRA_SPECS.
32398 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32400 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32402 * config/arc/simdext.md (vst64_insn): Update pattern.
32403 (vld32wh_insn): Likewise.
32404 (vld32wl_insn): Likewise.
32405 (vld64_insn): Likewise.
32406 (vld32_insn): Likewise.
32408 2017-03-28 Marek Polacek <polacek@redhat.com>
32411 * fold-const.c (fold_comparison): Use protected_set_expr_location
32412 instead of SET_EXPR_LOCATION.
32414 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32416 * tree.c (add_expr): Avoid name lookup warning.
32418 2017-03-27 Jeff Law <law@redhat.com>
32420 PR tree-optimization/80216
32421 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32422 function name. Limit recursion depth.
32423 (record_temporary_equivalences): Corresponding changes.
32425 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32427 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32430 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32433 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32435 * cfgcleanup.c (reg_note_cfa_p): New array.
32436 (insns_have_identical_cfa_notes): New function.
32437 (old_insns_match_p): Don't cross-jump in between /f
32438 and non-/f instructions. If both i1 and i2 are frame related,
32439 verify all CFA notes, their order and content.
32441 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32444 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32445 HImode and SImode with zero extend to DImode to one insn.
32446 (bswap<mode>2_extenddi): Likewise.
32447 (bswapsi2_extenddi): Likewise.
32448 (bswaphi2_extendsi): Likewise.
32449 (bswaphi2): Combine bswap HImode and SImode into one insn.
32450 Separate memory insns from swapping register.
32451 (bswapsi2): Likewise.
32452 (bswap<mode>2): Likewise.
32453 (bswaphi2_internal): Delete, no longer used.
32454 (bswapsi2_internal): Likewise.
32455 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32456 store, and gpr<-gpr swap insns.
32457 (bswap<mode>2_store): Likewise.
32458 (bswaphi2_reg): Register only splitter, combine with the splitter.
32459 (bswaphi2 splitter): Likewise.
32460 (bswapsi2_reg): Likewise.
32461 (bswapsi2 splitter): Likewise.
32462 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32463 the insns into load, store, and register/register insns.
32464 (bswapdi2_ldbrx): Likewise.
32465 (bswapdi2_load): Likewise.
32466 (bswapdi2_store): Likewise.
32467 (bswapdi2_reg): Likewise.
32469 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32471 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32472 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32474 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32477 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32480 special handling for target option conflicts between dform
32481 options (-mpower9-dform, -mpower9-dform-vector,
32482 -mpower9-dform-scalar) and -mno-direct-move.
32484 2017-03-27 Richard Biener <rguenther@suse.de>
32486 PR tree-optimization/80181
32487 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32489 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32491 * config/arc/predicates.md (move_double_src_operand): Replace the
32492 call to move_double_src_operand with a call to address_operand.
32494 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32496 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32497 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32498 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32500 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32502 * config/arc/predicates.md (long_immediate_loadstore_operand):
32503 Consider scaled addresses cases.
32505 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32507 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32508 restored when in interrupt.
32509 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32510 doesn't have delay slot.
32512 2017-03-27 Richard Biener <rguenther@suse.de>
32515 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32516 inlined thunk clones.
32518 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32521 * asan.c (instrument_derefs): Copy over last operand from
32522 original COMPONENT_REF to the new COMPONENT_REF with
32523 DECL_BIT_FIELD_REPRESENTATIVE.
32524 * ubsan.c (instrument_object_size): Likewise.
32526 2017-03-27 Richard Biener <rguenther@suse.de>
32528 PR tree-optimization/80170
32529 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32530 sure DR/SCEV didnt fold in constants we do not see when looking
32531 at the reference base alignment.
32533 2017-03-27 Richard Biener <rguenther@suse.de>
32535 PR middle-end/80171
32536 * gimple-fold.c (fold_ctor_reference): Properly guard against
32537 NULL return value from canonicalize_constructor_val.
32539 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32542 * config/i386/i386.c (ix86_expand_builtin)
32543 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32544 flags reg setting and flags reg using instructions.
32545 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32546 clobbering instructions to zero extend op2.
32548 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32550 * doc/install.texi (Configuration) <--with-aix-soname>:
32551 Update link to AIX ld.
32553 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32555 PR rtl-optimization/80160
32556 PR rtl-optimization/80159
32557 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32558 reg_alternate_class into account.
32560 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32563 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32564 to consider in curr_insn_transform.
32566 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32568 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32569 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32570 and emit_mode_inner.
32572 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32574 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32575 argument to the overloaded builtin variants. Use the new flag to
32576 deprecate certain builtin variants.
32577 * config/s390/s390-builtin-types.def: Add new builtin types.
32578 * config/s390/s390-builtins.h: Support new flags field for
32579 overloaded builtins.
32580 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32581 (s390_macro_to_expand): Enable vector float data type.
32582 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32583 builtins by incrementing the __VEC__ version number.
32584 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32586 (s390_resolve_overloaded_builtin): Emit error messages depending
32587 on the builtin flags.
32588 * config/s390/s390.c (s390_expand_builtin): Support additional
32589 flags argument. Change error message to match the messages
32590 emitted in s390-c.c.
32591 * config/s390/s390.md: New UNSPEC_* constants.
32592 (op_type): Add new instruction types.
32593 * config/s390/vecintrin.h: Add new builtins and test data class
32595 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32596 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32597 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32598 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32599 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32600 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32602 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32603 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32604 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32605 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32607 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32608 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32609 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32610 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32612 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32613 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32614 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32615 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32616 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32617 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32618 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32620 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32621 ("vec_scatter_element<V_HW_4:mode>_DI")
32622 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32623 ("vec_fpint<mode>", "vflls")
32624 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32625 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32626 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32627 ("*vec_cmphe<mode>_cc"): ... these.
32629 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32630 mode constant instead of magic value.
32632 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32634 * config/s390/s390.c (s390_expand_vec_compare): Support other
32635 vector floating point modes than just V2DF.
32636 (s390_expand_vcond): Likewise.
32637 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32638 (s390_cannot_change_mode_class): Prevent mode changes between TF
32639 and V1TF in vector registers.
32640 * config/s390/s390.md (DF, SF): New mode attributes.
32641 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32642 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32643 SFmode support for VRs.
32644 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32646 (VFT, VF_HW): New mode iterators.
32647 (vw, sdx): New mode attributes.
32648 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32649 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32650 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32651 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32652 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32653 also the new vector floating point modes. Renaming to ...
32655 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32656 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32657 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32658 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32659 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32660 ("vec_unordered<mode>"): ... these.
32662 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32663 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32664 ("*vec_extendv2df"): New insn definitions.
32666 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32668 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32669 ("mulditi3_2", "*muldi3_sign"): New patterns.
32670 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32671 rename the pattern definition.
32673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32675 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32677 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32681 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32682 instruction if possible.
32683 * config/s390/vector.md (vec_halfnumelts): New mode
32685 ("*vec_vllezlf<mode>"): New pattern.
32687 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32689 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32690 ("popcountv4si2", "popcountv2di2"): Rename to ...
32691 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32692 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32694 ("popcount<mode>2_vxe"): New pattern.
32696 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32698 * common/config/s390/s390-common.c (processor_flags_table): Add
32700 * config.gcc: Add arch12.
32701 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32702 Default to arch12 for unknown CPU model numbers.
32703 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32704 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32705 PROCESSOR_max sanity check.
32706 * config/s390/s390-opts.h (enum processor_type): Add
32708 * config/s390/s390.c (processor_table): Add arch12.
32709 (s390_expand_builtin): Add check for B_VXE flag.
32710 (s390_issue_rate): Add PROCESSOR_ARCH12.
32711 (s390_get_sched_attrmask): Likewise.
32712 (s390_get_unit_mask): Likewise.
32713 (s390_sched_score): Enable z13 scheduling for arch12.
32714 (s390_sched_reorder): Likewise.
32715 (s390_sched_variable_issue): Likewise.
32716 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32718 (s390_tune_attr): Use z13 scheduling also for arch12.
32719 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32720 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32721 (TARGET_VXE_P): New macros.
32722 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32723 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32724 * config/s390/s390.opt: Add arch12 as processor_type.
32726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32728 * config/s390/s390.md
32729 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32730 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32731 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32733 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32734 Rename expanders to ...
32736 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32737 ("fixuns_truncdddi2_emu"): ... these.
32739 ("fixuns_trunc<mode>si2_emu"): New expander.
32741 ("*fixuns_truncdfdi2_z13"): Rename to ...
32742 ("*fixuns_truncdfdi2_vx"): ... this.
32744 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32746 * config/s390/2964.md: Remove the single element vector compare
32747 instructions which are no longer used.
32748 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32750 (s390_canonicalize_comparison): Remove handling of DFmode
32752 (s390_expand_vec_compare_scalar): Remove function.
32753 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32754 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32756 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32760 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32762 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32763 will anyway by matched by mov<mode>_64dfp.
32765 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32767 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32768 vlef/vstef. Add missing operand to vleif.
32770 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32772 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32773 pair for all vector types with 64 bit elements.
32774 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32775 * config/s390/vector.md (V_HW_64): ... here.
32776 (V_128_NOSINGLE): New mode iterator.
32777 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32778 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32779 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32780 ("*vec_load_pairv2di"): Change to ...
32781 ("*vec_load_pair<mode>"): ... this one.
32783 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32785 * config/s390/constraints.md: Add comments.
32786 (jKK): Reject element sizes > 8 bytes.
32787 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32789 * config/s390/s390.md: Add the s_operand checks formerly in
32790 s390_split_ok_p to various splitters where they are still
32792 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32793 for 128 bit vectors. Plus two splitters.
32795 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32797 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32800 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32803 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32804 error if the boundary argument is not constant.
32806 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32808 PR rtl-optimization/80112
32809 * loop-doloop.c (doloop_condition_get): Don't check condition
32810 if cmp isn't SET with IF_THEN_ELSE src.
32812 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32814 PR tree-optimization/80158
32815 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32816 replacing a candidate statement, also replace it for the
32817 candidate's alternate interpretation.
32818 (replace_rhs_if_not_dup): Likewise.
32819 (replace_one_candidate): Likewise.
32821 2017-03-24 Richard Biener <rguenther@suse.de>
32823 PR tree-optimization/80167
32824 * graphite-isl-ast-to-gimple.c
32825 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32827 (translate_isl_ast_to_gimple::get_rename): Likewise.
32829 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32832 handling of certain combinations of target options, including the
32833 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32834 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32836 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32839 * config/arm/arm.md (*load_multiple): Add reload_completed to
32840 matching condition.
32842 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32843 Richard Biener <rguenth@suse.de>
32845 PR tree-optimization/79908
32846 PR tree-optimization/80136
32847 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32848 been cast away, gimplify_and_add suffices.
32850 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32852 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32854 2017-03-23 Richard Biener <rguenther@suse.de>
32856 PR tree-optimization/80032
32857 * gimplify.c (gimple_push_cleanup): Forced unconditional
32858 cleanups still have to go to the conditional_cleanups
32861 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32863 PR tree-optimization/80072
32864 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32866 (next_operand_entry_id): Change type to unsigned int.
32867 (sort_by_operand_rank): Make sure to return the right return value
32868 even if unsigned fields are bigger than INT_MAX.
32869 (struct oecount): Change cnt and id type to unsigned int.
32870 (oecount_hasher::equal): Formatting fix.
32871 (oecount_cmp): Make sure to return the right return value
32872 even if unsigned fields are bigger than INT_MAX.
32873 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32876 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32877 TREE_READONLY on result if writing it more than once.
32880 * doc/invoke.texi (-fsanitize=thread): Document that with
32881 -fnon-call-exceptions atomics are not able to throw
32885 * tsan.c: Include tree-eh.h.
32886 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32887 maybe_clean_or_replace_eh_stmt where needed.
32888 (instrument_memory_accesses): Add cfg_changed argument.
32889 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32890 if it returned true.
32891 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32893 PR rtl-optimization/63191
32894 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32895 wrapper function, moved the whole old content into ...
32896 (ix86_delegitimize_address_1): ... this. New inline function.
32897 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32898 true as last argument instead of ix86_delegitimize_address.
32900 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32902 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32903 cortexa57_branch_cost.
32905 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32907 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32909 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32912 * doc/md.texi (Constraints): Document wA constraint.
32913 * config/rs6000/constraints.md (wA): New.
32914 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32915 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32916 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32917 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32919 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32922 * gimplify.c (is_oacc_declared): New function.
32923 (oacc_default_clause): Use it to set default flags for acc declared
32924 variables inside parallel regions.
32925 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32926 declared variables.
32927 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32928 declare attribute to any decl as necessary.
32930 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32933 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32934 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32935 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32936 (arm_arch_lpae): This.
32937 * config/arm/arm.c (arm_arch7ve): Rename into ...
32938 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32939 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32942 2017-03-22 Martin Liska <mliska@suse.cz>
32945 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32946 error message instead of an ICE.
32948 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32950 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32952 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32954 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32956 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32959 2017-03-21 Martin Sebor <msebor@redhat.com>
32961 * doc/extend.texi: Use "cannot" instead of "can't."
32962 * doc/hostconfig.texi: Same.
32963 * doc/install.texi: Same.
32964 * doc/invoke.texi: Same.
32965 * doc/loop.texi: Same.
32966 * doc/md.texi: Same.
32967 * doc/objc.texi: Same.
32968 * doc/rtl.texi: Same.
32969 * doc/tm.texi: Same.
32970 * doc/tm.texi.in: Same.
32971 * doc/trouble.texi: Same.
32973 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32976 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32977 (collect_checksum_attributes): Set it.
32978 (die_checksum_ordered): Use it.
32980 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32982 PR tree-optimization/79908
32983 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32984 change: For a VA_ARG whose LHS has been cast away, use
32985 force_gimple_operand to construct the side effects.
32987 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32989 PR translation/80001
32990 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32991 more amenable to translation.
32992 (oacc_loop_auto_partitions): Likewise.
32994 2017-03-21 Marek Polacek <polacek@redhat.com>
32995 Martin Sebor <msebor@redhat.com>
32997 PR tree-optimization/80109
32998 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32999 on INTEGRAL_TYPE_P.
33001 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33002 Segher Boessenkool <segher@kernel.crashing.org>
33005 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33006 check reg_used_between_p between insn and one of succ or succ2
33007 depending on if succ is artificial insn not inserted into insn
33010 2017-03-21 Martin Liska <mliska@suse.cz>
33012 PR gcov-profile/80081
33013 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33014 * doc/gcc.texi: Include gcov-dump stuff.
33015 * doc/gcov-dump.texi: New file.
33017 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33019 PR rtl-optimization/79150
33020 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33021 conditional jump, if the jump is the last insn of the loop.
33023 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33024 Richard Biener <rguenth@suse.de>
33026 PR tree-optimization/79908
33027 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33028 been cast away, use force_gimple_operand to construct the side
33031 2017-03-21 Martin Liska <mliska@suse.cz>
33033 PR libfortran/79956
33034 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33037 2017-03-21 Brad Spengler <spender@grsecurity.net>
33040 * plugin.c (htab_hash_plugin): New function.
33041 (add_new_plugin): Use it and adjust.
33042 (parse_plugin_arg_opt): Adjust.
33043 (init_one_plugin): Likewise.
33045 2017-03-21 Richard Biener <rguenther@suse.de>
33047 PR tree-optimization/80032
33048 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33049 if set force the cleanup to happen unconditionally.
33050 (gimplify_target_expr): Push inserted clobbers with force_uncond
33051 to avoid them being removed by control-dependent DCE.
33053 2017-03-21 Richard Biener <rguenther@suse.de>
33055 PR tree-optimization/80122
33056 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33057 va_arg_pack_len when the inlined call stmt requires pack
33059 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33061 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33064 * tsan.c (instrument_builtin_call): If the memory model argument
33065 is not a constant, assume it is valid.
33068 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33071 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33073 PR rtl-optimization/79910
33074 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33075 if its dest is used by an insn before I2 (other than the combined
33076 insns themselves, which are properly handled already).
33078 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33081 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33083 * combine.c (record_used_regs): New static function.
33084 (try_combine): Handle situations where there is an additional
33085 instruction between I2 and I3 which needs to have a LOG_LINK
33089 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33091 * combine.c (try_combine): Delete redundant i1 test. Call
33092 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33094 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33097 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33098 alternatives 13/14.
33100 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33102 PR tree-optimization/80054
33103 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33104 the optimization if a PHI or any of its arguments is not dominated
33105 by the candidate's basis. Use gphi* rather than gimple* as
33107 (replace_profitable_candidates): Clean up a gimple* variable that
33108 should be a gphi* variable.
33110 2017-03-20 Martin Sebor <msebor@redhat.com>
33113 * doc/extend.texi (attribute constructor): Document present limitation.
33115 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33118 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33119 __POWER9_VECTOR__ #ifdef control, change template definition to
33120 use Power9-specific built-in function.
33121 (vec_any_eq): Likewise.
33122 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33123 to control outcomes from this test.
33124 (vector_ae_<mode>p): For VEC_F modes, likewise.
33126 2017-03-20 Ian Lance Taylor <iant@google.com>
33128 * config/i386/i386.c (ix86_function_regparm): Save an extra
33129 register for -fsplit-stack with DECL_STATIC_CHAIN.
33131 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33134 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33135 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33137 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33139 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33141 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33144 2017-03-20 Marek Polacek <polacek@redhat.com>
33147 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33149 2017-03-20 Richard Biener <rguenther@suse.de>
33151 PR tree-optimization/80113
33152 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33153 allocate extra SSA name for PHI def.
33154 (add_close_phis_to_outer_loops): Likewise.
33155 (add_close_phis_to_merge_points): Likewise.
33156 (copy_loop_close_phi_args): Likewise.
33157 (copy_cond_phi_nodes): Likewise.
33159 2017-03-20 Martin Liska <mliska@suse.cz>
33161 PR middle-end/79753
33162 * tree-chkp.c (chkp_build_returned_bound): Do not build
33163 returned bounds for a LHS that's not a BOUNDED_P type.
33165 2017-03-20 Martin Liska <mliska@suse.cz>
33169 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33170 COMPLEX_CST and VECTOR_CST.
33172 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33175 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33176 target operand. A new splitter adds the clobber statement in case
33177 the target operand is dead anyway.
33179 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33181 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33182 to age-old versions of binutils and glibc.
33184 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33186 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33188 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33190 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33192 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33194 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33195 requirement for binutils 2.13.
33197 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33199 * combine.c (try_combine): Delete redundant i1 test. Call
33200 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33202 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33204 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33205 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33207 <riscv64-*-elf>: Re-arrange section
33208 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33209 <riscv32-*-linux>: Likewise.
33210 <riscv64-*-elf>: Likewise
33211 <riscv64-*-linux>: Likewise.
33213 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33216 * aarch64.opt(verbose-cost-dump): Fix typo.
33218 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33221 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33222 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33224 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33226 * reload.c (find_reloads): When reloading a nonoffsettable address,
33227 use RELOAD_OTHER for it and its address reloads.
33229 PR rtl-optimization/79910
33230 * combine.c (record_used_regs): New static function.
33231 (try_combine): Handle situations where there is an additional
33232 instruction between I2 and I3 which needs to have a LOG_LINK
33235 2017-03-17 Jeff Law <law@redhat.com>
33237 PR tree-optimization/71437
33238 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33239 conditional in the hash table first.
33240 (vrp_dom_walker::before_dom_children): Extract condition from
33241 ASSERT_EXPR. Record condition, its inverion and any implied
33242 conditions as well.
33244 2017-03-17 Marek Polacek <polacek@redhat.com>
33245 Markus Trippelsdorf <markus@trippelsdorf.de>
33247 PR tree-optimization/80079
33248 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33251 2017-03-17 Richard Biener <rguenther@suse.de>
33253 PR middle-end/80075
33254 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33255 Properly verify the LHS before the RHS possibly claims to be
33257 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33260 2017-03-17 Martin Jambor <mjambor@suse.cz>
33262 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33263 (List of -O2 options): Likewise.
33264 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33267 2017-03-17 Tom de Vries <tom@codesourcery.com>
33269 * gcov-dump.c (print_usage): Print bug_report_url.
33271 2017-03-17 Richard Biener <rguenther@suse.de>
33273 PR middle-end/80050
33274 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33275 (parser::peek): Likewise.
33277 2017-03-17 Richard Biener <rguenther@suse.de>
33279 PR tree-optimization/80048
33280 * sese.c (free_sese_info): Properly release rename_map and
33281 copied_bb_map elements.
33283 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33285 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33286 Add linked-list forward and backlinks. Insert on
33287 construction, remove on destruction.
33288 (class pass_store_merging): Add m_stores_head field.
33289 (pass_store_merging::terminate_and_process_all_chains):
33290 Iterate over m_stores_head list.
33291 (pass_store_merging::terminate_all_aliasing_chains):
33293 (pass_store_merging::execute): Check for debug stmts first.
33294 Push new chains onto the m_stores_head stack.
33296 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33299 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33300 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33301 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33303 2017-03-16 Jeff Law <law@redhat.com>
33305 PR tree-optimization/71437
33306 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33307 member function. Implementation moved into after_dom_children
33308 member function and into the threader's thread_outgoing_edges
33310 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33311 some code into new thread_outgoing_edges.
33312 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33313 definition. Simplify marker handling (do it here). Assume we always
33314 have the available expression and the const/copies tables.
33315 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33317 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33318 * tree-vrp.c (equiv_stack): No longer file scoped.
33319 (vrp_dom_walker): New class.
33320 (vrp_dom_walker::before_dom_children): New member function.
33321 (vrp_dom_walker::after_dom_children): Likewise.
33322 (identify_jump_threads): Setup domwalker. Use it rather than
33323 walking edges in a random order by hand. Simplify setup/finalization.
33324 (finalize_jump_threads): Remove.
33325 (vrp_finalize): Do not call identify_jump_threads here.
33326 (execute_vrp): Do it here instead and call thread_through_all_blocks
33329 PR tree-optimization/71437
33330 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33332 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33334 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33335 (dom_opt_dom_walker::thread_across_edge): Remove
33336 handle_dominating_asserts argument. All callers changed.
33337 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33338 changes. Remove calls to lhs_of_dominating_assert. Other
33339 uses of handle_dominating_asserts turn into unconditional code
33340 (simplify_control_stmt_condition_1): Likewise.
33341 (simplify_control_stmt_condition): Likewise.
33342 (thread_through_normal_block, thread_across_edge): Likewise.
33343 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33344 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33345 object if it is not an SSA_NAME.
33346 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33347 before calling into the VRP specific simplifiers.
33348 (identify_jump_threads): Remove handle_dominating_asserts
33351 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33354 * tree-diagnostic.c (default_tree_printer): No longer static.
33355 * tree-diagnostic.h (default_tree_printer): New prototype.
33357 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33359 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33360 Change ins into fmov.
33362 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33364 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33365 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33366 Use h_con constraint for operand 1.
33367 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33368 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33370 2017-03-15 Jeff Law <law@redhat.com>
33372 PR tree-optimization/71437
33373 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33374 (record_temporary_equivalences): Use it.
33376 PR tree-optimization/71437
33377 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33378 tree-ssa-scopedtables.
33379 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33380 (record_conditions, record_cond, vuse_eq): Likewise.
33381 (record_edge_info): Adjust to API tweak of record_conditions.
33382 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33383 (record_temporary_equivalences, optimize_stmt): Likewise.
33384 (eliminate_redundant_computations): Likewise.
33385 (record_equivalences_from_stmt): Likewise.
33386 * tree-ssa-scopedtables.c: Include options.h and params.h.
33387 (vuse_eq): New function, moved from tree-ssa-dom.c
33388 (build_and_record_new_cond): Likewise.
33389 (record_conditions): Likewise. Accept vector of conditions rather
33390 than edge_equivalence structure for first argument.
33391 for the first argument.
33392 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33393 from tree-ssa-dom.c.
33394 (avail_exprs_stack::record_cond): Likewise.
33395 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33396 from tree-ssa-dom.c.
33397 (avail_exprs_stack): Add new member functions lookup_avail_expr
33399 (record_conditions): Declare.
33401 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33404 * lra-constraints.c (process_alt_operands): Increase reject for
33405 reloading an input/output operand.
33407 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33410 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33411 insns to convert from signed/unsigned char/short to IEEE 128-bit
33413 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33415 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33418 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33419 subreg of inner mode for values already in registers.
33421 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33423 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33424 iteration reg is used after the loop.
33426 2017-03-14 Martin Sebor <msebor@redhat.com>
33428 PR tree-optimization/79800
33429 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33430 precision in negative-positive range.
33431 (format_floating): Call non-const overload with adjusted precision.
33433 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33436 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33439 2017-03-14 Martin Sebor <msebor@redhat.com>
33441 PR middle-end/80020
33442 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33443 * builtins.def (aligned_alloc): Use it.
33446 * Makefile.in (GTFILES): Add calls.c.
33447 * calls.c: Include "gt-calls.h".
33449 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33451 PR rtl-optimization/79728
33452 * regs.h (struct target_regs): New field
33453 x_contains_allocatable_regs_of_mode.
33454 (contains_allocatable_regs_of_mode): New macro.
33455 * reginfo.c (init_reg_sets_1): Initialize it, and change
33456 contains_reg_of_mode so it includes global regs as well.
33457 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33458 rather than contains_regs_of_mode.
33460 2017-03-14 Martin Liska <mliska@suse.cz>
33462 * doc/invoke.texi: Document options that can't be combined with
33463 -fcheck-pointer-bounds.
33465 2017-03-14 Martin Liska <mliska@suse.cz>
33467 PR middle-end/79831
33468 * doc/invoke.texi (-Wchkp): Document the option.
33470 2017-03-14 Martin Liska <mliska@suse.cz>
33472 * Makefile.in: Install gcov-dump.
33474 2017-03-14 Martin Liska <mliska@suse.cz>
33476 * multiple_target.c (expand_target_clones): Bail out for
33477 an invalid attribute.
33479 2017-03-14 Richard Biener <rguenther@suse.de>
33481 * alias.c (struct alias_set_entry): Pack properly.
33482 * cfgloop.h (struct loop): Likewise.
33483 * cse.c (struct set): Likewise.
33484 * ipa-utils.c (struct searchc_env): Likewise.
33485 * loop-invariant.c (struct invariant): Likewise.
33486 * lra-remat.c (struct cand): Likewise.
33487 * recog.c (struct change_t): Likewise.
33488 * rtl.h (struct address_info): Likewise.
33489 * symbol-summary.h (function_summary): Likewise.
33490 * tree-loop-distribution.c (struct partition): Likewise.
33491 * tree-object-size.c (struct object_size_info): Likewise.
33492 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33493 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33494 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33495 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33496 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33497 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33498 (struct _stmt_vec_info): Likewise.
33500 2017-03-14 Martin Liska <mliska@suse.cz>
33503 * multiple_target.c (create_dispatcher_calls): Check that
33504 a target can create a function dispatcher.
33506 2017-03-14 Martin Liska <mliska@suse.cz>
33509 * multiple_target.c (expand_target_clones): Drop local.local
33510 flag for default implementation.
33512 2017-03-14 Richard Biener <rguenther@suse.de>
33514 PR tree-optimization/80030
33515 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33517 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33519 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33520 gcc_fallthrough() instead of __attribute__((fallthrough));
33522 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33524 * doc/gcc.texi: Remove "up" link to (DIR).
33525 * doc/gccint.texi: Ditto.
33527 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33529 * doc/install.texi (Specific) <avr>: Remove reference to
33532 2017-03-13 Jeff Law <law@redhat.com>
33534 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33535 attribute rather than comments.
33537 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33538 match_scratch operand is highest.
33540 2017-03-13 Martin Liska <mliska@suse.cz>
33542 PR middle-end/78339
33543 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33544 is a CHKP clone, use original declaration.
33546 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33548 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33549 (arc_conditional_register_usage): Use a different allocation order
33550 when optimizing for size.
33551 * common/config/arc/arc-common.c (arc_option_optimization_table):
33552 Section anchors default on when optimizing for size.
33554 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33556 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33558 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33560 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33561 * config/arc/arc.md (cpu_facility): Add cd variant.
33562 (*movqi_insn): Add code density variant.
33563 (*movhi_insn): Likewise.
33564 (*movqi_insn): Likewise.
33565 (*addsi3_mixed): Likewise.
33566 (subsi3_insn): Likewise.
33568 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33570 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33572 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33574 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33575 expressions with MINUS and UNARY ops.
33577 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33580 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33582 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33583 between vec_select and vector argument.
33584 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33585 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33586 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33587 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33588 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33589 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33591 2017-03-13 Richard Biener <rguenther@suse.de>
33594 * params.def (vect-max-peeling-for-alignment): Fix typo.
33596 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33598 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33599 issue that only occurred with binutils below 2.18.
33601 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33603 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33604 refer to binutils 2.11/2.12 minimum.
33606 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33608 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33609 ftp.kernel.org and simplify binutils requirement.
33611 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33613 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33615 (Optimize Options): Ditto. Also remove redundancy.
33617 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33619 PR translation/79848
33620 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33622 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33623 to G_ to avoid double translation.
33625 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33627 PR translation/79923
33628 * auto-profile.c (get_combined_location): Convert leading
33629 character of diagnostics to lower case and remove trailing period.
33630 (read_profile): Likewise for various diagnostics.
33631 * config/arm/arm.c (arm_option_override): Remove trailing period
33632 from various diagnostics.
33633 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33634 (msp430_expand_delay_cycles): Likewise.
33636 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33639 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33640 full command-line argument, rather than just "str".
33641 (aarch64_validate_march): Likewise.
33642 (aarch64_validate_mtune): Likewise.
33644 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33646 PR rtl-optimization/78911
33647 * lra-assigns.c (must_not_spill_p): New function.
33648 (spill_for): Use it.
33650 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33652 PR tree-optimization/79981
33653 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33654 ATOMIC_COMPARE_EXCHANGE ifn result.
33655 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33656 IFN_ATOMIC_COMPARE_EXCHANGE.
33658 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33661 * opts.c (parse_sanitizer_options): Add missing question mark to
33662 "did you mean" message.
33664 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33666 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33668 (VMULEUH_UNS): Likewise.
33669 (VMULOUB_UNS): Likewise.
33670 (VMULOUH_UNS): Likewise.
33671 * config/rs6000/rs6000.c (builtin_function_type): Remove
33672 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33674 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33677 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33678 x with result of extra_parsing_for_operand_code_0.
33679 (function_reader::extra_parsing_for_operand_code_0): Convert
33680 return type from void to rtx, returning x. When reading
33681 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33682 larger size containing struct block_symbol.
33684 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33687 -mfloat128-hardware without -m64.
33689 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33692 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33693 entries to the case statement that marks unsigned arguments to
33694 overloaded functions.
33696 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33698 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33699 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33701 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33704 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33705 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33707 2017-03-10 Martin Liska <mliska@suse.cz>
33711 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33712 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33715 2017-03-10 Olivier Hainque <hainque@adacore.com>
33717 * tree-switch-conversion (array_value_type): Start by resetting
33718 candidate type to it's main variant.
33720 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33722 PR rtl-optimization/79909
33723 * combine.c (try_combine): Use simplify_replace_rtx on individual
33724 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33725 of the whole CALL_INSN_FUNCTION_USAGE.
33727 PR tree-optimization/79972
33728 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33729 get_range_info on SSA_NAMEs. Formatting fixes.
33731 2017-03-10 Richard Biener <rguenther@suse.de>
33732 Jakub Jelinek <jakub@redhat.com>
33734 PR tree-optimization/77975
33735 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33736 edge to be constant.
33737 (get_val_for): For constant x return it. Formatting fix.
33738 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33739 would use the same bases as the current one.
33741 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33743 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33744 instead of vec_select for V1TImode.
33745 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33747 (VSX_LE_128): Add V1TI to this mode iterator.
33748 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33749 (*vsx_le_perm_store_<mode>): Likewise.
33750 (pre-reload splitter for VSX stores): Likewise.
33751 (post-reload splitter for VSX stores): Likewise.
33752 (*vsx_xxpermdi2_le_<mode>): Likewise.
33753 (*vsx_lxvd2x2_le_<mode>): Likewise.
33754 (*vsx_stxvd2x2_le_<mode>): Likewise.
33756 2017-03-09 Michael Eager <eager@eagercon.com>
33758 Correct failures with --enable-checking=yes,rtl.
33760 * config/microblaze/microblaze.c (microblaze_expand_shift):
33761 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33762 test for const0_rtx.
33763 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33764 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33766 2017-03-09 Richard Biener <rguenther@suse.de>
33768 PR tree-optimization/79977
33769 * graphite-scop-detection.c (scop_detection::merge_sese):
33770 Handle the case of extra exits to blocks dominating the entry.
33772 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33774 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33777 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33779 PR rtl-optimization/79949
33780 * lra-constraints.c (process_alt_operands): Check memory when
33781 trying to predict a cycle. Print about the overall increase.
33783 2017-03-09 Richard Biener <rguenther@suse.de>
33785 PR middle-end/79971
33786 * gimple-expr.c (useless_type_conversion_p): Preserve
33787 TYPE_SATURATING for fixed-point types.
33789 2017-03-09 Richard Biener <rguenther@suse.de>
33792 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33793 alignment of BLKmode params.
33795 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33798 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33799 (VALL_NO_V2Q): Likewise.
33801 * config/aarch64/aarch64-simd.md
33802 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33804 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33805 VALL_NO_V2Q mode iterator.
33806 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33808 2017-03-09 Martin Liska <mliska@suse.cz>
33810 PR tree-optimization/79631
33811 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33812 tree_int_cst_sign_bit just for INTEGER constants.
33814 2017-03-09 Martin Liska <mliska@suse.cz>
33818 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33821 2017-03-09 Marek Polacek <polacek@redhat.com>
33824 * tree.c (inchash::add_expr): Handle TREE_VEC.
33826 2017-03-09 Martin Liska <mliska@suse.cz>
33829 (chkp_narrow_size_and_offset): New function.
33830 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33831 (void chkp_parse_bit_field_ref): New function.
33832 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33833 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33835 2017-03-09 Martin Liska <mliska@suse.cz>
33838 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33839 (chkp_find_bounds_1): Remove gcc_unreachable.
33841 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33844 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33845 BUILT_IN_SYNC*, determine the access type from the size suffix and
33846 always build a MEM_REF with that type. Handle forgotten
33847 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33850 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33851 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33852 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33853 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33854 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33855 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33856 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33857 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33858 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33859 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33860 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33861 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33862 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33863 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33864 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33865 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33866 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33867 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33868 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33869 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33870 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33871 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33872 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33873 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33874 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33875 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33876 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33877 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33878 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33879 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33880 definitions outside of __OPTIMIZE__ guarded section.
33883 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33884 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33885 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33886 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33889 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33891 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33892 ("vfenez<mode>"): Add missing constraints.
33894 2017-03-08 Martin Sebor <msebor@redhat.com>
33897 * config/nds32/nds32.c (nds32_option_override):
33898 Fix misspelled diagnostic.
33900 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33903 * gimplify.c (gimplify_omp_for): Replace index var in outer
33904 taskloop statement with an artificial variable and add
33905 OMP_CLAUSE_PRIVATE clause for it.
33907 2017-03-08 Richard Biener <rguenther@suse.de>
33909 PR tree-optimization/79955
33910 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33911 for accesses that are completely outside of the variable.
33913 2017-03-08 Andrew Haley <aph@redhat.com>
33915 PR tree-optimization/79943
33916 * tree-ssa-loop-split.c (compute_new_first_bound): When
33917 calculating the new upper bound, (END-BEG) should be added, not
33920 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33922 * config/avr/avr.md (setmemhi): Make sure match_dup
33923 operand number comes before match_scratch.
33925 2017-03-08 Richard Biener <rguenther@suse.de>
33927 PR tree-optimization/79920
33928 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33929 with ncopies == 1 to ...
33930 (vect_transform_slp_perm_load): ... here. Properly compute
33931 all element loads by iterating VF times over the group. Do
33932 not handle ncopies (computed in a broken way) in
33933 vect_create_mask_and_perm.
33935 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33938 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33939 is a uniform vector, use uniform_vector_p return value instead of
33940 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33942 2017-03-07 Marek Polacek <polacek@redhat.com>
33944 PR middle-end/79809
33945 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33946 (alloca_call_type): Likewise.
33948 2017-03-07 Martin Liska <mliska@suse.cz>
33950 * gcov.c (process_args): Put comment to correct location.
33952 2017-03-07 Martin Liska <mliska@suse.cz>
33954 PR middle-end/68270
33955 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33956 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33957 (chkp_narrow_bounds_for_field): Likewise.
33958 (chkp_parse_array_and_component_ref): Pass one more argument to
33961 2017-03-07 Richard Biener <rguenther@suse.de>
33963 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33966 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33968 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33969 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33971 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33974 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33975 to end of description.
33976 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33978 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33980 PR rtl-optimization/79901
33981 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33983 (*avx512f_<code><mode>3<mask_name>): ... this.
33984 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33985 iterator instead of VI8_AVX2_AVX512BW.
33987 PR rtl-optimization/79901
33988 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33989 min/max expander, expand it using expand_vec_cond_expr.
33992 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33995 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33998 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33999 to void * for PCH reasons.
34000 * dwarf2out.c (output_loc_operands, output_die): Cast
34001 v.val_vec.array to unsigned char *.
34003 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34006 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34009 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34011 PR rtl-optimization/79571
34012 * lra-constraints.c (process_alt_operands): Calculate static
34013 reject and subtract it from overall when only addresses will be
34016 2017-03-06 Julia Koval <julia.koval@intel.com>
34019 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34020 incoming stack boundary to 128 for 64-bit targets.
34022 2017-03-06 Richard Biener <rguenther@suse.de>
34024 PR tree-optimization/79894
34025 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34026 to NULL after folding it.
34028 2017-03-06 Richard Biener <rguenther@suse.de>
34030 PR tree-optimization/79824
34031 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34032 check disabling peeling for gaps.
34034 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34036 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34037 attributes): Document gettimeofday.
34039 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34041 * config/s390/s390.c (s390_option_override_internal): Set
34042 PARAM_MIN_VECT_LOOP_BOUND
34044 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34046 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34047 * config/s390/s390.md: Likewise.
34049 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34052 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34053 (<avx2_avx512>_perm<mode>): Rename to ...
34054 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34056 (<avx512>_perm<mode>_mask): Rename to ...
34057 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34059 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34060 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34061 instead of VI8F_256_512.
34062 (avx512f_perm<mode>): New define_expand.
34063 (avx512f_perm<mode>_mask): Likewise.
34064 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34065 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34067 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34069 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34070 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34072 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34074 2017-03-06 Martin Liska <mliska@suse.cz>
34077 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34078 when having a SSA NAME w/o VAR_DECL assigned to it.
34080 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34082 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34083 msa_dpsub_<su>_d): Fix MODE for vec_select.
34085 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34087 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34089 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34091 2017-03-06 Richard Biener <rguenther@suse.de>
34093 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34094 * plugin.c (register_plugin_info): Likewise.
34095 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34097 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34099 * config/i386/sse.md (sse_storehps, sse_storelps,
34100 avx_<castmode><avxsizesuffix>_<castmode>,
34101 avx512f_<castmode><avxsizesuffix>_<castmode>,
34102 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34103 in condition that at least one operand is not a MEM.
34105 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34107 PR middle-end/79805
34108 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34109 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34111 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34112 gimple_call_nothrow_p flag based on whether original builtin can throw.
34113 If it can, emit following stmts on the fallthrough edge.
34114 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34115 don't create new bb if inserting just debug stmts on the edge, try to
34116 insert them on the fallthru bb or just reset debug stmts.
34118 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34121 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34122 restore recog_data (including the operand rtxes inside it) around
34123 the call to get_insn_template.
34125 2017-03-03 Martin Sebor <msebor@redhat.com>
34127 PR tree-optimization/79699
34128 * context.c (context::~context): Free MPFR caches to avoid
34129 a memory leak on program exit.
34131 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34133 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34134 Use wide_int::ulow () instead of .elt (0).
34136 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34138 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34139 (*pushxf): Limit oF constraint to 32bit targets and add oC
34140 constraint for 64bit targets.
34141 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34142 (*pushdf): Change rmF constraint to rmC.
34144 2017-03-03 Martin Liska <mliska@suse.cz>
34146 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34147 Remove unused variable.
34149 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34152 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34153 is a memory operand, increase num_memory.
34154 (ix86_expand_args_builtin): Likewise.
34156 2017-03-03 Jan Hubicka <jh@suse.cz>
34159 * ipa-devirt.c (maybe_record_node): Properly handle
34160 __cxa_pure_virtual visibility.
34162 2017-03-03 Martin Liska <mliska@suse.cz>
34164 PR tree-optimization/79803
34165 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34167 (pass_loop_prefetch::execute): Disabled optimization if an
34168 assumption about L1 cache size is not met.
34170 2017-03-03 Martin Liska <mliska@suse.cz>
34172 PR rtl-optimization/79574
34173 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34174 (hash_scan_set): Likewise.
34175 (dump_hash_table): Likewise.
34176 (hoist_code): Likewise.
34178 2017-03-03 Richard Biener <rguenther@suse.de>
34180 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34181 in place of uhigh ().
34182 (fixed_convert_from_real): Likewise.
34184 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34187 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34189 2017-03-03 Richard Biener <rguenther@suse.de>
34191 PR middle-end/79818
34192 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34193 TYPE_OVERFLOW_UNDEFINED check.
34195 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34197 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34199 (vector_ae_<mode>_p): Likewise.
34200 (vector_nez_<mode>_p): Likewise.
34201 (vector_ne_v2di_p): Likewise.
34202 (vector_ae_v2di_p): Likewise.
34203 (vector_ne_<mode>_p): Likewise.
34204 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34206 (vsx_tsqrt<mode>2_fe): Likewise.
34208 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34211 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34213 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34215 PR rtl-optimization/79780
34216 * cprop.c (one_cprop_pass): When second and further conditional trap
34217 in a single basic block is turned into an unconditional trap, turn it
34218 into a deleted note to avoid RTL verification failures.
34220 2017-03-02 Richard Biener <rguenther@suse.de>
34222 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34224 2017-03-02 Richard Biener <rguenther@suse.de>
34226 PR tree-optimization/79345
34228 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34229 param and abort the walk, returning -1 if it is hit.
34230 (walk_aliased_vdefs): Take a limit param and pass it on.
34231 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34232 defaulting to 0 and return a signed int.
34233 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34234 (check_defs): New helper.
34235 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34236 about uninitialized memory.
34237 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34238 bogus uninitialized warning.
34239 (fixed_convert_from_real): Likewise.
34241 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34243 PR tree-optimization/66768
34244 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34245 iv_use if base object can't be determined.
34247 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34249 PR tree-optimization/79345
34250 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34251 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34252 (get_pattern_stats): Initialize it.
34253 * genemit.c (gen_expand): Verify match_scratch numbers come after
34254 match_operand/match_dup numbers.
34255 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34256 match_scratch numbers.
34257 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34259 * config/s390/s390.md (trunctdsd2): Likewise.
34261 2017-03-02 Richard Biener <rguenther@suse.de>
34263 * wide-int.h (wide_int_storage::operator=): Implement in terms
34266 2017-03-02 Richard Biener <rguenther@suse.de>
34268 PR tree-optimization/79777
34269 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34270 the to insert expression to sth existing.
34272 2017-03-01 Martin Sebor <msebor@redhat.com>
34274 PR middle-end/79692
34275 * gimple-ssa-sprintf.c
34276 (directive::known_width_and_precision): New function.
34277 (format_integer): Use it.
34278 (get_mpfr_format_length): Consider the full range of precision
34279 when computing %g output with the # flag. Set the likely byte
34280 count to 3 rather than 1 when precision is indeterminate.
34281 (format_floating): Correct the lower bound of precision.
34283 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34285 * doc/invoke.texi: Document default code model for 64-bit Linux.
34287 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34290 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34291 udiv rather than div since input pattern is unsigned.
34293 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34295 * config/i386/i386.c (print_reg): Warn for values of
34296 unsupported size in integer register.
34298 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34301 * config/rs6000/predicates.md (current_file_function_operand): Do
34302 not allow self calls to be local if the function is replaceable.
34304 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34307 * config/rs6000/altivec.h (vec_ctz and others): Change the
34308 preprocessor macro that controls conditional compilation from
34309 _ARCH_PWR9 to __POWER9_VECTOR__.
34310 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34311 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34312 control (instead of _ARCH_PWR9 control) so that template
34313 definition uses power9-specific function.
34314 (vec_any_eq): Likewise.
34315 (vec_all_ne): Change macro definition to use a power9-specific
34316 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34317 _ARCH_PWR9 control).
34318 (vec_any_eq) Likewise.
34319 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34320 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34321 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34322 support for xvcmpnedp instruction.
34323 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34324 macro expansion so that Power9 implementation of vec_all_ne does
34325 not use the AltiVec predicate framework.
34326 (VCMPNEH_P): Likewise.
34327 (VCMPNEW_P): Likewise.
34328 (VCMPNED_P): Likewise.
34329 (VCMPNEFP_P): Likewise.
34330 (VCMPNEDP_P): Likewise.
34331 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34332 implementation of vec_any_eq to not use AltiVec predicate
34334 (VCMPAEH_P): Likewise.
34335 (VCMPAEW_P): Likewise.
34336 (VCMPAED_P): Likewise.
34337 (VCMPAEFP_P): Likewise.
34338 (VCMPAEDP_P): Likewise.
34339 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34340 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34341 not use the AltiVec predicate framework.
34342 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34343 of vec_any_eq to not use AltiVec predicate framework.
34344 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34345 support for predefined __POWER9_VECTOR__ macro to indicate that
34346 Power9 instruction selection is enabled.
34347 (altivec_overloaded_builtins): Remove extraneous
34348 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34349 function argument types RS6000_BTI_bool_V16QI and
34350 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34351 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34352 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34353 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34354 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34355 Power9 for implementations of vec_cmpne. Change the signature for
34356 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34357 (representing vec_all_ne) to remove the previously described first
34358 argument of type RS6000_BTI_INTSI, as this was an artifact of
34359 reliance on the AltiVec predicate framework, which is no longer
34360 used in the implementation of these functions. Add
34361 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34362 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34363 since, unlike the AltiVec predicate framework implementation, we
34364 do not share function descriptors between vec_alle and vec_anyeq.
34365 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34366 set of modes that receive special treatment even when
34367 TARGET_P9_VECTOR is true. The special treatment emits code that
34368 does not depend on Power9 instructions.
34369 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34370 define_expand to not rely on AltiVec predicate framework.
34371 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34373 (vector_ne_v2di_p): Change this define_expand to not rely on
34374 AltiVec predicate framework.
34375 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34377 (vector_ne_<mode>_p): Change this define_expand to not rely on
34378 AltiVec predicate framework.
34379 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34381 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34382 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34383 define_insn pattern.
34384 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34385 define_insn pattern because the xvcmpne<VSs>. instruction is not
34387 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34388 instruction is not supported.
34390 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34392 * config/nvptx/nvptx.c: Include intl.h.
34394 2017-03-01 Martin Jambor <mjambor@suse.cz>
34397 * ipa-prop.h (ipa_bits): Removed field known.
34398 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34399 to pointers. Adjusted their comments to warn about their sharing.
34400 (ipcp_transformation_summary): Change bits to a vector of pointers.
34401 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34402 (ipa_get_ipa_bits_for_value): Declare.
34403 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34404 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34405 (ipa_bits_hash_table): Likewise.
34406 (ipa_vr_ggc_hash_traits): Likewise.
34407 (ipa_vr_hash_table): Likewise.
34408 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34409 being pointers and vr_known being removed.
34410 (ipa_set_jf_unknown): Likewise.
34411 (ipa_get_ipa_bits_for_value): New function.
34412 (ipa_set_jfunc_bits): Likewise.
34413 (ipa_get_value_range): New overloaded functions.
34414 (ipa_set_jfunc_vr): Likewise.
34415 (ipa_compute_jump_functions_for_edge): Use the above functions to
34416 construct bits and vr parts of jump functions.
34417 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34418 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34420 (ipcp_grow_transformations_if_necessary): Also allocate
34421 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34423 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34424 them. Fix too long lines.
34425 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34426 vr_known being removed.
34427 (ipa_read_jump_function): Use new setter functions to construct bits
34428 and vr parts of jump functions or set them to NULL.
34429 (write_ipcp_transformation_info): Adjust for bits being pointers.
34430 (read_ipcp_transformation_info): Likewise.
34431 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34433 Include gt-ipa-prop.h.
34434 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34436 (ipcp_store_bits_results): Likewise.
34437 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34438 Do not write to existing jump functions but use a temporary instead.
34440 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34443 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34444 attempt to use its first operand as BIT_FIELD_REF base.
34446 2017-03-01 Richard Biener <rguenther@suse.de>
34448 PR middle-end/79721
34449 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34450 interpolating formula in wrapping arithmetic.
34451 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34453 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34455 PR tree-optimization/79734
34456 * tree-vect-generic.c (expand_vector_condition): Optimize
34457 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34458 Handle VEC_COND_EXPR where comparison has different inner width from
34459 type's inner width.
34461 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34463 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34464 markup, and similar issues. Remove @opindex entries for things
34465 that aren't options. Add missing -mmpy-option entries.
34467 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34469 PR tree-optimization/79737
34470 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34471 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34472 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34473 instead of byte_size. Formatting fix.
34474 (shift_bytes_in_array_right): Formatting fix.
34476 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34479 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34480 condition on optimize for the leaf function test.
34482 2017-02-28 Martin Liska <mliska@suse.cz>
34485 * read-rtl-function.c (function_reader::handle_unknown_directive):
34486 Bail out when one uses -flto.
34488 2017-02-28 Martin Liska <mliska@suse.cz>
34490 * common.opt: Replace space with tabular for options of <number>
34492 * config/i386/i386.opt: Show <number> value for
34493 -mlarge-data-threshold.
34494 * opts.c (print_filtered_help): Do not display number in hexadecimal
34497 2017-02-28 Martin Liska <mliska@suse.cz>
34499 * common.opt: Fix --help=option -Q for options which are of
34502 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34504 * config/i386/i386.c (print_reg): Error out for values
34505 of 8-bit size in invalid integer register.
34507 2017-02-28 Martin Sebor <msebor@redhat.com>
34509 PR tree-optimization/79691
34510 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34512 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34515 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34516 gcc_unreachable with output_operand_lossage.
34518 2017-02-28 Richard Biener <rguenther@suse.de>
34520 PR tree-optimization/79740
34521 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34523 (visit_nary_op): Insert the nary into the hashtable if we
34524 pattern-matched sth.
34525 * tree-ssa-pre.c (eliminate_insert): Robustify.
34527 2017-02-28 Richard Biener <rguenther@suse.de>
34529 PR middle-end/79731
34530 * fold-const.c (decode_field_reference): Reject out-of-bound
34533 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34535 * config/i386/i386.c: Include intl.h.
34536 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34537 instead of just cond ? "..." : "...".
34538 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34539 * coverage.c (read_counts_file): Likewise.
34540 * omp-offload.c: Include intl.h.
34541 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34542 of just cond ? "..." : "...".
34543 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34544 of just cond ? "..." : "...".
34546 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34549 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34551 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34552 'tune for' CPU name.
34553 * config/arm/arm-cpu-data.h: Regenerated.
34555 2017-02-28 Richard Biener <rguenther@suse.de>
34557 PR tree-optimization/79732
34558 * tree-inline.c (expand_call_inline): Do not shadow var.
34560 2017-02-28 Richard Biener <rguenther@suse.de>
34562 PR tree-optimization/79723
34563 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34564 address-space properly.
34566 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34568 * doc/optinfo.texi (Optimization groups): Fix option used for
34570 * doc/invoke.texi (-fopt-info): Document "omp".
34571 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34572 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34573 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34574 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34575 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34577 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34579 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34580 OPTGROUP_OMP with "omp".
34582 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34585 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34586 for arithmetic shift of unsigned V2DI.
34588 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34590 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34591 arc/linux.h headers.
34592 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34593 (LINK_SPEC): Likewise.
34594 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34595 (EXTRA_SPECS): Likewise.
34596 (STARTFILE_SPEC): Likewise.
34597 (ENDFILE_SPEC): Likewise.
34598 (LIB_SPEC): Likewise.
34599 (TARGET_SDATA_DEFAULT): Likewise.
34600 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34601 (MULTILIB_DEFAULTS): Likewise.
34602 (DWARF2_UNWIND_INFO): Likewise.
34603 * config/arc/big.h: New file.
34604 * config/arc/elf.h: Likewise.
34605 * config/arc/linux.h: Likewise.
34606 * config/arc/t-uClibc: Remove.
34608 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34610 PR tree-optimization/77536
34611 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34612 (tree_transform_and_unroll_loop): Use above function to compute the
34613 estimated niter of unrolled loop and use it when scaling profile.
34614 Also use count info rather than frequency if it's non-zero.
34615 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34616 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34617 (vect_transform_loop): Call above function.
34619 2017-02-27 Richard Biener <rguenther@suse.de>
34621 PR tree-optimization/45397
34622 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34623 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34624 (visit_nary_op): Add pattern matching for CSEing sign-changed
34625 or truncated operations with wider ones.
34627 2017-02-27 Richard Biener <rguenther@suse.de>
34629 PR tree-optimization/79690
34630 * tree-vect-stmts.c (vectorizable_store): Use vector type
34631 built from the DR with address-space.
34633 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34635 * doc/invoke.texi (Optimize Options): Refine the description
34636 of asan-use-after-return.
34638 2017-02-25 Alan Modra <amodra@gmail.com>
34640 PR rtl-optimization/79584
34641 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34642 base, not ad->base_term, the reg within base. Remove assertion
34643 that ad->base == ad->base_term. Replace gen_int_mode using
34644 bogus mode with const0_rtx.
34646 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34648 PR middle-end/79396
34649 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34650 FMA_EXPR like tcc_binary or tcc_unary.
34652 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34655 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34657 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34658 (output_loc_operands): Handle DW_OP_call_ref and
34659 DW_OP_GNU_variable_value.
34660 (struct variable_value_struct): New type.
34661 (struct variable_value_hasher): Likewise.
34662 (variable_value_hash): New variable.
34663 (string_types): Remove.
34664 (copy_loc_descr): New function.
34665 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34666 (prepend_loc_descr_to_each): New function.
34667 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34668 instead of add_loc_descr_to_each if the first argument is single
34669 location list and the second has multiple.
34670 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34671 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34672 when looking for variable value which doesn't have other location info.
34673 (loc_list_from_tree): Formatting fix.
34674 (gen_array_type_die): Simplify DW_AT_string_length handling.
34675 (adjust_string_types): Remove.
34676 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34677 string_types. Call resolve_variable_values.
34678 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34679 (resolve_addr_in_expr): Likewise. Add A argument.
34680 (copy_deref_exprloc): Remove deref argument. Adjust for the
34681 original expression being DW_OP_GNU_variable_value with optionally
34682 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34683 optionally after it.
34684 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34685 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34686 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34687 (variable_value_hasher::hash, variable_value_hasher::equal): New
34689 (resolve_variable_value_in_expr, resolve_variable_value,
34690 resolve_variable_values, note_variable_value_in_expr,
34691 note_variable_value): New functions.
34692 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34695 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34698 * opts.h (handle_generated_option): Add GENERATED_P argument.
34699 * opts-common.c (handle_option): Adjust function comment.
34700 (handle_generated_option): Add GENERATED_P argument, pass it to
34702 (control_warning_option): Pass false to handle_generated_option
34704 * opts.c (maybe_default_option): Pass true to handle_generated_option
34706 * optc-gen.awk: Likewise.
34708 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34710 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34711 a REG, look at the REG it is a SUBREG of.
34712 (splitter for cmpeqsi_t): Ditto.
34714 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34716 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34717 the special USEs with the pattern of the insn, not the insn itself.
34719 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34722 * doc/invoke.texi: Document -mload-store-pairs.
34724 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34725 Sandra Loosemore <sandra@codesourcery.com>
34727 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34728 argument isn't a CONST_INT.
34729 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34730 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34731 (nios2_validate_compare): Bypass alternate compare logic if *op2
34732 is not a CONST_INT.
34733 (ldstwm_operation_p): Return false if first_base is not a REG or
34734 if first_offset is not a CONST_INT.
34736 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34738 * config/cris/cris.md: Use correct operand in a define_peephole2.
34740 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34742 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34746 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34747 this_insn if it is an INSN or JUMP_INSN.
34748 (force_offsettable): Look at base, not at addr.
34749 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34750 on things that aren't necessarily CONST_INTs.
34752 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34754 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34755 -mfpmath=sse is the default also for x86-32 targets with SSE2
34756 instruction set when @option{-ffast-math} is enabled
34758 2017-02-24 Jeff Law <law@redhat.com>
34760 PR rtl-optimizatoin/79286
34761 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34764 2017-02-24 Richard Biener <rguenther@suse.de>
34766 PR tree-optimization/79389
34767 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34770 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34772 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34773 function comment to reflect reality.
34774 (loop_exits_before_overflow): Fix typo in function description.
34776 2017-02-24 Richard Biener <rguenther@suse.de>
34778 PR tree-optimization/79389
34779 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34780 properly that a threading opportunity exists. Detect conditional
34781 copy/constant propagation opportunities.
34783 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34785 * config/visium/visium.md (type): Add trap.
34786 (b): New mode attribute.
34787 (*btst): Rename into...
34788 (*btst<mode>): ...this and adjust.
34789 (*cbranchsi4_btst_insn): Rename into...
34790 (*cbranch<mode>4_btst_insn): ...this and adjust.
34791 (trap): New define_insn.
34793 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34795 PR tree-optimization/79389
34796 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34797 (noce_reversed_cond_code): New function.
34798 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34799 reversed_comparison_code. Formatting fix.
34800 (noce_try_store_flag): Test rev_cond != NULL in addition to
34801 reversed_comparison_code.
34802 (noce_try_store_flag_constants): Likewise.
34803 (noce_try_store_flag_mask): Likewise.
34804 (noce_try_addcc): Use rev_cond if non-NULL instead of
34805 reversed_comparison_code.
34806 (noce_try_cmove_arith): Likewise. Formatting fixes.
34807 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34808 (noce_find_if_block): Initialize rev_cond.
34809 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34810 instead of false as last argument never attempt to reverse it
34813 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34815 PR tree-optimization/79663
34816 * tree-predcom.c (combine_chains): Process refs in reverse order
34817 only for ZERO length chains, and add explaining comment.
34819 2017-02-23 Jeff Law <law@redhat.com>
34821 PR tree-optimization/79578
34822 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34823 in call to operand_equal_p.
34825 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34828 * config/i386/cpuid.h: Fix another undefined behavior.
34830 2017-02-23 Richard Biener <rguenther@suse.de>
34832 PR tree-optimization/79683
34833 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34834 vector types for data-refs.
34836 2017-02-23 Martin Liska <mliska@suse.cz>
34838 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34840 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34842 PR middle-end/79665
34843 * internal-fn.c (get_range_pos_neg): Moved to ...
34844 * tree.c (get_range_pos_neg): ... here. No longer static.
34845 * tree.h (get_range_pos_neg): New prototype.
34846 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34847 are known to be in between 0 and signed maximum inclusive, try to
34848 expand both unsigned and signed divmod and use the cheaper one from
34851 2017-02-22 Jeff Law <law@redhat.com>
34853 PR tree-optimization/79578
34854 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34855 to compare base operands.
34857 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34860 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34861 gpc_reg_operand instead of fpr_reg_operand.
34863 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34865 * config/mips/mips.c (mips_return_in_memory): Force FP
34866 vector types to be returned in memory for o32 ABI.
34868 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34870 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34871 instead of DW_TAG_member for static data member declarations and don't
34872 set no_linkage_name for static inline data members.
34873 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34876 2017-02-22 Martin Liska <mliska@suse.cz>
34878 * doc/invoke.texi: Replace inequality signs with square brackets
34881 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34884 * lra-constraints.c (simplify_operand_subreg): Handle
34885 WORD_REGISTER_OPERATIONS targets.
34887 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34890 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34891 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34892 elimination by swapping fld*.
34894 2017-02-22 Richard Biener <rguenther@suse.de>
34896 PR tree-optimization/79673
34897 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34898 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34899 irrelevant address-space qualifiers and avoiding a
34900 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34902 2017-02-22 Richard Biener <rguenther@suse.de>
34904 PR tree-optimization/79666
34905 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34906 to not symbolically negate if that may introduce undefined
34909 2017-02-22 Martin Liska <mliska@suse.cz>
34912 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34913 * data-streamer-out.c (streamer_write_gcov_count_stream):
34915 * value-prof.c (stream_out_histogram_value): Make assert more
34916 precise based on type of counter.
34918 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34921 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34922 Use nonimmediate_operand instead of memory_operand for operand 1.
34923 (float-extend standard_x87sse_constant_load splitter): Ditto.
34925 2017-02-21 Jeff Law <law@redhat.com>
34927 PR tree-optimization/79621
34928 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34929 blocks with edges to themselves.
34931 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34934 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34935 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34936 Use gimple_call_builtin_p.
34939 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34940 on temporarily removed DEBUG_INSNs.
34942 PR tree-optimization/79649
34943 * tree-loop-distribution.c (classify_partition): Give up on
34944 non-generic address space loads/stores.
34946 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34948 * doc/loop.texi (Loop manipulation): Remove nonexistent
34949 tree_ssa_loop_version from the documentation.
34950 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34952 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34955 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34956 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34957 * config/rs6000/rs6000.c: Include except.h.
34958 (rs6000_expand_split_stack_prologue): Call
34959 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34961 2017-02-21 Martin Jambor <mjambor@suse.cz>
34964 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34965 have been analyzed.
34967 2017-02-21 Martin Jambor <mjambor@suse.cz>
34969 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34970 for backward compatibility only.
34971 * doc/invoke.texi (Option Summary): Remove all references to
34972 -fipa-cp-alignment.
34974 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34978 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34980 * lra-constraints.c (curr_insn_transform): Handle
34981 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34983 2017-02-21 Martin Liska <mliska@suse.cz>
34985 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34987 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34989 PR translation/79638
34990 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34992 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34995 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34996 (arm_function_ok_for_sibcall): Return false for an indirect call by
34997 descriptor if all the argument registers are used.
34998 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34999 alignment of the function.
35001 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35003 PR tree-optimization/61441
35004 * simplify-rtx.c (simplify_const_unary_operation): For
35005 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35006 the sNaN unmodified.
35008 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35010 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35011 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35012 instead of SYSTEM_HEADER_DIR.
35014 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35015 Martin Liška <mliska@suse.cz>
35017 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35018 Fix typos and grammar, use active voice, and clarify.
35020 2017-02-20 Marek Polacek <polacek@redhat.com>
35022 PR middle-end/79537
35023 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35026 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35028 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35031 * config/i386/i386.c (ix86_expand_builtin): Handle
35032 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35033 ix86_builtins_isa[fcode].isa as a requirement of those
35034 flags and any other flag in the bitmask.
35035 (ix86_init_mmx_sse_builtins): Use 0 instead of
35036 ~OPTION_MASK_ISA_64BIT as mask.
35037 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35038 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35039 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35040 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35042 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35045 * lra-constraints.c (split_reg): Check requested split mode
35046 is supported by the register.
35048 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35050 * lra-constraints.c (simplify_operand_subreg): Remove early
35053 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35056 * lra-constraints.c (curr_insn_transform): Tighten condition
35057 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35059 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35062 * lra-constraints.c (curr_insn_transform): Handle
35063 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35065 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35068 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35070 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35072 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35075 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35078 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35080 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35083 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35085 * final.c (last_columnnum, override_columnnum): New variables.
35086 (final_start_function): Set last_columnnum, pass it to begin_prologue
35087 hook and pass 0 to dwarf2out_begin_prologue.
35088 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35089 to source_line debug hook.
35090 (notice_source_line): Compute last_columnnum and for debug_column_info
35091 return true on column changes.
35092 * debug.h (struct gcc_debug_hooks): Add column argument to
35093 source_line and begin_prologue hooks.
35094 (debug_nothing_int_charstar_int_bool): Remove prototype.
35095 (debug_nothing_int_int_charstar,
35096 debug_nothing_int_int_charstar_int_bool): New prototypes.
35097 (dwarf2out_begin_prologue): Add column argument.
35098 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35099 begin_prologue hooks.
35100 (debug_nothing_int_charstar_int_bool): Remove.
35101 (debug_nothing_int_int_charstar,
35102 debug_nothing_int_int_charstar_int_bool): New functions.
35103 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35104 through to dwarf2out_source_line.
35105 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35106 (dwarf2out_source_line): Add column argument, emit it if requested.
35107 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35109 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35110 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35111 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35112 through to dwarf2out_begin_prologue.
35113 (vmsdbgout_source_line): Add column argument, pass it through to
35114 dwarf2out_source_line.
35115 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35116 dbxout_source_line caller.
35117 (dbxout_source_line): Add column argument.
35119 * common.opt (gno-column-info, gcolumn-info): New options.
35120 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35121 (check_die): Also test for multiple DW_AT_decl_column attributes.
35122 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35123 DW_AT_decl_column if requested.
35124 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35126 (gen_variable_die): Likewise.
35127 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35128 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35131 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35132 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35133 (ix86_handle_option): Handle OPT_m3dnowa.
35134 * doc/invoke.texi (-m3dnowa): Document.
35135 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35136 -m3dnowa instead of -m3dnow -march=athlon.
35139 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35140 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35142 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35145 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35146 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35147 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35148 generator for vsx_xxpermdi_<mode>_be.
35149 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35150 force big-endian semantics.
35151 (vsx_xxpermdi_<mode>_be): New define_expand with same
35152 implementation as previous version of vsx_xxpermdi_<mode>.
35154 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35156 PR tree-optimization/79327
35157 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35158 variable, its initialization and use.
35160 2017-02-17 Julia Koval <julia.koval@intel.com>
35162 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35163 (OPTION_MASK_ISA_PKU_UNSET): New.
35164 (ix86_handle_option): Handle -mrdpid.
35165 * config/i386/cpuid.h (bit_RDPID): New.
35166 * config/i386/driver-i386.c (host_detect_local_cpu):
35167 Detect RDPID feature.
35168 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35169 * config/i386/i386-c.c (ix86_target_macros_internal):
35171 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35172 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35173 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35174 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35175 * config/i386/i386.md (define_insn "rdpid"): New.
35176 * config/i386/i386.opt Add -mrdpid.
35177 * config/i386/immintrin.h (_rdpid_u32): New.
35179 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35181 PR rtl-optimization/79541
35182 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35183 instead of transforming it into USE.
35185 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35187 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35188 If HONOR_SNANS (SFmode) force the input to a register.
35189 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35190 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35191 an frsp or similar insn.
35193 2017-02-17 Martin Liska <mliska@suse.cz>
35195 PR rtl-optimization/79577
35196 * params.def (selsched-max-sched-times): Increase minimum to 1.
35198 2017-02-17 Martin Liska <mliska@suse.cz>
35200 PR rtl-optimization/79574
35201 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35203 2017-02-17 Martin Liska <mliska@suse.cz>
35205 PR tree-optimization/79529
35206 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35207 ssa_defined_default_def_p to handle cases which are implicitly
35209 * tree-ssa.c (ssa_defined_default_def_p): New function.
35210 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35211 which are implicitly defined.
35212 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35214 2017-02-17 Richard Biener <rguenther@suse.de>
35216 PR middle-end/79576
35217 * params.def (max-ssa-name-query-depth): Limit to 10.
35219 2017-02-17 Richard Biener <rguenther@suse.de>
35221 PR tree-optimization/79552
35222 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35225 2017-02-17 Richard Biener <rguenther@suse.de>
35228 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35230 2017-02-17 Marek Polacek <polacek@redhat.com>
35232 PR middle-end/79536
35233 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35234 (fold_negate_expr): New wrapper.
35236 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35238 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35239 Correct terminology and de-emphasize pre-standard behavior.
35241 2017-02-16 Alan Modra <amodra@gmail.com>
35243 PR rtl-optimization/79286
35244 * ira.c (def_dominates_uses): New function.
35245 (update_equiv_regs): Don't create an equivalence for insns that
35246 may trap where the register def does not dominate the use.
35248 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35250 PR rtl-optimization/78127
35251 * lra.c (lra): Call lra_eliminate before finish the loop after
35254 2017-02-16 Richard Biener <rguenther@suse.de>
35256 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35258 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35259 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35260 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35261 (isl_val_int_from_wi): New function.
35262 (extract_affine_gmp): Rename to ...
35263 (extract_affine_wi): ... this, take a widest_int.
35264 (extract_affine_int): Just wrap extract_affine_wi.
35265 (add_param_constraints): Use isl_val_int_from_wi.
35266 (add_loop_constraints): Likewise, and extract_affine_wi.
35268 2017-02-15 Jeff Law <law@redhat.com>
35270 PR middle-end/79521
35271 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35272 ira_init_register_move_cost_if_necessary.
35274 2017-02-15 Martin Sebor <msebor@redhat.com>
35276 PR middle-end/32003
35277 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35278 removed in a prior commit.
35280 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35282 PR tree-optimization/79347
35283 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35284 counters during peeling.
35286 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35288 * Makefile.in (site.exp): Remove "set ISLVER".
35290 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35293 * real.c (real_from_integer): Call real_convert even for decimal.
35295 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35298 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35300 2017-02-14 Andrew Pinski <apinski@cavium.com>
35302 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35303 cores and change the partno/implementer to be correct.
35304 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35305 the 'B" as the implementer.
35306 * config/aarch64/aarch64-tune.md: Regenerate.
35308 2017-02-14 Carl Love <cel@us.ibm.com>
35310 * config/rs6000/rs6000.c: Add case statement entry to make the
35311 xvcvuxdsp built-in argument unsigned.
35312 * config/rs6000/vsx.md: Fix the source and return operand types so they
35313 match the instruction definitions from the ISA document. Fix typo
35314 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35317 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35320 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35321 member early_clobber_alts.
35322 * lra-lives.c (reg_early_clobber_p): New.
35323 (process_bb_lives): Use it.
35324 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35325 (debug_operand_data): Initialize early_clobber_alts.
35326 (setup_operand_alternative): Set up early_clobber_alts.
35327 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35328 alternatives to new_insn_reg.
35329 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35331 (lra_update_insn_regno_info): Pass the new arg.
35333 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35335 PR middle-end/79505
35336 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35337 (new_oacc_loop_raw): Don't clear already cleared fields.
35340 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35341 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35342 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35344 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35347 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35349 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35352 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35353 the extra instruction to the right place to store 128-bit constant
35356 2017-02-14 Martin Sebor <msebor@redhat.com>
35358 PR middle-end/79448
35359 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35360 warning for strings of unknown length.
35362 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35364 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35366 2017-02-14 Jeff Law <law@redhat.com>
35369 * ira-costs.c (scan_one_insn): Initialize register move costs
35370 for pseudos seen in USE/CLOBBER insns.
35372 PR tree-optimization/79095
35373 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35374 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35375 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35376 if the operands are known to be not equal, then the resulting range
35378 (intersect_ranges): If the new range is ~[0,0] and the old range is
35379 wide, then prefer ~[0,0].
35380 * tree-vrp.c (overflow_comparison_p_1): New function.
35381 (overflow_comparison_p): New function.
35382 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35383 if NAME is used in an overflow test.
35384 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35385 overflow check that can be expressed as an equality test, then adjust
35386 ops to be that equality test.
35388 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35390 * config/s390/s390-builtin-types.def: Remove flags argument.
35391 * config/s390/s390.c (s390_init_builtins): Likewise.
35393 2017-02-14 Martin Liska <mliska@suse.cz>
35395 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35396 vector. Fix trailing white spaces.
35398 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35400 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35403 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35405 PR rtl-optimization/68664
35406 * config/arm/arm.c (arm_sched_can_speculate_insn):
35407 New function. Declare prototype.
35408 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35410 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35412 PR rtl-optimization/68664
35413 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35415 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35417 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35419 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35420 max skip bytes for function, loop and jump.
35422 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35424 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35425 ABS_EXPR for gimple dump.
35427 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35430 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35432 PR tree-optimization/79408
35433 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35434 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35435 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35436 also if rhs1 is INTEGER_CST.
35438 2017-02-14 Richard Biener <rguenther@suse.de>
35440 PR middle-end/79432
35441 * tree-into-ssa.c (insert_phi_nodes): When the function can
35442 have abnormal edges rewrite SSA names with broken use-def
35443 dominance out of SSA and register them for PHI insertion.
35445 2017-02-13 Martin Sebor <msebor@redhat.com>
35447 PR middle-end/79496
35448 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35449 clearing info.nowrite flag when snprintf size argument is a range.
35451 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35453 * cprop.c (cprop_jump): Add missing space in string literal.
35454 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35455 (get_constraint_for_component_ref): Likewise.
35456 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35457 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35458 * lra-constraints.c (process_alt_operands): Likewise.
35459 * ipa-inline.c (inline_small_functions): Likewise.
35460 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35461 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35462 * trans-mem.c (diagnose_tm_1_op): Likewise.
35463 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35464 (grid_parallel_clauses_gridifiable): Likewise.
35466 * config/nvptx/mkoffload.c (process): Add space in between
35469 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35470 "MOD4_SSE_REGS" and "ALL_REGS".
35472 * spellcheck.c (test_data): Add , in between "foo" and "food".
35474 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35477 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35478 boundary crossing check and subsequent code generation agree.
35480 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35482 * config/aarch64/aarch64.c (has_memory_op): Delete.
35483 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35486 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35488 PR rtl-optimization/79388
35489 PR rtl-optimization/79450
35490 * combine.c (distribute_notes): When removing TEM_INSN for which
35491 corresponding dest has last value recorded, invalidate that last
35494 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35496 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35497 of explicit '@'. Add missing assembly comment marker on branch costs
35500 2017-02-13 Nathan Sidwell <nathan@acm.org>
35502 * gengtype-lex.l (<in_struct>): Add '/'.
35504 2017-02-13 Martin Liska <mliska@suse.cz>
35507 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35509 2017-02-13 Richard Biener <rguenther@suse.de>
35511 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35513 * configure: Re-generate.
35514 * config.in: Likewise.
35515 * graphite-dependences.c: Simplify as if
35516 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35517 * graphite-isl-ast-to-gimple.c: Likewise.
35518 * graphite-optimize-isl.c: Likewise.
35519 * graphite-poly.c: Likewise.
35520 * graphite-sese-to-poly.c: Likewise.
35521 * graphite.h: Likewise.
35522 * toplev.c: Include isl/version.h and use isl_version () for
35523 printing the ISL version.
35524 * doc/install.texi: Update ISL requirement.
35526 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35528 * doc/standards.texi (Standards): Update reference to
35531 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35533 * doc/extend.texi (Named Address Spaces): sourceware.org now
35535 * doc/install.texi (Binaries): Ditto.
35538 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35540 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35541 standard terminology "stringize"/"stringizing" throughout.
35542 * doc/cppinternals.texi: Likewise.
35544 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35546 * doc/extend.texi: Fix some spelling mistakes and typos.
35547 * doc/invoke.texi: Likewise.
35549 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35552 * params.def (inline-min-speedup) Change from 10 to 8.
35554 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35556 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35559 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35562 * ipa-inline-analysis.c (get_minimal_bb): New function.
35563 (record_modified): Use it.
35564 (remap_edge_change_prob): Handle also ancestor functions.
35566 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35568 * doc/contrib.texi (Contributors): Remove broken link into
35569 the Mauve CVS repository.
35571 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35573 PR middle-end/79454
35574 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35575 result computation whenever lhs doesn't have vector mode, not
35576 just when it has BLKmode.
35578 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35580 * doc/makefile.texi (profiledbootstrap): Refer to the
35581 installation instructions only in textual form.
35583 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35586 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35588 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35590 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35591 (Specific): Update mingw-w64 reference.
35593 (Specific): Remove broken link to Renesas RX processor.
35595 2017-02-10 Richard Biener <rguenther@suse.de>
35597 * toplev.c (process_options): Do not mention obsolete graphite
35598 options when printing sorry message about missing graphite support.
35599 Mention -floop-nest-optimize.
35601 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35603 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35604 (vtst_p16): Likewise.
35605 (vtstq_p8): Likewise.
35606 (vtstq_p16): Likewise.
35608 (vtstq_p64): Likewise.
35609 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35610 (vset_lane_p64): New.
35611 (vsetq_lane_p64): New.
35613 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35615 PR tree-optimization/79411
35616 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35617 stmt operands are SSA_NAMEs used in abnormal phis.
35618 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35621 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35624 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35627 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35629 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35631 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35633 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35637 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35638 not arbitrary TREE_CONSTANT.
35641 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35642 "omp declare target link" attribute unless is_global_var.
35643 * omp-offload.c (find_link_var_op): Likewise.
35645 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35646 Chung-Lin Tang <cltang@codesourcery.com>
35648 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35650 (gimplify_adjust_omp_clauses): Don't delete TILE.
35651 (gimplify_omp_for): Deal with TILE.
35652 * internal-fn.c (expand_GOACC_TILE): New function.
35653 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35655 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35656 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35658 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35659 avoid DIV for outermost collapse var.
35660 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35661 Remove out of date comments, fix whitespace.
35662 * omp-general.c (omp_extract_for_data): Deal with tiling.
35663 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35664 adjust OLF_DIM_BASE value.
35665 (struct omp_for_data): Add tiling field.
35666 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35667 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35668 for auto loops. Remove default auto determining, moved to
35669 oacc_loop_fixed_partitions.
35670 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35671 stmts, add e_mask field.
35672 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35673 (oacc_thread_numbers): Use oacc_dim_call.
35674 (oacc_xform_tile): New.
35675 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35676 (finish_oacc_loop): Adjust for ifns vector.
35677 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35678 add case for GOACC_TILE fns.
35679 (oacc_loop_xform_loop): Delete.
35680 (oacc_loop_process): Iterate over call list directly, and add
35681 handling for GOACC_TILE fns.
35682 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35684 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35685 vector partitioning to outer loops. Assign 2 partitions to loops
35686 when available. Add TILE handling.
35687 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35688 (execite_oacc_device_lower): Process GOACC_TILE fns,
35689 ignore unknown specs.
35690 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35691 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35692 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35694 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35696 * configure.ac (ACX_BUGURL): Update.
35697 * configure: Regenerate.
35699 2017-02-09 Richard Biener <rguenther@suse.de>
35701 PR tree-optimization/69823
35702 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35703 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35705 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35707 * config/arc/arc-c.def: Add __NPS400__ definition.
35708 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35709 (TARGET_NPS400): Define.
35711 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35713 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35715 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35716 pointer, arch_info.
35717 (arc_cpu_types): Fill the arch_info field with a pointer into the
35718 arc_arch_types table.
35719 (arc_selected_cpu): Declare.
35720 * config/arc/arc.c (arc_selected_cpu): Make global.
35721 (arc_selected_arch): Delete.
35722 (arc_base_cpu): Delete.
35723 (arc_override_options): Remove references to deleted variables,
35724 update access to arch information.
35725 (ARC_OPT): Update access to arch information.
35726 (ARC_OPTX): Likewise.
35727 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35728 (TARGET_ARC600): Update access to arch information.
35729 (TARGET_ARC601): Likewise.
35730 (TARGET_ARC700): Likewise.
35731 (TARGET_EM): Likewise.
35732 (TARGET_HS): Likewise.
35733 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35736 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35739 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35740 condition/operands for integer GE/LE/GEU/LEU operations.
35742 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35744 PR translation/79397
35745 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35748 2017-02-08 Martin Jambor <mjambor@suse.cz>
35751 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35752 whether allocation happened.
35753 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35754 nothing was allocated.
35756 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35758 PR tree-optimization/79408
35759 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35760 constant, but SSA_NAME with a known integer range, use the minimum
35761 of that range instead of op1 to determine if modulo can be replaced
35762 with its first operand.
35764 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35766 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35768 2017-02-08 Richard Biener <rguenther@suse.de>
35770 PR tree-optimization/71824
35771 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35772 Check all loops contained in the merged region.
35774 2017-02-07 Andrew Pinski <apinski@cavium.com>
35776 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35778 2017-02-07 Andrew Pinski <apinski@cavium.com>
35780 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35781 (thunderxt88): Likewise.
35782 (thunderxt81): Disable LSE and change v8.1 to v8.
35783 (thunderxt83): Likewise.
35785 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35786 Richard Biener <rguenther@suse.de>
35788 PR middle-end/79399
35789 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35790 type from int to size_t.
35791 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35793 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35795 PR rtl-optimization/79386
35796 * cprop.c (bypass_conditional_jumps): Initialize
35797 bypass_last_basic_block already before splitting bbs after
35798 unconditional traps...
35799 (bypass_conditional_jumps): ... rather than here.
35802 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35803 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35804 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35805 fix -masm=intel patterns.
35807 2017-02-07 Richard Biener <rguenther@suse.de>
35809 PR tree-optimization/79256
35810 PR middle-end/79278
35811 * builtins.c (get_object_alignment_2): Use min_align_of_type
35812 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35813 and ADJUST_FIELD_ALIGN.
35815 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35817 * doc/tm.texi: Regenerate.
35818 * stor-layout.c (layout_decl): Adjust.
35819 (update_alignment_for_field): Likewise.
35820 (place_field): Likewise.
35821 (min_align_of_type): Likewise.
35822 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35823 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35824 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35825 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35826 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35827 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35828 * config/i386/i386.c (x86_field_alignment): Likewise.
35829 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35830 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35831 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35832 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35833 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35834 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35838 2017-01-30 Richard Biener <rguenther@suse.de>
35840 PR tree-optimization/79256
35841 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35842 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35845 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35847 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35848 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35849 builtins to SImode and emit a zero-extend, if necessary.
35851 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35853 * docs/invoke.texi (RISC-V Options): Alphabetize.
35855 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35857 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35860 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35862 * config/riscv/riscv.c: New file.
35863 * common/config/riscv/riscv-common.c: Likewise.
35864 * config.gcc: Likewise.
35865 * config/riscv/constraints.md: Likewise.
35866 * config/riscv/elf.h: Likewise.
35867 * config/riscv/generic.md: Likewise.
35868 * config/riscv/linux.h: Likewise.
35869 * config/riscv/multilib-generator: Likewise.
35870 * config/riscv/peephole.md: Likewise.
35871 * config/riscv/pic.md: Likewise.
35872 * config/riscv/predicates.md: Likewise.
35873 * config/riscv/riscv-builtins.c: Likewise.
35874 * config/riscv/riscv-c.c: Likewise.
35875 * config/riscv/riscv-ftypes.def: Likewise.
35876 * config/riscv/riscv-modes.def: Likewise.
35877 * config/riscv/riscv-opts.h: Likewise.
35878 * config/riscv/riscv-protos.h: Likewise.
35879 * config/riscv/riscv.h: Likewise.
35880 * config/riscv/riscv.md: Likewise.
35881 * config/riscv/riscv.opt: Likewise.
35882 * config/riscv/sync.md: Likewise.
35883 * config/riscv/t-elf-multilib: Likewise.
35884 * config/riscv/t-linux: Likewise.
35885 * config/riscv/t-linux-multilib: Likewise.
35886 * config/riscv/t-riscv: Likewise.
35887 * configure.ac: Likewise.
35888 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35889 Waterman as RISC-V maintainers.
35890 * doc/install.texi: Add RISC-V entries.
35891 * doc/invoke.texi: Add RISC-V options section.
35892 * doc/md.texi: Add RISC-V constraints section.
35893 * configure: Regenerated.
35895 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35898 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35899 false values to be constant vectors with all 0 or all 1 bits set.
35900 (vcondu<mode><mode>): Likewise.
35901 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35903 (fpmask_comparison_operator): Update comment.
35904 (vecint_comparison_operator): New predicate.
35905 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35906 vector conditionals when the true and false values are constant
35907 vectors with all 0 bits or all 1 bits set.
35909 2017-02-06 Martin Sebor <msebor@redhat.com>
35911 PR tree-optimization/79376
35912 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35914 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35916 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35917 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35918 to simplify split condition.
35920 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35922 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35923 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35926 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35928 PR rtl-optimization/68664
35929 * target.def (can_speculate_insn): New hook.
35930 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35931 * doc/tm.texi: Regenerate.
35932 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35933 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35934 (rs6000_sched_can_speculate_insn): New function.
35936 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35938 PR tree-optimization/79284
35939 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35940 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35941 vectorizable_mask_load_store, vectorizable_operation,
35942 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35943 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35944 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35945 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35946 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35947 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35948 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35949 is_gimple_assign (stmt). Replace another such test with
35950 is_gimple_assign (stmt).
35952 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35955 * config/avr/avr.c (rtl-iter.h): Include it.
35956 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35957 (avr_legitimate_combined_insn): ...and implementation.
35959 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35961 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35962 * config/s390/s390.c (s390_const_operand_ok)
35963 (s390_canonicalize_comparison, s390_extract_part)
35964 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35965 (s390_contiguous_bitmask_p, s390_rtx_costs)
35966 (legitimize_pic_address): Likewise.
35967 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35968 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35969 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35970 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35971 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35973 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35975 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35976 REGNO($0) == REGNO($1).
35978 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35980 * config/s390/linux.h(SIZE_TYPE): Add comment.
35982 2017-02-06 Julian Brown <julian@codesourcery.com>
35983 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35984 Virendra Pathak <virendra.pathak@broadcom.com>
35986 * config/aarch64/aarch64-cores.def: Change the scheduler
35988 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35989 * config/aarch64/thunderx2t99.md: New file.
35991 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35993 * doc/standards.texi (Go Language): Update link to language
35996 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35998 * tree-eh.c (lower_resx): Sanitize profile.
35999 (cleanup_empty_eh_move_lp): Likewise.
36001 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36004 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36006 * cfgloopmanip.h (loop_version): Update prototype.
36007 * modulo-sched.c (sms_schedule): Update call of loop_version.
36008 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36009 * tree-parloops.c (gen_parallel_loop): Likewise.
36010 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36011 * tree-ssa-loop-split.c (split_loop): Likewise.
36012 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36013 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36015 2017-02-05 Martin Liska <mliska@suse.cz>
36018 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36020 (print_operand_address): Initialize a struct to zero.
36022 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36024 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36025 garbage collector only in textual form.
36027 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36029 * doc/extend.texi (x86 specific memory model extensions for
36030 transactional memory): Simplify a phrase.
36032 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36035 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36036 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36037 (atomic_storedi_1): Likewise.
36039 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36041 PR tree-optimization/79338
36042 * tree-parloops.c (gather_scalar_reductions): Don't call
36043 vect_analyze_loop_form for loop->inner before destroying loop's
36046 2017-02-03 Martin Sebor <msebor@redhat.com>
36048 PR tree-optimization/79327
36049 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36050 when precision has resulted in leading zeros.
36051 (format_integer): Adjust the likely counter to assume an unknown
36052 argument that may be zero is non-zero.
36054 2017-02-03 Jason Merrill <jason@redhat.com>
36057 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36058 avoid copying non-taken branch.
36060 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36062 PR tree-optimization/79340
36063 * tree-vect-loop.c (vectorizable_reduction): Release
36064 vec_defs elements after safe_splicing them into other vectors.
36067 PR tree-optimization/79327
36068 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36069 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36071 (format_integer): Use wide_int_to_tree instead of build_int_cst
36072 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36073 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36074 of shortest and longest sequence.
36076 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36078 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36079 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36081 2017-02-03 Walter Lee <walt@tilera.com>
36084 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36085 after initial stackframe link reg save.
36086 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36088 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36091 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36092 wu for stxssp alternative.
36094 2017-02-03 Martin Sebor <msebor@redhat.com>
36096 PR tree-optimization/79352
36097 * gimple-fold.c (get_range_strlen): Add argument.
36098 (get_range_strlen): Change return type to bool.
36099 (get_maxval_strlen): Pass in a dummy argument.
36100 * gimple-fold.h (get_range_strlen): Change return type to bool.
36101 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36102 * tree.h (array_at_struct_end_p): Add argument.
36103 * tree.c (array_at_struct_end_p): Handle it.
36105 2017-02-03 Martin Liska <mliska@suse.cz>
36108 * multiple_target.c (create_dispatcher_calls): Redirect edge
36109 from a caller of a dispatcher.
36110 (expand_target_clones): Make the clones local.
36111 (ipa_target_clone): Do both target clones and resolvers.
36112 (ipa_dispatcher_calls): Remove the pass.
36113 (pass_dispatcher_calls::gate): Likewise.
36114 (make_pass_dispatcher_calls): Likewise.
36115 * passes.def (pass_target_clone): Put as very first IPA early
36118 2017-02-03 Martin Liska <mliska@suse.cz>
36120 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36121 in case of a function with ifunc attribute.
36123 2017-02-03 Martin Liska <mliska@suse.cz>
36125 * cgraph.c (cgraph_node::dump): Dump function version info.
36126 * symtab.c (symtab_node::dump_base): Add missing new line.
36128 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36130 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36131 (ifcombine_ifandif): Use it.
36133 2017-02-03 Martin Liska <mliska@suse.cz>
36135 * doc/invoke.texi: Document default value for
36136 use-after-scope-direct-emission-threshold.
36138 2017-02-03 Martin Liska <mliska@suse.cz>
36140 PR tree-optimization/79339
36141 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36142 (format_floating): Likewise.
36144 2017-02-03 Martin Liska <mliska@suse.cz>
36147 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36149 (ipa_node_params_t::remove): Likewise.
36150 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36151 initialization from removed ipa_node_params_t::insert.
36152 (ipa_node_params::~ipa_node_params): Move from removed
36153 ipa_node_params_t::release.
36154 * symbol-summary.h (symbol_summary::m_released): New member.
36155 Do not release a summary twice. Do not allow to call finalizer
36156 for types of a summary that live in GGC memory.
36158 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36160 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36163 2017-02-02 Martin Sebor <msebor@redhat.com>
36165 PR middle-end/79275
36166 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36167 (format_string): Tighten up the range of output for non-constant
36168 strings and correct the expected range for wide non-constant strings.
36170 2017-02-02 Martin Sebor <msebor@redhat.com>
36172 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36174 PR middle-end/32003
36175 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36177 (-fdump-tree-@var): Add to index and document how to come up
36178 with pass-specific option and dump file names.
36179 (-fdump-passes): Clarify where to look for output.
36181 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36183 PR middle-end/77445
36184 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36185 statistics of the analyzed path; allow threading for speed when
36186 any of BBs along the path are optimized for speed.
36188 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36190 PR middle-end/78468
36191 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36192 settings of the virtual registers.
36195 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36197 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36198 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36201 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36203 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36204 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36206 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36208 * config/s390/s390.md: Add missing comments with the expanded
36210 * config/s390/vector.md: Likewise.
36211 * config/s390/vx-builtins.md: Likewise.
36213 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36216 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36217 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36218 conditions on a single line.
36220 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36222 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36223 __S390_VX__ to __VX__.
36225 2017-02-01 Andrew Pinski <apinski@cavium.com>
36227 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36228 stmt_info to record_stmt_cost.
36229 (vect_get_known_peeling_cost): Pass stmt_info if known to
36231 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36232 cpu_vector_cost field into
36233 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36234 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36235 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36236 splitting of scalar_stmt_cost and vec_stmt_cost.
36237 (thunderx_vector_cost): Likewise.
36238 (cortexa57_vector_cost): LIkewise.
36239 (exynosm1_vector_cost): Likewise.
36240 (xgene1_vector_cost): Likewise.
36241 (thunderx2t99_vector_cost): Improve after the splitting of the two
36243 (aarch64_builtin_vectorization_cost): Update for the splitting of
36244 scalar_stmt_cost and vec_stmt_cost.
36246 2017-02-01 Torvald Riegel <triegel@redhat.com>
36247 Richard Henderson <rth@redhat.com>
36249 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36250 conditional on existance of a fast atomic load.
36251 * optabs-query.c (can_atomic_load_p): New function.
36252 * optabs-query.h (can_atomic_load_p): Declare it.
36253 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36254 no fast atomic load is available for the particular size of access.
36255 (expand_atomic_compare_and_swap): Likewise.
36256 (expand_atomic_load): Likewise.
36257 (expand_atomic_store): Likewise.
36258 (expand_atomic_fetch_op): Likewise.
36259 * testsuite/lib/target-supports.exp
36260 (check_effective_target_sync_int_128): Remove x86 because it provides
36261 no fast atomic load.
36262 (check_effective_target_sync_int_128_runtime): Likewise.
36264 2017-02-01 Richard Biener <rguenther@suse.de>
36266 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36267 (graphite_transform_loops): Provide opt-info for optimized nests.
36268 * tree-parloop.c (parallelize_loops): Provide opt-info for
36269 parallelized loops.
36271 2017-02-01 Richard Biener <rguenther@suse.de>
36273 PR middle-end/79315
36274 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36275 was not set before.
36277 2017-02-01 Richard Biener <rguenther@suse.de>
36279 PR tree-optimization/71824
36280 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36281 Verify the loops are valid in the merged SESE region.
36282 (scop_detection::can_represent_loop_1): Check analyzing the
36283 evolution of the number of iterations in the region succeeds.
36285 2017-01-31 Ian Lance Taylor <iant@golang.org>
36287 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36288 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36290 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36292 PR preprocessor/79210
36293 * input.c (get_substring_ranges_for_loc): Replace line_width
36294 assertion with error-handling.
36296 2017-01-31 Richard Biener <rguenther@suse.de>
36298 PR tree-optimization/77318
36299 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36300 (create_pw_aff_from_tree): Take loop parameter.
36301 (add_condition_to_pbb): Pass loop of the condition to
36302 create_pw_aff_from_tree.
36304 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36306 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36307 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36309 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36313 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36315 (convert_int_to_float128): Likewise.
36316 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36317 (convert_int_to_float128): Likewise.
36318 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36319 (UNSPEC_IEEE128_CONVERT): Likewise.
36320 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36321 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36322 Use local variables for IBM extended format.
36323 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36324 (fix_trunc<mode>si2_fprs): Likewise.
36325 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36326 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36327 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36328 to know that we can now have integers of all sizes in vector
36330 (fix<uns>_<mode>di2_hw): Likewise.
36331 (float<uns>_<mode>si2_hw): Likewise.
36332 (fix_<mode>si2_hw): Likewise.
36333 (fixuns_<mode>si2_hw): Likewise.
36334 (float<uns>_<mode>di2_hw): Likewise.
36335 (float_<mode>di2_hw): Likewise.
36336 (float_<mode>si2_hw): Likewise.
36337 (floatuns_<mode>di2_hw): Likewise.
36338 (floatuns_<mode>si2_hw): Likewise.
36339 (xscvqp<su>wz_<mode>): Delete, no longer used.
36340 (xscvqp<su>dz_<mode>): Likewise.
36341 (xscv<su>dqp_<mode>): Likewise.
36342 (ieee128_mfvsrd_64bit): Likewise.
36343 (ieee128_mfvsrd_32bit): Likewise.
36344 (ieee128_mfvsrwz): Likewise.
36345 (ieee128_mtvsrw): Likewise.
36346 (ieee128_mtvsrd_64bit): Likewise.
36347 (ieee128_mtvsrd_32bit): Likewise.
36349 2017-01-31 Martin Liska <mliska@suse.cz>
36352 * ipa-prop.c (ipa_free_all_node_params): Call release method
36353 instead of ~sumbol_summary to not to trigger double times
36356 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36358 PR tree-optimization/71691
36359 * bitmap.h (class auto_bitmap): New.
36360 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36361 is_maybe_undefined instead of ssa_undefined_value_p.
36363 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36365 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36366 __S390_ARCH_LEVEL__ to __ARCH__.
36368 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36370 PR tree-optimization/79267
36371 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36372 if should_remove_lhs_p is true.
36374 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36377 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36378 (add_alignment_attribute): New.
36379 (base_type_die): Add alignment attribute.
36380 (subrange_type_die): Likewise.
36381 (modified_type_die): Likewise.
36382 (gen_array_type_die): Likewise.
36383 (gen_descr_array_type_die: Likewise.
36384 (gen_enumeration_type_die): Likewise.
36385 (gen_subprogram_die): Likewise.
36386 (gen_variable_die): Likewise.
36387 (gen_field_die): Likewise.
36388 (gen_ptr_to_mbr_type_die): Likewise.
36389 (gen_struct_or_union_type_die): Likewise.
36390 (gen_subroutine_type_die): Likewise.
36391 (gen_typedef_die): Likewise.
36392 (base_type_cmp): Compare alignment attribute.
36394 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36397 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36398 (setb_unsigned) New pattern for setb with CCUNS.
36399 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36400 subfc./subfe sequence to avoid overflow problems. Generate a
36401 shorter sequence with cmpld/setb for power9.
36402 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36403 for generating subfc. instruction.
36404 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36405 now uses this instruction.
36407 2017-01-30 Ian Lance Taylor <iant@google.com>
36410 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36411 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36413 2017-01-30 Martin Sebor <msebor@redhat.com>
36415 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36416 Move constant to the right of a relational operator.
36417 (get_mpfr_format_length, format_character, format_string): Ditto.
36418 (should_warn_p, maybe_warn): Same.
36420 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36422 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36425 * asan.c (get_translation_unit_decl): Remove function.
36426 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36428 2017-01-30 Martin Liska <mliska@suse.cz>
36430 PR gcov-profile/79259
36431 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36432 -fprofile-generate.
36434 2017-01-30 Martin Liska <mliska@suse.cz>
36437 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36438 Initialize variables with NULL value.
36440 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36443 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36445 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36447 2017-01-30 Richard Biener <rguenther@suse.de>
36449 PR tree-optimization/79276
36450 * tree-vrp.c (process_assert_insertions): Properly adjust common
36451 when removing a duplicate.
36453 2017-01-30 Richard Biener <rguenther@suse.de>
36455 PR tree-optimization/79256
36456 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36457 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36459 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36461 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36464 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36465 ("*r<noxa>sbg_<mode>_sll_bitmask")
36466 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36467 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36468 Use contiguous_bitmask_nowrap_operand.
36470 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36473 * config/rs6000/altivec.h (vec_xl): Revise #define.
36474 (vec_xst): Likewise.
36476 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36478 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36480 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36482 PR rtl-optimization/79194
36483 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36484 traps before call to bypass_conditional_jumps.
36486 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36488 PR tree-optimization/71374
36489 * lra-constraints.c (check_conflict_input_operands): New.
36490 (match_reload): Use it.
36492 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36495 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36496 account to calculate conflict_set.
36498 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36500 PR rtl-optimization/78559
36501 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36502 other_insn in combine.
36504 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36506 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36507 uint16_type_node for BT_UINT16.
36509 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36511 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36512 "RTL Tests" to menu.
36513 (GIMPLE Tests): New node.
36514 (RTL Tests): New node.
36516 2017-01-27 Richard Biener <rguenther@suse.de>
36518 PR tree-optimization/79245
36519 * tree-loop-distribution.c (distribute_loop): Apply cost
36520 modeling also to detected patterns.
36522 2017-01-27 Richard Biener <rguenther@suse.de>
36524 PR tree-optimization/71433
36525 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36526 (compare_assert_loc): New function.
36527 (process_assert_insertions): Sort and optimize assert locations
36528 to remove duplicates and push down identical assertions on
36529 edges to their destination block.
36531 2017-01-27 Richard Biener <rguenther@suse.de>
36533 PR tree-optimization/79244
36534 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36535 out SSA names even if abnormal.
36537 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36539 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36540 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36541 instead of MPFR_RNDN.
36543 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36546 * arm.c (arm_option_override): Don't call build_target_option_node
36547 until after doing all option overrides.
36548 (arm_valid_target_attribute_tree): Likewise.
36550 2017-01-27 Martin Liska <mliska@suse.cz>
36552 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36553 for {cd}tors and C++ {cd}tors.
36555 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36557 * config/s390/s390.md ("*setmem_long_and")
36558 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36560 2017-01-26 Martin Sebor <msebor@redhat.com>
36562 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36565 2017-01-26 Martin Sebor <msebor@redhat.com>
36567 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36568 HAVE_DFmode before using XFmode or DFmode.
36569 (parse_directive): Avoid using the z length modifier to avoid
36570 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36572 PR middle-end/78703
36573 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36574 to accept adjustment as an array.
36575 (get_int_range): New function.
36576 (struct directive): Make width and prec arrays.
36577 (directive::set_width, directive::set_precision): Call get_int_range.
36578 (format_integer, format_floating): Handle width and precision ranges.
36579 (format_string, parse_directive): Same.
36581 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36584 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36585 comdat_type_p set, just clone them, but keep the children in the
36589 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36590 which have direct callers with -fvar-tracking-assignments enabled
36592 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36593 inside of type units.
36595 2017-01-26 Martin Sebor <msebor@redhat.com>
36597 PR middle-end/78703
36598 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36600 (struct format_result): Replace number_chars, number_chars_min,
36601 and number_chars_max with a single member of struct result_range.
36603 (format_result::operator+=): Adjust.
36604 (struct fmtresult): Remove bounded. Handle likely and unlikely
36606 (fmtresult::adjust_for_width_or_precision): New function.
36607 (fmtresult:type_max_digits): New function.
36608 (bytes_remaining): Handle likely and unlikely counters.
36609 (min_bytes_remaining): Remove.
36610 (format_percent): Simplify.
36611 (format_integer, format_floating): Set likely and unlikely counters.
36612 (get_string_length, format_character, format_string): Same.
36613 (format_plain, should_warn_p): New function.
36614 (maybe_warn): Call should_warn_p. Update diagnostic messages
36615 and handle those for all directives, including plain strings.
36616 (format_directive): Handle likely and unlikely counters.
36617 Remove unnecessary quoting from diagnostics. Add an informational
36619 (add_bytes): Remove.
36620 (pass_sprintf_length::compute_format_length): Simplify.
36621 (try_substitute_return_value): Handle likely and unlikely counters.
36623 2017-01-26 Carl Love <cel@us.ibm.com>
36625 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36626 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36628 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36631 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36632 endianess for subregs into account.
36633 * lra-constraints.c (lra_constraints): Do risky transformations
36634 always on the first iteration.
36635 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36636 last_call_used_reg_set.
36637 (process_bb_lives): Define and use last_call_used_reg_set.
36638 * lra.c (lra): Always continue after lra_constraints on the first
36641 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36643 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36645 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36647 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36649 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36650 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36651 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36652 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36653 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36654 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36655 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36656 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36657 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36659 2017-01-26 Marek Polacek <polacek@redhat.com>
36662 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36663 for the third operand.
36665 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36667 PR middle-end/79236
36668 * omp-low.c (struct omp_context): Add simt_stmt field.
36669 (scan_omp_for): Return omp_context *.
36670 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36671 context to the _simt_ SIMD stmt.
36672 (lower_omp_for): For combined SIMD with sibling _simt_
36673 SIMD, make sure to use the same decls in _looptemp_
36674 clauses as in the sibling.
36676 2017-01-26 David Sherwood <david.sherwood@arm.com>
36678 PR middle-end/79212
36679 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36682 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36685 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36686 emit fld b; fld a; if possible.
36688 * brig-builtins.def: Update copyright years.
36689 * config/arm/arm_acle_builtins.def: Update copyright years.
36691 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36694 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36695 constraint instead of o for the stxsd instruction.
36697 2017-01-25 Carl Love <cel@us.ibm.com>
36699 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36700 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36702 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36704 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36706 2017-01-25 Richard Biener <rguenther@suse.de>
36708 PR tree-optimization/69264
36709 * target.def (vector_alignment_reachable): Improve documentation.
36710 * doc/tm.texi: Regenerate.
36711 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36713 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36714 earlier changes with respect to TYPE_USER_ALIGN.
36715 (vector_alignment_reachable_p): Likewise. Improve dumping.
36717 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36720 * config/arm/arm.md (xordi3): Force constant operand into a register
36723 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36725 * doc/invoke.texi (-fstore-merging): Correct default optimization
36726 levels at which it is enabled.
36727 (-O): Move -fstore-merging from list to...
36730 2017-01-25 Richard Biener <rguenther@suse.de>
36733 * omp-expand.c: Include debug.h.
36734 (expand_omp_taskreg): Make sure to generate early debug before
36735 outlining anything from a function.
36736 (expand_omp_target): Likewise.
36737 (grid_expand_target_grid_body): Likewise.
36739 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36742 * asan.c (get_translation_unit_decl): New function.
36743 (asan_add_global): Extract modules file name from globals
36744 TRANSLATION_UNIT_DECL name.
36746 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36749 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36750 for long calls with APCS frame and VFP.
36752 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36754 * cfg.c (original_copy_tables_initialized_p): New function.
36755 * cfg.h (original_copy_tables_initialized_p): New decl.
36756 * cfgrtl.c (relink_block_chain): Guard the call to
36757 free_original_copy_tables with a call to
36758 original_copy_tables_initialized_p.
36759 * cgraph.h (symtab_node::native_rtl_p): New decl.
36760 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36761 (symtab_node::needed_p): Don't assert for early assembly output
36762 for __RTL functions.
36763 (cgraph_node::finalize_function): Set "force_output" for __RTL
36765 (cgraph_node::analyze): Bail out early for __RTL functions.
36766 (analyze_functions): Update assertion to support __RTL functions.
36767 (cgraph_node::expand): Bail out early for __RTL functions.
36768 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36770 * function.h (struct function): Update comment for field
36772 * gimple-expr.c: Include "tree-pass.h".
36773 (gimple_has_body_p): Return false for __RTL functions.
36774 * Makefile.in (OBJS): Add run-rtl-passes.o.
36775 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36777 (gcc::pass_manager::get_clean_slate): New accessor.
36778 * passes.c: Include "insn-addr.h".
36779 (should_skip_pass_p): Add logging. Update logic for running
36780 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36781 property-provider override so it is only done for gimple passes.
36783 (skip_pass): New function.
36784 (execute_one_pass): Call skip_pass when skipping passes.
36785 * read-md.c (md_reader::read_char): Support filtering
36786 the input to a subset of line numbers.
36787 (md_reader::md_reader): Initialize fields
36788 m_first_line and m_last_line.
36789 (md_reader::read_file_fragment): New function.
36790 * read-md.h (md_reader::read_file_fragment): New decl.
36791 (md_reader::m_first_line): New field.
36792 (md_reader::m_last_line): New field.
36793 * read-rtl-function.c (function_reader::create_function): Only
36794 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36795 curr_properties. Set DECL_INITIAL to a dummy block.
36796 (read_rtl_function_body_from_file_range): New function.
36797 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36799 * run-rtl-passes.c: New file.
36800 * run-rtl-passes.h: New file.
36802 2017-01-24 Jeff Law <law@redhat.com>
36804 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36807 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36809 PR tree-optimization/79159
36810 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36811 (record_nonwrapping_iv): Improve boundary using above function if no
36812 value range information.
36814 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36815 Martin Jambor <mjambor@suse.cz>
36817 * brig-builtins.def: New file.
36818 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36819 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36820 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36821 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36822 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36823 * builtin-types.def (BT_INT8): New.
36824 (BT_INT16): Likewise.
36825 (BT_UINT8): Likewise.
36826 (BT_UINT16): Likewise.
36827 (BT_FN_ULONG): Likewise.
36828 (BT_FN_UINT_INT): Likewise.
36829 (BT_FN_UINT_ULONG): Likewise.
36830 (BT_FN_UINT_LONG): Likewise.
36831 (BT_FN_UINT_PTR): Likewise.
36832 (BT_FN_ULONG_PTR): Likewise.
36833 (BT_FN_INT8_FLOAT): Likewise.
36834 (BT_FN_INT16_FLOAT): Likewise.
36835 (BT_FN_UINT32_FLOAT): Likewise.
36836 (BT_FN_UINT16_FLOAT): Likewise.
36837 (BT_FN_UINT8_FLOAT): Likewise.
36838 (BT_FN_UINT64_FLOAT): Likewise.
36839 (BT_FN_UINT16_UINT32): Likewise.
36840 (BT_FN_UINT32_UINT16): Likewise.
36841 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36842 (BT_FN_INT_PTR_INT): Likewise.
36843 (BT_FN_UINT_PTR_UINT): Likewise.
36844 (BT_FN_LONG_PTR_LONG): Likewise.
36845 (BT_FN_ULONG_PTR_ULONG): Likewise.
36846 (BT_FN_VOID_UINT64_UINT64): Likewise.
36847 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36848 (BT_FN_INT8_INT8_INT8): Likewise.
36849 (BT_FN_INT16_INT16_INT16): Likewise.
36850 (BT_FN_INT_INT_INT): Likewise.
36851 (BT_FN_UINT_FLOAT_UINT): Likewise.
36852 (BT_FN_FLOAT_UINT_UINT): Likewise.
36853 (BT_FN_ULONG_UINT_UINT): Likewise.
36854 (BT_FN_ULONG_UINT_PTR): Likewise.
36855 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36856 (BT_FN_UINT_UINT_UINT): Likewise.
36857 (BT_FN_VOID_UINT_PTR): Likewise.
36858 (BT_FN_UINT_UINT_PTR: Likewise.
36859 (BT_FN_UINT32_UINT64_PTR): Likewise.
36860 (BT_FN_INT_INT_UINT_UINT): Likewise.
36861 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36862 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36863 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36864 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36865 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36866 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36867 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36868 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36869 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36870 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36871 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36872 * doc/frontends.texi: List BRIG FE.
36873 * doc/install.texi (Testing): Add BRIG tesring requirements.
36874 * doc/invoke.texi (Overall Options): Mention BRIG.
36875 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36877 2017-01-24 Richard Biener <rguenther@suse.de>
36879 PR translation/79208
36880 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36882 2017-01-24 Martin Jambor <mjambor@suse.cz>
36885 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36886 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36887 and known_contexts.
36889 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36891 PR middle-end/79123
36892 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36893 casts from signed to unsigned really don't have a range.
36895 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36897 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36898 GMP_RNDx for compatiblity.
36900 2017-01-24 Martin Liska <mliska@suse.cz>
36903 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36904 that would prevent us to call alloca with -1 as argument.
36906 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36908 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36909 Avoid -Wformat-security warning.
36911 2017-01-23 Andrew Pinski <apinski@cavium.com>
36913 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36916 2017-01-23 Martin Sebor <msebor@redhat.com>
36918 PR middle-end/78703
36919 * gimple-ssa-sprintf.c (warn_level): New global.
36920 (format_integer): Use it here and throughout the rest of the file.
36921 Use the same switch to compute sign as base.
36922 (maybe_warn): New function.
36923 (format_directive): Factor out warnings into maybe_warn.
36924 Add debugging output. Use warn_level.
36925 (add_bytes): Use warn_level.
36926 (pass_sprintf_length::compute_format_length): Add debugging output.
36927 (try_substitute_return_value): Same.
36928 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36930 PR middle-end/78703
36931 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36932 (struct fmtresult, format_integer, format_floating): Adjust.
36933 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36934 (get_string_length, format_string,format_directive): Same.
36935 (pass_sprintf_length::compute_format_length): Same.
36936 (try_substitute_return_value): Simplify slightly.
36938 PR middle-end/78703
36939 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36940 (fmtresult::operator+=): Outlined.
36941 (struct fmtresult): Add ctors.
36942 (struct conversion_spec): Rename...
36943 (struct directive): ...to this. Add and remove data members.
36944 (directive::set_width, directive::set_precision): New functions.
36945 (format_percent): Use fmtresult ctor.
36946 (get_width_and_precision): Remove.
36947 (format_integer): Make naming changes. Avoid computing width and
36949 (format_floating): Same. Adjust indentation.
36950 (format_character, format_none): New functions.
36951 (format_string): Moved character handling to format_character.
36952 (format_directive): Remove arguments, change return type.
36953 (parse_directive): New function.
36954 (pass_sprintf_length::compute_format_length): Move directive
36955 parsing to parse_directive.
36957 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36959 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36960 (assign_assembler_name_if_needed): ... this.
36961 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36962 (assign_assembler_name_if_needed): ... this.
36963 (free_lang_data_in_cgraph): Adjust callers.
36964 * cgraphunit.c (cgraph_node::analyze): Likewise.
36965 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36968 2017-01-23 Richard Biener <rguenther@suse.de>
36970 PR tree-optimization/79088
36971 PR tree-optimization/79188
36972 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36973 resetting loop bounds after last path deletion. Reset loop
36974 bounds of the target loop, make code match the comments.
36975 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36976 Make sure loops need no fixups.
36978 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36980 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36981 exponent support with double type for first argument.
36982 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36983 type returned by __builtin_vec_extract_sig,
36984 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36985 functions from "vector int" to "vector unsigned int" or from
36986 "vector long long int" to "vector unsigned long long int".
36987 Changed type returned by __builtin_vec_extract_exp,
36988 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36989 functions from "vector int" to "vector unsigned int" or from
36990 "vector long long int" to "vector unsigned long long int".
36991 Changed return type of __builtin_vec_test_data_class,
36992 __builtin_vec_test_data_class_sp, and
36993 __builtin_vec_test_data_class_dp from "vector int" to
36994 "vector bool int" or from "vector long long int" to "vector bool
36995 long long int" and changed second argument type from "unsigned
36996 int" to "int". Added new overloaded function forms "vector float
36997 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36998 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36999 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37000 double, vector unsigned long long int)" and "vector double
37001 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37002 long int)". Changed return type of
37003 __builtin_scalar_test_data_class and
37004 __builtin_scalar_test_data_class_sp and
37005 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37006 int" and changed second argument from "unsigned int" to "int".
37007 Changed type returned by __builtin_scalar_test_neg,
37008 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37009 from "int" to "bool int". Added new overloaded function form
37010 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37011 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37012 exponent double-precision with floating point first argument.
37013 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37014 documentation of scalar_test_data_class, scalar_test_neg,
37015 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37016 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37017 vec_test_data_class built-in functions to reflect refinements in
37018 their type signatures.
37020 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37022 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37024 (aarch64_elf_asm_destructor): Likewise.
37026 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37028 PR rtl-optimization/78634
37029 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37030 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37031 * ifcvt.c (noce_try_cmove): Add missing cost check.
37033 PR rtl-optimization/71724
37034 * combine.c (if_then_else_cond): Look for situations where it is
37035 beneficial to undo the work of one of the recursive calls.
37037 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37039 PR tree-optimization/70754
37040 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37041 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37042 combined stmt before it if not NULL.
37043 (combine_chains): Process refs reversely and compute dominance point
37046 2017-01-23 Martin Liska <mliska@suse.cz>
37048 PR tree-optimization/79196
37049 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37050 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37052 (strlen_optimize_stmt): Call the renamed function.
37054 2017-01-23 Michael Matz <matz@suse.de>
37056 PR tree-optimization/78384
37057 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37059 2017-01-23 Richard Biener <rguenther@suse.de>
37061 PR tree-optimization/79186
37062 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37063 both incoming edges before moving an assert.
37065 2017-01-23 Martin Jambor <mjambor@suse.cz>
37067 * ipa-prop.c (load_from_param_1): Removed.
37068 (load_from_unmodified_param): Bits from load_from_param_1 put back
37070 (load_from_param): Removed.
37071 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37072 with stmt. Reverted back to use of load_from_unmodified_param.
37074 2017-01-23 Martin Jambor <mjambor@suse.cz>
37077 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37078 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37079 field a pointer to garbage collected vector, mark lattices and
37080 ipcp_orig_node with GTY((skip)).
37081 (ipa_get_param_count): Adjust to descriptors being a pointer.
37082 (ipa_get_param): Likewise.
37083 (ipa_get_type): Likewise.
37084 (ipa_get_param_move_cost): Likewise.
37085 (ipa_set_param_used): Likewise.
37086 (ipa_get_controlled_uses): Likewise.
37087 (ipa_set_controlled_uses): Likewise.
37088 (ipa_is_param_used): Likewise.
37089 (ipa_node_params_t): Move into garbage collector. New methods insert
37091 (ipa_node_params_sum): Annotate wth GTY(()).
37092 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37094 (ipa_load_from_parm_agg): Adjust declaration.
37095 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37096 * ipa-profile.c (ipa_profile): Likewise.
37097 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37098 (ipa_populate_param_decls): Make descriptors parameter garbage
37100 (ipa_dump_param): Adjust to descriptors being a pointer.
37101 (ipa_alloc_node_params): Likewise.
37102 (ipa_initialize_node_params): Likewise.
37103 (load_from_param_1): Make descriptors parameter garbage collected.
37104 (load_from_unmodified_param): Likewise.
37105 (load_from_param): Likewise.
37106 (ipa_load_from_parm_agg): Likewise.
37107 (ipa_node_params::~ipa_node_params): Removed.
37108 (ipa_free_all_node_params): Remove call to delete operator.
37109 (ipa_node_params_t::insert): New.
37110 (ipa_node_params_t::remove): Likewise.
37111 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37112 copy known_csts and known_contexts vectors.
37113 (ipa_read_node_info): Adjust to descriptors being a pointer.
37114 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37116 (ipcp_transform_function): Make descriptors variable garbage
37119 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37121 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37122 * config/i386/avx512dqintrin.h: Ditto.
37123 * config/i386/avx512fintrin.h: Ditto.
37124 * config/i386/i386.c: Handle new builtins.
37125 * config/i386/i386-builtin.def: Add new builtins.
37126 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37127 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37129 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37130 Martin Liska <mliska@suse.cz>
37132 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37133 * asan.c (asan_expand_poison_ifn): Support stores and use
37134 appropriate ASAN report function.
37135 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37136 * internal-fn.def (ASAN_POISON_USE): Declare.
37137 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37138 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37139 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37140 ASAN_POISON calls w/o LHS.
37141 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37142 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37143 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37144 * gimplify.c (asan_poison_variables): Add attribute
37145 use_after_scope_memory to variables that really needs to live
37147 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37148 having the attribute.
37150 2017-01-23 Martin Liska <mliska@suse.cz>
37152 * asan.c (create_asan_shadow_var): New function.
37153 (asan_expand_poison_ifn): Likewise.
37154 * asan.h (asan_expand_poison_ifn): New declaration.
37155 * internal-fn.c (expand_ASAN_POISON): Likewise.
37156 * internal-fn.def (ASAN_POISON): New builtin.
37157 * sanopt.c (pass_sanopt::execute): Expand
37158 asan_expand_poison_ifn.
37159 * tree-inline.c (copy_decl_for_dup_finish): Make function
37161 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37162 * tree-ssa.c (is_asan_mark_p): New function.
37163 (execute_update_addresses_taken): Rewrite local variables
37164 (identified just by use-after-scope as addressable) into SSA.
37166 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37168 * doc/install.texi (Specific): opensource.apple.com uses https
37169 now. Remove trailing slash.
37171 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37173 * README.Portability: Remove note on an Irix compatibility issue.
37175 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37177 * gcov.c (INCLUDE_ALGORITHM): Define.
37178 (INCLUDE_VECTOR): Define.
37179 No longer include <vector> and <algorithm> directly.
37181 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37183 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37185 * doc/invoke.texi (Code Gen Options): Ditto.
37187 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37190 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37192 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37194 rtl-optimization/79125
37195 * cprop.c (local_cprop_pass): Handle cases where we make an
37196 unconditional trap.
37198 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37202 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37203 read from, for big endian.
37205 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37207 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37208 register pauth builtins for LP64 only.
37210 2017-01-20 Marek Polacek <polacek@redhat.com>
37213 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37216 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37218 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37220 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37221 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37222 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37224 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37227 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37228 in big-endian mode when they are not a single duplicated value.
37230 2017-01-20 Richard Biener <rguenther@suse.de>
37232 * BASE-VER: Bump to 7.0.1.
37234 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37236 * omp-low.c (omplow_simd_context): New struct. Use it...
37237 (lower_rec_simd_input_clauses): ...here and...
37238 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37239 references to idx, lane, max_vf, is_simt.
37241 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37243 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37246 2017-01-20 Martin Jambor <mjambor@suse.cz>
37248 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37249 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37251 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37252 (GTFILES): Rename hsa.c to hsa-common.c.
37253 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37254 * hsa-dump.c: Likewise.
37255 * hsa-gen.c: Likewise.
37256 * hsa-regalloc.c: Likewise.
37257 * ipa-hsa.c: Likewise.
37258 * omp-expand.c: Likewise.
37259 * omp-low.c: Likewise.
37260 * toplev.c: Likewise.
37262 2017-01-20 Marek Polacek <polacek@redhat.com>
37265 * doc/invoke.texi: Document -Wduplicated-branches.
37266 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37267 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37268 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37269 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37270 return 0 only when not OEP_LEXICOGRAPHIC.
37271 (fold_build_cleanup_point_expr): Use the expression
37272 location when building CLEANUP_POINT_EXPR.
37273 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37274 * tree.c (add_expr): Handle error_mark_node.
37276 2017-01-20 Martin Liska <mliska@suse.cz>
37279 * tree-profile.c (init_ic_make_global_vars): Do not call
37281 (gimple_init_gcov_profiler): Likewise.
37283 2017-01-20 Martin Liska <mliska@suse.cz>
37286 * cgraph.h (maybe_create_reference): Remove argument and
37288 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37290 * ipa-cp.c (create_specialized_node): Likewise.
37291 * symtab.c (symtab_node::maybe_create_reference): Handle
37292 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37294 2017-01-20 Martin Liska <mliska@suse.cz>
37296 * read-rtl-function.c (function_reader::create_function): Use
37297 build_decl instread of build_decl_stat.
37299 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37301 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37302 * config/i386/avx512dqintrin.h: Ditto.
37303 * config/i386/avx512fintrin.h: Ditto.
37304 * config/i386/i386-builtin-types.def: Add new types.
37305 * config/i386/i386.c: Handle new types.
37306 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37307 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37308 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37309 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37310 (__builtin_ia32_kshiftridi): New.
37311 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37313 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37317 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37318 define to rs6000_init_stack_protect_guard.
37319 (rs6000_init_stack_protect_guard): New function.
37321 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37322 Yunqiang Su <yunqiang.su@imgtec.com>
37324 * config.gcc (supported_defaults): Add madd4.
37325 (with_madd4): Add validation.
37326 (all_defaults): Add madd4.
37327 * config/mips/mips.opt (mmadd4): New option.
37328 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37330 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37332 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37333 (ISA_HAS_FUSED_MADD4): Likewise.
37334 * doc/invoke.texi (-mmadd4): Document the new option.
37335 * doc/install.texi (--with-madd4): Document the new option.
37337 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37339 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37340 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37341 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37342 (aarch64_init_pauth_hint_builtins): New.
37343 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37344 (aarch64_expand_builtin): Expand new builtins.
37346 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37348 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37349 * combine-stack-adj.c (no_unhandled_cfa): Handle
37350 REG_CFA_TOGGLE_RA_MANGLE.
37351 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37352 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37353 info for return address signing.
37354 (aarch64_expand_epilogue): Likewise.
37356 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37358 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37359 * config/aarch64/aarch64-protos.h
37360 (aarch64_return_address_signing_enabled): New declaration.
37361 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37363 (aarch64_expand_prologue): Sign return address before it's pushed onto
37365 (aarch64_expand_epilogue): Authenticate return address fetched from
37367 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37368 (aarch64_attributes): New function attributes for "sign-return-address".
37369 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37370 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37371 ("*do_return"): Generate combined instructions according to key index.
37372 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37373 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37375 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37376 * config/aarch64/aarch64.opt (msign-return-address=): New.
37377 * doc/extend.texi (AArch64 Function Attributes): Documents
37378 "sign-return-address=".
37379 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37381 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37383 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37384 overall option summary.
37386 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37388 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37389 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37390 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37391 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37393 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37395 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37396 -mpower9-minmax by default for -mcpu=power9.
37397 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37398 128-bit floating point.
37400 2017-01-20 Alan Modra <amodra@gmail.com>
37402 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37403 optimizing for size.
37405 2017-01-20 Alan Modra <amodra@gmail.com>
37408 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37409 for strcmp and strncmp from corresponding builtin decl.
37411 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37413 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37414 instead of i386/rtems-64.h.
37415 * config/i386/rtems-64.h: Remove.
37417 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37421 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37423 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37425 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37427 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37428 Change int to HOST_WIDE_INT.
37429 * config/aarch64/aarch64-protos.h
37430 (aarch64_simd_gen_const_vector_dup): Likewise.
37431 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37433 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37435 * langhooks-def.h (lhd_type_for_size): New decl.
37436 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37437 * langhooks.c (lhd_type_for_size): New function, taken from
37440 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37442 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37443 define_bypass for CR latency.
37444 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37445 (power9-alu2): Add define_bypass for CR latency.
37447 (power9-mul): Update insn latency.
37448 (power9-mul-compare): Update insn latency, bypass latency and remove
37451 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37453 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37455 * config/aarch64/aarch64.md
37456 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37457 aarch64_nopcrelative_literal_loads.
37458 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37460 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37462 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37463 TARGET_LOONGSON_3A.
37464 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37466 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37469 * config.gcc (supported_defaults): Add lxc1-sxc1.
37470 (with_lxc1_sxc1): Add validation.
37471 (all_defaults): Add lxc1-sxc1.
37472 * config/mips/mips.opt (mlxc1-sxc1): New option.
37473 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37475 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37476 __mips_no_lxc1_sxc1.
37477 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37478 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37479 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37481 2017-01-19 Richard Biener <rguenther@suse.de>
37483 PR tree-optimization/72488
37484 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37485 sure to restore SSA info.
37486 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37488 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37490 PR rtl-optimization/79121
37491 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37492 of the inner type when shifting an extended value.
37494 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37497 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37498 interposable aliases.
37500 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37503 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37504 Use the evmergelohi instruction.
37505 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37506 (mov_sitf_e500_subreg8_2_be): Likewise.
37507 (mov_sitf_e500_subreg12_2_le): Likewise.
37508 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37509 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37510 (mov_sitf_e500_subreg8_2_le): Likewise.
37511 (mov_sitf_e500_subreg12_2_be): Likewise.
37513 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37515 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37516 attribute from vecsimple to vecperm.
37517 (altivec_vbpermq2): Likewise.
37519 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37522 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37524 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37525 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37526 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37527 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37528 case where N arg is SIZE_MAX.
37529 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37530 (cmpstrsi): Add pattern.
37532 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37534 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37535 __builtin_vec_revb builtins.
37536 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37537 built-in functions to support generation of the ISA 3.0 XXBR<x>
37538 vector byte reverse instructions.
37539 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37540 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37541 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37542 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37543 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37544 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37545 (P9V_BUILTIN_VEC_REVB): Likewise.
37546 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37547 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37548 (p9_xxbrq_v16qi): Likewise.
37549 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37550 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37551 (p9_xxbrh_v8hi): Likewise.
37552 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37553 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37554 vec_revb built-in functions.
37556 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37558 PR rtl-optimization/78952
37559 * config/i386/i386.md (any_extract): New code iterator.
37560 (*insvqi_2): Use any_extract for source operand.
37561 (*insvqi_3): Use any_shiftrt for source operand.
37563 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37565 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37567 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37569 2017-01-18 Matthias Klose <doko@ubuntu.com>
37571 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37573 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37575 * config/rs6000/altivec.h (vec_bperm): Change #define.
37576 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37577 (altivec_vbpermq2): New define_insn.
37578 (altivec_vbpermd): Likewise.
37579 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37580 function interface.
37581 (VBPERMD): Likewise.
37582 (VBPERM): New polymorphic function interface.
37583 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37584 Add entries for P9V_BUILTIN_VEC_VBPERM.
37585 * doc/extend.texi: Add interfaces for vec_bperm.
37587 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37589 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37590 first letter of error messages.
37591 (s390_resolve_overloaded_builtin): Likewise.
37592 * config/s390/s390.c (s390_expand_builtin): Likewise.
37593 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37594 (s390_valid_target_attribute_inner_p): Likewise.
37595 * config/s390/s390.md ("tabort"): Likewise.
37597 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37599 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37600 (ISA_AVOID_DIV_HILO): New macro.
37601 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37602 (ISA_HAS_DDIV): Likewise.
37604 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37606 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37608 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37610 * doc/invoke.texi (fabi-version): Spelling fix.
37612 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37615 * doc/invoke.texi (fabi-version): Mention mangling fix for
37618 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37621 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37623 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37626 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37627 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37629 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37631 (stack_protect_test): Ditto.
37632 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37633 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37635 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37636 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37637 -mstack-protector-guard-offset=.
37638 (RS/6000 and PowerPC Options): Ditto.
37640 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37642 * config/i386/i386.h (MASK_CLASS_P): New define.
37643 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37644 there are no registers from different register sets also when
37645 mask registers are used. Update function comment.
37646 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37647 to (*k/*r) and (*k/*km) alternatives.
37649 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37651 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37652 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37653 (EH_RETURN_HANDLER_RTX): New define.
37654 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37655 Force frame pointer in EH return functions.
37656 (aarch64_expand_epilogue): Add barrier for eh_return.
37657 (aarch64_final_eh_return_addr): Remove.
37658 (aarch64_eh_return_handler_rtx): New function.
37659 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37661 (aarch64_eh_return_handler_rtx): New prototype.
37663 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37665 * config/rs6000/altivec.h (vec_rlmi): New #define.
37666 (vec_vrlnm): Likewise.
37667 (vec_rlnm): Likewise.
37668 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37669 (UNSPEC_VRLNM): Likewise.
37670 (VIlong): New mode iterator.
37671 (altivec_vrl<VI_char>mi): New define_insn.
37672 (altivec_vrl<VI_char>nm): Likewise.
37673 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37675 (VRLDNM): Likewise.
37676 (RLNM): New polymorphic function entry.
37677 (VRLWMI): New monomorphic function entry.
37678 (VRLDMI): Likewise.
37679 (RLMI): New polymorphic function entry.
37680 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37681 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37682 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37685 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37688 * dwarf2out.c (field_byte_offset): Restore the
37689 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37690 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37691 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37694 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37697 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37699 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37702 * dwarf2out.c (add_data_member_location_attribute): For constant
37703 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37704 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37705 DW_AT_byte_size attributes.
37707 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37709 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37710 after forcing to constant memory when the code model is medium.
37712 2017-01-17 Julia Koval <julia.koval@intel.com>
37715 * config/i386/avx512fintrin.h
37716 (_mm512_i32gather_ps): Change __addr type to void const*.
37717 (_mm512_mask_i32gather_ps): Ditto.
37718 (_mm512_i32gather_pd): Ditto.
37719 (_mm512_mask_i32gather_pd): Ditto.
37720 (_mm512_i64gather_ps): Ditto.
37721 (_mm512_mask_i64gather_ps): Ditto.
37722 (_mm512_i64gather_pd): Ditto.
37723 (_mm512_mask_i64gather_pd): Ditto.
37724 (_mm512_i32gather_epi32): Ditto.
37725 (_mm512_mask_i32gather_epi32): Ditto.
37726 (_mm512_i32gather_epi64): Ditto.
37727 (_mm512_mask_i32gather_epi64): Ditto.
37728 (_mm512_i64gather_epi32): Ditto.
37729 (_mm512_mask_i64gather_epi32): Ditto.
37730 (_mm512_i64gather_epi64): Ditto.
37731 (_mm512_mask_i64gather_epi64): Ditto.
37732 (_mm512_i32scatter_ps): Change __addr type to void*.
37733 (_mm512_mask_i32scatter_ps): Ditto.
37734 (_mm512_i32scatter_pd): Ditto.
37735 (_mm512_mask_i32scatter_pd): Ditto.
37736 (_mm512_i64scatter_ps): Ditto.
37737 (_mm512_mask_i64scatter_ps): Ditto.
37738 (_mm512_i64scatter_pd): Ditto.
37739 (_mm512_mask_i64scatter_pd): Ditto.
37740 (_mm512_i32scatter_epi32): Ditto.
37741 (_mm512_mask_i32scatter_epi32): Ditto.
37742 (_mm512_i32scatter_epi64): Ditto.
37743 (_mm512_mask_i32scatter_epi64): Ditto.
37744 (_mm512_i64scatter_epi32): Ditto.
37745 (_mm512_mask_i64scatter_epi32): Ditto.
37746 (_mm512_i64scatter_epi64): Ditto.
37747 (_mm512_mask_i64scatter_epi64): Ditto.
37748 * config/i386/avx512pfintrin.h
37749 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37750 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37751 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37752 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37753 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37754 (_mm512_prefetch_i32scatter_ps): Ditto.
37755 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37756 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37757 (_mm512_prefetch_i64scatter_pd): Ditto.
37758 (_mm512_prefetch_i64scatter_ps): Ditto.
37759 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37760 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37761 * config/i386/avx512vlintrin.h
37762 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37763 (_mm_mmask_i32gather_ps): Ditto.
37764 (_mm256_mmask_i32gather_pd): Ditto.
37765 (_mm_mmask_i32gather_pd): Ditto.
37766 (_mm256_mmask_i64gather_ps): Ditto.
37767 (_mm_mmask_i64gather_ps): Ditto.
37768 (_mm256_mmask_i64gather_pd): Ditto.
37769 (_mm_mmask_i64gather_pd): Ditto.
37770 (_mm256_mmask_i32gather_epi32): Ditto.
37771 (_mm_mmask_i32gather_epi32): Ditto.
37772 (_mm256_mmask_i32gather_epi64): Ditto.
37773 (_mm_mmask_i32gather_epi64): Ditto.
37774 (_mm256_mmask_i64gather_epi32): Ditto.
37775 (_mm_mmask_i64gather_epi32): Ditto.
37776 (_mm256_mmask_i64gather_epi64): Ditto.
37777 (_mm_mmask_i64gather_epi64): Ditto.
37778 (_mm256_i32scatter_ps): Change __addr type to void*.
37779 (_mm256_mask_i32scatter_ps): Ditto.
37780 (_mm_i32scatter_ps): Ditto.
37781 (_mm_mask_i32scatter_ps): Ditto.
37782 (_mm256_i32scatter_pd): Ditto.
37783 (_mm256_mask_i32scatter_pd): Ditto.
37784 (_mm_i32scatter_pd): Ditto.
37785 (_mm_mask_i32scatter_pd): Ditto.
37786 (_mm256_i64scatter_ps): Ditto.
37787 (_mm256_mask_i64scatter_ps): Ditto.
37788 (_mm_i64scatter_ps): Ditto.
37789 (_mm_mask_i64scatter_ps): Ditto.
37790 (_mm256_i64scatter_pd): Ditto.
37791 (_mm256_mask_i64scatter_pd): Ditto.
37792 (_mm_i64scatter_pd): Ditto.
37793 (_mm_mask_i64scatter_pd): Ditto.
37794 (_mm256_i32scatter_epi32): Ditto.
37795 (_mm256_mask_i32scatter_epi32): Ditto.
37796 (_mm_i32scatter_epi32): Ditto.
37797 (_mm_mask_i32scatter_epi32): Ditto.
37798 (_mm256_i32scatter_epi64): Ditto.
37799 (_mm256_mask_i32scatter_epi64): Ditto.
37800 (_mm_i32scatter_epi64): Ditto.
37801 (_mm_mask_i32scatter_epi64): Ditto.
37802 (_mm256_i64scatter_epi32): Ditto.
37803 (_mm256_mask_i64scatter_epi32): Ditto.
37804 (_mm_i64scatter_epi32): Ditto.
37805 (_mm_mask_i64scatter_epi32): Ditto.
37806 (_mm256_i64scatter_epi64): Ditto.
37807 (_mm256_mask_i64scatter_epi64): Ditto.
37808 (_mm_i64scatter_epi64): Ditto.
37809 (_mm_mask_i64scatter_epi64): Ditto.
37810 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37811 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37812 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37813 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37814 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37815 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37816 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37817 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37818 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37819 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37820 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37821 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37822 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37823 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37824 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37825 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37826 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37827 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37828 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37829 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37830 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37831 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37832 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37833 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37834 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37835 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37836 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37837 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37838 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37839 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37840 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37841 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37842 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37843 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37844 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37845 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37846 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37847 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37848 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37849 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37850 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37851 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37852 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37853 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37854 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37855 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37856 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37857 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37858 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37859 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37860 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37861 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37862 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37863 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37864 definitions accordingly.
37866 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37867 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37870 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37873 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37876 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37877 hard regs for inner regno.
37879 2017-01-17 Martin Liska <mliska@suse.cz>
37882 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37883 assumption and add comment.
37885 2017-01-17 Nathan Sidwell <nathan@acm.org>
37887 * ipa-visibility.c (localize_node): New function, broken out of ...
37888 (function_and_variable_visibility): ... here. Call it.
37890 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37892 PR middle-end/77445
37893 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37894 correctly set frequency of oudgoing edge.
37895 (duplicate_thread_path): Fix profile updating.
37897 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37900 * configure.ac: Add GCC_BASE_VER.
37901 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37902 version from BASE-VER file.
37903 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37904 (gcc.o): Depend on $(BASEVER).
37905 * common.opt (dumpfullversion): New option.
37906 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37907 * doc/invoke.texi: Document -dumpfullversion.
37908 * doc/install.texi: Document --with-gcc-major-version-only.
37909 * configure: Regenerated.
37911 2017-01-17 Richard Biener <rguenther@suse.de>
37913 PR tree-optimization/71433
37914 * tree-vrp.c (register_new_assert_for): Merge same asserts
37915 on all incoming edges.
37916 (process_assert_insertions_for): Handle insertions at the
37919 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37921 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37922 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37924 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37927 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37930 2017-01-17 Alan Modra <amodra@gmail.com>
37933 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37934 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37935 symbolic stack limit when pic.
37937 2017-01-16 Martin Sebor <msebor@redhat.com>
37939 PR tree-optimization/78608
37940 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37942 2017-01-16 Jeff Law <law@redhat.com>
37945 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37946 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37947 for several include directories that may be relative to sysroot.
37948 * config/i386/x-mingw32 (gplus_includedir): Define.
37949 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37950 (native_system_includedir): Likewise.
37951 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37952 override if TARGET_SYSTEM_ROOT is defined.
37953 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37955 PR tree-optimization/79090
37956 PR tree-optimization/33562
37957 PR tree-optimization/61912
37958 PR tree-optimization/77485
37959 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37960 and computed trims into the dump file.
37962 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37964 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37966 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37969 * gimplify.c (gimplify_init_constructor): If want_value and
37970 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37974 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37975 sequence. Formatting fixes.
37976 (doloop_optimize): Formatting fixes.
37979 * gcc.c (debug_level_greater_than_spec_func): New function.
37980 (static_spec_functions): Add debug-level-gt spec function.
37981 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37983 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37984 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37985 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37986 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37987 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37988 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37990 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37992 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37993 QImode fixups to general and mask registers only.
37995 2017-01-16 Carl Love <cel@us.ibm.com>
37997 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37998 for built-in functions
37999 vector signed char vec_nabs (vector signed char)
38000 vector signed short vec_nabs (vector signed short)
38001 vector signed int vec_nabs (vector signed int)
38002 vector signed long long vec_nabs (vector signed long long)
38003 vector float vec_nabs (vector float)
38004 vector double vec_nabs (vector double)
38005 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38007 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38008 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38009 * doc/extend.texi: Update the documentation file for the new built-in
38012 2017-01-16 Martin Sebor <msebor@redhat.com>
38014 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38017 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38019 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38020 UNSPEC_VSX__XXSPLTD to require special splat handling.
38022 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38025 * system.h: Poison strndup.
38027 2017-01-16 Alan Modra <amodra@gmail.com>
38030 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38033 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38035 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38037 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38039 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38040 call recog here. Assert that INSN_CODE (insn) is non-negative.
38042 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38045 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38047 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38048 in the currently scheduled RTL fragment.
38050 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38052 PR rtl-optimization/78751
38053 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38056 2017-01-14 Jeff Law <law@redhat.com>
38058 PR tree-optimization/79090
38059 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38060 variable length stores.
38061 (compute_trims): Delete dead assignment to *trim_tail.
38062 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38065 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38067 PR rtl-optimization/78626
38068 PR rtl-optimization/78727
38069 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38070 of a block, and split such blocks after everything else is finished.
38072 2017-01-14 Alan Modra <amodra@gmail.com>
38075 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38076 target legitimate_combined_insn.
38077 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38078 (rs6000_legitimate_combined_insn): New function.
38079 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38081 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38082 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38083 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38085 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38087 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38089 2017-01-13 Jeff Law <law@redhat.com>
38091 PR tree-optimization/33562
38092 PR tree-optimization/61912
38093 PR tree-optimization/77485
38094 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38096 (delete_dead_assignment): Likewise.
38097 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38098 statement to delete_dead_call and delete_dead_assignment.
38100 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38103 * substring-locations.c (format_warning_va): Strengthen case 1 so
38104 that both endpoints of the substring must be within the format
38105 range for just the substring to be printed.
38107 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38109 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38110 * config/i386/i386.c (ix86_target_string): Add missing options
38111 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38112 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38113 flags_other and ix86_target_other to flags2_other. Display unknown
38115 (ix86_valid_target_attribute_inner_p): Add missing options and
38116 reorder options by implied ISAs, as in ix86_target_string.
38118 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38120 * hash-table.h (hash_table::too_empty_p): New function.
38121 (hash_table::expand): Use it.
38122 (hash_table::traverse): Likewise.
38123 (hash_table::empty_slot): Use sizeof (value_type) instead of
38124 sizeof (PTR) to convert bytes to elements. Shrink the table
38125 if the current size is excessive for the current number of
38128 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38130 * ira-costs.c (record_reg_classes): Break from the inner loop
38131 early once alt_fail is known to be true. Update outer loop
38132 handling accordingly.
38134 2017-01-13 Jeff Law <law@redhat.com>
38136 * tree-ssa-dse.c (decrement_count): New function.
38137 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38138 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38139 when we know the partially dead statement is a mem* function.
38141 PR tree-optimization/61912
38142 PR tree-optimization/77485
38143 * tree-ssa-dse.c: Include expr.h.
38144 (maybe_trim_constructor_store): New function.
38145 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38147 PR tree-optimization/33562
38148 PR tree-optimization/61912
38149 PR tree-optimization/77485
38150 * doc/invoke.texi: Document new dse-max-object-size param.
38151 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38152 * tree-ssa-dse.c: Include params.h.
38153 (dse_store_status): New enum.
38154 (initialize_ao_ref_for_dse): New, partially extracted from
38156 (valid_ao_ref_for_dse, normalize_ref): New.
38157 (setup_live_bytes_from_ref, compute_trims): Likewise.
38158 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38159 (maybe_trim_partially_dead_store): Likewise.
38160 (maybe_trim_complex_store): Likewise.
38161 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38162 Track what bytes live from the original store. Return tri-state
38163 for dead, partially dead or live.
38164 (dse_dom_walker): Add constructor, destructor and new private members.
38165 (delete_dead_call, delete_dead_assignment): New extracted from
38167 (dse_optimize_stmt): Make a member of dse_dom_walker.
38168 Use initialize_ao_ref_for_dse.
38170 PR tree-optimization/33562
38171 PR tree-optimization/61912
38172 PR tree-optimization/77485
38173 * sbitmap.h (bitmap_count_bits): Prototype.
38174 (bitmap_clear_range, bitmap_set_range): Likewise.
38175 * sbitmap.c (bitmap_clear_range): New function.
38176 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38178 2017-01-13 Martin Liska <mliska@suse.cz>
38181 * function.c (set_cfun): Add new argument force.
38182 * function.h (set_cfun): Likewise.
38183 * ipa-inline-transform.c (inline_call): Use the function when
38184 strict alising from is dropped for function we inline to.
38186 2017-01-13 Richard Biener <rguenther@suse.de>
38188 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38189 for dumping GIMPLE INTEGER_CSTs.
38191 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38193 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38194 to 201112L since C++17.
38196 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38199 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38200 if -fsanitize=kernel-address is present.
38202 2017-01-13 Richard Biener <rguenther@suse.de>
38204 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38205 as _Literal ( type ) number in case usual suffixes do not
38206 preserve all information.
38208 2017-01-13 Richard Biener <rguenther@suse.de>
38210 PR tree-optimization/77283
38211 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38212 and ssa-iterators.h.
38213 (is_feasible_trace): Implement a cost model based on joiner
38216 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38219 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38220 char or short to __float128/_Float128 directly.
38222 2017-01-12 Martin Sebor <msebor@redhat.com>
38224 to -Wformat-overflow.
38225 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38226 (min_bytes_remaining): Same.
38227 (get_string_length): Same.
38228 (format_string): Same.
38229 (format_directive): Same.
38231 (pass_sprintf_length::handle_gimple_call): Same.
38233 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38235 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38236 info.nowrite calls with no lhs that can't throw. Return bool
38237 whether gsi_remove has been called or not.
38238 (pass_sprintf_length::handle_gimple_call): Return bool whether
38239 try_substitute_return_value called gsi_remove. Formatting fix.
38240 (pass_sprintf_length::execute): Don't use gsi_remove if
38241 handle_gimple_call returned true.
38244 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38245 be removed due to side-effects, don't remove following barrier nor
38246 turn the successor edge into fallthru edge.
38248 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38251 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38252 element-reversing loads and stores as not swappable.
38254 2017-01-12 Nathan Sidwell <nathan@acm.org>
38255 Nicolai Stange <nicstange@gmail.com>
38257 * combine.c (try_combine): Don't ignore result of overlap checking
38258 loop. Combine overlap & asm check into single loop.
38260 2017-01-12 Richard Biener <rguenther@suse.de>
38262 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38263 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38265 2017-01-12 Richard Biener <rguenther@suse.de>
38267 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38268 and TS_TARGET_OPTION directly derive from TS_BASE.
38269 * tree-core.h (tree_optimization_option): Derive from tree_base.
38270 (tree_target_option): Likewise.
38272 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38274 * config/i386/i386.c (memory_address_length): Increase len
38275 only when rip_relative_addr_p returns false.
38277 2017-01-11 Julia Koval <julia.koval@intel.com>
38279 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38280 (OPTION_MASK_ISA_SGX_SET): New.
38281 (ix86_handle_option): Handle OPT_msgx.
38282 * config.gcc: Added sgxintrin.h.
38283 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38284 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38285 * config/i386/i386.c (ix86_target_string): Add -msgx.
38287 (ix86_option_override_internal): Handle new options.
38288 (ix86_valid_target_attribute_inner_p): Add sgx.
38289 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38290 * config/i386/i386.opt: Add msgx.
38291 * config/i386/sgxintrin.h: New file.
38292 * config/i386/x86intrin.h: Add sgxintrin.h.
38294 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38297 * fold-const.c (maybe_nonzero_address): Return 1 for function
38299 (tree_single_nonzero_warnv_p): Don't handle function local objects
38303 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38306 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38309 * opts.c: Include "spellcheck.h"
38310 (struct string_fragment): New struct.
38311 (struct edit_distance_traits<const string_fragment &>): New
38313 (get_closest_sanitizer_option): New function.
38314 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38316 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38318 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38320 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38321 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38322 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38323 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38324 for initial die_offset if dwarf_split_debug_info.
38325 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38326 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38328 (output_skeleton_debug_sections): Formatting fix. Use
38329 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38330 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38332 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38334 * config/arm/cortex-a53.md: Add bypasses for
38335 cortex_a53_r2f_cvt.
38336 (cortex_a53_r2f): Only use for transfers.
38337 (cortex_a53_f2r): Likewise.
38338 (cortex_a53_r2f_cvt): Add reservation for conversions.
38339 (cortex_a53_f2r_cvt): Likewise.
38341 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38343 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38344 to all inlined functions, change static to extern.
38346 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38349 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38351 (arm_assemble_integer): Likewise.
38353 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38355 * config.gcc: Use new awk script to check CPU, FPU and architecture
38356 parameters for --with-... options.
38357 * config/arm/parsecpu.awk: New file
38358 * config/arm/arm-cpus.in: New file.
38359 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38361 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38363 * config/arm/t-arm: Update dependency rules.
38364 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38365 of processing .def files.
38366 * config/arm/genopt.sh: Deleted.
38367 * config/arm/gentune.sh: Deleted.
38368 * config/arm/arm-cores.def: Deleted.
38369 * config/arm/arm-arches.def: Deleted.
38370 * config/arm/arm-fpus.def: Deleted.
38371 * config/arm/arm-tune.md: Regenerated.
38372 * config/arm/arm-tables.opt: Regenerated.
38373 * config/arm/arm-cpu.h: New generated file.
38374 * config/arm/arm-cpu-data.h: New generated file.
38375 * config/arm/arm-cpu-cdata.h: New generated file.
38377 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38380 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38382 (input_varpool_node): Unpack dynamically_initialized bit.
38384 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38386 PR rtl-optimization/79032
38387 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38388 the alignment of the adjusted memory reference against that of MODE,
38389 instead of the alignment of the original memory reference.
38391 2017-01-11 Martin Jambor <mjambor@suse.cz>
38393 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38395 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38396 decorated functions.
38398 2017-01-11 Richard Biener <rguenther@suse.de>
38400 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38401 set range/nonnull info for PHI results. Do not set it on
38402 stmts marked for removal.
38404 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38406 * expr.c (store_field): In the bitfield case, fetch the return value
38407 from the registers before applying a single big-endian adjustment.
38408 Always do a final load for a BLKmode value not larger than a word.
38410 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38413 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38414 that we correctly handle column numbers greater than
38415 LINE_MAP_MAX_COLUMN_NUMBER.
38417 2017-01-10 Martin Sebor <msebor@redhat.com>
38419 PR middle-end/78245
38420 * gimple-ssa-sprintf.c (get_destination_size): Call
38421 {init,fini}object_sizes.
38422 * tree-object-size.c (addr_object_size): Adjust.
38423 (pass_through_call): Adjust.
38424 (pass_object_sizes::execute): Adjust.
38425 * tree-object-size.h (fini_object_sizes): Declare.
38427 2017-01-10 Martin Sebor <msebor@redhat.com>
38429 PR tree-optimization/78775
38430 * builtins.c (get_size_range): Move...
38431 * calls.c: ...to here.
38432 (alloc_max_size): Accept zero argument.
38433 (operand_signed_p): Remove.
38434 (maybe_warn_alloc_args_overflow): Call get_size_range.
38435 * calls.h (get_size_range): Declare.
38437 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38439 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38440 from TI's devices.csv file as of September 2016.
38441 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38443 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38445 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38446 * doc/invoke.texi: Likewise.
38447 * doc/md.texi: Likewise.
38448 * doc/objc.texi: Likewise.
38450 2017-01-10 Joshua Conner <joshconner@google.com>
38452 * config/arm/fuchsia-elf.h: New file.
38453 * config/fuchsia.h: New file.
38454 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38455 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38457 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38459 2016-01-10 Richard Biener <rguenther@suse.de>
38461 PR tree-optimization/79034
38462 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38463 Propagate out degenerate PHIs in the joiner.
38465 2017-01-10 Martin Liska <mliska@suse.cz>
38467 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38468 (sort_congruence_classes_by_decl_uid): Likewise.
38469 (sort_congruence_class_groups_by_decl_uid): Likewise.
38470 (sem_item_optimizer::merge_classes): Sort class, groups in these
38471 classes and members in the groups by DECL_UID of declarations.
38472 This would make merge operations stable.
38474 2017-01-10 Martin Liska <mliska@suse.cz>
38476 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38477 usage of m_classes_vec.
38478 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38479 (sem_item_optimizer::get_group_by_hash): Likewise.
38480 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38481 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38482 (sem_item_optimizer::verify_classes): Likewise.
38483 (sem_item_optimizer::process_cong_reduction): Likewise.
38484 (sem_item_optimizer::dump_cong_classes): Likewise.
38485 (sem_item_optimizer::merge_classes): Likewise.
38486 * ipa-icf.h (congruence_class_hash): Rename from
38487 congruence_class_group_hash. Remove declaration of m_classes_vec.
38489 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38491 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38492 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38493 * config.gcc: Add avx512vpopcntdqintrin.h.
38494 * config/i386/avx512vpopcntdqintrin.h: New.
38495 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38496 * config/i386/i386-builtin-types.def: Add new types.
38497 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38498 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38499 __builtin_ia32_vpopcountq_v8di_mask): New.
38500 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38501 __AVX512VPOPCNTDQ__.
38502 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38503 (PTA_AVX512VPOPCNTDQ): Define.
38504 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38505 TARGET_AVX512VPOPCNTDQ_P): Define.
38506 * config/i386/i386.opt: Add mavx512vpopcntdq.
38507 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38508 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38510 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38512 PR middle-end/77484
38513 * predict.def (PRED_CALL): Set to 67.
38515 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38517 * expr.c (store_field): In the bitfield case, if the value comes from
38518 a function call and is of an aggregate type returned in registers, do
38519 not modify the field mode; extract the value in all cases if the mode
38520 is BLKmode and the size is not larger than a word.
38522 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38525 * config/i386/cpuid.h: Fix undefined behavior.
38527 2017-01-04 Jeff Law <law@redhat.com>
38529 PR tree-optimization/79007
38530 PR tree-optimization/67955
38531 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38532 conservative for pt.null when flag_non_call_exceptions is on.
38534 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38536 PR translation/79019
38537 PR translation/79020
38538 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38539 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38540 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38542 * config/avr/avr.opt (maccumulate-args): Likewise.
38543 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38544 * common.opt (freport-bug): Likewise.
38545 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38546 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38547 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38548 translatable string.
38549 * config/i386/i386.c (function_value_32): Likewise.
38550 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38551 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38553 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38554 * common/config/msp430/msp430-common.c (msp430_handle_option):
38556 * symtab.c (symtab_node::verify_base): Likewise.
38557 * opts.c (set_debug_level): Likewise.
38558 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38560 missing whitespace to translatable strings.
38561 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38562 * config/sh/superh.h: Likewise.
38563 * config/i386/xopintrin.h: Likewise.
38564 * config/i386/znver1.md: Likewise.
38565 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38566 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38567 * double-int.h (struct double_int): Likewise.
38568 * double-int.c (div_and_round_double): Likewise.
38569 * wide-int.cc: Likewise.
38570 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38571 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38572 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38573 (crossjumps_occurred): ... this.
38574 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38577 PR tree-optimization/78899
38578 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38579 returning bool return struct loop *, NULL for failure and the new
38581 (versionable_outer_loop_p): Don't version outer loop if it has
38582 dont_vectorized bit set.
38583 (tree_if_conversion): When versioning outer loop, ensure
38584 tree_if_conversion is performed also on the inner loop of the
38585 non-vectorizable outer loop copy.
38586 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38587 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38588 prevent vectorization of it.
38589 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38590 the outer loop vectorization of the non-scalar version is attempted
38591 before vectorization of the inner loop in scalar version. If
38592 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38593 vectorization of its inner loop.
38594 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38595 has 2 inner loops, rename also on edges from bb whose single pred
38596 is outer_loop->header. Fix typo in function comment.
38598 2017-01-09 Martin Sebor <msebor@redhat.com>
38601 * asan.c (asan_emit_stack_protection): Increase local buffer size
38602 to avoid snprintf truncation warning.
38604 2017-01-09 Andrew Pinski <apinski@cavium.com>
38606 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38607 to reference thunderx2t99 for the tuning structure
38608 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38610 (thunderx2t99_extra_costs): This.
38611 * config/aarch64/aarch64-tune.md: Regenerate.
38612 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38613 (vulcan_addrcost_table): This.
38614 (vulcan_regmove_cost): Rename to ...
38615 (thunderx2t99_regmove_cost): This.
38616 (vulcan_vector_cost): Rename to ...
38617 (thunderx2t99_vector_cost): this.
38618 (vulcan_branch_cost): Rename to ...
38619 (thunderx2t99_branch_cost): This.
38620 (vulcan_tunings): Rename to ...
38621 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38622 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38624 2017-01-09 Martin Jambor <mjambor@suse.cz>
38628 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38629 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38630 (propagate_vr_accross_jump_function): Use the above function for all
38631 value range computations for pass-through jump functions and type
38632 converasion from explicit value range values.
38633 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38634 parameters from TYPE_ARG_TYPES.
38635 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38636 (ipa_write_node_info): Stream type of the actual argument.
38637 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38639 2017-01-09 Martin Liska <mliska@suse.cz>
38642 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38643 (lookup_compiler): Do not show error message with have_E.
38645 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38647 PR tree-optimization/78938
38648 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38649 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38650 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38651 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38654 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38656 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38659 2017-01-09 Richard Biener <rguenther@suse.de>
38661 PR tree-optimization/78997
38662 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38663 name condition properly.
38665 2017-01-09 Richard Biener <rguenther@suse.de>
38668 * dwarf2out.c (is_cxx): New overload with context.
38669 (is_naming_typedef_decl): Use it.
38671 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38673 * invoke.texi (Option Summary): Correct spacing in option lists
38674 and add line breaks to fix over-long lines.
38676 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38678 PR middle-end/17660
38680 * extend.texi (Common Variable Attributes): Add xref to GCC
38681 Internals manual to explain mode attribute keywords.
38683 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38686 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38687 and Preprocessor Options.
38688 (Options for Linking): Document -pthread here....
38689 (RS/6000 and PowerPC Options): ...not here.
38690 (Solaris 2 Options): ...or here.
38691 * doc/cppopts.texi: Document -pthread.
38693 2017-01-08 Martin Sebor <msebor@redhat.com>
38695 PR middle-end/77708
38696 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38697 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38698 New member functions.
38699 (format_directive): Used them.
38701 (pass_sprintf_length::handle_gimple_call): Same.
38702 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38703 to avoid truncation for any argument.
38704 (extract_affine_mul): Same.
38705 * tree.c (get_file_function_name): Same.
38707 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38709 PR middle-end/77484
38710 * predict.def (PRED_INDIR_CALL): Set to 86.
38712 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38714 PR preprocessor/54124
38715 * doc/cppopts.texi: Reformat -d subtable to list the full name
38716 of the options. Add cross-reference to the docs for the general
38717 compiler -d options.
38718 * doc/invoke.texi (Developer Options): Add cross-reference to the
38719 preprocessor-specific -d option documentation.
38721 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38723 PR preprocessor/13498
38724 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38725 redudant material, and reflect new command-line options.
38726 (System Headers): Likewise.
38728 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38730 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38731 -isystem, and -idirafter. Copy-edit.
38732 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38733 default for -ftrack-macro-expansion. Delete obsolete and
38734 badly-formatted implementation details about -fdebug-cpp output.
38735 * doc/cppwarnopts.texi: Copy-edit.
38737 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38740 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38741 that the transition from a max line width >= 1<<10 to narrower
38742 lines works correctly.
38744 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38746 * doc/options.texi (PerFunction): New.
38747 * opt-functions.awk (switch_flags): Map both Optimization and
38748 PerFunction to CL_OPTIMIZATION.
38749 * opth-gen.awk: Test for PerFunction flag along with
38751 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38752 it only when the latter is present. Skip those that don't in
38753 the hash function generator.
38754 * common.opt (fvar-tracking): Mark as PerFunction instead of
38756 (fvar-tracking-assignments): Likewise.
38757 (fvar-tracking-assignments-toggle): Likewise.
38758 (fvar-tracking-uninit): Likewise.
38760 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38762 PR translation/79018
38763 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38766 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38769 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38770 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38771 TARGET_LONG_JUMP_TABLE_OFFSETS.
38772 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38773 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38774 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38775 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38776 * config/m68k/m68k.md (tablejump expander): Likewise.
38777 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38778 TARGET_LONG_JUMP_TABLE_OFFSETS.
38779 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38780 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38782 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38783 David Holsgrove <david.holsgrove@xilinx.com>
38785 * common/config/microblaze/microblaze-common.c
38786 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38787 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38789 * config/microblaze/microblaze.c (microblaze_must_save_register)
38790 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38792 (microblaze_eh_return): New function.
38793 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38794 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38795 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38796 * config/microblaze/microblaze.md (eh_return): New pattern.
38798 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38800 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38801 GCC_DIAGNOSTIC_STRINGIFY): Define.
38803 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38805 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38807 * config/arm/arm.md (<mcrr>): New.
38809 * config/arm/arm.c (arm_arch5te): New.
38810 (arm_option_override): Set arm_arch5te.
38811 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38813 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38814 (arm_mcrr_qualifiers): ... this. New.
38815 (MRRC_QUALIFIERS): Define to...
38816 (arm_mrrc_qualifiers): ... this. New.
38817 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38819 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38820 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38821 (MRRCI, mrrc, MRRC): New.
38822 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38823 VUNSPEC_MRRC2): New.
38825 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38827 * config/arm/arm.md (<mcr>): New.
38829 * config/arm/arm.c (arm_coproc_builtin_available): Add
38830 support for mcr, mrc, mcr2 and mrc2.
38831 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38832 (arm_mcr_qualifiers): ... this. New.
38833 (MRC_QUALIFIERS): Define to ...
38834 (arm_mrc_qualifiers): ... this. New.
38835 (MCR_QUALIFIERS): Define to ...
38836 (arm_mcr_qualifiers): ... this. New.
38837 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38839 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38840 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38841 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38842 VUNSPEC_MRC2): New.
38844 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38846 * config/arm/arm.md (*ldc): New.
38850 * config/arm/arm.c (arm_coproc_builtin_available): Add
38851 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38852 (arm_coproc_ldc_stc_legitimate_address): New.
38853 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38854 'qualifier_const_pointer'.
38855 (LDC_QUALIFIERS): Define to...
38856 (arm_ldc_qualifiers): ... this. New.
38857 (STC_QUALIFIERS): Define to...
38858 (arm_stc_qualifiers): ... this. New.
38859 * config/arm/arm-protos.h
38860 (arm_coproc_ldc_stc_legitimate_address): New.
38861 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38862 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38863 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38864 stc2, stcl, stc2l): New.
38865 * config/arm/constraints.md (Uz): New.
38866 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38867 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38868 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38869 VUNSPEC_STC2L): New.
38871 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38873 * config/arm/arm.md (<cdp>): New.
38874 * config/arm/arm.c (neon_const_bounds): Rename this ...
38875 (arm_const_bounds): ... this.
38876 (arm_coproc_builtin_available): New.
38877 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38878 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38879 (CDP_QUALIFIERS): Define to...
38880 (arm_cdp_qualifiers): ... this. New.
38882 (arm_expand_builtin_args): Add case for 6 arguments.
38883 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38884 (arm_const_bounds): ... this.
38885 (arm_coproc_builtin_available): New.
38886 * config/arm/arm_acle.h (__arm_cdp): New.
38888 * config/arm/arm_acle_builtins.def (cdp): New.
38890 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38891 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38892 'arm_const_bounds'.
38893 * config/arm/types.md (coproc): New.
38894 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38895 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38896 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38897 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38899 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38901 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38902 (UBINOP_QUALIFIERS): New.
38904 (acle_builtin_data): New. Change comment.
38905 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38906 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38907 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38908 arm_acle_builtins.def.
38909 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38910 (arm_init_acle_builtins): New.
38911 (CRC32_BUILTIN): Remove.
38912 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38913 crc32cb, crc32ch and crc32cw.
38914 (arm_init_crc32_builtins): Remove.
38915 (arm_init_builtins): Use arm_init_acle_builtins rather
38916 than arm_init_crc32_builtins.
38917 (arm_expand_acle_builtin): New.
38918 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38919 * config/arm/arm_acle_builtins.def: New.
38921 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38923 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38924 (arm_builtin_datum): ... this.
38925 (arm_init_neon_builtin): Rename to ...
38926 (arm_init_builtin): ... this. Add a new parameters PREFIX
38927 and USE_SIG_IN_NAME.
38928 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38929 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38930 'arm_builtin_datum'.
38931 (arm_init_vfp_builtins): Likewise.
38932 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38933 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38934 (arm_expand_neon_args): Rename to ...
38935 (arm_expand_builtin_args): ... this. Rename builtin_arg
38936 enum values and differentiate between ARG_BUILTIN_MEMORY
38937 and ARG_BUILTIN_NEON_MEMORY.
38938 (arm_expand_neon_builtin_1): Rename to ...
38939 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38940 values, arm_expand_builtin_args and add bool parameter NEON.
38941 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38942 (arm_expand_vfp_builtin): Likewise.
38943 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38945 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38947 PR middle-end/77484
38948 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38949 * predict.c (tree_estimate_probability_bb): Reverse direction of
38950 polymorphic call predictor.
38952 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38954 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38955 (determine_pass_name_match): ...this new function and...
38956 (should_skip_pass_p): ...this new function.
38958 2017-01-06 Nathan Sidwell <nathan@acm.org>
38960 * ipa-visibility.c (function_and_variable_visibility): Reformat
38961 comments and long lines. Remove extrneous if.
38962 * symtab.c (symtab_node::make_decl_local): Fix code format.
38963 (symtab_node::set_section_for_node): Fix comment typo.
38965 2017-01-06 Martin Liska <mliska@suse.cz>
38968 * lra-constraints.c: Rename invariant to lra_invariant.
38969 * predict.c (set_even_probabilities): Initialize e to NULL.
38971 2017-01-05 Martin Sebor <msebor@redhat.com>
38973 PR tree-optimization/78910
38974 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38975 (format_integer): Correct off-by-one error in the handling
38976 of precision with negative numbers in signed conversions..
38978 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38980 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38982 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38984 PR tree-optimization/71016
38985 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38986 factor_out_conditional_conversion. Formatting fix.
38987 (factor_out_conditional_conversion): Add cond_stmt argument.
38988 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38989 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38992 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38994 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38995 read-rtl-function.o, and selftest-rtl.o.
38996 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38997 (selftest::aarch64_test_loading_full_dump): New function.
38998 (selftest::aarch64_run_selftests): New function.
38999 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39000 selftest::aarch64_run_selftests.
39001 * config/i386/i386.c
39002 (selftest::ix86_test_loading_dump_fragment_1): New function.
39003 (selftest::ix86_test_loading_call_insn): New function.
39004 (selftest::ix86_test_loading_full_dump): New function.
39005 (selftest::ix86_test_loading_unspec): New function.
39006 (selftest::ix86_run_selftests): Call the new functions.
39007 * emit-rtl.c (maybe_set_max_label_num): New function.
39008 * emit-rtl.h (maybe_set_max_label_num): New decl.
39009 * function.c (instantiate_decls): Guard call to
39010 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39011 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39013 * gensupport.c (gen_reader::gen_reader): Pass "false"
39014 for new "compact" param of rtx_reader.
39015 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39016 rather than an empty string for NULL strings.
39017 * read-md.c: Potentially include config.h rather than bconfig.h.
39018 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39019 (have_error): New global, copied from errors.c.
39020 (md_reader::read_name): Rename to...
39021 (md_reader::read_name_1): ...this, adding "out_loc" param,
39022 and converting "missing name or number" to returning false, rather
39024 (md_reader::read_name): Reimplement in terms of read_name_1.
39025 (md_reader::read_name_or_nil): New function.
39026 (md_reader::read_string): Handle "(nil)" by returning NULL.
39027 (md_reader::md_reader): Add new param "compact".
39028 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39029 (md_reader::read_file): New method.
39030 * read-md.h (md_reader::md_reader): Add new param "compact".
39031 (md_reader::read_file): New method.
39032 (md_reader::is_compact): New accessor.
39033 (md_reader::read_name): Convert return type from void to file_location.
39034 (md_reader::read_name_or_nil): New decl.
39035 (md_reader::read_name_1): New decl.
39036 (md_reader::m_compact): New field.
39037 (noop_reader::noop_reader): Pass "false" for new "compact" param
39039 (rtx_reader::rtx_reader): Add new "compact" param.
39040 (rtx_reader::read_rtx_operand): Make virtual and convert return
39041 type from void to rtx.
39042 (rtx_reader::read_until): New decl.
39043 (rtx_reader::handle_any_trailing_information): New virtual function.
39044 (rtx_reader::postprocess): New virtual function.
39045 (rtx_reader::finalize_string): New virtual function.
39046 (rtx_reader::m_in_call_function_usage): New field.
39047 (rtx_reader::m_reuse_rtx_by_id): New field.
39048 * read-rtl-function.c: New file.
39049 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39050 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39051 (selftest::verify_three_block_rtl_cfg): New decl.
39052 * read-rtl-function.h: New file.
39053 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39054 For host, include function.h, memmodel.h, and emit-rtl.h.
39055 (one_time_initialization): New function.
39056 (struct compact_insn_name): New struct.
39057 (compact_insn_names): New array.
39058 (find_code): Handle insn codes in compact dumps.
39059 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39060 (bind_subst_iter_and_attr): Likewise.
39061 (add_condition_to_string): Likewise.
39062 (add_condition_to_rtx): Likewise.
39063 (apply_attribute_uses): Likewise.
39064 (add_current_iterators): Likewise.
39065 (apply_iterators): Likewise.
39066 (initialize_iterators): Guard usage of apply_subst_iterator with
39067 #ifdef GENERATOR_FILE.
39068 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39069 (md_reader::read_mapping): Likewise.
39070 (add_define_attr_for_define_subst): Likewise.
39071 (add_define_subst_attr): Likewise.
39072 (read_subst_mapping): Likewise.
39073 (check_code_iterator): Likewise.
39074 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39076 (one_time_initialization): New function.
39077 (rtx_reader::read_until): New method.
39078 (read_flags): New function.
39079 (parse_reg_note_name): New function.
39080 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39081 Handle reuse_rtx ids.
39082 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39083 Add parsing support for RTL dumps, mirroring the special-cases in
39084 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39085 values, and calling handle_any_trailing_information.
39086 (rtx_reader::read_rtx_operand): Convert return type from void
39087 to rtx, returning return_rtx. Handle case 'e'. Call
39088 finalize_string on XSTR and XTMPL fields.
39089 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39090 "(nil)" values were omitted. Call the postprocess vfunc on the
39092 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39093 class ctor. Initialize m_in_call_function_usage. Call
39094 one_time_initialization.
39095 * rtl-tests.c (selftest::test_uncond_jump): Call
39096 set_new_first_and_last_insn.
39097 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39098 * selftest-rtl.c: New file.
39099 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39100 (selftest::get_insn_by_uid): New decl.
39101 * selftest-run-tests.c (selftest::run_tests): Call
39102 read_rtl_function_c_tests.
39103 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39104 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39107 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39109 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39110 operands in a special way. Assert that pos+len <= mode precision.
39112 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39114 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39115 3 argument Alias with unlimited for the negative form.
39116 (fno-vect-cost-model): Removed.
39118 2017-01-05 Martin Liska <mliska@suse.cz>
39120 * hsa-gen.c (gen_hsa_divmod): New function.
39121 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39123 2017-01-05 Martin Liska <mliska@suse.cz>
39126 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39129 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39131 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39132 small constant length operands.
39134 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39136 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39137 between loop iterations.
39139 2017-01-05 Martin Liska <mliska@suse.cz>
39142 * gimplify.c (gimplify_decl_expr): Compare to
39143 asan_poisoned_variables instread of checking flags.
39144 (gimplify_target_expr): Likewise.
39145 (gimplify_expr): Likewise.
39146 (gimplify_function_tree): Conditionally initialize
39147 asan_poisoned_variables.
39149 2017-01-04 Jeff Law <law@redhat.com>
39151 PR tree-optimizatin/78812
39152 * rtl.h (contains_mem_rtx_p): Prototype.
39153 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39154 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39155 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39156 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39157 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39159 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39161 * input.c (assert_char_at_range): Default-initialize actual_range.
39163 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39165 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39166 to match the caller.
39168 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39170 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39171 insns after final jump in test to emit dummy move.
39173 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39175 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39176 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39178 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39180 * multiple_target.c (create_dispatcher_calls): Init e_next.
39181 * tree-ssa-loop-split.c (split_loop): Init border.
39182 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39185 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39190 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39191 (altivec_register_operand): Do not return true if the operand
39192 contains a SUBREG mixing SImode and SFmode.
39193 (vsx_register_operand): Likewise.
39194 (vsx_reg_sfsubreg_ok): New predicate.
39195 (vfloat_operand): Do not return true if the operand contains a
39196 SUBREG mixing SImode and SFmode.
39197 (vint_operand): Likewise.
39198 (vlogical_operand): Likewise.
39199 (gpc_reg_operand): Likewise.
39200 (int_reg_operand): Likewise.
39201 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39202 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39203 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39205 (rs6000_emit_move_si_sf_subreg): New helper function.
39206 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39207 fixup SUBREGs involving SImode and SFmode.
39208 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39209 numbers for the new peephole2 optimization.
39210 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39211 the GLIBC math library that do AND/IOR/XOR operations on single
39212 precision floating point.
39213 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39214 target macros to say whether we need to avoid SUBREGs mixing
39216 (TARGET_ALLOW_SF_SUBREG): Likewise.
39217 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39218 (UNSPEC_SI_FROM_SF): Likewise.
39219 (iorxor): Change spacing.
39220 (and_ior_xor): New iterator for AND, IOR, and XOR.
39221 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39222 (movdi_from_sf_zero_ext): Likewise.
39223 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39224 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39225 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39226 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39227 (fms<mode>4): Likewise.
39228 (fnma<mode>4): Likewise.
39229 (fnms<mode>4): Likewise.
39230 (nfma<mode>4): Likewise.
39231 (nfms<mode>4): Likewise.
39233 2017-01-04 Marek Polacek <polacek@redhat.com>
39236 * doc/invoke.texi: Document -Wpointer-compare.
39238 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39240 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39243 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39244 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39247 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39250 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39251 documentation of the powerpc_popcntb_ok attribute.
39252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39253 code to issue warning messages if a requested CPU configuration is
39254 not supported by the binary (assembler and loader) toolchain.
39255 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39256 made to define a built-in function that has been disabled.
39257 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39258 made to define a built-in function that has been disabled.
39259 (altivec_init_builtins): Add comment explaining why definition
39260 of the DST built-in functions is not preceded by an assertion
39261 check. Add assertions to prevent ICE if attempts are made to
39262 define an altivec predicate or an abs* built-in function that has
39264 (htm_init_builtins): Add comment explaining why definition of the
39265 htm built-in functions is not preceded by an assertion check.
39267 2017-01-04 Jeff Law <law@redhat.com>
39269 PR tree-optimizatin/67955
39270 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39271 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39272 the points-to solution does not include pt_null. Use DECL_PT_UID
39275 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39277 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39278 Use gen_int_mode instead of gen_lopwart for const_int operands.
39280 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39282 PR tree-optimization/71563
39283 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39284 out of range value - has low bits known to be zero.
39286 2017-01-04 Alan Modra <amodra@gmail.com>
39288 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39289 * configure: Regenerate.
39290 * config.in: Regenerate.
39292 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39295 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39296 a substring of the message, but strcmp with the whole message. Ifdef
39297 ENABLE_NLS, translate the message first using dgettext.
39299 2017-01-03 Jeff Law <law@redhat.com>
39301 PR tree-optimizatin/78856
39302 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39303 (mark_threaded_blocks): Remove code to truncate thread paths that
39304 cross multiple loop headers. Instead invalidate the cached loop
39305 iteration information and handle case of a thread path walking
39306 into an irreducible region.
39308 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39311 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39312 assertions. Add support for doing the signbit if the IEEE 128-bit
39313 floating point value is in a GPR.
39314 * config/rs6000/rs6000.md (Fsignbit): Delete.
39315 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39316 Update the length attribute if the value is in a GPR.
39317 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39318 the sign or zero extension instruction, since the value is always 0/1.
39319 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39322 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39323 extracting SImode to a GPR register so that we can generate a
39324 store, limit the vector to be in a traditional Altivec register
39325 for the vextuwrx instruction.
39327 2017-01-03 Ian Lance Taylor <iant@google.com>
39329 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39331 2017-01-03 Martin Sebor <msebor@redhat.com>
39333 PR tree-optimization/78696
39334 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39335 precision. Use MPFR for %f for greater fidelity. Correct handling
39337 (pass_sprintf_length::compute_format_length): Set width and precision
39338 specified by asrerisk to void_node for vararg functions.
39339 (try_substitute_return_value): Adjust dump output.
39341 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39343 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39345 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39347 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39348 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39349 -mlra/-mno-lra was passed to the compiler.
39351 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39353 PR rtl-optimization/65618
39354 * emit-rtl.c (try_split): Move initialization of "before" and
39355 "after" to just before the call to emit_insn_after_setloc.
39357 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39359 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39361 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39363 * dwarf2out.c (gen_enumeration_type_die): When
39364 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39366 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39368 PR tree-optimization/78965
39369 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39370 Change first argument from const call_info & to call_info &. For %n
39371 set info.nowrite to false.
39373 PR middle-end/78901
39374 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39375 possibly throwing calls.
39377 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39378 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39379 and fns handling, rather than in a separate case SSA_NAME.
39381 2017-01-02 Jeff Law <law@redhat.com>
39383 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39384 fixes for first_period and second_period variables.
39386 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39389 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39390 (*insvqi_1): New insn pattern.
39391 (*insvqi_1_mem_rex64): Ditto.
39392 (*insvqi_2): Ditto.
39393 (*insvqi_3): Rename from *insvqi.
39395 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39397 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39399 * doc/cfg.texi (Edges): Remove reference to Java.
39400 (Maintaining the CFG): Ditto.
39402 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39404 PR middle-end/77674
39405 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39406 transparent aliases.
39408 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39410 PR middle-end/77484
39411 * predict.def (PRED_CALL): Update hitrate.
39412 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39413 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39414 into direct/indirect/polymorphic variants.
39416 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39418 Update copyright years.
39420 * gcc.c (process_command): Update copyright notice dates.
39421 * gcov-dump.c (print_version): Ditto.
39422 * gcov.c (print_version): Ditto.
39423 * gcov-tool.c (print_version): Ditto.
39424 * gengtype.c (create_file): Ditto.
39425 * doc/cpp.texi: Bump @copying's copyright year.
39426 * doc/cppinternals.texi: Ditto.
39427 * doc/gcc.texi: Ditto.
39428 * doc/gccint.texi: Ditto.
39429 * doc/gcov.texi: Ditto.
39430 * doc/install.texi: Ditto.
39431 * doc/invoke.texi: Ditto.
39433 Copyright (C) 2017 Free Software Foundation, Inc.
39435 Copying and distribution of this file, with or without modification,
39436 are permitted in any medium without royalty provided the copyright
39437 notice and this notice are preserved.