1 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6 * tree.c (build_vector): Likewise.
7 (build_vector_from_ctor): Update accordingly.
8 (build_vector_from_val): Likewise.
9 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11 * tree-vect-generic.c (add_rshift): Likewise.
12 (expand_vector_divmod): Likewise.
13 (optimize_vector_constructor): Likewise.
14 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15 (vect_transform_slp_perm_load): Likewise.
16 (vect_schedule_slp_instance): Likewise.
17 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
18 (vectorizable_call): Likewise.
19 (vect_gen_perm_mask_any): Likewise. Add elements in order.
20 * expmed.c (make_tree): Likewise.
21 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
22 a vector passed to build_vector.
23 (fold_convert_const): Likewise.
24 (exact_inverse): Likewise.
25 (fold_ternary_loc): Likewise.
26 (fold_relational_const): Likewise.
27 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
28 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
29 (const_unop): Likewise. Store the reduction accumulator in a
30 variable rather than an array.
31 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
32 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
33 the new vector, rather than constructing it after the input arrays.
34 (native_interpret_vector): Use auto_vec<tree> when building
35 a vector passed to build_vector. Add elements in order.
36 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
37 auto_vec<tree> when building a vector passed to build_vector.
38 (vect_create_epilog_for_reduction): Likewise.
39 (vectorizable_induction): Likewise.
40 (get_initial_def_for_reduction): Likewise. Fix indentation of
42 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
44 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
45 passed to build_vector.
47 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
48 Alan Hayward <alan.hayward@arm.com>
49 David Sherwood <david.sherwood@arm.com>
51 * tree-core.h (tree_base::u): Add an "nelts" field.
52 (tree_vector): Use VECTOR_CST_NELTS as the length.
53 * tree.c (tree_size): Likewise.
54 (make_vector): Initialize VECTOR_CST_NELTS.
55 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
56 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
58 * expr.c (const_vector_mask_from_tree): Consistently use "units"
59 as the number of units, setting it from VECTOR_CST_NELTS.
60 (const_vector_from_tree): Likewise.
61 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
62 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
63 (fold_negate_expr_1): Likewise.
64 (fold_convert_const): Likewise.
65 (const_binop): Likewise. Differentiate the number of output and
67 (const_unop): Likewise.
68 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
69 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
70 in cases that did the opposite.
72 2017-09-14 Richard Biener <rguenther@suse.de>
74 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
77 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
79 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
81 2017-09-14 Jakub Jelinek <jakub@redhat.com>
84 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
85 if and where to split a bb, except for splitting before debug insn
86 sequences followed by non-label real insn. Delete debug insns
87 in between basic blocks.
89 * combine.c (make_compound_operation_int): Formatting fixes.
91 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
92 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
93 * config/netbsd.h (LINK_EH_SPEC): Likewise.
94 * config/sol2.h (LINK_EH_SPEC): Likewise.
95 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
96 * config/s390/linux.h (LINK_SPEC): Likewise.
97 * config/freebsd.h (LINK_EH_SPEC): Likewise.
98 * config/openbsd.h (LINK_EH_SPEC): Likewise.
99 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
100 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
101 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
102 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
103 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
104 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
106 2017-09-13 Jakub Jelinek <jakub@redhat.com>
108 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
110 (ENDFILE_LINUX_SPEC): Likewise.
111 (LINK_EH_SPEC): Likewise.
112 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
113 (LINK_OS_LINUX_SPEC32): Likewise.
114 (LINK_OS_LINUX_SPEC64): Likewise.
115 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
116 (LINK_OS_LINUX_SPEC): Likewise.
118 2017-09-13 Martin Liska <mliska@suse.cz>
121 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
122 CASE_HIGH is NULL_TREE.
124 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
125 Alan Hayward <alan.hayward@arm.com>
126 David Sherwood <david.sherwood@arm.com>
128 * target.def (secondary_memory_needed): New hook.
129 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
130 instead of SECONDARY_MEMORY_NEEDED.
131 (secondary_memory_needed_mode): Likewise.
132 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
133 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
134 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
135 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
136 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
137 * doc/tm.texi: Regenerate.
138 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
139 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
140 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
141 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
142 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
143 * config/i386/i386.c (inline_secondary_memory_needed): Put the
144 mode argument first and change the reg_class arguments to reg_class_t.
145 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
146 Make static. Update the call to inline_secondary_memory_needed.
147 (ix86_register_move_cost): Update the call to
148 inline_secondary_memory_needed.
149 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
150 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
152 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
153 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
155 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
156 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
157 * config/mips/mips.c (mips_secondary_memory_needed): Make static
158 and match hook interface. Add comment from mips.h.
159 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
160 * config/mmix/mmix.md (truncdfsf2): Refer to
161 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
163 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
164 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
165 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
166 (pa_secondary_memory_needed): New function.
167 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
168 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
169 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
170 (pdp11_secondary_memory_needed): Make static and match hook interface.
171 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
172 * config/powerpcspe/powerpcspe-protos.h
173 (rs6000_secondary_memory_needed_ptr): Delete.
174 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
176 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
177 (rs6000_option_override_internal): Assign to
178 targetm.secondary_memory_needed rather than
179 rs6000_secondary_memory_needed_ptr.
180 (rs6000_secondary_memory_needed): Match hook interface.
181 (rs6000_debug_secondary_memory_needed): Likewise.
182 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
183 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
184 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
185 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
186 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
187 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
189 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
190 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
191 (rs6000_option_override_internal): Assign to
192 targetm.secondary_memory_needed rather than
193 rs6000_secondary_memory_needed_ptr.
194 (rs6000_secondary_memory_needed): Match hook interface.
195 (rs6000_debug_secondary_memory_needed): Likewise.
196 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
197 * config/s390/s390.c (s390_secondary_memory_needed): New function.
198 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
199 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
200 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
201 (sparc_secondary_memory_needed): New function.
202 * lra-constraints.c (check_and_process_move): Refer to
203 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
205 (curr_insn_transform): Likewise.
206 (process_alt_operands): Use targetm.secondary_memory_needed
207 instead of TARGET_SECONDARY_MEMORY_NEEDED.
208 (check_secondary_memory_needed_p): Likewise.
209 (choose_split_class): Likewise.
210 * reload.c: Unconditionally include code that was previously
211 conditional on SECONDARY_MEMORY_NEEDED.
212 (push_secondary_reload): Use targetm.secondary_memory_needed
213 instead of TARGET_SECONDARY_MEMORY_NEEDED.
214 (push_reload): Likewise.
215 * reload1.c: Unconditionally include code that was previously
216 conditional on SECONDARY_MEMORY_NEEDED.
217 (choose_reload_regs): Use targetm.secondary_memory_needed
218 instead of TARGET_SECONDARY_MEMORY_NEEDED.
219 (gen_reload): Likewise.
220 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
222 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * target.def (secondary_memory_needed_mode): New hook:
227 * targhooks.c (default_secondary_memory_needed_mode): Declare.
228 * targhooks.h (default_secondary_memory_needed_mode): New function.
229 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
230 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
231 * doc/tm.texi: Regenerate.
232 * lra-constraints.c (check_and_process_move): Use
233 targetm.secondary_memory_needed_mode instead of
234 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
235 (curr_insn_transform): Likewise.
236 * reload.c (get_secondary_mem): Likewise.
237 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
238 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
240 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
241 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
242 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
243 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
244 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
246 * config/powerpcspe/powerpcspe-protos.h
247 (rs6000_secondary_memory_needed_mode): Delete.
248 * config/powerpcspe/powerpcspe.c
249 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
250 (rs6000_secondary_memory_needed_mode): Make static.
251 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
252 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
254 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
256 (rs6000_secondary_memory_needed_mode): Make static.
257 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
258 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
259 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
260 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
261 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
263 (sparc_secondary_memory_needed_mode): New function.
264 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
266 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
268 * config/aarch64/constraints.md (Umq): New constraint.
269 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
271 (mov<mode>): Update condition.
273 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
275 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
276 when bitposition is the same.
278 2017-09-13 Richard Biener <rguenther@suse.de>
280 * dwarf2out.c (output_die_symbol): Remove.
281 (output_die): Do not output a DIEs symbol.
283 2017-09-13 Richard Biener <rguenther@suse.de>
286 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
287 default-def to avoid breaking iterator update with the weird
288 interaction with cgraph_update_edges_for_call_stmt_node.
290 2017-09-13 Richard Biener <rguenther@suse.de>
292 * tree-cfg.c (verify_gimple_assign_binary): Add verification
293 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
294 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
295 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
297 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
299 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
300 Disable pc relative literal load irrespective of
301 TARGET_FIX_ERR_A53_84341 for default.
303 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
305 * config/sparc/sparc.c (output_return): Output the source location of
306 the insn in the delay slot, if any.
307 (output_sibcall): Likewise.
309 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
312 * common.opt (-static-pie): New alias.
313 (shared): Negate static-pie.
314 (-no-pie): Update help text.
316 (static-pie): New option.
317 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
319 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
320 (LINK_EH_SPEC): Likewise.
321 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
322 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
323 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
324 * gcc.c (LINK_COMMAND_SPEC): Likewise.
325 (init_gcc_specs): Likewise.
326 (init_spec): Likewise.
327 (display_help): Update help message for -pie.
328 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
331 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
333 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
334 (movdi_aarch64): Likewise.
335 (movti_aarch64): Likewise.
337 2017-09-12 Simon Wright <simon@pushface.org>
340 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
341 calculation of the minor version, always output as 0.
343 2017-09-12 Jakub Jelinek <jakub@redhat.com>
346 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
347 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
348 on it early, rather than manual conversion late. For
349 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
350 instead of performing manual conversion.
352 2017-09-12 Carl Love <cel@us.ibm.com>
354 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
355 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
357 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
358 VMULOSW): Add definitions.
359 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
360 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
361 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
362 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
363 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
365 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
367 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
369 (movti_aarch64): Likewise.
370 (movdf_aarch64): Likewise.
371 (movtf_aarch64): Likewise.
372 (load_pairdi): Likewise.
373 (store_pairdi): Likewise.
374 (load_pairdf): Likewise.
375 (store_pairdf): Likewise.
376 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
377 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
378 (ldr_got_small_<mode>): Likewise.
379 (ldr_got_small_28k_<mode>): Likewise.
380 (ldr_got_tiny): Likewise.
381 * config/aarch64/iterators.md (ldst_sz): New.
382 (ldpstp_sz): Likewise.
383 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
385 (thunderx_load): Split load_8 to load_16.
386 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
388 (thunderx2t99_storepair_basic): Split store_8 to store_16.
389 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
390 (xgene1_store_pair): Split store_8 to store_16.
391 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
392 (falkor_st_0_st_sd): Split store_8 to store_16.
394 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
396 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
397 and store1/2/3/4 to store_4/8/12/16.
398 * config/aarch64/aarch64.md: Update for rename.
399 * config/arm/arm.md: Likewise.: Likewise.
400 * config/arm/arm.c: Likewise.
401 * config/arm/thumb1.md: Likewise.
402 * config/arm/thumb2.md: Likewise.
403 * config/arm/vfp.md: Likewise.
404 * config/arm/arm-generic.md: Likewise.
405 * config/arm/arm1020e.md: Likewise.
406 * config/arm/arm1026ejs.md: Likewise.
407 * config/arm/arm1136jfs.md: Likewise.
408 * config/arm/arm926ejs.md: Likewise.
409 * config/arm/cortex-a15.md: Likewise.
410 * config/arm/cortex-a17.md: Likewise.
411 * config/arm/cortex-a5.md: Likewise.
412 * config/arm/cortex-a53.md: Likewise.
413 * config/arm/cortex-a57.md: Likewise.
414 * config/arm/cortex-a7.md: Likewise.
415 * config/arm/cortex-a8.md: Likewise.
416 * config/arm/cortex-a9.md: Likewise.
417 * config/arm/cortex-m4.md: Likewise.
418 * config/arm/cortex-m7.md: Likewise.
419 * config/arm/cortex-r4.md: Likewise.
420 * config/arm/exynos-m1.md: Likewise.
421 * config/arm/fa526.md: Likewise.
422 * config/arm/fa606te.md: Likewise.
423 * config/arm/fa626te.md: Likewise.
424 * config/arm/fa726te.md: Likewise.
425 * config/arm/fmp626.md: Likewise.
426 * config/arm/iwmmxt.md: Likewise.
427 * config/arm/ldmstm.md: Likewise.
428 * config/arm/marvell-pj4.md: Likewise.
429 * config/arm/xgene1.md: Likewise.
430 * config/aarch64/thunderx.md: Likewise.
431 * config/aarch64/thunderx2t99.md: Likewise.
432 * config/aarch64/falkor.md: Likewise.
434 2017-09-12 Martin Liska <mliska@suse.cz>
436 * attribs.c (private_lookup_attribute): New function.
437 * attribs.h (private_lookup_attribute): Declared here.
438 (lookup_attribute): Called from this place.
440 2017-09-12 Richard Biener <rguenther@suse.de>
442 PR tree-optimization/82157
443 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
444 stmts with side-effects.
446 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
450 * target.def (hard_regno_nregs): New hook.
451 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
452 * targhooks.h (default_hard_regno_nregs): Declare.
453 * targhooks.c (default_hard_regno_nregs): New function.
454 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
455 (TARGET_HARD_REGNO_NREGS): ...this hook.
456 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
457 (CLASS_MAX_NREGS): Likewise.
458 * doc/tm.texi: Regenerate.
459 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
460 instead of HARD_REGNO_NREGS.
461 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
462 HARD_REGNO_NREGS in the comment.
463 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
464 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
465 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
466 Return an unsigned int.
467 (TARGET_HARD_REGNO_NREGS): Redefine.
468 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
469 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
470 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
471 (arc_hard_regno_nregs): New function.
472 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
473 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
474 (arm_hard_regno_nregs): New function.
475 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
476 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
477 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
478 (TARGET_HARD_REGNO_NREGS): Redefine.
479 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
480 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
481 (HARD_REGNO_NREGS): Delete.
482 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
483 (cr16_hard_regno_nregs): New function.
484 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
485 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
486 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
487 (cris_hard_regno_nregs): New function.
488 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
489 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
490 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
491 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
492 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
493 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
494 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
495 (frv_hard_regno_nregs): Make static. Take and return an
497 (frv_class_max_nregs): Remove outdated copy of documentation.
498 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
499 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
500 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
501 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
502 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
503 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
504 (TARGET_HARD_REGNO_NREGS): Redefine.
505 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
506 (CLASS_MAX_NREGS): Update comment.
507 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
508 (ia64_hard_regno_nregs): New function.
509 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
510 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
511 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
512 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
513 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
515 (m32c_hard_regno_nregs): Likewise. Make static.
516 (TARGET_HARD_REGNO_NREGS): Redefine.
517 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
518 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
519 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
520 (m68k_hard_regno_nregs): New function.
521 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
522 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
523 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
524 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
525 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
526 Take and return an unsigned int.
527 (TARGET_HARD_REGNO_NREGS): Redefine.
528 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
529 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
530 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
531 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
532 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
533 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
534 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
535 (msp430_hard_regno_nregs): Make static. Take and return an
537 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
538 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
539 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
540 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
541 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
542 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
543 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
544 (TARGET_HARD_REGNO_NREGS): Redefine.
545 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
546 (PA_HARD_REGNO_NREGS): ...this.
547 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
548 (PA_HARD_REGNO_NREGS): ...this.
549 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
550 (pa_hard_regno_nregs): New function.
551 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
552 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
553 (pdp11_hard_regno_nregs): New function.
554 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
555 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
556 (rs6000_hard_regno_nregs_hook): New function.
557 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
558 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
559 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
560 Take and return an unsigned int. Move earlier in file.
561 (TARGET_HARD_REGNO_NREGS): Redefine.
562 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
563 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
564 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
565 (rl78_hard_regno_nregs): Make static. Take and return an
567 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
568 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
569 (rs6000_hard_regno_nregs_hook): New function.
570 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
571 * config/rx/rx.c (rx_hard_regno_nregs): New function.
572 (TARGET_HARD_REGNO_NREGS): Redefine.
573 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
574 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
575 instead of HARD_REGNO_NREGS.
576 (s390_hard_regno_nregs): New function.
577 (s390_hard_regno_mode_ok): Add comment from s390.h.
578 (TARGET_HARD_REGNO_NREGS): Redefine.
579 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
580 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
581 (sh_hard_regno_nregs): New function.
582 (sh_pass_in_reg_p): Use it.
583 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
584 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
585 (sparc_hard_regno_nregs): New function.
586 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
587 * config/spu/spu.c (spu_hard_regno_nregs): New function.
588 (spu_function_arg_advance): Use it, supplying a valid register number.
589 (TARGET_HARD_REGNO_NREGS): Redefine.
590 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
591 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
592 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
593 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
594 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
595 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
596 (CLASS_MAX_NREGS): Remove copy of old documentation.
597 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
598 (visium_hard_regno_nregs): New function.
599 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
600 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
601 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
602 xtensa_hard_regno_nregs): New function.
603 * system.h (HARD_REGNO_NREGS): Poison.
605 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
607 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
608 hard_regno_nregs instead of HARD_REGNO_NREGS.
609 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
610 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
611 (c6x_expand_epilogue): Likewise.
612 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
613 (frv_read_iacc_argument): Likewise.
614 * config/sh/sh.c: Include regs.h.
615 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
616 (regs_used): Likewise.
617 (output_stack_adjust): Likewise.
618 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
619 * expmed.c: Include regs.h.
620 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
621 * ree.c: Include regs.h.
622 (combine_reaching_defs): Use hard_regno_nregs instead of
624 (add_removable_extension): Likewise.
626 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
628 * regs.h (hard_regno_nregs): Turn into a function.
629 (end_hard_regno): Update accordingly.
630 * caller-save.c (setup_save_areas): Likewise.
631 (save_call_clobbered_regs): Likewise.
632 (replace_reg_with_saved_mem): Likewise.
633 (insert_restore): Likewise.
634 (insert_save): Likewise.
635 * combine.c (can_change_dest_mode): Likewise.
636 (move_deaths): Likewise.
637 (distribute_notes): Likewise.
638 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
639 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
640 (rs6000_split_multireg_move): Likewise.
641 (rs6000_register_move_cost): Likewise.
642 (rs6000_memory_move_cost): Likewise.
643 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
644 (rs6000_split_multireg_move): Likewise.
645 (rs6000_register_move_cost): Likewise.
646 (rs6000_memory_move_cost): Likewise.
647 * cselib.c (cselib_reset_table): Likewise.
648 (cselib_lookup_1): Likewise.
649 * emit-rtl.c (set_mode_and_regno): Likewise.
650 * function.c (aggregate_value_p): Likewise.
651 * ira-color.c (setup_profitable_hard_regs): Likewise.
652 (check_hard_reg_p): Likewise.
653 (calculate_saved_nregs): Likewise.
654 (assign_hard_reg): Likewise.
655 (improve_allocation): Likewise.
656 (calculate_spill_cost): Likewise.
657 * ira-emit.c (modify_move_list): Likewise.
658 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
659 (ira_hard_reg_in_set_p): Likewise.
660 * ira.c (setup_reg_mode_hard_regset): Likewise.
661 (clarify_prohibited_class_mode_regs): Likewise.
662 (check_allocation): Likewise.
663 * lra-assigns.c (find_hard_regno_for_1): Likewise.
664 (lra_setup_reg_renumber): Likewise.
665 (setup_try_hard_regno_pseudos): Likewise.
666 (spill_for): Likewise.
667 (assign_hard_regno): Likewise.
668 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
669 * lra-constraints.c (in_class_p): Likewise.
670 (lra_constraint_offset): Likewise.
671 (simplify_operand_subreg): Likewise.
672 (lra_constraints): Likewise.
673 (split_reg): Likewise.
674 (split_if_necessary): Likewise.
675 (invariant_p): Likewise.
676 (inherit_in_ebb): Likewise.
677 * lra-lives.c (process_bb_lives): Likewise.
678 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
679 (get_hard_regs): Likewise.
680 (do_remat): Likewise.
681 * lra-spills.c (assign_spill_hard_regs): Likewise.
682 * mode-switching.c (create_pre_exit): Likewise.
683 * postreload.c (reload_combine_recognize_pattern): Likewise.
684 * recog.c (peep2_find_free_register): Likewise.
685 * regcprop.c (kill_value_regno): Likewise.
686 (set_value_regno): Likewise.
687 (copy_value): Likewise.
688 (maybe_mode_change): Likewise.
689 (find_oldest_value_reg): Likewise.
690 (copyprop_hardreg_forward_1): Likewise.
691 * regrename.c (check_new_reg_p): Likewise.
692 (regrename_do_replace): Likewise.
693 * reload.c (push_reload): Likewise.
694 (combine_reloads): Likewise.
695 (find_dummy_reload): Likewise.
696 (operands_match_p): Likewise.
697 (find_reloads): Likewise.
698 (find_equiv_reg): Likewise.
699 (reload_adjust_reg_for_mode): Likewise.
700 * reload1.c (count_pseudo): Likewise.
701 (count_spilled_pseudo): Likewise.
702 (find_reg): Likewise.
703 (clear_reload_reg_in_use): Likewise.
704 (free_for_value_p): Likewise.
705 (allocate_reload_reg): Likewise.
706 (choose_reload_regs): Likewise.
707 (reload_adjust_reg_for_temp): Likewise.
708 (emit_reload_insns): Likewise.
709 (delete_output_reload): Likewise.
710 * rtlanal.c (subreg_get_info): Likewise.
711 * sched-deps.c (sched_analyze_reg): Likewise.
712 * sel-sched.c (init_regs_for_mode): Likewise.
713 (mark_unavailable_hard_regs): Likewise.
714 (choose_best_reg_1): Likewise.
715 (verify_target_availability): Likewise.
716 * valtrack.c (dead_debug_insert_temp): Likewise.
717 * var-tracking.c (track_loc_p): Likewise.
718 (emit_note_insn_var_location): Likewise.
719 * varasm.c (make_decl_rtl): Likewise.
720 * reginfo.c (choose_hard_reg_mode): Likewise.
721 (init_reg_modes_target): Refer directly to
722 this_target_regs->x_hard_regno_nregs.
724 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
726 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
727 instead of hard_regno_nregs.
729 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
731 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
732 end_hard_regno instead of hard_regno_nregs.
733 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
734 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
735 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
736 * ira-color.c (improve_allocation): Likewise.
737 * lra-assigns.c (find_hard_regno_for_1): Likewise.
738 * lra-lives.c (mark_regno_live): Likewise.
739 (mark_regno_dead): Likewise.
740 * lra-remat.c (operand_to_remat): Likewise.
741 * lra.c (collect_non_operand_hard_regs): Likewise.
742 * postreload.c (reload_combine_note_store): Likewise.
743 (move2add_valid_value_p): Likewise.
744 * reload.c (regno_clobbered_p): Likewise.
746 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
748 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
750 * config/v850/v850.c (v850_reorg): Likewise.
751 * reload.c (refers_to_regno_for_reload_p): Likewise.
752 (find_equiv_reg): Likewise.
753 * reload1.c (reload_reg_reaches_end_p): Likewise.
755 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
757 * caller-save.c (add_used_regs): Use REG_NREGS instead of
759 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
760 * config/arm/arm.c (output_move_neon): Likewise.
761 (arm_attr_length_move_neon): Likewise.
762 (neon_split_vcombine): Likewise.
763 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
764 (c6x_mark_reg_written): Likewise.
765 (c6x_dwarf_register_span): Likewise.
766 * config/i386/i386.c (ix86_save_reg): Likewise.
767 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
768 (rws_access_reg): Likewise.
769 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
770 * mode-switching.c (create_pre_exit): Likewise.
771 * ree.c (combine_reaching_defs): Likewise.
772 (add_removable_extension): Likewise.
773 * regcprop.c (find_oldest_value_reg): Likewise.
774 (copyprop_hardreg_forward_1): Likewise.
775 * reload.c (reload_inner_reg_of_subreg): Likewise.
776 (push_reload): Likewise.
777 (combine_reloads): Likewise.
778 (find_dummy_reload): Likewise.
779 (reload_adjust_reg_for_mode): Likewise.
780 * reload1.c (find_reload_regs): Likewise.
781 (forget_old_reloads_1): Likewise.
782 (reload_reg_free_for_value_p): Likewise.
783 (reload_adjust_reg_for_temp): Likewise.
784 (emit_reload_insns): Likewise.
785 (delete_output_reload): Likewise.
786 * sel-sched.c (choose_best_reg_1): Likewise.
787 (choose_best_pseudo_reg): Likewise.
789 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
790 Alan Hayward <alan.hayward@arm.com>
791 David Sherwood <david.sherwood@arm.com>
793 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
794 * target.def (slow_unaligned_access): New hook.
795 * targhooks.h (default_slow_unaligned_access): Declare.
796 * targhooks.c (default_slow_unaligned_access): New function.
797 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
798 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
799 * doc/tm.texi: Regenerate.
800 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
801 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
802 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
804 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
805 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
807 (rs6000_slow_unaligned_access): New function.
808 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
809 (expand_block_compare): Likewise.
810 (expand_strn_compare): Likewise.
811 (rs6000_rtx_costs): Likewise.
812 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
813 (riscv_slow_unaligned_access): Likewise.
814 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
815 (riscv_slow_unaligned_access_p): ...this and make static.
816 (riscv_option_override): Update accordingly.
817 (riscv_slow_unaligned_access): New function.
818 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
819 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
820 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
821 (rs6000_slow_unaligned_access): New function.
822 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
823 (rs6000_rtx_costs): Likewise.
824 * config/rs6000/rs6000-string.c (expand_block_compare)
825 (expand_strn_compare): Use targetm.slow_unaligned_access instead
826 of SLOW_UNALIGNED_ACCESS.
827 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
828 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
829 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
830 of SLOW_UNALIGNED_ACCESS.
831 * expmed.c (simple_mem_bitfield_p): Likewise.
832 * expr.c (alignment_for_piecewise_move): Likewise.
833 (emit_group_load_1): Likewise.
834 (emit_group_store): Likewise.
835 (copy_blkmode_from_reg): Likewise.
836 (emit_push_insn): Likewise.
837 (expand_assignment): Likewise.
838 (store_field): Likewise.
839 (expand_expr_real_1): Likewise.
840 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
841 * lra-constraints.c (simplify_operand_subreg): Likewise.
842 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
843 * gimple-ssa-store-merging.c: Likewise in block comment at start
845 * tree-ssa-strlen.c: Include target.h.
846 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
847 of SLOW_UNALIGNED_ACCESS.
848 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
850 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
852 PR rtl-optimization/82185
853 * expmed.c (emit_store_flag_int): Only test tem if it has been
856 2017-09-12 Richard Biener <rguenther@suse.de>
859 * match.pd ((FTYPE) N CMP CST): Fix typo.
861 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
863 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
865 (mips_near_type_p): Add 'short_call' attribute as a synonym
867 * doc/extend.texi (short_call): Document new function attribute.
869 2017-09-12 Jakub Jelinek <jakub@redhat.com>
872 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
873 assertion check that in the condition.
874 (get_atomic_generic_size): Likewise. Before testing if parameter
875 has pointer type, if it has array type, call for C++
876 default_conversion to perform array-to-pointer conversion.
878 2017-09-12 Richard Biener <rguenther@suse.de>
880 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
881 for operations we cannot scalarize.
883 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
885 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
886 vectors heap vectors. Clean up comments.
887 Make visited_bbs a reference.
888 (profitable_jump_thread_path): Make GC
889 vectors heap vectors. Clean up comments.
891 (convert_and_register_jump_thread_path): Make GC vectors heap
893 (check_subpath_and_update_thread_path): Same. Clean up comments.
894 Make visited_bbs a reference.
895 (handle_phi): Abstract common code to to
896 register_jump_thread_path_if_profitable.
897 Rename VAR_BB to DEF_BB.
899 Make GC vectors heap vectors.
900 Make visited_bbs a reference.
901 (handle_assignment): Same.
902 (register_jump_thread_path_if_profitable): New.
903 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
905 Make GC vectors heap vectors. Clean up comments.
906 Make visited_bbs a reference.
907 (find_jump_threads_backwards): Make visited_bbs live in the stack.
908 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
911 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
914 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
915 words of E_DImode object are reachable by xtensa_uimm8x4 access.
917 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
919 Revert r251800 and r251799.
921 2017-09-11 Martin Jambor <mjambor@suse.cz>
924 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
925 arguments in advance.
926 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
927 use it to find predecessor edges.
928 (naive_outof_ssa): Collect vector of predecessors.
930 2017-09-08 Jason Merrill <jason@redhat.com>
932 PR c++/70029 - ICE with ref-qualifier and -flto
933 * langhooks.h (struct lang_hooks_for_types): Add
934 copy_lang_qualifiers.
935 * attribs.c (build_type_attribute_qual_variant): Use it.
936 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
938 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
939 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
941 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
944 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
945 (*mulsi3_sp64): New instruction.
946 (mulsi3): New expander.
948 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
950 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
952 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
954 * sancov.c: Include memmodel.h.
956 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
959 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
962 2017-09-07 Carl Love <cel@us.ibm.com>
964 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
965 the sldi instruction.
967 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
969 * sancov.c: Include tm_p.h.
971 2017-09-07 Jakub Jelinek <jakub@redhat.com>
974 * output.h (switch_to_other_text_partition): New declaration.
975 * varasm.c (switch_to_other_text_partition): New function.
976 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
977 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
978 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
979 to the other text partition before emitting LCL label and switch back
980 after emitting the word after it.
982 2017-09-07 Richard Biener <rguenther@suse.de>
984 * passes.def (pass_split_crit_edges): Remove instance before PRE.
985 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
986 critical edges here, after loop init.
987 (pass_data_pre): Remove PROP_no_crit_edges flags.
988 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
989 for valueization of call args to avoid leaking VN_TOP.
990 (visit_use): Assert we do not visit default defs.
991 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
992 Use error_mark_node to more easily detect leaking VN_TOP.
993 All default-defs are varying, not VN_TOP. Mark them visited.
994 (run_scc_vn): Make code match comment.
996 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
998 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
999 OPTION_MASK_FLOAT128_KEYWORD.
1000 (POWERPC_MASKS): Likewise.
1001 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1002 support for the -mfloat128-type option, and make -mfloat128
1003 default on PowerPC Linux systems. Define or undefine
1004 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1005 Define __float128 to be __ieee128 if IEEE 128-bit support is
1006 enabled, or undefine it.
1007 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1008 Delete defining __FLOAT128_TYPE__.
1009 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1010 -mfloat128-type option and make -mfloat128 default on PowerPC
1012 (TARGET_FLOAT128_TYPE): Likewise.
1013 (-mfloat128-type): Likewise.
1014 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1015 Delete the -mfloat128-type option and make -mfloat128 default on
1016 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1017 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1018 128-bit floating point is enabled. Change tests from using
1019 -mfloat128-type to -mfloat128.
1020 (rs6000_mangle_type): Use the correct mangling for the __float128
1021 type even if normal long double is restricted to 64-bits.
1022 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1024 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1025 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1026 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1028 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1029 documentation for -mfloat128.
1031 2017-09-06 Olivier Hainque <hainque@adacore.com>
1033 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1035 2017-09-06 Wish Wu <wishwu007@gmail.com>
1036 Jakub Jelinek <jakub@redhat.com>
1038 * asan.c (initialize_sanitizer_builtins): Add
1039 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1040 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1041 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1042 BT_FN_VOID_UINT64_PTR variables.
1043 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1044 (BT_FN_VOID_UINT16_UINT16): Likewise.
1045 (BT_FN_VOID_UINT32_UINT32): Likewise.
1046 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1047 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1048 (BT_FN_VOID_UINT64_PTR): Likewise.
1049 * common.opt (flag_sanitize_coverage): New variable.
1050 (fsanitize-coverage=trace-pc): Remove.
1051 (fsanitize-coverage=): Add.
1052 * flag-types.h (enum sanitize_coverage_code): New enum.
1053 * fold-const.c (fold_range_test): Disable non-short-circuit
1054 optimization if flag_sanitize_coverage.
1055 (fold_truth_andor): Likewise.
1056 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1057 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1058 (coverage_sanitizer_opts): New array.
1059 (get_closest_sanitizer_option): Add OPTS argument, handle also
1060 OPT_fsanitize_coverage_.
1061 (parse_sanitizer_options): Adjusted to also handle
1062 OPT_fsanitize_coverage_.
1063 (common_handle_option): Add OPT_fsanitize_coverage_.
1064 * sancov.c (instrument_comparison, instrument_switch): New function.
1065 (sancov_pass): Add trace-cmp support.
1066 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1067 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1068 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1069 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1070 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1071 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1072 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1073 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1074 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1075 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1077 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1079 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1080 error. Only quit immediately if parsing is complete.
1081 (BEGIN): Initialize fatal_err and parse_done.
1082 (begin fpu, end fpu): Check number of arguments.
1083 (begin arch, end arch): Likewise.
1084 (begin cpu, end cpu): Likewise.
1085 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1086 (optalias): Likewise.
1088 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1090 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1091 * config/arm/arm-isa.h: Delete. Move definitions to ...
1092 * arm-cpus.in: ... here. Use new feature and fgroup values.
1093 * config/arm/arm.c (arm_option_override): Use lower case for feature
1095 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1096 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1097 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1099 (print_isa_bits_for): New function.
1100 (gen_isa): New function.
1101 (gen_comm_data): Use print_isa_bits_for.
1102 (define feature): New keyword.
1103 (define fgroup): New keyword.
1104 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1105 (arm-isa.h): Add rule to generate file.
1106 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1107 case for feature bit names.
1109 2017-09-06 Richard Biener <rguenther@suse.de>
1111 * tree-ssa-pre.c (NECESSARY): Remove.
1112 (create_expression_by_pieces): Do not touch pass-local flags.
1113 (insert_into_preds_of_block): Likewise.
1114 (do_pre_regular_insertion): Likewise.
1115 (eliminate_insert): Likewise.
1116 (eliminate_dom_walker::before_dom_children): Likewise.
1117 (fini_eliminate): Do not look at inserted_exprs.
1118 (mark_operand_necessary): Remove.
1119 (remove_dead_inserted_code): Replace with simple work-list
1120 algorithm based on inserted_exprs and SSA uses.
1121 (pass_pre::execute): Re-order fini_eliminate and
1122 remove_dead_inserted_code.
1124 2017-09-06 Olivier Hainque <hainque@adacore.com>
1126 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1127 for VxWorks 7. Adjust surrounding comments.
1129 2017-09-06 Richard Biener <rguenther@suse.de>
1131 * gimple-ssa-strength-reduction.c
1132 (find_candidates_dom_walker::before_dom_children): Also allow
1135 2017-09-06 Richard Biener <rguenther@suse.de>
1137 PR tree-optimization/82108
1138 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1139 for gap in the non-permutation SLP case.
1141 2017-09-06 Martin Jambor <mjambor@suse.cz>
1143 PR tree-optimization/82078
1144 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1145 add_access_to_work_queue...
1146 (build_accesses_from_assign): ...here.
1147 (propagate_all_subaccesses): Make sure racc is the group
1148 representative, if there is one.
1150 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1153 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1156 2017-09-06 Richard Biener <rguenther@suse.de>
1158 * gimple-ssa-strength-reduction.c
1159 (find_candidates_dom_walker::before_doom_children): Use a
1160 type and not a mode check.
1162 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1165 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1167 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1169 2017-09-05 Jeff Law <law@redhat.com>
1171 PR tree-optimization/64910
1172 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1173 cases where we have 3 or more operands.
1175 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1178 * omp-low.c (lower_omp_for): Recompute tree invariant if
1179 gimple_omp_for_initial/final is ADDR_EXPR.
1182 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1183 into gimple val before gimplification fo the COND_EXPR.
1185 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1187 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1188 REGION_COPY argument.
1189 (thread_through_all_blocks): Remove unused argument to
1190 duplicate_thread_path.
1192 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1197 Take a scalar_mode rather than a machine_mode.
1198 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1199 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1200 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1201 (aarch64_gen_adjusted_ldpstp): Likewise.
1202 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1204 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1205 Alan Hayward <alan.hayward@arm.com>
1206 David Sherwood <david.sherwood@arm.com>
1208 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1209 Take a scalar_int_mode instead of a machine_mode.
1210 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1211 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1212 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1213 (aarch64_simd_attr_length_rglist): Delete.
1214 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1215 a scalar_int_mode instead of a machine_mode.
1216 (aarch64_add_offset): Likewise.
1217 (aarch64_internal_mov_immediate): Likewise
1218 (aarch64_add_constant_internal): Likewise.
1219 (aarch64_add_constant): Likewise.
1220 (aarch64_movw_imm): Likewise.
1221 (aarch64_rtx_arith_op_extract_p): Likewise.
1222 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1223 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1224 Remove assert that the mode isn't a vector.
1225 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1226 (aarch64_expand_mov_immediate): Update calls after above changes.
1227 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1228 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1229 (aarch64_move_imm): Likewise.
1230 (aarch64_can_const_movi_rtx_p): Likewise.
1231 (aarch64_strip_extend): Likewise.
1232 (aarch64_extr_rtx_p): Likewise.
1233 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1234 a CONST_INT when the mode parameter is VOIDmode.
1235 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1237 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1239 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1240 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1241 (bitwise_type_for_mode): Update accordingly.
1243 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1245 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1246 * stor-layout.c (mode_for_size_tree): Likewise.
1247 (mode_for_array): Update accordingly.
1248 (layout_decl): Likewise.
1249 (compute_record_mode): Likewise. Only set the mode once.
1251 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1253 * target.def (get_mask_mode): Change return type to opt_mode.
1255 * doc/tm.texi: Regenerate.
1256 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1257 * targhooks.c (default_get_mask_mode): Likewise.
1258 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1259 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1260 targetm.get_mask_mode.
1261 * tree.c (build_truth_vector_type): Likewise.
1263 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1265 * machmode.h (mode_for_vector): Return an opt_mode.
1266 * stor-layout.c (mode_for_vector): Likewise.
1267 (mode_for_int_vector): Update accordingly.
1268 (layout_type): Likewise.
1269 * config/i386/i386.c (emit_memmov): Likewise.
1270 (ix86_expand_set_or_movmem): Likewise.
1271 (ix86_expand_vector_init): Likewise.
1272 (ix86_get_mask_mode): Likewise.
1273 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1275 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1276 * expmed.c (extract_bit_field_1): Likewise.
1277 * expr.c (expand_expr_real_2): Likewise.
1278 * optabs-query.c (can_vec_perm_p): Likewise.
1279 (can_vec_mask_load_store_p): Likewise.
1280 * optabs.c (expand_vec_perm): Likewise.
1281 * targhooks.c (default_get_mask_mode): Likewise.
1282 * tree-vect-stmts.c (vectorizable_store): Likewise.
1283 (vectorizable_load): Likewise.
1284 (get_vectype_for_scalar_type_and_size): Likewise.
1286 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1288 * machmode.h (mode_for_int_vector): New function.
1289 * stor-layout.c (mode_for_int_vector): Likewise.
1290 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1291 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1292 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1293 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1294 (s390_expand_vcond): Likewise.
1296 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1298 * machmode.h (opt_machine_mode): New type.
1299 (opt_mode<T>): Allow construction from anything that can be
1301 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1302 (mode_for_size): Return an opt_machine_mode.
1303 * stor-layout.c (mode_for_size): Likewise.
1304 (mode_for_size_tree): Update call accordingly.
1305 (bitwise_mode_for_mode): Likewise.
1306 (make_fract_type): Likewise.
1307 (make_accum_type): Likewise.
1308 * caller-save.c (replace_reg_with_saved_mem): Update call
1310 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1311 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1312 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1313 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1314 * expmed.c (extract_bit_field_1): Likewise.
1315 * reload.c (get_secondary_mem): Likewise.
1316 * varasm.c (assemble_integer): Likewise.
1317 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1322 * machmode.h (decimal_float_mode_for_size): New function.
1323 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1324 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1325 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1326 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1327 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1328 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1330 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1332 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1333 (get_builtin_sync_mode): Likewise.
1334 (expand_ifn_atomic_compare_exchange): Likewise.
1335 (expand_builtin_atomic_clear): Likewise.
1336 (expand_builtin_atomic_test_and_set): Likewise.
1337 (fold_builtin_atomic_always_lock_free): Likewise.
1338 * calls.c (compute_argument_addresses): Likewise.
1339 (emit_library_call_value_1): Likewise.
1340 (store_one_arg): Likewise.
1341 * combine.c (combine_instructions): Likewise.
1342 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1343 * config/arm/arm.c (arm_function_value): Likewise.
1344 (aapcs_allocate_return_reg): Likewise.
1345 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1346 * config/i386/i386.c (construct_container): Likewise.
1347 (ix86_gimplify_va_arg): Likewise.
1348 (ix86_expand_sse_cmp): Likewise.
1349 (emit_memmov): Likewise.
1350 (emit_memset): Likewise.
1351 (expand_small_movmem_or_setmem): Likewise.
1352 (ix86_expand_pextr): Likewise.
1353 (ix86_expand_pinsr): Likewise.
1354 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1355 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1357 * config/mips/mips.c (mips_function_value_1) Likewise.
1358 (mips_block_move_straight): Likewise.
1359 (mips_expand_ins_as_unaligned_store): Likewise.
1360 * config/powerpcspe/powerpcspe.c
1361 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1362 (rs6000_darwin64_record_arg_flush): Likewise.
1363 * config/rs6000/rs6000.c
1364 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1365 (rs6000_darwin64_record_arg_flush): Likewise.
1366 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1367 (sparc_function_value_1): Likewise.
1368 * config/spu/spu.c (adjust_operand): Likewise.
1369 (spu_emit_branch_or_set): Likewise.
1370 (arith_immediate_p): Likewise.
1371 * emit-rtl.c (gen_lowpart_common): Likewise.
1372 * expr.c (expand_expr_real_1): Likewise.
1373 * function.c (assign_parm_setup_block): Likewise.
1374 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1375 * reload1.c (alter_reg): Likewise.
1376 * stor-layout.c (mode_for_vector): Likewise.
1377 (layout_type): Likewise.
1379 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1381 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1382 (spu_convert_move): Likewise.
1383 * lower-subreg.c (resolve_simple_move): Likewise.
1385 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1388 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1389 define_insn to a define_expand.
1390 (altivec_vsum2sws_direct): New define_insn.
1391 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1393 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1395 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1398 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1403 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1404 realign stack if stack alignment needed is less than incoming
1407 2017-09-05 Marek Polacek <polacek@redhat.com>
1410 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1413 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1415 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1417 2017-09-05 Richard Biener <rguenther@suse.de>
1419 PR tree-optimization/82084
1420 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1422 2017-09-05 Richard Biener <rguenther@suse.de>
1424 PR tree-optimization/82102
1425 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1427 2017-09-05 Martin Liska <mliska@suse.cz>
1429 PR tree-optimization/82032
1430 * tree-cfg.c (generate_range_test): New function.
1431 * tree-cfg.h (generate_range_test): Declared here.
1432 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1433 (cleanup_control_expr_graph): Use it.
1434 * tree-switch-conversion.c (try_switch_expansion): Remove
1436 (emit_case_nodes): Use generate_range_test.
1438 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1441 * config/i386/i386.md (*<btsc><mode>_mask): Add
1442 TARGET_USE_BT to insn constraint.
1443 (*btr<mode>_mask): Ditto.
1445 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1447 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1448 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1450 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1453 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1454 TARGET_NEON and TARGET_IWMMXT.
1455 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1456 TARGET_NEON and TARGET_IWMMXT.
1457 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1459 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1461 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1462 (ix86_rewrite_tls_address): Ditto.
1463 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1464 (ix86_rewrite_tls_address_1): Ditto.
1465 (ix86_rewrite_tls_address): Ditto.
1466 * config/i386/predicates.md (tls_address_pattern): New predicate.
1467 * config/i386/i386.md (TLS address splitter): New splitter.
1469 2017-09-04 Richard Biener <rguenther@suse.de>
1471 PR tree-optimization/82084
1472 * fold-const.h (can_native_encode_string_p): Declare.
1473 * fold-const.c (can_native_encode_string_p): Factor out from ...
1474 (native_encode_string): ... here.
1475 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1476 vectorizing stores from constants we later cannot handle.
1478 2017-09-04 Marek Polacek <polacek@redhat.com>
1481 * doc/invoke.texi: Update -Wtautological-compare documentation.
1483 2017-09-04 Jeff Law <law@redhat.com>
1485 PR tree-optimization/64910
1486 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1487 swap the first and last operand if the last is a constant.
1489 2017-09-04 Marek Polacek <polacek@redhat.com>
1492 * convert.c (do_narrow): When sanitizing signed integer overflows,
1493 bail out for signed types.
1494 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1496 2017-09-04 Richard Biener <rguenther@suse.de>
1498 PR tree-optimization/82060
1499 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1500 Move devirtualization after stmt folding and before EH/AB/noreturn
1501 cleanup to get the stmt refs canonicalized. Use a bool instead
1502 of gimple_modified_p since that doesn't work for NOPs. Schedule
1503 NOPs generated by folding for removal.
1505 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * coretypes.h (pad_direction): New enum.
1510 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1511 (FUNCTION_ARG_PADDING): Likewise.
1512 * target.def (function_arg_padding): New hook.
1513 * targhooks.h (default_function_arg_padding): Declare.
1514 * targhooks.c (default_function_arg_padding): New function.
1515 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1516 (TARGET_FUNCTION_ARG_PADDING): ...this.
1517 * doc/tm.texi: Regenerate.
1518 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1519 instead of direction.
1520 (compute_argument_addresses): Likewise.
1521 (load_register_parameters): Likewise.
1522 (emit_library_call_value_1): Likewise.
1523 (store_one_arg): Use targetm.calls.function_arg_padding instead
1524 of FUNCTION_ARG_PADDING.
1525 (must_pass_in_stack_var_size_or_pad): Likewise.
1526 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1527 (emit_group_store): Likewise.
1528 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1529 instead of FUNCTION_ARG_PADDING.
1530 (emit_push_insn): Likewise, and propagate enum change throughout
1532 * function.h (direction): Delete.
1533 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1535 * function.c (assign_parm_find_stack_rtl): Likewise.
1536 (assign_parm_setup_block_p): Likewise.
1537 (assign_parm_setup_block): Likewise.
1538 (gimplify_parameters): Likewise.
1539 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1540 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1542 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1543 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1544 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1545 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1546 (aarch64_function_arg_padding): ...this new function.
1547 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1548 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1549 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1550 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1551 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1552 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1553 (arm_pad_arg_upward): Replace with...
1554 (arm_function_arg_padding): ...this new function.
1555 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1557 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1558 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1559 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1560 (ia64_hpux_function_arg_padding): Replace with...
1561 (ia64_function_arg_padding): ...this new function. Use pad_direction
1562 instead of direction. Check for TARGET_HPUX.
1563 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1564 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1565 (iq2000_function_arg_padding): New function.
1566 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
1567 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
1568 (mips_function_arg_padding): ...this new function.
1569 (mips_pad_reg_upward): Update accordingly.
1570 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1571 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
1572 targetm.calls.function_arg_padding.
1573 (FUNCTION_ARG_PADDING): Delete.
1574 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1575 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
1576 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1577 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
1578 (nios2_block_reg_padding): Return pad_direction instead of direction.
1579 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
1580 instead of direction.
1581 (nios2_function_arg_padding): Likewise. Make static.
1582 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1583 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
1584 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
1585 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
1586 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1587 (pa_function_arg_padding): Make static. Return pad_direction instead
1589 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
1590 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1591 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
1592 instead of direction. Use targetm.calls.function_arg_padding.
1593 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
1594 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1595 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
1596 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
1597 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
1599 (function_arg_padding): Rename to...
1600 (rs6000_function_arg_padding): ...this. Make static. Return
1601 pad_direction instead of direction.
1602 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1603 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1604 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1605 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1606 instead of direction. Use targetm.calls.function_arg_padding.
1607 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1608 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1609 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1610 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1611 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1612 (function_arg_padding): Rename to...
1613 (rs6000_function_arg_padding): ...this. Make static. Return
1614 pad_direction instead of direction.
1615 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1616 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1617 * config/s390/s390.c (s390_function_arg_padding): New function.
1618 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1619 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1620 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1621 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1622 (function_arg_padding): Rename to...
1623 (sparc_function_arg_padding): ...this. Make static. Return
1624 pad_direction instead of direction.
1625 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1626 * config/spu/spu.c (spu_function_arg_padding): New function.
1627 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1628 * system.h (FUNCTION_ARG_PADDING): Poison.
1630 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1631 Alan Hayward <alan.hayward@arm.com>
1632 David Sherwood <david.sherwood@arm.com>
1634 * target.def (modes_tieable_p): New hook.
1635 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1636 (TARGET_MODES_TIEABLE_P): ...this.
1637 * doc/tm.texi.in: Regenerate.
1638 * hooks.h (hook_bool_mode_mode_true): Declare.
1639 * hooks.c (hook_bool_mode_mode_true): New function.
1640 * combine.c (subst): Use targetm.modes_tieable_p instead of
1642 * dse.c (find_shift_sequence): Likewise.
1643 * expmed.c (extract_low_bits): Likewise.
1644 * lower-subreg.c: Include target.h.
1645 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1647 * rtlanal.c (rtx_cost): Likewise.
1648 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1649 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1650 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1651 (TARGET_MODES_TIEABLE_P): Redefine.
1652 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1653 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1654 (TARGET_MODES_TIEABLE_P): Redefine.
1655 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1656 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1657 (arc_modes_tieable_p): New function.
1658 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1659 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1660 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1661 (arm_modes_tieable_p): Make static.
1662 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1663 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1664 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1665 (TARGET_MODES_TIEABLE_P): Redefine.
1666 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1667 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1668 (TARGET_MODES_TIEABLE_P): Redefine.
1669 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1670 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1671 (cr16_modes_tieable_p): New function.
1672 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1673 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1674 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1675 (TRULY_NOOP_TRUNCATION): Update comment.
1676 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1677 (TRULY_NOOP_TRUNCATION): Update comment.
1678 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1679 (frv_modes_tieable_p): New function.
1680 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1681 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1682 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1683 (TARGET_MODES_TIEABLE_P): Redefine.
1684 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1685 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1686 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1687 (TARGET_MODES_TIEABLE_P): Redefine.
1688 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1689 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1690 (ia64_modes_tieable_p): New function.
1691 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1692 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1693 (iq2000_modes_tieable_p): New function.
1694 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1695 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1696 (lm32_modes_tieable_p): New function.
1697 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1698 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1699 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1700 (TARGET_MODES_TIEABLE_P): Redefine.
1701 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1702 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1703 (m32r_modes_tieable_p): New function.
1704 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1705 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1706 (m68k_modes_tieable_p): New function.
1707 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1708 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1709 (mcore_modes_tieable_p): New function.
1710 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1711 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1713 (TARGET_MODES_TIEABLE_P): Redefine.
1714 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1715 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1716 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1717 (TARGET_MODES_TIEABLE_P): Redefine.
1718 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1719 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1720 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1721 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1722 (mn10300_modes_tieable_p): ...this and make static.
1723 (TARGET_MODES_TIEABLE_P): Redefine.
1724 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1725 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1726 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1727 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1728 (msp430_modes_tieable_p): Make static.
1729 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1730 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1731 (TARGET_MODES_TIEABLE_P): Redefine.
1732 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1733 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1734 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1735 (TARGET_MODES_TIEABLE_P): Redefine.
1736 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1737 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1738 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1739 (TARGET_MODES_TIEABLE_P): Redefine.
1740 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1741 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1742 (pdp11_modes_tieable_p): New function.
1743 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1744 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1745 (rs6000_modes_tieable_p): New function.
1746 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1747 * config/powerpcspe/powerpcspe.md: Update comment.
1748 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1749 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1750 (TARGET_MODES_TIEABLE_P): Redefine.
1751 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1752 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1753 (rl78_modes_tieable_p): New function.
1754 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1755 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1756 (rs6000_modes_tieable_p): New function.
1757 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1758 * config/rs6000/rs6000.md: Update comment.
1759 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1760 * config/rx/rx.c (rx_modes_tieable_p): New function.
1761 (TARGET_MODES_TIEABLE_P): Redefine.
1762 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1763 * config/s390/s390.c (s390_modes_tieable_p): New function.
1764 (TARGET_MODES_TIEABLE_P): Redefine.
1765 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1766 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1767 (sh_modes_tieable_p): New function.
1768 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1769 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1770 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1771 (sparc_modes_tieable_p): Make static.
1772 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1773 * config/spu/spu.c (spu_modes_tieable_p): New function.
1774 (TARGET_MODES_TIEABLE_P): Redefine.
1775 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1776 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1777 (TARGET_MODES_TIEABLE_P): Redefine.
1778 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1779 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1780 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1781 * config/v850/v850.c (v850_modes_tieable_p): New function.
1782 (TARGET_MODES_TIEABLE_P): Redefine.
1783 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1784 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1785 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1786 (visium_modes_tieable_p): New function.
1787 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1788 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1789 (xtensa_modes_tieable_p): New function.
1790 * system.h (MODES_TIEABLE_P): Poison.
1792 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1793 Alan Hayward <alan.hayward@arm.com>
1794 David Sherwood <david.sherwood@arm.com>
1796 * target.def (hard_regno_mode_ok): New hook.
1797 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1798 (TARGET_HARD_REGNO_MODE_OK): ...this.
1799 * doc/tm.texi.in: Regenerate.
1800 * hooks.h (hook_bool_uint_mode_true): Declare.
1801 * hooks.c (hook_bool_uint_mode_true): New function.
1802 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1804 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1805 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1806 instead of HARD_REGNO_MODE_OK.
1807 * caller-save.c: Include target.h.
1808 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1810 * combine.c (can_combine_p): Likewise.
1811 (combinable_i3pat): Likewise.
1812 (can_change_dest_mode): Likewise.
1813 * expr.c (init_expr_target): Likewise.
1814 (convert_move): Likewise.
1815 (convert_modes): Likewise.
1816 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1817 (setup_prohibited_mode_move_regs): Likewise.
1818 * ira.h (target_ira): Likewise.
1819 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1820 * lra-constraints.c (process_alt_operands): Likewise.
1821 (split_reg): Likewise.
1822 * recog.c (peep2_find_free_register): Likewise.
1823 * ree.c (combine_reaching_defs): Likewise.
1824 * regcprop.c (maybe_mode_change): Likewise.
1825 * reginfo.c (init_reg_sets_1): Likewise.
1826 (choose_hard_reg_mode): Likewise.
1827 (simplifiable_subregs): Likewise.
1828 * regrename.c (check_new_reg_p): Likewise.
1829 * reload.c (find_valid_class): Likewise.
1830 (find_valid_class_1): Likewise.
1831 (reload_inner_reg_of_subreg): Likewise.
1832 (push_reload): Likewise.
1833 (combine_reloads): Likewise.
1834 (find_dummy_reload): Likewise.
1835 (find_reloads): Likewise.
1836 * reload1.c (find_reg): Likewise.
1837 (set_reload_reg): Likewise.
1838 (allocate_reload_reg): Likewise.
1839 (choose_reload_regs): Likewise.
1840 (reload_adjust_reg_for_temp): Likewise.
1841 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1842 (simplify_subreg_regno): Likewise.
1843 * sel-sched.c (init_regs_for_mode): Likewise.
1844 * varasm.c (make_decl_rtl): Likewise.
1845 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1846 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1848 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1849 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1851 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1852 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1853 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1854 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1855 (arc_mode_class): Delete.
1856 (HARD_REGNO_MODE_OK): Delete.
1857 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1858 (arc_hard_regno_mode_ok): Rename old array to...
1859 (arc_hard_regno_mode_ok_modes): ...this.
1860 (arc_conditional_register_usage): Update accordingly.
1861 (arc_mode_class): Make static.
1862 (arc_hard_regno_mode_ok): New function.
1863 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1864 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1865 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1866 (arm_hard_regno_mode_ok): Make static.
1867 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1869 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1870 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1871 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1873 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1874 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1875 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1876 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1877 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1878 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1879 * config/bfin/predicates.md (valid_reg_operand): Use
1880 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1881 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1882 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1884 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1885 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1886 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1887 (cr16_hard_regno_mode_ok): Make static and return a bool.
1888 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1889 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1890 (cris_hard_regno_mode_ok): New function.
1891 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1892 (epiphany_mode_class): Delete.
1893 (HARD_REGNO_MODE_OK): Delete.
1894 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1895 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1896 (hard_regno_mode_ok): Rename to...
1897 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1898 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1899 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1901 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1902 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1903 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1904 (frv_hard_regno_mode_ok): Make static and return a bool.
1905 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1907 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1908 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1909 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1910 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1913 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1914 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1915 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1918 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1919 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1920 (ia64_hard_regno_mode_ok): New function.
1921 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1922 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1923 (iq2000_hard_regno_mode_ok): New function.
1924 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1925 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1926 (lm32_hard_regno_mode_ok): New function.
1927 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1928 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1929 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1930 instead of HARD_REGNO_MODE_OK.
1931 (m32c_hard_regno_ok): Rename to...
1932 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1933 (m32c_cannot_change_mode_class): Update accordingly.
1934 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1935 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1936 (m32r_mode_class): Delete.
1937 (HARD_REGNO_MODE_OK): Delete.
1938 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1939 (m32r_hard_regno_mode_ok): Rename to...
1940 (m32r_hard_regno_modes): ...this.
1941 (m32r_mode_class): Make static.
1942 (m32r_hard_regno_mode_ok): New function.
1943 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1944 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1945 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1946 (m68k_hard_regno_mode_ok): Make static.
1947 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1948 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1949 (mcore_hard_regno_mode_ok): New function.
1950 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1951 (HARD_REGNO_MODE_OK): Delete.
1952 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1954 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1955 (microblaze_hard_regno_mode_ok): New function.
1956 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1957 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1958 (mips_hard_regno_mode_ok): Delete.
1959 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1960 (mips_hard_regno_mode_ok_p): ...this and make static.
1961 (mips_hard_regno_mode_ok_p): Rename to...
1962 (mips_hard_regno_mode_ok_uncached): ...this.
1963 (mips_hard_regno_mode_ok): New function.
1964 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1965 of HARD_REGNO_MODE_OK.
1966 (mips_option_override): Update after above name changes.
1967 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1968 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1969 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1970 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1971 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1973 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1974 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1975 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1976 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1977 (msp430_hard_regno_mode_ok): Make static and return a bool.
1978 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1979 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1980 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1982 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1983 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1984 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1985 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1986 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1987 (PA_HARD_REGNO_MODE_OK): ...this
1988 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1989 (PA_HARD_REGNO_MODE_OK): ...this.
1990 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1991 (pa_hard_regno_mode_ok): New function.
1992 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1993 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1994 (pdp11_hard_regno_mode_ok): New function.
1995 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1996 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1998 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2001 (rs6000_hard_regno_mode_ok): Rename to...
2002 (rs6000_hard_regno_mode_ok_uncached): ...this.
2003 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2004 (rs6000_hard_regno_mode_ok): New function.
2005 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2006 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2007 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2008 (riscv_hard_regno_mode_ok): ...this and make static.
2009 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2010 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2011 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2012 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2013 (rl78_hard_regno_mode_ok): Make static and return bool.
2014 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2015 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2017 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2019 (rs6000_hard_regno_mode_ok): Rename to...
2020 (rs6000_hard_regno_mode_ok_uncached): ...this.
2021 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2022 (rs6000_hard_regno_mode_ok): New function.
2023 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2024 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2025 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2026 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2027 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2028 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2029 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2030 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2031 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2032 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2033 (sh_hard_regno_mode_ok): Make static.
2034 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2035 instead of HARD_REGNO_MODE_OK.
2036 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2037 (sparc_mode_class): Delete.
2038 (HARD_REGNO_MODE_OK): Delete.
2039 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2040 (hard_regno_mode_classes): Make static.
2041 (sparc_mode_class): Likewise.
2042 (sparc_hard_regno_mode_ok): New function.
2043 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2044 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2045 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2048 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2049 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2050 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2051 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2053 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2054 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2055 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2056 (visium_hard_regno_mode_ok): New function.
2057 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2058 instead of HARD_REGNO_MODE_OK.
2059 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2060 (HARD_REGNO_MODE_OK): Delete.
2061 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2062 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2063 (xtensa_option_override): Update accordingly.
2064 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2065 (xtensa_hard_regno_mode_ok): New function.
2066 * system.h (HARD_REGNO_MODE_OK): Poison.
2068 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2069 Alan Hayward <alan.hayward@arm.com>
2070 David Sherwood <david.sherwood@arm.com>
2072 * target.def (hard_regno_call_part_clobbered): New hook.
2073 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2074 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2075 * doc/tm.texi: Regenerate.
2076 * hooks.h (hook_bool_uint_mode_false): Declare.
2077 * hooks.c (hook_bool_uint_mode_false): New function.
2078 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2079 * cselib.c (cselib_process_insn): Use
2080 targetm.hard_regno_call_part_clobbered instead of
2081 HARD_REGNO_CALL_PART_CLOBBERED.
2082 * ira-conflicts.c (ira_build_conflicts): Likewise.
2083 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2084 * lra-constraints.c (need_for_call_save_p): Likewise.
2085 * lra-lives.c: Include target.h.
2086 (check_pseudos_live_through_calls): Use
2087 targetm.hard_regno_call_part_clobbered instead of
2088 HARD_REGNO_CALL_PART_CLOBBERED.
2089 * regcprop.c: Include target.h.
2090 (copyprop_hardreg_forward_1): Use
2091 targetm.hard_regno_call_part_clobbered instead of
2092 HARD_REGNO_CALL_PART_CLOBBERED.
2093 * reginfo.c (choose_hard_reg_mode): Likewise.
2094 * regrename.c (check_new_reg_p): Likewise.
2095 * reload.c (find_equiv_reg): Likewise.
2096 * reload1.c (emit_reload_insns): Likewise.
2097 * sched-deps.c (deps_analyze_insn): Likewise.
2098 * sel-sched.c (init_regs_for_mode): Likewise.
2099 (mark_unavailable_hard_regs): Likewise.
2100 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2101 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2102 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2104 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2105 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2106 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2108 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2110 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2111 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2112 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2114 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2115 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2116 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2118 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2119 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2121 * config/powerpcspe/powerpcspe.c
2122 (rs6000_hard_regno_call_part_clobbered): New function.
2123 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2124 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2125 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2127 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2128 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2129 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2131 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2132 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2133 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2135 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2136 Alan Hayward <alan.hayward@arm.com>
2137 David Sherwood <david.sherwood@arm.com>
2139 * rtl.h (subreg_memory_offset): Declare.
2140 * emit-rtl.c (subreg_memory_offset): New function.
2141 * expmed.c (store_bit_field_1): Use it.
2142 * expr.c (undefined_operand_subword_p): Likewise.
2143 * simplify-rtx.c (simplify_subreg): Likewise.
2145 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2147 PR rtl-optimization/57448
2150 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2151 using atomic_load pattern.
2152 (expand_atomic_store): Likewise.
2154 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2157 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2158 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2159 handling. Use replace_call_with_value with NULL instead of
2160 gsi_replace, unlink_stmt_vdef and release_defs.
2162 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2165 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2167 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2170 * rtl.h (emit_library_call_value_1): Declare.
2171 (emit_library_call): Replace declaration with a series of overloads.
2172 Remove the parameter count argument.
2173 (emit_library_call_value): Likewise.
2174 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2175 with an "rtx_mode_t *".
2176 (emit_library_call_value): Delete.
2177 (emit_library_call): Likewise.
2178 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2179 (asan_emit_allocas_unpoison): Likewise.
2180 * builtins.c (expand_builtin_powi): Likewise.
2181 (expand_asan_emit_allocas_unpoison): Likewise.
2182 * cfgexpand.c (expand_main_function): Likewise.
2183 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2184 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2185 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2186 * config/arm/arm.c (arm_trampoline_init): Likewise.
2187 (arm_call_tls_get_addr): Likewise.
2188 (arm_expand_divmod_libfunc): Likewise.
2189 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2190 (smulsi3_highpart): Likewise.
2191 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2192 (c6x_expand_compare): Likewise.
2193 (c6x_expand_movmem): Likewise.
2194 * config/frv/frv.c (frv_trampoline_init): Likewise.
2195 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2196 (ix86_expand_divmod_libfunc): Likewise.
2197 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2198 (ia64_expand_compare): Likewise.
2199 (ia64_profile_hook): Likewise.
2200 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2201 (nonlocal_goto): Likewise.
2202 (restore_stack_nonlocal): Likewise.
2203 * config/m32r/m32r.c (block_move_call): Likewise.
2204 (m32r_trampoline_init): Likewise.
2205 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2206 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2207 (m68k_call_m68k_read_tp): Likewise.
2208 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2209 (microblaze_expand_divide): Likewise.
2210 * config/mips/mips.h (mips_args): Likewise.
2211 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2212 (MIPS_ICACHE_SYNC): Likewise.
2213 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2214 (nios2_trampoline_init): Likewise.
2215 * config/pa/pa.c (hppa_tls_call): Likewise.
2216 (pa_trampoline_init): Likewise.
2217 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2218 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2219 (expand_strn_compare): Likewise.
2220 (rs6000_generate_compare): Likewise.
2221 (rs6000_expand_float128_convert): Likewise.
2222 (output_profile_hook): Likewise.
2223 (rs6000_trampoline_init): Likewise.
2224 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2225 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2226 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2227 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2228 (rs6000_generate_compare): Likewise.
2229 (rs6000_expand_float128_convert): Likewise.
2230 (output_profile_hook): Likewise.
2231 (rs6000_trampoline_init): Likewise.
2232 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2233 * config/sh/sh.c (sh_trampoline_init): Likewise.
2234 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2235 (sparc_emit_float_lib_cmp): Likewise.
2236 (sparc32_initialize_trampoline): Likewise.
2237 (sparc64_initialize_trampoline): Likewise.
2238 (sparc_profile_hook): Likewise.
2239 * config/spu/spu.c (ea_load_store): Likewise.
2240 * config/spu/spu.md (floatunssidf2): Likewise.
2241 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2242 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2243 * config/visium/visium.c (expand_block_move_4): Likewise.
2244 (expand_block_move_2): Likewise.
2245 (expand_block_move_1): Likewise.
2246 (expand_block_set_4): Likewise.
2247 (expand_block_set_2): Likewise.
2248 (expand_block_set_1): Likewise.
2249 (visium_trampoline_init): Likewise.
2250 (visium_profile_hook): Likewise.
2251 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2252 (xtensa_setup_frame_addresses): Likewise.
2253 (xtensa_trampoline_init): Likewise.
2254 * except.c (sjlj_emit_function_enter): Likewise.
2255 (sjlj_emit_function_exit): Likewise.
2256 * explow.c (allocate_dynamic_stack_space): Likewise.
2257 (probe_stack_range): Likewise.
2258 * expr.c (convert_mode_scalar): Likewise.
2259 * optabs.c (expand_binop): Likewise.
2260 (expand_twoval_binop_libfunc): Likewise.
2261 (expand_unop): Likewise.
2262 (prepare_cmp_insn): Likewise.
2263 (prepare_float_lib_cmp): Likewise.
2264 (expand_float): Likewise.
2265 (expand_fix): Likewise.
2266 (expand_fixed_convert): Likewise.
2267 (maybe_emit_sync_lock_test_and_set): Likewise.
2268 (expand_atomic_compare_and_swap): Likewise.
2269 (expand_mem_thread_fence): Likewise.
2270 (expand_atomic_fetch_op): Likewise.
2272 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2274 * doc/generic.texi (OpenACC): Adjust URL.
2275 * doc/invoke.texi (C Dialect Options): Ditto.
2277 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2279 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2280 predicate for operand 1. Add (m,<S>) constraint.
2281 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2282 Prevent memory operand 1 with register operand 2.
2284 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2286 PR rtl-optimization/82024
2287 * combine.c (try_combine): If the combination result is a PARALLEL,
2288 and we only need to retain the SET in there that would be placed
2289 at I2, check that we can place that at I3 instead, before doing so.
2291 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2294 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2296 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2297 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2300 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2301 Jeff Law <law@redhat.com>
2303 * varasm.c (bss_initializer_p): Do not put constants into .bss
2304 (categorize_decl_for_section): Handle bss_initializer_p returning
2305 false when DECL_INITIAL is NULL.
2307 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2310 * config/s390/s390.c (s390_can_inline_p): New function.
2312 2017-09-01 Jeff Law <law@redhat.com>
2314 PR tree-optimization/82052
2315 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2316 Always initialize the returned slot after a hash table miss
2317 when INSERT is true.
2319 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2321 * config/s390/s390.md (mem_signal_fence): Remove.
2322 * doc/md.texi (mem_signal_fence): Remove.
2323 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2325 * target-insns.def (mem_signal_fence): Remove.
2327 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2330 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2333 * asan.c (create_odr_indicator): Strip name encoding from assembler
2334 name before appending it after __odr_asan_.
2336 2017-09-01 Martin Liska <mliska@suse.cz>
2338 PR tree-optimization/82059
2339 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2340 frequency only when an edge is redirected.
2342 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2344 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2345 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2346 (arc_conditional_register_usage): Remove ARC600 lp_count
2348 (arc_file_start): Emit Tag_ARC_CPU_variation.
2349 (arc_can_use_doloop_p): New conditions to use ZOLs.
2350 (hwloop_fail): New function.
2351 (hwloop_optimize): Likewise.
2352 (hwloop_pattern_reg): Likewise.
2353 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2354 (arc_reorg_loops): New function, calls reorg_loops.
2355 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2356 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2358 (arc_loop_hazard): Remove function, functionality moved into
2360 (arc_hazard): Remove arc_loop_hazard call.
2361 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2362 into hwloop_optimize.
2363 (arc_label_align): Remove ZOL handling.
2364 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2365 * config/arc/arc.md (doloop_begin): Remove pattern.
2366 (doloop_begin_i): Likewise.
2367 (doloop_end_i): Likewise.
2368 (doloop_fallback): Likewise.
2369 (doloop_fallback_m): Likewise.
2370 (doloop_end): Reimplement expand.
2371 (arc_lp): New pattern for LP instruction.
2372 (loop_end): New pattern.
2373 (loop_fail): Likewise.
2374 (decrement_and_branch_until_zero): Likewise.
2375 * config/arc/arc.opt (mlpc-width): New option.
2376 * doc/invoke.texi (mlpc-width): Document option.
2378 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2380 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2381 (arc_ccfsm_advance): Fix checking for delay slots.
2382 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2384 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2386 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2388 (movhi_insn): Update store instruction constraint which are saving
2389 6-bit short immediates.
2390 (movsi_insn): Consider also short scaled load operations.
2391 (zero_extendhisi2_i): Use Usd constraint instead of T.
2392 (extendhisi2_i): Add q constraint.
2393 (arc_clzsi2): Add type and length attributes.
2394 (arc_ctzsi2): Likewise.
2395 * config/arc/constraints.md (Usc): Update constraint, the
2396 assembler can parse two relocations for a single instruction.
2398 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2400 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2401 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2403 2017-08-31 Olivier Hainque <hainque@adacore.com>
2405 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2406 match as powerpc-wrs-vxworks*.
2408 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2410 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2411 register constraint for by-element operand.
2412 (aarch64_mls_elt_merge<mode>): Likewise.
2414 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2416 * config/arc/arc.c (arc_can_follow_jump): Check for short
2419 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2421 * config.gcc: Use g.opt for arc.
2422 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2423 functionality moved to ...
2424 (legitimate_scaled_address_p): New function, ...here.
2425 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2426 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2427 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2429 (arc_override_options): Handle G option.
2430 (arc_output_pic_addr_const): Correct function definition.
2431 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2432 (arc_decl_anon_ns_mem_p): Delete.
2433 (arc_in_small_data_p): Overhaul this function to take into
2434 consideration the value given via G option.
2435 (arc_rewrite_small_data_1): Renamed and corrected old
2436 arc_rewrite_small_data function.
2437 (arc_rewrite_small_data): New function.
2438 (small_data_pattern): Don't use pic_offset_table_rtx.
2439 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2440 * config/arc/simdext.md (movmisalignv2hi): Use
2441 prepare_move_operands function.
2443 (movmisalign*): Likewise.
2444 * doc/invoke.texi (ARC options): Document -G option.
2446 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2448 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2450 * config/arc/arc.c (arc_print_operand): Output scalled address for
2451 sdata whenever is possible.
2452 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2453 load/stores are available.
2454 (compact_sda_memory_operand): Check for the alignment required by
2455 code density instructions.
2456 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2458 * config/arc/constraints.md (Usd): Update constraint.
2459 (Us0): New constraint.
2460 (Usc): Update constraint.
2462 2017-08-31 Richard Biener <rguenther@suse.de>
2465 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2468 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2470 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2471 Resize type_signature.
2473 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2474 Alan Hayward <alan.hayward@arm.com>
2475 David Sherwood <david.sherwood@arm.com>
2477 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2478 subregs whose inner modes can be stored in GPRs.
2479 (aarch64_classify_index): Likewise.
2481 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2482 Alan Hayward <alan.hayward@arm.com>
2483 David Sherwood <david.sherwood@arm.com>
2485 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2486 (V_INT_EQUIV): ...this.
2487 (v_cmp_result): Rename to...
2488 (v_int_equiv): ...this.
2489 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2490 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2491 (copysign<mode>3): Likewise.
2492 (aarch64_simd_bsl<mode>_internal): Likewise.
2493 (aarch64_simd_bsl<mode>): Likewise.
2494 (vec_cmp<mode><mode>): Likewise.
2495 (vcond<mode><mode>): Likewise.
2496 (vcond<v_cmp_mixed><mode>): Likewise.
2497 (vcondu<mode><v_cmp_mixed>): Likewise.
2498 (aarch64_cm<optab><mode>): Likewise.
2499 (aarch64_cmtst<mode>): Likewise.
2500 (aarch64_fac<optab><mode>): Likewise.
2501 (vec_perm_const<mode>): Likewise.
2502 (vcond_mask_<mode><v_cmp_result>): Rename to...
2503 (vcond_mask_<mode><v_int_equiv>): ...this.
2504 (vec_cmp<mode><v_cmp_result>): Rename to...
2505 (vec_cmp<mode><v_int_equiv>): ...this.
2507 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2508 Alan Hayward <alan.hayward@arm.com>
2509 David Sherwood <david.sherwood@arm.com>
2511 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2513 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2514 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2515 (UNSPEC_LD4_DREG): New unspecs.
2516 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2517 (aarch64_ld2<mode>_dreg_be): Replace with...
2518 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2520 (aarch64_ld3<mode>_dreg_le)
2521 (aarch64_ld3<mode>_dreg_be): Replace with...
2522 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2524 (aarch64_ld4<mode>_dreg_le)
2525 (aarch64_ld4<mode>_dreg_be): Replace with...
2526 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2529 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2531 PR tree-optimization/81987
2532 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2533 insert an initializer in a location not dominated by the stride
2536 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2538 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2539 on the entire header of the finally block in the fallthru case.
2541 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2543 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2545 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2547 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2548 rs6000_emit_move_from_cr and call renamed function.
2549 (rs6000_emit_prologue): Call renamed functions.
2550 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2551 movesi_from_cr, remove volatile CRs.
2553 2017-08-30 Jon Beniston <jon@beniston.com>
2554 Richard Biener <rguenther@suse.de>
2556 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2557 of VECTOR_MODE_P check.
2558 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2559 element vector types.
2561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2563 * df.h (df_read_modify_subreg_p): Remove in favor of...
2564 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2565 const_rtx instead of an rtx.
2566 * cprop.c (local_cprop_find_used_regs): Update accordingly.
2567 * df-problems.c (df_word_lr_mark_ref): Likewise.
2568 * ira-lives.c (mark_pseudo_reg_live): Likewise.
2569 (mark_pseudo_reg_dead): Likewise.
2570 (mark_ref_dead): Likewise.
2571 * reginfo.c (init_subregs_of_mode): Likewise.
2572 * sched-deps.c (sched_analyze_1): Likewise.
2573 * df-scan.c (df_def_record_1): Likewise.
2574 (df_uses_record): Likewise.
2575 (df_read_modify_subreg_p): Remove in favor of...
2576 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
2577 const_rtx instead of an rtx.
2579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2580 Alan Hayward <alan.hayward@arm.com>
2581 David Sherwood <david.sherwood@arm.com>
2583 * rtl.h (partial_subreg_p): New function.
2584 * caller-save.c (save_call_clobbered_regs): Use it.
2585 * calls.c (expand_call): Likewise.
2586 * combine.c (combinable_i3pat): Likewise.
2587 (simplify_set): Likewise.
2588 (make_extraction): Likewise.
2589 (make_compound_operation_int): Likewise.
2590 (gen_lowpart_or_truncate): Likewise.
2591 (force_to_mode): Likewise.
2592 (make_field_assignment): Likewise.
2593 (reg_truncated_to_mode): Likewise.
2594 (record_truncated_value): Likewise.
2595 (move_deaths): Likewise.
2596 * cse.c (record_jump_cond): Likewise.
2597 (cse_insn): Likewise.
2598 * cselib.c (cselib_lookup_1): Likewise.
2599 * expmed.c (extract_bit_field_using_extv): Likewise.
2600 * function.c (assign_parm_setup_reg): Likewise.
2601 * ifcvt.c (noce_convert_multiple_sets): Likewise.
2602 * ira-build.c (create_insn_allocnos): Likewise.
2603 * lra-coalesce.c (merge_pseudos): Likewise.
2604 * lra-constraints.c (match_reload): Likewise.
2605 (simplify_operand_subreg): Likewise.
2606 (curr_insn_transform): Likewise.
2607 * lra-lives.c (process_bb_lives): Likewise.
2608 * lra.c (new_insn_reg): Likewise.
2609 (lra_substitute_pseudo): Likewise.
2610 * regcprop.c (mode_change_ok): Likewise.
2611 (maybe_mode_change): Likewise.
2612 (copyprop_hardreg_forward_1): Likewise.
2613 * reload.c (push_reload): Likewise.
2614 (find_reloads): Likewise.
2615 (find_reloads_subreg_address): Likewise.
2616 * reload1.c (alter_reg): Likewise.
2617 (eliminate_regs_1): Likewise.
2618 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2620 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2622 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2623 back to if statements, including unpack.
2625 2017-08-30 Martin Liska <mliska@suse.cz>
2628 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2630 (func_checker::compare_asm_inputs_outputs): ... this function.
2631 (func_checker::compare_gimple_asm): Use the function to compare
2632 also ASM constrains.
2633 * ipa-icf-gimple.h: Rename the function.
2635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2636 Alan Hayward <alan.hayward@arm.com>
2637 David Sherwood <david.sherwood@arm.com>
2639 * coretypes.h (complex_mode): New type.
2640 * gdbhooks.py (build_pretty_printer): Handle it.
2641 * machmode.h (complex_mode): New class.
2642 (complex_mode::includes_p): New function.
2643 (is_complex_int_mode): Likewise.
2644 (is_complex_float_mode): Likewise.
2645 * genmodes.c (get_mode_class): Handle complex mode classes.
2646 * function.c (expand_function_end): Use is_complex_int_mode.
2648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2649 Alan Hayward <alan.hayward@arm.com>
2650 David Sherwood <david.sherwood@arm.com>
2652 * coretypes.h (scalar_mode_pod): New typedef.
2653 * gdbhooks.py (build_pretty_printer): Handle it.
2654 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2655 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2656 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2657 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2661 Alan Hayward <alan.hayward@arm.com>
2662 David Sherwood <david.sherwood@arm.com>
2664 * machmode.h (mode_for_vector): Take a scalar_mode instead
2666 * stor-layout.c (mode_for_vector): Likewise.
2667 * explow.c (promote_mode): Use as_a <scalar_mode>.
2668 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2671 Alan Hayward <alan.hayward@arm.com>
2672 David Sherwood <david.sherwood@arm.com>
2674 * target.def (preferred_simd_mode): Take a scalar_mode
2675 instead of a machine_mode.
2676 * targhooks.h (default_preferred_simd_mode): Likewise.
2677 * targhooks.c (default_preferred_simd_mode): Likewise.
2678 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2679 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2680 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2681 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2682 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2683 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2684 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2685 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2687 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2688 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2689 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2690 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2691 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2692 * doc/tm.texi: Regenerate.
2693 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2697 Alan Hayward <alan.hayward@arm.com>
2698 David Sherwood <david.sherwood@arm.com>
2700 * target.def (scalar_mode_supported_p): Take a scalar_mode
2701 instead of a machine_mode.
2702 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2703 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2704 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2705 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2706 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2707 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2708 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2709 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2710 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2711 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2712 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2713 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2714 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2715 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2717 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2718 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2719 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2720 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2721 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2723 * doc/tm.texi: Regenerate.
2725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2726 Alan Hayward <alan.hayward@arm.com>
2727 David Sherwood <david.sherwood@arm.com>
2729 * coretypes.h (opt_scalar_mode): New typedef.
2730 * gdbhooks.py (build_pretty_printers): Handle it.
2731 * machmode.h (mode_iterator::get_2xwider): Add overload for
2733 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2735 * expr.c (convert_mode_scalar): Likewise.
2736 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2737 * optabs.c (expand_float): Likewise.
2738 (expand_fix): Likewise.
2739 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2742 Alan Hayward <alan.hayward@arm.com>
2743 David Sherwood <david.sherwood@arm.com>
2745 * optabs.c (expand_float): Explicitly check for scalars before
2746 using a branching expansion.
2747 (expand_fix): Likewise.
2749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2750 Alan Hayward <alan.hayward@arm.com>
2751 David Sherwood <david.sherwood@arm.com>
2753 * expr.c (convert_mode): Split scalar handling out into...
2754 (convert_mode_scalar): ...this new function. Treat the modes
2757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2758 Alan Hayward <alan.hayward@arm.com>
2759 David Sherwood <david.sherwood@arm.com>
2761 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2763 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2766 Alan Hayward <alan.hayward@arm.com>
2767 David Sherwood <david.sherwood@arm.com>
2769 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2770 rather than a machine_mode.
2771 (fixed_from_string): Likewise.
2772 (fixed_convert): Likewise.
2773 (fixed_convert_from_int): Likewise.
2774 (fixed_convert_from_real): Likewise.
2775 (real_convert_from_fixed): Likewise.
2776 * fixed-value.c (fixed_from_double_int): Likewise.
2777 (fixed_from_string): Likewise.
2778 (fixed_convert): Likewise.
2779 (fixed_convert_from_int): Likewise.
2780 (fixed_convert_from_real): Likewise.
2781 (real_convert_from_fixed): Likewise.
2782 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2789 of separate mode class checks. Do not allow vector modes here.
2790 (immed_wide_int_const): Use as_a <scalar_mode>.
2791 * explow.c (trunc_int_for_mode): Likewise.
2792 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2793 (wi::shwi): Likewise.
2794 (wi::min_value): Likewise.
2795 (wi::max_value): Likewise.
2796 * dwarf2out.c (loc_descriptor): Likewise.
2797 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2801 Alan Hayward <alan.hayward@arm.com>
2802 David Sherwood <david.sherwood@arm.com>
2804 * tree.h (SCALAR_TYPE_MODE): New macro.
2805 * expr.c (expand_expr_addr_expr_1): Use it.
2806 (expand_expr_real_2): Likewise.
2807 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2808 (fold_convert_const_fixed_from_int): Likewise.
2809 (fold_convert_const_fixed_from_real): Likewise.
2810 (native_encode_fixed): Likewise
2811 (native_encode_complex): Likewise
2812 (native_encode_vector): Likewise.
2813 (native_interpret_fixed): Likewise.
2814 (native_interpret_real): Likewise.
2815 (native_interpret_complex): Likewise.
2816 (native_interpret_vector): Likewise.
2817 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2818 (simd_clone_adjust_argument_types): Likewise.
2819 (simd_clone_init_simd_arrays): Likewise.
2820 (simd_clone_adjust): Likewise.
2821 * stor-layout.c (layout_type): Likewise.
2822 * tree.c (build_minus_one_cst): Likewise.
2823 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2824 * tree-inline.c (estimate_move_cost): Likewise.
2825 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2826 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2827 (vectorizable_reduction): Likewise.
2828 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2829 (vect_recog_mixed_size_cond_pattern): Likewise.
2830 (check_bool_pattern): Likewise.
2831 (adjust_bool_pattern): Likewise.
2832 (search_type_for_mask_1): Likewise.
2833 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2834 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2835 (vectorizable_load): Likewise.
2836 (vectorizable_store): Likewise.
2837 * ubsan.c (ubsan_encode_value): Likewise.
2838 * varasm.c (output_constant): Likewise.
2840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2841 Alan Hayward <alan.hayward@arm.com>
2842 David Sherwood <david.sherwood@arm.com>
2844 * coretypes.h (scalar_mode): New class.
2845 * machmode.h (scalar_mode): Likewise.
2846 (scalar_mode::includes_p): New function.
2847 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2848 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2849 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2850 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2851 * expmed.c (store_bit_field_1): Likewise.
2852 (extract_bit_field_1): Likewise.
2853 * expr.c (write_complex_part): Likewise.
2854 (read_complex_part): Likewise.
2855 (emit_move_complex_push): Likewise.
2856 (expand_expr_real_2): Likewise.
2857 * function.c (assign_parm_setup_reg): Likewise.
2858 (assign_parms_unsplit_complex): Likewise.
2859 * optabs.c (expand_binop): Likewise.
2860 * rtlanal.c (subreg_get_info): Likewise.
2861 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2862 * varasm.c (output_constant_pool_2): Likewise.
2864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2865 Alan Hayward <alan.hayward@arm.com>
2866 David Sherwood <david.sherwood@arm.com>
2868 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2870 (expmed_mult_highpart_optab): Likewise.
2871 (expmed_mult_highpart): Likewise.
2873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2874 Alan Hayward <alan.hayward@arm.com>
2875 David Sherwood <david.sherwood@arm.com>
2877 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2878 instead of a machine_mode.
2879 (builtin_memset_read_str): Likewise.
2880 * builtins.c (c_readstr): Likewise.
2881 (builtin_memcpy_read_str): Likewise.
2882 (builtin_strncpy_read_str): Likewise.
2883 (builtin_memset_read_str): Likewise.
2884 (builtin_memset_gen_str): Likewise.
2885 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2886 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2887 instead of a machine_mode.
2888 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2890 (make_extraction): Likewise.
2891 (try_widen_shift_mode): Take and return scalar_int_modes instead
2893 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2894 a scalar_int_mode instead of a machine_mode.
2895 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2896 (avr_addr_space_pointer_mode): Likewise.
2897 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2898 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2899 (msp430_unwind_word_mode): Likewise.
2900 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2901 (spu_addr_space_pointer_mode): Likewise.
2902 (spu_addr_space_address_mode): Likewise.
2903 (spu_libgcc_cmp_return_mode): Likewise.
2904 (spu_libgcc_shift_count_mode): Likewise.
2905 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2906 (rl78_addr_space_pointer_mode): Likewise.
2907 (fl78_unwind_word_mode): Likewise.
2908 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2910 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2911 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2912 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2913 (mips_valid_pointer_mode): Likewise.
2914 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2915 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2916 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2918 (ft32_addr_space_address_mode): Likewise.
2919 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2920 scalar_int_mode instead of a machine_mode.
2921 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2923 (m32c_addr_space_address_mode): Likewise.
2924 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2925 (rs6000_eh_return_filter_mode): Likewise.
2926 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2927 (rs6000_eh_return_filter_mode): Likewise.
2928 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2929 (s390_libgcc_shift_count_mode): Likewise.
2930 (s390_unwind_word_mode): Likewise.
2931 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2933 * target.def (mode_rep_extended): Likewise.
2934 (valid_pointer_mode): Likewise.
2935 (addr_space.valid_pointer_mode): Likewise.
2936 (eh_return_filter_mode): Return a scalar_int_mode rather than
2938 (libgcc_cmp_return_mode): Likewise.
2939 (libgcc_shift_count_mode): Likewise.
2940 (unwind_word_mode): Likewise.
2941 (addr_space.pointer_mode): Likewise.
2942 (addr_space.address_mode): Likewise.
2943 * doc/tm.texi: Regenerate.
2944 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2946 (do_jump): Use scalar_int_mode for local variables.
2947 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2948 rather than a machine_mode.
2949 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2950 (scompare_loc_descriptor_wide): Likewise.
2951 (scompare_loc_descriptor_narrow): Likewise.
2952 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2954 * except.c (sjlj_emit_dispatch_table): Likewise.
2955 (expand_builtin_eh_copy_values): Likewise.
2956 * explow.c (convert_memory_address_addr_space_1): Likewise.
2957 Take a scalar_int_mode rather than a machine_mode.
2958 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2959 than a machine_mode.
2960 (memory_address_addr_space): Use scalar_int_mode for local variables.
2961 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2962 rather than a machine_mode.
2963 * expmed.c (mask_rtx): Likewise.
2964 (init_expmed_one_conv): Likewise.
2965 (expand_mult_highpart_adjust): Likewise.
2966 (extract_high_half): Likewise.
2967 (expmed_mult_highpart_optab): Likewise.
2968 (expmed_mult_highpart): Likewise.
2969 (expand_smod_pow2): Likewise.
2970 (expand_sdiv_pow2): Likewise.
2971 (emit_store_flag_int): Likewise.
2972 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2974 (extract_low_bits): Likewise.
2975 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2977 * expr.c (pieces_addr::adjust): Likewise.
2978 (can_store_by_pieces): Likewise.
2979 (store_by_pieces): Likewise.
2980 (clear_by_pieces_1): Likewise.
2981 (expand_expr_addr_expr_1): Likewise.
2982 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2983 (expand_expr_real_1): Likewise.
2984 (try_casesi): Likewise.
2985 * final.c (shorten_branches): Likewise.
2986 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2987 type of "mode" to machine_mode.
2988 * internal-fn.c (expand_arith_overflow_result_store): Take a
2989 scalar_int_mode rather than a machine_mode.
2990 (expand_mul_overflow): Use scalar_int_mode for local variables.
2991 * loop-doloop.c (doloop_modify): Likewise.
2992 (doloop_optimize): Likewise.
2993 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2994 than a machine_mode.
2995 (expand_doubleword_shift_condmove): Likewise.
2996 (expand_doubleword_shift): Likewise.
2997 (expand_doubleword_clz): Likewise.
2998 (expand_doubleword_popcount): Likewise.
2999 (expand_doubleword_parity): Likewise.
3000 (expand_absneg_bit): Use scalar_int_mode for local variables.
3001 (prepare_float_lib_cmp): Likewise.
3002 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3003 rather than a machine_mode.
3004 (convert_memory_address_addr_space): Likewise.
3005 (get_mode_bounds): Likewise.
3006 (get_address_mode): Return a scalar_int_mode rather than a
3008 * rtlanal.c (get_address_mode): Likewise.
3009 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3010 than a machine_mode.
3011 * targhooks.c (default_mode_rep_extended): Likewise.
3012 (default_valid_pointer_mode): Likewise.
3013 (default_addr_space_valid_pointer_mode): Likewise.
3014 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3015 than a machine_mode.
3016 (default_libgcc_cmp_return_mode): Likewise.
3017 (default_libgcc_shift_count_mode): Likewise.
3018 (default_unwind_word_mode): Likewise.
3019 (default_addr_space_pointer_mode): Likewise.
3020 (default_addr_space_address_mode): Likewise.
3021 * targhooks.h (default_eh_return_filter_mode): Likewise.
3022 (default_libgcc_cmp_return_mode): Likewise.
3023 (default_libgcc_shift_count_mode): Likewise.
3024 (default_unwind_word_mode): Likewise.
3025 (default_addr_space_pointer_mode): Likewise.
3026 (default_addr_space_address_mode): Likewise.
3027 (default_mode_rep_extended): Take a scalar_int_mode rather than
3029 (default_valid_pointer_mode): Likewise.
3030 (default_addr_space_valid_pointer_mode): Likewise.
3031 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3033 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3034 rather than a machine_mode.
3035 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3036 for local variables.
3037 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3038 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3039 than a machine_mode.
3041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3046 the mode argument to scalar_int_mode.
3047 (do_jump_by_parts_zero_rtx): Likewise.
3048 (do_jump_by_parts_equality_rtx): Likewise.
3049 (do_jump_by_parts_greater): Take a mode argument.
3050 (do_jump_by_parts_equality): Likewise.
3051 (do_jump_1): Update calls accordingly.
3053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3054 Alan Hayward <alan.hayward@arm.com>
3055 David Sherwood <david.sherwood@arm.com>
3057 * is-a.h (safe_dyn_cast): New function.
3058 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3059 (jump_table_for_label): Likewise.
3060 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3061 instead of an rtx_insn *.
3062 (shorten_branches): Use dyn_cast instead of LABEL_P and
3063 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3064 rtx_jump_table_data::get_data_mode.
3065 (final_scan_insn): Likewise.
3067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3068 Alan Hayward <alan.hayward@arm.com>
3069 David Sherwood <david.sherwood@arm.com>
3071 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3072 trying to combine a full-register integer set with a subreg
3075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3076 Alan Hayward <alan.hayward@arm.com>
3077 David Sherwood <david.sherwood@arm.com>
3079 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3080 that is always either address_mode or pointer_mode.
3082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3083 Alan Hayward <alan.hayward@arm.com>
3084 David Sherwood <david.sherwood@arm.com>
3086 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3087 when the two are known to be equal.
3089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3090 Alan Hayward <alan.hayward@arm.com>
3091 David Sherwood <david.sherwood@arm.com>
3093 * simplify-rtx.c (simplify_const_unary_operation): Use
3094 is_a <scalar_int_mode> instead of checking for a nonzero
3095 precision. Forcibly convert op_mode to a scalar_int_mode
3096 in that case. More clearly differentiate the operand and
3097 result modes and use the former when deciding what the value
3098 of a count-bits operation should be. Use is_int_mode instead
3099 of checking for a MODE_INT. Remove redundant check for whether
3100 this mode has a zero precision.
3102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3103 Alan Hayward <alan.hayward@arm.com>
3104 David Sherwood <david.sherwood@arm.com>
3106 * optabs.c (widen_leading): Change the type of the mode argument
3107 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3108 (widen_bswap): Likewise.
3109 (expand_parity): Likewise.
3110 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3111 (expand_ffs): Likewise.
3112 (epand_unop): Check for scalar integer modes before calling the
3115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3116 Alan Hayward <alan.hayward@arm.com>
3117 David Sherwood <david.sherwood@arm.com>
3119 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3121 (expand_expr_real_1): Update call accordingly.
3123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3124 Alan Hayward <alan.hayward@arm.com>
3125 David Sherwood <david.sherwood@arm.com>
3127 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3128 value_mode arguments. Use scalar_int_mode internally.
3129 (store_bit_field_1): Rename the new integer mode from imode
3130 to op0_mode and use it instead of GET_MODE (op0). Update calls
3131 to store_split_bit_field, store_bit_field_using_insv and
3132 store_fixed_bit_field.
3133 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3134 Use scalar_int_mode internally. Use a bit count rather than a mode
3135 when calculating the largest bit size for get_best_mode.
3136 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3137 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3138 Remove assertion that OP0 has a scalar integer mode.
3139 (store_split_bit_field): Add op0_mode and value_mode arguments.
3140 Update calls to extract_fixed_bit_field.
3141 (extract_bit_field_using_extv): Add an op0_mode argument.
3142 Use scalar_int_mode internally.
3143 (extract_bit_field_1): Rename the new integer mode from imode to
3144 op0_mode and use it instead of GET_MODE (op0). Update calls to
3145 extract_split_bit_field, extract_bit_field_using_extv and
3146 extract_fixed_bit_field.
3147 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3148 to extract_split_bit_field and extract_fixed_bit_field_1.
3149 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3150 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3152 (extract_split_bit_field): Add an op0_mode argument. Update call
3153 to extract_fixed_bit_field.
3155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3160 * explow.c (hard_function_value): Likewise.
3161 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3162 convert_to_mode call outside the loop.
3163 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3164 for the mode iterator. Require the mode specified by max_pieces
3166 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3168 (copy_blkmode_to_reg): Likewise.
3169 (set_storage_via_setmem): Likewise.
3170 * optabs.c (prepare_cmp_insn): Likewise.
3171 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3172 * stor-layout.c (finish_bitfield_representative): Likewise.
3174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3179 * expr.c (convert_move): Use them.
3180 (convert_modes): Likewise.
3181 (store_expr_with_bounds): Likewise.
3183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3184 Alan Hayward <alan.hayward@arm.com>
3185 David Sherwood <david.sherwood@arm.com>
3187 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3188 parameter for the mode of "x". Remove the "known_x", "known_mode"
3189 and "known_ret" arguments. Change the type of the mode argument
3191 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3192 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3193 (reg_num_sign_bit_copies_for_combine): Likewise.
3194 * rtlanal.c (nonzero_bits1): Likewise.
3195 (num_sign_bit_copies1): Likewise.
3196 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3197 (reg_num_sign_bit_copies_general): Likewise.
3198 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3199 (reg_nonzero_bits_general): Likewise.
3201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3206 than in subroutines. Return 1 for non-integer modes.
3207 (cached_num_sign_bit_copies): Change the type of the mode parameter
3209 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3210 classes. Handle CONST_INT_P first and then check whether X also
3211 has a scalar integer mode. Check the same thing for inner registers
3212 of a SUBREG and for values that are being extended or truncated.
3214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3215 Alan Hayward <alan.hayward@arm.com>
3216 David Sherwood <david.sherwood@arm.com>
3218 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3219 in subroutines. Return the mode mask for non-integer modes.
3220 (cached_nonzero_bits): Change the type of the mode parameter
3222 (nonzero_bits1): Likewise. Remove early exit for other mode
3223 classes. Handle CONST_INT_P first and then check whether X
3224 also has a scalar integer mode.
3226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3227 Alan Hayward <alan.hayward@arm.com>
3228 David Sherwood <david.sherwood@arm.com>
3230 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3231 Return a scalar_int_mode and assert that the size is greater than
3233 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3234 (op_by_pieces_d::op_by_pieces_d): Likewise.
3235 (op_by_pieces_d::run): Likewise.
3236 (can_store_by_pieces): Likewise.
3238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3239 Alan Hayward <alan.hayward@arm.com>
3240 David Sherwood <david.sherwood@arm.com>
3242 * combine.c (extract_left_shift): Add a mode argument and update
3244 (make_compound_operation_int): Change the type of the mode parameter
3245 to scalar_int_mode and update the call to extract_left_shift.
3247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3248 Alan Hayward <alan.hayward@arm.com>
3249 David Sherwood <david.sherwood@arm.com>
3251 * combine.c (simplify_and_const_int): Change the type of the mode
3252 parameter to scalar_int_mode.
3253 (simplify_and_const_int_1): Likewise. Update recursive call.
3255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3256 Alan Hayward <alan.hayward@arm.com>
3257 David Sherwood <david.sherwood@arm.com>
3259 * combine.c (simplify_compare_const): Check that the mode is a
3260 scalar_int_mode (rather than VOIDmode) before testing its
3262 (simplify_comparison): Move COMPARISON_P handling out of the
3263 loop and restrict the latter part of the loop to scalar_int_modes.
3264 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3265 and when considering SUBREG_REGs. Use is_int_mode instead of
3266 checking GET_MODE_CLASS against MODE_INT.
3268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3269 Alan Hayward <alan.hayward@arm.com>
3270 David Sherwood <david.sherwood@arm.com>
3272 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3273 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3274 shift_unit_mode and for modes involved in scalar shifts.
3276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3277 Alan Hayward <alan.hayward@arm.com>
3278 David Sherwood <david.sherwood@arm.com>
3280 * combine.c (force_int_to_mode): New function, split out from...
3281 (force_to_mode): ...here. Keep xmode up-to-date and use it
3282 instead of GET_MODE (x).
3284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3285 Alan Hayward <alan.hayward@arm.com>
3286 David Sherwood <david.sherwood@arm.com>
3288 * optabs-query.h (extraction_insn::struct_mode): Change type to
3289 opt_scalar_int_mode and update comment.
3290 (extraction_insn::field_mode): Change type to scalar_int_mode.
3291 (extraction_insn::pos_mode): Likewise.
3292 * combine.c (make_extraction): Update accordingly.
3293 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3294 (get_optab_extraction_insn): Likewise.
3295 * recog.c (simplify_while_replacing): Likewise.
3296 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3297 parameter to opt_scalar_int_mode.
3299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3300 Alan Hayward <alan.hayward@arm.com>
3301 David Sherwood <david.sherwood@arm.com>
3303 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3304 to a scalar_int_mode instead of a machine_mode.
3305 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3306 (get_best_mode): Return a boolean and use a pointer argument to store
3307 the selected mode. Replace the limit mode parameter with a bit limit.
3308 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3309 for the values returned by bit_field_mode_iterator::next_mode.
3310 (store_bit_field): Update call to get_best_mode.
3311 (store_fixed_bit_field): Likewise.
3312 (extract_fixed_bit_field): Likewise.
3313 * expr.c (optimize_bitfield_assignment_op): Likewise.
3314 * fold-const.c (optimize_bit_field_compare): Likewise.
3315 (fold_truth_andor_1): Likewise.
3316 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3317 Update for new type of m_mode.
3318 (get_best_mode): As above.
3320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3321 Alan Hayward <alan.hayward@arm.com>
3322 David Sherwood <david.sherwood@arm.com>
3324 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3325 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3326 (store_bit_field): Check is_a <scalar_int_mode> before calling
3327 strict_volatile_bitfield_p.
3328 (extract_bit_field): Likewise.
3330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3331 Alan Hayward <alan.hayward@arm.com>
3332 David Sherwood <david.sherwood@arm.com>
3334 * target.def (cstore_mode): Return a scalar_int_mode.
3335 * doc/tm.texi: Regenerate.
3336 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3337 * targhooks.h (default_cstore_mode): Likewise.
3338 * targhooks.c (default_cstore_mode): Likewise, using a forced
3340 * expmed.c (emit_cstore): Expect the target of the cstore to be
3343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3344 Alan Hayward <alan.hayward@arm.com>
3345 David Sherwood <david.sherwood@arm.com>
3347 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3349 (niter_desc): Likewise mode.
3350 (iv_analyze): Add a mode parameter.
3352 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3353 and change its type to scalar_int_mode.
3354 * loop-iv.c: Update commentary at head of file.
3355 (iv_constant): Pass the mode paraeter before the rtx it describes
3356 and change its type to scalar_int_mode. Remove VOIDmode handling.
3357 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3358 (iv_extend): Likewise.
3359 (shorten_into_mode): Likewise.
3360 (iv_add): Use scalar_int_mode.
3361 (iv_mult): Likewise.
3362 (iv_shift): Likewise.
3363 (canonicalize_iv_subregs): Likewise.
3364 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3365 it describes and change its mode to scalar_int_mode. Also change
3366 the type of the returned inner_mode to scalar_int_mode.
3367 (get_biv_step): Likewise, turning outer_mode from a pointer
3368 into a direct parameter. Update call to get_biv_step_1.
3369 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3370 iv_constant and get_biv_step.
3371 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3372 and change its type to scalar_int_mode. Don't initialise iv->mode
3373 to VOIDmode and remove later checks for its still being VOIDmode.
3374 Update calls to iv_analyze_op and iv_analyze_expr. Check
3375 is_a <scalar_int_mode> when changing the mode under consideration.
3376 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3377 Update call to iv_analyze_expr.
3378 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3379 inner register is not also a scalar_int_mode. Update call to
3381 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3382 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3383 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3384 separate mode class checks. Update calls to iv_analyze. Remove
3385 fix-up of VOIDmodes after iv_analyze_biv.
3386 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3387 don't have a scalar_int_mode. Update call to biv_p.
3389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3390 Alan Hayward <alan.hayward@arm.com>
3391 David Sherwood <david.sherwood@arm.com>
3393 * cfgexpand.c (convert_debug_memory_address): Use
3394 as_a <scalar_int_mode>.
3395 * combine.c (expand_compound_operation): Likewise.
3396 (make_extraction): Likewise.
3397 (change_zero_ext): Likewise.
3398 (simplify_comparison): Likewise.
3399 * cse.c (cse_insn): Likewise.
3400 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3401 (mem_loc_descriptor): Likewise.
3402 (loc_descriptor): Likewise.
3403 * expmed.c (init_expmed_one_mode): Likewise.
3404 (synth_mult): Likewise.
3405 (emit_store_flag_1): Likewise.
3406 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3407 of a comparison with size.
3408 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3409 (reduce_to_bit_field_precision): Likewise.
3410 * function.c (expand_function_end): Likewise.
3411 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3412 * loop-doloop.c (doloop_modify): Likewise.
3413 * optabs.c (expand_binop): Likewise.
3414 (expand_unop): Likewise.
3415 (expand_copysign_absneg): Likewise.
3416 (prepare_cmp_insn): Likewise.
3417 (maybe_legitimize_operand): Likewise.
3418 * recog.c (const_scalar_int_operand): Likewise.
3419 * rtlanal.c (get_address_mode): Likewise.
3420 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3421 (simplify_cond_clz_ctz): Likewise.
3422 * tree-nested.c (get_nl_goto_field): Likewise.
3423 * tree.c (build_vector_type_for_mode): Likewise.
3424 * var-tracking.c (use_narrower_mode): Likewise.
3426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3427 Alan Hayward <alan.hayward@arm.com>
3428 David Sherwood <david.sherwood@arm.com>
3430 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3431 * builtins.c (expand_builtin_signbit): Use it.
3432 * cfgexpand.c (expand_debug_expr): Likewise.
3433 * dojump.c (do_jump): Likewise.
3434 (do_compare_and_jump): Likewise.
3435 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3436 * expmed.c (make_tree): Likewise.
3437 * expr.c (expand_expr_real_2): Likewise.
3438 (expand_expr_real_1): Likewise.
3439 (try_casesi): Likewise.
3440 * fold-const-call.c (fold_const_call_ss): Likewise.
3441 * fold-const.c (unextend): Likewise.
3442 (extract_muldiv_1): Likewise.
3443 (fold_single_bit_test): Likewise.
3444 (native_encode_int): Likewise.
3445 (native_encode_string): Likewise.
3446 (native_interpret_int): Likewise.
3447 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3448 * internal-fn.c (expand_addsub_overflow): Likewise.
3449 (expand_neg_overflow): Likewise.
3450 (expand_mul_overflow): Likewise.
3451 (expand_arith_overflow): Likewise.
3452 * match.pd: Likewise.
3453 * stor-layout.c (layout_type): Likewise.
3454 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3455 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3456 * tree-ssanames.c (get_range_info): Likewise.
3457 * tree-switch-conversion.c (array_value_type) Likewise.
3458 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3459 (vect_recog_divmod_pattern): Likewise.
3460 (vect_recog_mixed_size_cond_pattern): Likewise.
3461 * tree-vrp.c (extract_range_basic): Likewise.
3462 (simplify_float_conversion_using_ranges): Likewise.
3463 * tree.c (int_fits_type_p): Likewise.
3464 * ubsan.c (instrument_bool_enum_load): Likewise.
3465 * varasm.c (mergeable_string_section): Likewise.
3466 (narrowing_initializer_constant_valid_p): Likewise.
3467 (output_constant): Likewise.
3469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3470 Alan Hayward <alan.hayward@arm.com>
3471 David Sherwood <david.sherwood@arm.com>
3473 * machmode.h (NARROWEST_INT_MODE): New macro.
3474 * expr.c (alignment_for_piecewise_move): Use it instead of
3475 GET_CLASS_NARROWEST_MODE (MODE_INT).
3476 (push_block): Likewise.
3477 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3479 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3482 Alan Hayward <alan.hayward@arm.com>
3483 David Sherwood <david.sherwood@arm.com>
3485 * postreload.c (move2add_valid_value_p): Change the type of the
3486 mode parameter to scalar_int_mode.
3487 (move2add_use_add2_insn): Add a mode parameter and use it instead
3489 (move2add_use_add3_insn): Likewise.
3490 (reload_cse_move2add): Update accordingly.
3492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3493 Alan Hayward <alan.hayward@arm.com>
3494 David Sherwood <david.sherwood@arm.com>
3496 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3498 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3499 * optabs.c (expand_unop): Likewise.
3501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3502 Alan Hayward <alan.hayward@arm.com>
3503 David Sherwood <david.sherwood@arm.com>
3505 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3506 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3507 (popcount_loc_descriptor): Likewise.
3508 (bswap_loc_descriptor): Likewise.
3509 (rotate_loc_descriptor): Likewise.
3510 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3511 calling the functions above.
3513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3514 Alan Hayward <alan.hayward@arm.com>
3515 David Sherwood <david.sherwood@arm.com>
3517 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3519 (try_combine): Likewise.
3520 (simplify_if_then_else): Likewise.
3521 * cse.c (cse_insn): Likewise.
3522 * dwarf2out.c (mem_loc_descriptor): Likewise.
3523 * emit-rtl.c (gen_lowpart_common): Likewise.
3524 * simplify-rtx.c (simplify_truncation): Likewise.
3525 (simplify_binary_operation_1): Likewise.
3526 (simplify_const_relational_operation): Likewise.
3527 (simplify_ternary_operation): Likewise.
3528 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3531 Alan Hayward <alan.hayward@arm.com>
3532 David Sherwood <david.sherwood@arm.com>
3534 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3535 * reload.c (push_reload): Likewise.
3536 (find_reloads): Likewise.
3538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3539 Alan Hayward <alan.hayward@arm.com>
3540 David Sherwood <david.sherwood@arm.com>
3542 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3543 (make_compound_operation_int): Likewise.
3544 (change_zero_ext): Likewise.
3545 * expr.c (convert_move): Likewise.
3546 (convert_modes): Likewise.
3547 * fwprop.c (forward_propagate_subreg): Likewise.
3548 * loop-iv.c (get_biv_step_1): Likewise.
3549 * optabs.c (widen_operand): Likewise.
3550 * postreload.c (move2add_valid_value_p): Likewise.
3551 * recog.c (simplify_while_replacing): Likewise.
3552 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3553 (simplify_binary_operation_1): Likewise. Remove redundant
3554 mode equality check.
3556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3557 Alan Hayward <alan.hayward@arm.com>
3558 David Sherwood <david.sherwood@arm.com>
3560 * combine.c (combine_simplify_rtx): Add checks for
3561 is_a <scalar_int_mode>.
3562 (simplify_if_then_else): Likewise.
3563 (make_field_assignment): Likewise.
3564 (simplify_comparison): Likewise.
3565 * ifcvt.c (noce_try_bitop): Likewise.
3566 * loop-invariant.c (canonicalize_address_mult): Likewise.
3567 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3570 Alan Hayward <alan.hayward@arm.com>
3571 David Sherwood <david.sherwood@arm.com>
3573 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
3574 is_a <scalar_int_mode> instead of != BLKmode.
3576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3577 Alan Hayward <alan.hayward@arm.com>
3578 David Sherwood <david.sherwood@arm.com>
3580 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
3581 instead of != VOIDmode.
3582 * combine.c (if_then_else_cond): Likewise.
3583 (change_zero_ext): Likewise.
3584 * dwarf2out.c (mem_loc_descriptor): Likewise.
3585 (loc_descriptor): Likewise.
3586 * rtlanal.c (canonicalize_condition): Likewise.
3587 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
3589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3590 Alan Hayward <alan.hayward@arm.com>
3591 David Sherwood <david.sherwood@arm.com>
3593 * simplify-rtx.c (simplify_binary_operation_1): Use
3594 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
3596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * wide-int.h (int_traits<unsigned char>) New class.
3601 (int_traits<unsigned short>) Likewise.
3602 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
3603 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3605 * combine.c (set_nonzero_bits_and_sign_copies): Use
3606 is_a <scalar_int_mode>.
3607 (find_split_point): Likewise.
3608 (combine_simplify_rtx): Likewise.
3609 (simplify_logical): Likewise.
3610 (expand_compound_operation): Likewise.
3611 (expand_field_assignment): Likewise.
3612 (make_compound_operation): Likewise.
3613 (extended_count): Likewise.
3614 (change_zero_ext): Likewise.
3615 (simplify_comparison): Likewise.
3616 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3617 (ucompare_loc_descriptor): Likewise.
3618 (minmax_loc_descriptor): Likewise.
3619 (mem_loc_descriptor): Likewise.
3620 (loc_descriptor): Likewise.
3621 * expmed.c (init_expmed_one_mode): Likewise.
3622 * lra-constraints.c (lra_constraint_offset): Likewise.
3623 * optabs.c (prepare_libcall_arg): Likewise.
3624 * postreload.c (move2add_note_store): Likewise.
3625 * reload.c (operands_match_p): Likewise.
3626 * rtl.h (load_extend_op): Likewise.
3627 * rtlhooks.c (gen_lowpart_general): Likewise.
3628 * simplify-rtx.c (simplify_truncation): Likewise.
3629 (simplify_unary_operation_1): Likewise.
3630 (simplify_binary_operation_1): Likewise.
3631 (simplify_const_binary_operation): Likewise.
3632 (simplify_const_relational_operation): Likewise.
3633 (simplify_subreg): Likewise.
3634 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3635 * var-tracking.c (adjust_mems): Likewise.
3636 (prepare_call_arguments): Likewise.
3638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3639 Alan Hayward <alan.hayward@arm.com>
3640 David Sherwood <david.sherwood@arm.com>
3642 * machmode.h (is_int_mode): New fuction.
3643 * combine.c (find_split_point): Use it.
3644 (combine_simplify_rtx): Likewise.
3645 (simplify_if_then_else): Likewise.
3646 (simplify_set): Likewise.
3647 (simplify_shift_const_1): Likewise.
3648 (simplify_comparison): Likewise.
3649 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3650 * cse.c (notreg_cost): Likewise.
3651 (cse_insn): Likewise.
3652 * cselib.c (cselib_lookup_1): Likewise.
3653 * dojump.c (do_jump_1): Likewise.
3654 (do_compare_rtx_and_jump): Likewise.
3655 * dse.c (get_call_args): Likewise.
3656 * dwarf2out.c (rtl_for_decl_init): Likewise.
3657 (native_encode_initializer): Likewise.
3658 * expmed.c (emit_store_flag_1): Likewise.
3659 (emit_store_flag): Likewise.
3660 * expr.c (convert_modes): Likewise.
3661 (store_field): Likewise.
3662 (expand_expr_real_1): Likewise.
3663 * fold-const.c (fold_read_from_constant_string): Likewise.
3664 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3665 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3666 * optabs.c (expand_binop): Likewise.
3667 (expand_unop): Likewise.
3668 (expand_abs_nojump): Likewise.
3669 (expand_one_cmpl_abs_nojump): Likewise.
3670 * simplify-rtx.c (mode_signbit_p): Likewise.
3671 (val_signbit_p): Likewise.
3672 (val_signbit_known_set_p): Likewise.
3673 (val_signbit_known_clear_p): Likewise.
3674 (simplify_relational_operation_1): Likewise.
3675 * tree.c (vector_type_mode): Likewise.
3677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3678 Alan Hayward <alan.hayward@arm.com>
3679 David Sherwood <david.sherwood@arm.com>
3681 * machmode.h (smallest_mode_for_size): Fix formatting.
3682 (smallest_int_mode_for_size): New function.
3683 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3684 instead of smallest_mode_for_size.
3685 * combine.c (make_extraction): Likewise.
3686 * config/arc/arc.c (arc_expand_movmem): Likewise.
3687 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3688 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3689 * config/s390/s390.c (s390_expand_insv): Likewise.
3690 * config/sparc/sparc.c (assign_int_registers): Likewise.
3691 * config/spu/spu.c (spu_function_value): Likewise.
3692 (spu_function_arg): Likewise.
3693 * coverage.c (get_gcov_type): Likewise.
3694 (get_gcov_unsigned_t): Likewise.
3695 * dse.c (find_shift_sequence): Likewise.
3696 * expmed.c (store_bit_field_1): Likewise.
3697 * expr.c (convert_move): Likewise.
3698 (store_field): Likewise.
3699 * internal-fn.c (expand_arith_overflow): Likewise.
3700 * optabs-query.c (get_best_extraction_insn): Likewise.
3701 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3702 * stor-layout.c (layout_type): Likewise.
3703 (initialize_sizetypes): Likewise.
3704 * targhooks.c (default_get_mask_mode): Likewise.
3705 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3708 Alan Hayward <alan.hayward@arm.com>
3709 David Sherwood <david.sherwood@arm.com>
3711 * machmode.h (opt_mode::else_blk): New function.
3712 (int_mode_for_mode): Declare.
3713 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3714 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3716 * cfgexpand.c (expand_debug_expr): Likewise.
3717 * combine.c (gen_lowpart_or_truncate): Likewise.
3718 (gen_lowpart_for_combine): Likewise.
3719 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3720 * config/avr/avr.c (avr_to_int_mode): Likewise.
3721 (avr_out_plus_1): Likewise.
3722 (avr_out_plus): Likewise.
3723 (avr_out_round): Likewise.
3724 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3725 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3726 (s390_expand_vcond): Likewise.
3727 * config/spu/spu.c (spu_split_immediate): Likewise.
3728 (spu_expand_mov): Likewise.
3729 * dse.c (get_stored_val): Likewise.
3730 * expmed.c (store_bit_field_1): Likewise.
3731 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3733 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3734 (extract_low_bits): Likewise.
3735 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3736 handling rather than repeating the check.
3737 (emit_group_store): Likewise.
3738 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3739 * optabs.c (expand_absneg_bit): Likewise.
3740 (expand_copysign_absneg): Likewise.
3741 (expand_copysign_bit): Likewise.
3742 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3743 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3744 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3745 * var-tracking.c (prepare_call_arguments): Likewise.
3746 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3747 int_mode_for_mode instead of mode_for_size.
3748 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3751 Alan Hayward <alan.hayward@arm.com>
3752 David Sherwood <david.sherwood@arm.com>
3754 * machmode.h (int_mode_for_size): New function.
3755 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3756 instead of mode_for_size.
3757 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3759 * combine.c (expand_field_assignment): Use int_mode_for_size
3760 instead of mode_for_size.
3761 (make_extraction): Likewise.
3762 (simplify_shift_const_1): Likewise.
3763 (simplify_comparison): Likewise.
3764 * dojump.c (do_jump): Likewise.
3765 * dwarf2out.c (mem_loc_descriptor): Likewise.
3766 * emit-rtl.c (init_derived_machine_modes): Likewise.
3767 * expmed.c (flip_storage_order): Likewise.
3768 (convert_extracted_bit_field): Likewise.
3769 * expr.c (copy_blkmode_from_reg): Likewise.
3770 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3771 * internal-fn.c (expand_mul_overflow): Likewise.
3772 * lower-subreg.c (simple_move): Likewise.
3773 * optabs-libfuncs.c (init_optabs): Likewise.
3774 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3775 * tree.c (vector_type_mode): Likewise.
3776 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3777 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3778 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3779 * tree-vect-stmts.c (vectorizable_load): Likewise.
3780 (vectorizable_store): Likewise.
3782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3783 Alan Hayward <alan.hayward@arm.com>
3784 David Sherwood <david.sherwood@arm.com>
3786 * coretypes.h (pod_mode): New type.
3787 (scalar_int_mode_pod): New typedef.
3788 * machmode.h (pod_mode): New class.
3789 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3790 * genmodes.c (emit_mode_int_n): Update accordingly.
3791 * lower-subreg.h (target_lower_subreg): Change type to
3792 scalar_int_mode_pod.
3793 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3794 scalar_int_mode_pod.
3796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3797 Alan Hayward <alan.hayward@arm.com>
3798 David Sherwood <david.sherwood@arm.com>
3800 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3801 machine_mode to scalar_int_mode.
3802 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3803 (rs6000_option_override_internal): Remove cast to int.
3804 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3805 machine_mode to scalar_int_mode.
3806 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3807 (rs6000_option_override_internal): Remove cast to int.
3808 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3809 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3811 * config/s390/s390.c (s390_expand_builtin): Likewise.
3812 * coretypes.h (scalar_int_mode): New type.
3813 (opt_scalar_int_mode): New typedef.
3814 * machmode.h (scalar_int_mode): New class.
3815 (scalar_int_mode::includes_p): New function.
3816 (byte_mode): Change type to scalar_int_mode.
3817 (word_mode): Likewise.
3818 (ptr_mode): Likewise.
3819 * emit-rtl.c (byte_mode): Likewise.
3820 (word_mode): Likewise.
3821 (ptr_mode): Likewise.
3822 (init_derived_machine_modes): Update accordingly.
3823 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3824 and MODE_PARTIAL_INT.
3825 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3826 opt_scalar_int_mode.
3828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3829 Alan Hayward <alan.hayward@arm.com>
3830 David Sherwood <david.sherwood@arm.com>
3832 * target.def (libgcc_floating_mode_supported_p): Take a
3834 * doc/tm.texi: Regenerate.
3835 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3837 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3838 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3842 Alan Hayward <alan.hayward@arm.com>
3843 David Sherwood <david.sherwood@arm.com>
3845 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3846 * doc/tm.texi: Regenerate.
3847 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3848 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3849 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3850 * targhooks.h (default_floatn_mode): Likewise.
3851 * targhooks.c (default_floatn_mode): Likewise.
3852 * tree.c (build_common_tree_nodes): Update accordingly.
3854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3855 Alan Hayward <alan.hayward@arm.com>
3856 David Sherwood <david.sherwood@arm.com>
3858 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3859 (mode_iterator::iterate_p): Likewise.
3860 (mode_iterator::get_wider): Likewise.
3861 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3864 Alan Hayward <alan.hayward@arm.com>
3865 David Sherwood <david.sherwood@arm.com>
3867 * coretypes.h (opt_scalar_float_mode): New typedef.
3868 * machmode.h (float_mode_for_size): New function.
3869 * emit-rtl.c (double_mode): Delete.
3870 (init_emit_once): Use float_mode_for_size.
3871 * stor-layout.c (layout_type): Likewise.
3872 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3875 Alan Hayward <alan.hayward@arm.com>
3876 David Sherwood <david.sherwood@arm.com>
3878 * output.h (assemble_real): Take a scalar_float_mode.
3879 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3880 * config/arm/arm.md (consttable_4): Likewise.
3881 (consttable_8): Likewise.
3882 (consttable_16): Likewise.
3883 * config/mips/mips.md (consttable_float): Likewise.
3884 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3885 * varasm.c (assemble_real): Take a scalar_float_mode.
3886 (output_constant_pool_2): Update accordingly.
3887 (output_constant): Likewise.
3889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3890 Alan Hayward <alan.hayward@arm.com>
3891 David Sherwood <david.sherwood@arm.com>
3893 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3894 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3895 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3896 (native_encode_real): Likewise.
3897 (native_interpret_real): Likewise.
3898 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3899 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3902 Alan Hayward <alan.hayward@arm.com>
3903 David Sherwood <david.sherwood@arm.com>
3905 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3906 <scalar_float_mode>. Simplify.
3907 (gen_extend_conv_libfunc): Likewise.
3909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3910 Alan Hayward <alan.hayward@arm.com>
3911 David Sherwood <david.sherwood@arm.com>
3913 * coretypes.h (scalar_float_mode): New type.
3914 * machmode.h (mode_traits::from_int): Use machine_mode if
3915 USE_ENUM_MODES is defined.
3916 (is_a): New function.
3918 (dyn_cast): Likewise.
3919 (scalar_float_mode): New class.
3920 (scalar_float_mode::includes_p): New function.
3921 (is_float_mode): Likewise.
3922 * gdbhooks.py (MachineModePrinter): New class.
3923 (build_pretty_printer): Use it for scalar_float_mode.
3924 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3925 (format_helper::format_helper): Turn into a template.
3926 * genmodes.c (get_mode_class): New function.
3927 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3928 or machine_mode if none.
3929 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3930 as_a <scalar_float_mode>.
3931 * dwarf2out.c (mem_loc_descriptor): Likewise.
3932 (insert_float): Likewise.
3933 (add_const_value_attribute): Likewise.
3934 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3935 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3936 (expand_unop): Update accordingly.
3937 (expand_abs_nojump): Likewise.
3938 (expand_copysign_absneg): Take a scalar_float_mode.
3939 (expand_copysign_bit): Likewise.
3940 (expand_copysign): Update accordingly.
3942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3943 Alan Hayward <alan.hayward@arm.com>
3944 David Sherwood <david.sherwood@arm.com>
3946 * coretypes.h (opt_mode): New class.
3947 * machmode.h (opt_mode): Likewise.
3948 (opt_mode::else_void): New function.
3949 (opt_mode::require): Likewise.
3950 (opt_mode::exists): Likewise.
3951 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3952 (GET_MODE_2XWIDER_MODE): Likewise.
3953 (mode_iterator::get_wider): Update accordingly.
3954 (mode_iterator::get_2xwider): Likewise.
3955 (mode_iterator::get_known_wider): Likewise, turning into a template.
3956 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3957 forcing a wider mode to exist.
3958 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3959 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3960 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3961 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3962 * lower-subreg.c (init_lower_subreg): Likewise.
3963 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3964 on the final iteration.
3965 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3966 a wider mode exists before asking for a move pattern.
3967 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3968 forcing a wider mode to exist.
3969 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3970 returning false if no such mode exists.
3971 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3972 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3973 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3974 Avoid checking for a MODE_INT if we already know the mode is not a
3976 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3977 forcing a wider mode to exist.
3978 (expmed_mult_highpart_optab): Likewise.
3979 (expmed_mult_highpart): Likewise.
3980 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3982 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3983 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3984 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3985 * internal-fn.c (expand_mul_overflow): Update use of
3986 GET_MODE_2XWIDER_MODE.
3987 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3988 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3989 GET_MODE_WIDER_MODE.
3990 (convert_plusminus_to_widen): Likewise.
3991 * tree-switch-conversion.c (array_value_type): Likewise.
3992 * var-tracking.c (emit_note_insn_var_location): Likewise.
3993 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3994 Return false inside rather than outside the loop if no wider mode
3996 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3997 and GET_MODE_2XWIDER_MODE
3998 (can_compare_p): Use else_void.
3999 * gdbhooks.py (OptMachineModePrinter): New class.
4000 (build_pretty_printer): Use it for opt_mode.
4002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4003 Alan Hayward <alan.hayward@arm.com>
4004 David Sherwood <david.sherwood@arm.com>
4006 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4007 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4010 Alan Hayward <alan.hayward@arm.com>
4011 David Sherwood <david.sherwood@arm.com>
4013 * machmode.h (mode_traits): New structure.
4014 (get_narrowest_mode): New function.
4015 (mode_iterator::start): Likewise.
4016 (mode_iterator::iterate_p): Likewise.
4017 (mode_iterator::get_wider): Likewise.
4018 (mode_iterator::get_known_wider): Likewise.
4019 (mode_iterator::get_2xwider): Likewise.
4020 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4021 (FOR_EACH_MODE): Likewise.
4022 (FOR_EACH_MODE_FROM): Likewise.
4023 (FOR_EACH_MODE_UNTIL): Likewise.
4024 (FOR_EACH_WIDER_MODE): Likewise.
4025 (FOR_EACH_2XWIDER_MODE): Likewise.
4026 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4027 * combine.c (simplify_comparison): Likewise
4028 * config/i386/i386.c (type_natural_mode): Likewise.
4029 * cse.c (cse_insn): Likewise.
4030 * dse.c (find_shift_sequence): Likewise.
4031 * emit-rtl.c (init_derived_machine_modes): Likewise.
4032 (init_emit_once): Likewise.
4033 * explow.c (hard_function_value): Likewise.
4034 * expmed.c (extract_fixed_bit_field_1): Likewise.
4035 (extract_bit_field_1): Likewise.
4036 (expand_divmod): Likewise.
4037 (emit_store_flag_1): Likewise.
4038 * expr.c (init_expr_target): Likewise.
4039 (convert_move): Likewise.
4040 (alignment_for_piecewise_move): Likewise.
4041 (widest_int_mode_for_size): Likewise.
4042 (emit_block_move_via_movmem): Likewise.
4043 (copy_blkmode_to_reg): Likewise.
4044 (set_storage_via_setmem): Likewise.
4045 (compress_float_constant): Likewise.
4046 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4047 * optabs-query.c (get_best_extraction_insn): Likewise.
4048 * optabs.c (expand_binop): Likewise.
4049 (expand_twoval_unop): Likewise.
4050 (expand_twoval_binop): Likewise.
4051 (widen_leading): Likewise.
4052 (widen_bswap): Likewise.
4053 (expand_parity): Likewise.
4054 (expand_unop): Likewise.
4055 (prepare_cmp_insn): Likewise.
4056 (prepare_float_lib_cmp): Likewise.
4057 (expand_float): Likewise.
4058 (expand_fix): Likewise.
4059 (expand_sfix_optab): Likewise.
4060 * postreload.c (move2add_use_add2_insn): Likewise.
4061 * reg-stack.c (reg_to_stack): Likewise.
4062 * reginfo.c (choose_hard_reg_mode): Likewise.
4063 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4064 * stor-layout.c (mode_for_size): Likewise.
4065 (smallest_mode_for_size): Likewise.
4066 (mode_for_vector): Likewise.
4067 (finish_bitfield_representative): Likewise.
4068 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4069 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4070 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4071 * var-tracking.c (prepare_call_arguments): Likewise.
4073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4074 Alan Hayward <alan.hayward@arm.com>
4075 David Sherwood <david.sherwood@arm.com>
4077 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4078 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4079 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4080 * machmode.h (mode_size): Move earlier in file.
4081 (mode_precision): Likewise.
4082 (mode_inner): Likewise.
4083 (mode_nunits): Likewise.
4084 (mode_unit_size): Likewise.
4085 (unit_unit_precision): Likewise.
4086 (mode_wider): Likewise.
4087 (mode_2xwider): Likewise.
4088 (machine_mode): New class.
4089 (mode_to_bytes): New function.
4090 (mode_to_bits): Likewise.
4091 (mode_to_precision): Likewise.
4092 (mode_to_inner): Likewise.
4093 (mode_to_unit_size): Likewise.
4094 (mode_to_unit_precision): Likewise.
4095 (mode_to_nunits): Likewise.
4096 (GET_MODE_SIZE): Use mode_to_bytes.
4097 (GET_MODE_BITSIZE): Use mode_to_bits.
4098 (GET_MODE_PRECISION): Use mode_to_precision.
4099 (GET_MODE_INNER): Use mode_to_inner.
4100 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4101 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4102 (GET_MODE_NUNITS): Use mode_to_nunits.
4103 * system.h (ALWAYS_INLINE): New macro.
4104 * config/powerpcspe/powerpcspe-c.c
4105 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4106 int for arg1_mode and arg2_mode.
4108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4109 Alan Hayward <alan.hayward@arm.com>
4110 David Sherwood <david.sherwood@arm.com>
4112 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4113 Prefix mode names with E_ in case statements.
4114 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4115 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4116 (aarch64_split_simd_move): Likewise.
4117 (aarch64_gen_storewb_pair): Likewise.
4118 (aarch64_gen_loadwb_pair): Likewise.
4119 (aarch64_gen_store_pair): Likewise.
4120 (aarch64_gen_load_pair): Likewise.
4121 (aarch64_get_condition_code_1): Likewise.
4122 (aarch64_constant_pool_reload_icode): Likewise.
4123 (get_rsqrte_type): Likewise.
4124 (get_rsqrts_type): Likewise.
4125 (get_recpe_type): Likewise.
4126 (get_recps_type): Likewise.
4127 (aarch64_gimplify_va_arg_expr): Likewise.
4128 (aarch64_simd_container_mode): Likewise.
4129 (aarch64_emit_load_exclusive): Likewise.
4130 (aarch64_emit_store_exclusive): Likewise.
4131 (aarch64_expand_compare_and_swap): Likewise.
4132 (aarch64_gen_atomic_cas): Likewise.
4133 (aarch64_emit_bic): Likewise.
4134 (aarch64_emit_atomic_swap): Likewise.
4135 (aarch64_emit_atomic_load_op): Likewise.
4136 (aarch64_evpc_trn): Likewise.
4137 (aarch64_evpc_uzp): Likewise.
4138 (aarch64_evpc_zip): Likewise.
4139 (aarch64_evpc_ext): Likewise.
4140 (aarch64_evpc_rev): Likewise.
4141 (aarch64_evpc_dup): Likewise.
4142 (aarch64_gen_ccmp_first): Likewise.
4143 (aarch64_gen_ccmp_next): Likewise.
4144 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4145 (alpha_emit_xfloating_libcall): Likewise.
4146 (emit_insxl): Likewise.
4147 (alpha_arg_type): Likewise.
4148 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4149 (arc_preferred_simd_mode): Likewise.
4150 (arc_secondary_reload): Likewise.
4151 (get_arc_condition_code): Likewise.
4152 (arc_print_operand): Likewise.
4153 (arc_legitimate_constant_p): Likewise.
4154 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4155 * config/arc/arc.md (casesi_load): Likewise.
4156 (casesi_compact_jump): Likewise.
4157 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4158 (cc_use_register): Likewise.
4159 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4160 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4161 (arm_init_iwmmxt_builtins): Likewise.
4162 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4163 (neon_expand_vector_init): Likewise.
4164 (arm_attr_length_move_neon): Likewise.
4165 (maybe_get_arm_condition_code): Likewise.
4166 (arm_emit_vector_const): Likewise.
4167 (arm_preferred_simd_mode): Likewise.
4168 (arm_output_iwmmxt_tinsr): Likewise.
4169 (thumb1_output_casesi): Likewise.
4170 (thumb2_output_casesi): Likewise.
4171 (arm_emit_load_exclusive): Likewise.
4172 (arm_emit_store_exclusive): Likewise.
4173 (arm_expand_compare_and_swap): Likewise.
4174 (arm_evpc_neon_vuzp): Likewise.
4175 (arm_evpc_neon_vzip): Likewise.
4176 (arm_evpc_neon_vrev): Likewise.
4177 (arm_evpc_neon_vtrn): Likewise.
4178 (arm_evpc_neon_vext): Likewise.
4179 (arm_validize_comparison): Likewise.
4180 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4181 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4182 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4183 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4184 (c6x_preferred_simd_mode): Likewise.
4185 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4186 (epiphany_rtx_costs): Likewise.
4187 * config/epiphany/predicates.md (proper_comparison_operator):
4189 * config/frv/frv.c (condexec_memory_operand): Likewise.
4190 (frv_emit_move): Likewise.
4191 (output_move_single): Likewise.
4192 (output_condmove_single): Likewise.
4193 (frv_hard_regno_mode_ok): Likewise.
4194 (frv_matching_accg_mode): Likewise.
4195 * config/h8300/h8300.c (split_adds_subs): Likewise.
4196 (h8300_rtx_costs): Likewise.
4197 (h8300_print_operand): Likewise.
4198 (compute_mov_length): Likewise.
4199 (output_logical_op): Likewise.
4200 (compute_logical_op_length): Likewise.
4201 (compute_logical_op_cc): Likewise.
4202 (h8300_shift_needs_scratch_p): Likewise.
4203 (output_a_shift): Likewise.
4204 (compute_a_shift_length): Likewise.
4205 (compute_a_shift_cc): Likewise.
4206 (expand_a_rotate): Likewise.
4207 (output_a_rotate): Likewise.
4208 * config/i386/i386.c (classify_argument): Likewise.
4209 (function_arg_advance_32): Likewise.
4210 (function_arg_32): Likewise.
4211 (function_arg_64): Likewise.
4212 (function_value_64): Likewise.
4213 (ix86_gimplify_va_arg): Likewise.
4214 (ix86_legitimate_constant_p): Likewise.
4215 (put_condition_code): Likewise.
4216 (split_double_mode): Likewise.
4217 (ix86_avx256_split_vector_move_misalign): Likewise.
4218 (ix86_expand_vector_logical_operator): Likewise.
4219 (ix86_split_idivmod): Likewise.
4220 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4221 (ix86_build_const_vector): Likewise.
4222 (ix86_build_signbit_mask): Likewise.
4223 (ix86_match_ccmode): Likewise.
4224 (ix86_cc_modes_compatible): Likewise.
4225 (ix86_expand_branch): Likewise.
4226 (ix86_expand_sse_cmp): Likewise.
4227 (ix86_expand_sse_movcc): Likewise.
4228 (ix86_expand_int_sse_cmp): Likewise.
4229 (ix86_expand_vec_perm_vpermi2): Likewise.
4230 (ix86_expand_vec_perm): Likewise.
4231 (ix86_expand_sse_unpack): Likewise.
4232 (ix86_expand_int_addcc): Likewise.
4233 (ix86_split_to_parts): Likewise.
4234 (ix86_vectorize_builtin_gather): Likewise.
4235 (ix86_vectorize_builtin_scatter): Likewise.
4236 (avx_vpermilp_parallel): Likewise.
4237 (inline_memory_move_cost): Likewise.
4238 (ix86_tieable_integer_mode_p): Likewise.
4239 (x86_maybe_negate_const_int): Likewise.
4240 (ix86_expand_vector_init_duplicate): Likewise.
4241 (ix86_expand_vector_init_one_nonzero): Likewise.
4242 (ix86_expand_vector_init_one_var): Likewise.
4243 (ix86_expand_vector_init_concat): Likewise.
4244 (ix86_expand_vector_init_interleave): Likewise.
4245 (ix86_expand_vector_init_general): Likewise.
4246 (ix86_expand_vector_set): Likewise.
4247 (ix86_expand_vector_extract): Likewise.
4248 (emit_reduc_half): Likewise.
4249 (ix86_emit_i387_round): Likewise.
4250 (ix86_mangle_type): Likewise.
4251 (ix86_expand_round_sse4): Likewise.
4252 (expand_vec_perm_blend): Likewise.
4253 (canonicalize_vector_int_perm): Likewise.
4254 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4255 (expand_vec_perm_1): Likewise.
4256 (expand_vec_perm_interleave3): Likewise.
4257 (expand_vec_perm_even_odd_pack): Likewise.
4258 (expand_vec_perm_even_odd_1): Likewise.
4259 (expand_vec_perm_broadcast_1): Likewise.
4260 (ix86_vectorize_vec_perm_const_ok): Likewise.
4261 (ix86_expand_vecop_qihi): Likewise.
4262 (ix86_expand_mul_widen_hilo): Likewise.
4263 (ix86_expand_sse2_abs): Likewise.
4264 (ix86_expand_pextr): Likewise.
4265 (ix86_expand_pinsr): Likewise.
4266 (ix86_preferred_simd_mode): Likewise.
4267 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4268 * config/i386/sse.md (*andnot<mode>3): Likewise.
4269 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4270 (*<code><mode>3): Likewise.
4271 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4272 (ia64_expand_atomic_op): Likewise.
4273 (ia64_arg_type): Likewise.
4274 (ia64_mode_to_int): Likewise.
4275 (ia64_scalar_mode_supported_p): Likewise.
4276 (ia64_vector_mode_supported_p): Likewise.
4277 (expand_vec_perm_broadcast): Likewise.
4278 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4279 (iq2000_function_arg_advance): Likewise.
4280 (iq2000_function_arg): Likewise.
4281 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4282 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4283 (m68k_libcall_value): Likewise.
4284 (m68k_function_value): Likewise.
4285 (sched_attr_op_type): Likewise.
4286 * config/mcore/mcore.c (mcore_output_move): Likewise.
4287 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4289 (microblaze_function_arg): Likewise.
4290 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4291 (mips_print_operand): Likewise.
4292 (mips_mode_ok_for_mov_fmt_p): Likewise.
4293 (mips_vector_mode_supported_p): Likewise.
4294 (mips_preferred_simd_mode): Likewise.
4295 (mips_expand_vpc_loongson_even_odd): Likewise.
4296 (mips_expand_vec_unpack): Likewise.
4297 (mips_expand_vi_broadcast): Likewise.
4298 (mips_expand_vector_init): Likewise.
4299 (mips_expand_vec_reduc): Likewise.
4300 (mips_expand_msa_cmp): Likewise.
4301 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4302 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4303 (cc_flags_for_mode): Likewise.
4304 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4305 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4306 (nds32_output_casesi_pc_relative): Likewise.
4307 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4308 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4309 (nvptx_gen_unpack): Likewise.
4310 (nvptx_gen_pack): Likewise.
4311 (nvptx_gen_shuffle): Likewise.
4312 (nvptx_gen_wcast): Likewise.
4313 (nvptx_preferred_simd_mode): Likewise.
4314 * config/pa/pa.c (pa_secondary_reload): Likewise.
4315 * config/pa/predicates.md (base14_operand): Likewise.
4316 * config/powerpcspe/powerpcspe-c.c
4317 (altivec_resolve_overloaded_builtin): Likewise.
4318 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4320 (rs6000_preferred_simd_mode): Likewise.
4321 (output_vec_const_move): Likewise.
4322 (rs6000_expand_vector_extract): Likewise.
4323 (rs6000_split_vec_extract_var): Likewise.
4324 (reg_offset_addressing_ok_p): Likewise.
4325 (rs6000_legitimate_offset_address_p): Likewise.
4326 (rs6000_legitimize_address): Likewise.
4327 (rs6000_emit_set_const): Likewise.
4328 (rs6000_const_vec): Likewise.
4329 (rs6000_emit_move): Likewise.
4330 (spe_build_register_parallel): Likewise.
4331 (rs6000_darwin64_record_arg_recurse): Likewise.
4332 (swap_selector_for_mode): Likewise.
4333 (spe_init_builtins): Likewise.
4334 (paired_init_builtins): Likewise.
4335 (altivec_init_builtins): Likewise.
4336 (do_load_for_compare): Likewise.
4337 (rs6000_generate_compare): Likewise.
4338 (rs6000_expand_float128_convert): Likewise.
4339 (emit_load_locked): Likewise.
4340 (emit_store_conditional): Likewise.
4341 (rs6000_output_function_epilogue): Likewise.
4342 (rs6000_handle_altivec_attribute): Likewise.
4343 (rs6000_function_value): Likewise.
4344 (emit_fusion_gpr_load): Likewise.
4345 (emit_fusion_p9_load): Likewise.
4346 (emit_fusion_p9_store): Likewise.
4347 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4348 (fusion_gpr_mem_load): Likewise.
4349 (fusion_addis_mem_combo_load): Likewise.
4350 (fusion_addis_mem_combo_store): Likewise.
4351 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4352 (fusion_gpr_mem_load): Likewise.
4353 (fusion_addis_mem_combo_load): Likewise.
4354 (fusion_addis_mem_combo_store): Likewise.
4355 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4357 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4358 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4359 (rs6000_preferred_simd_mode): Likewise.
4360 (output_vec_const_move): Likewise.
4361 (rs6000_expand_vector_extract): Likewise.
4362 (rs6000_split_vec_extract_var): Likewise.
4363 (reg_offset_addressing_ok_p): Likewise.
4364 (rs6000_legitimate_offset_address_p): Likewise.
4365 (rs6000_legitimize_address): Likewise.
4366 (rs6000_emit_set_const): Likewise.
4367 (rs6000_const_vec): Likewise.
4368 (rs6000_emit_move): Likewise.
4369 (rs6000_darwin64_record_arg_recurse): Likewise.
4370 (swap_selector_for_mode): Likewise.
4371 (paired_init_builtins): Likewise.
4372 (altivec_init_builtins): Likewise.
4373 (rs6000_expand_float128_convert): Likewise.
4374 (emit_load_locked): Likewise.
4375 (emit_store_conditional): Likewise.
4376 (rs6000_output_function_epilogue): Likewise.
4377 (rs6000_handle_altivec_attribute): Likewise.
4378 (rs6000_function_value): Likewise.
4379 (emit_fusion_gpr_load): Likewise.
4380 (emit_fusion_p9_load): Likewise.
4381 (emit_fusion_p9_store): Likewise.
4382 * config/rx/rx.c (rx_gen_move_template): Likewise.
4383 (flags_from_mode): Likewise.
4384 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4385 (s390_slb_comparison): Likewise.
4386 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4387 (s390_vector_mode_supported_p): Likewise.
4388 (s390_cc_modes_compatible): Likewise.
4389 (s390_match_ccmode_set): Likewise.
4390 (s390_canonicalize_comparison): Likewise.
4391 (s390_emit_compare_and_swap): Likewise.
4392 (s390_branch_condition_mask): Likewise.
4393 (s390_rtx_costs): Likewise.
4394 (s390_secondary_reload): Likewise.
4395 (__SECONDARY_RELOAD_CASE): Likewise.
4396 (s390_expand_cs): Likewise.
4397 (s390_preferred_simd_mode): Likewise.
4398 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4399 * config/sh/sh.c (sh_print_operand): Likewise.
4400 (dump_table): Likewise.
4401 (sh_secondary_reload): Likewise.
4402 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4403 * config/sh/sh.md (casesi_worker_1): Likewise.
4404 (casesi_worker_2): Likewise.
4405 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4406 (fcc_comparison_operator): Likewise.
4407 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4408 (emit_soft_tfmode_cvt): Likewise.
4409 (sparc_preferred_simd_mode): Likewise.
4410 (output_cbranch): Likewise.
4411 (sparc_print_operand): Likewise.
4412 (sparc_expand_vec_perm_bmask): Likewise.
4413 (vector_init_bshuffle): Likewise.
4414 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4415 (spu_vector_mode_supported_p): Likewise.
4416 (spu_expand_insv): Likewise.
4417 (spu_emit_branch_or_set): Likewise.
4418 (spu_handle_vector_attribute): Likewise.
4419 (spu_builtin_splats): Likewise.
4420 (spu_builtin_extract): Likewise.
4421 (spu_builtin_promote): Likewise.
4422 (spu_expand_sign_extend): Likewise.
4423 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4424 (tilegx_simd_int): Likewise.
4425 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4426 (tilepro_simd_int): Likewise.
4427 * config/v850/v850.c (const_double_split): Likewise.
4428 (v850_print_operand): Likewise.
4429 (ep_memory_offset): Likewise.
4430 * config/vax/vax.c (vax_rtx_costs): Likewise.
4431 (vax_output_int_move): Likewise.
4432 (vax_output_int_add): Likewise.
4433 (vax_output_int_subtract): Likewise.
4434 * config/visium/predicates.md (visium_branch_operator): Likewise.
4435 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4436 (visium_print_operand_address): Likewise.
4437 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4438 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4439 (xtensa_expand_conditional_branch): Likewise.
4440 (xtensa_copy_incoming_a7): Likewise.
4441 (xtensa_output_literal): Likewise.
4442 * dfp.c (decimal_real_maxval): Likewise.
4443 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4446 Alan Hayward <alan.hayward@arm.com>
4447 David Sherwood <david.sherwood@arm.com>
4449 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4450 (mode_nunits_inline): Likewise.
4451 (mode_inner_inline): Likewise.
4452 (mode_unit_size_inline): Likewise.
4453 (mode_unit_precision_inline): Likewise.
4454 (emit_insn_modes_h): Likewise. Also emit a #define of the
4456 (emit_mode_wider): Add an E_ prefix to mode names.
4457 (emit_mode_complex): Likewise.
4458 (emit_mode_inner): Likewise.
4459 (emit_mode_adjustments): Likewise.
4460 (emit_mode_int_n): Likewise.
4461 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4462 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4463 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4464 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4465 (CRC32_BUILTIN, ENTRY): Likewise.
4466 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4467 (aarch64_pop_regs): Likewise.
4468 (aarch64_process_components): Likewise.
4469 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4470 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4471 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4472 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4473 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4474 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4476 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4477 (pcmpistr): Likewise.
4478 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4479 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4480 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4482 * config/rl78/rl78.c (mduc_regs): Likewise.
4483 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4484 (htm_expand_builtin): Likewise.
4485 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4486 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4487 * config/xtensa/xtensa.c (print_operand): Likewise.
4488 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4489 (NUM_MODE_VECTOR_INT): Likewise.
4490 * genoutput.c (null_operand): Likewise.
4491 (output_operand_data): Likewise.
4492 * genrecog.c (print_parameter_value): Likewise.
4493 * lra.c (debug_operand_data): Likewise.
4495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4497 * dwarf2out.c (scompare_loc_descriptor_wide)
4498 (scompare_loc_descriptor_narrow): New functions, split out from...
4499 (scompare_loc_descriptor): ...here.
4500 * expmed.c (emit_store_flag_int): New function, split out from...
4501 (emit_store_flag): ...here.
4503 2017-08-30 Richard Biener <rguenther@suse.de>
4505 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4506 (dwarf2out_early_finish): Move setting of AT_pubnames from
4507 early debug output to early finish.
4509 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4511 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4512 and -mdata-region to the assembler.
4514 2017-08-30 Richard Biener <rguenther@suse.de>
4516 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4518 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4519 (dwarf2out_early_global_decl): Only generate a DIE for the
4520 abstract origin if it doesn't already exist or is a declaration DIE.
4521 (resolve_addr): Do not add the linkage name twice when
4522 generating a stub DIE for the DW_TAG_GNU_call_site target.
4524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4527 Use machine_mode rather than int for arg1_mode.
4529 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4532 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4533 that the second argument of the built-in functions to unpack
4534 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4535 switch statement instead a lot of if statements.
4536 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4537 Allow 64-bit values to be in Altivec registers as well as
4538 traditional floating point registers.
4539 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4541 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4543 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4544 MAX_REGS_PER_ADDRESS == 1.
4546 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4548 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4549 * config/i386/i386.c (ix86_option_override_internal): Simplify
4550 setting of opts->x_flag_entry.
4552 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4553 Jakub Jelinek <jakub@redhat.com>
4554 Richard Biener <rguenther@suse.de>
4556 PR tree-optimization/81503
4557 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4558 folded constant fits in the target type; reorder tests for clarity.
4560 2017-08-29 Martin Liska <mliska@suse.cz>
4562 * passes.def: Include pass_lower_switch.
4563 * stmt.c (dump_case_nodes): Remove and move to
4564 tree-switch-conversion.
4565 (case_values_threshold): Likewise.
4566 (expand_switch_as_decision_tree_p): Likewise.
4567 (emit_case_decision_tree): Likewise.
4568 (expand_case): Likewise.
4569 (balance_case_nodes): Likewise.
4570 (node_has_low_bound): Likewise.
4571 (node_has_high_bound): Likewise.
4572 (node_is_bounded): Likewise.
4573 (emit_case_nodes): Likewise.
4574 (struct simple_case_node): New struct.
4575 (add_case_node): Remove.
4576 (emit_case_dispatch_table): Use vector instead of case_list.
4577 (reset_out_edges_aux): Remove.
4578 (compute_cases_per_edge): Likewise.
4579 (expand_case): Build list of simple_case_node.
4580 (expand_sjlj_dispatch_table): Use it.
4581 * tree-switch-conversion.c (struct case_node): Moved from
4582 stmt.c and adjusted.
4583 (emit_case_nodes): Likewise.
4584 (node_has_low_bound): Likewise.
4585 (node_has_high_bound): Likewise.
4586 (node_is_bounded): Likewise.
4587 (case_values_threshold): Likewise.
4588 (reset_out_edges_aux): Likewise.
4589 (compute_cases_per_edge): Likewise.
4590 (add_case_node): Likewise.
4591 (dump_case_nodes): Likewise.
4592 (balance_case_nodes): Likewise.
4593 (expand_switch_as_decision_tree_p): Likewise.
4594 (emit_jump): Likewise.
4595 (emit_case_decision_tree): Likewise.
4596 (try_switch_expansion): Likewise.
4597 (do_jump_if_equal): Likewise.
4598 (emit_cmp_and_jump_insns): Likewise.
4599 (fix_phi_operands_for_edge): New function.
4600 (record_phi_operand_mapping): Likewise.
4601 (class pass_lower_switch): New pass.
4602 (pass_lower_switch::execute): New function.
4603 (make_pass_lower_switch): Likewise.
4604 (conditional_probability):
4605 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4606 * tree-pass.h: Add make_pass_lower_switch.
4608 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4611 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4614 2017-08-29 Richard Biener <rguenther@suse.de>
4616 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4617 we do not add a DW_AT_inline attribute twice.
4618 (gen_subprogram_die): Remove code setting DW_AT_inline on
4619 DECL_ABSTRACT_P nodes.
4621 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4623 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4624 calls to internal functions.
4625 (gimplify_modify_expr): Likewise.
4626 * tree-call-cdce.c (use_internal_fn): Likewise.
4627 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4628 (convert_to_divmod): Set the nothrow flag.
4629 * tree-if-conv.c (predicate_mem_writes): Likewise.
4630 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4631 (vectorizable_call): Likewise.
4632 (vectorizable_store): Likewise.
4633 (vectorizable_load): Likewise.
4634 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4635 (vect_recog_mask_conversion_pattern): Likewise.
4637 2017-08-29 Martin Liska <mliska@suse.cz>
4640 * gcc.c (driver_handle_option): Add new argument.
4641 * opts-common.c (handle_option): Pass
4642 target_option_override_hook.
4643 * opts-global.c (lang_handle_option): Add new option.
4644 (set_default_handlers): Add new argument.
4645 (decode_options): Likewise.
4646 * opts.c (target_handle_option): Likewise.
4647 (common_handle_option): Call target_option_override_hook.
4648 * opts.h (struct cl_option_handler_func): Add hook for
4649 target option override.
4650 (struct cl_option_handlers): Likewise.
4651 (set_default_handlers): Add new argument.
4652 (decode_options): Likewise.
4653 (common_handle_option): Likewise.
4654 (target_handle_option): Likewise.
4655 * toplev.c (toplev::main): Pass targetm.target_option.override
4658 2017-08-29 Richard Biener <rguenther@suse.de>
4659 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4661 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4662 life to the active subtree.
4664 2017-08-28 Jeff Law <law@redhat.com>
4666 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4667 derive_equivalences.
4668 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4670 (edge_info::derive_equivalences): New private member function
4672 * tree-ssa-dom.c (class edge_info): Changed from a struct
4673 to a class. Add ctor/dtor, methods and data members.
4674 (edge_info::edge_info): Renamed from allocate_edge_info.
4675 Initialize additional members.
4676 (edge_info::~edge_info): New.
4677 (free_dom_edge_info): Delete the edge info.
4678 (record_edge_info): Use new class & associated member functions.
4679 Tighten forms for testing for edge equivalences.
4680 (record_temporary_equivalences): Iterate over the simple
4681 equivalences rather than assuming there's only one per edge.
4682 (cprop_into_successor_phis): Iterate over the simple
4683 equivalences rather than assuming there's only one per edge.
4684 (optimize_stmt): Use operand_equal_p rather than pointer
4685 equality for mini-DSE code.
4687 2017-08-28 Nathan Sidwell <nathan@acm.org>
4689 * gcc.c (execute): Fold SIGPIPE handling into switch
4690 statement. Adjust internal error message.
4692 2017-08-28 Richard Biener <rguenther@suse.de>
4695 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4696 Do nothing for removed DIEs.
4698 2017-08-28 Richard Biener <rguenther@suse.de>
4700 PR tree-optimization/81977
4701 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4704 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4707 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4708 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4709 using targetm.gen_mem_thread_fence.
4711 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4714 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4715 predicate to register_operand. Reorder operands.
4716 (*btr<mode>): Ditto.
4717 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4718 (*btr<mode>_mask): Ditto.
4720 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4722 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4723 * config/rs6000/xmmintrin.h: New file.
4724 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4726 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4729 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4730 parameter and_insn and return it.
4731 (recombine_lvx_pattern): Insert a copy to ensure availability of
4732 the base register of the copied masking operation at the point of
4733 the instruction replacement.
4734 (recombine_stvx_pattern): Likewise.
4736 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4738 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4739 undocumented switches.
4740 (-mpower9-dform-vector): Likewise.
4741 (-mpower9-dform): Likewise.
4742 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4743 comments to delete references to -mpower9-dform* switches.
4744 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4745 Delete reference to -mpower9-dform* switches, test for
4746 -mpower9-vector instead.
4747 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4748 (OTHER_P9_VECTOR_MASKS): Likewise.
4749 (POWERPC_MASKS): Likewise.
4750 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4751 tests against -mpower9-dform* to -mpower9-vector. Delete code
4752 that checked for -mpower9-dform* consistancy with other options.
4753 Add test for -mpower9-misc to enable other power9 switches.
4754 (rs6000_init_hard_regno_mode_ok): Likewise.
4755 (rs6000_option_override_internal): Likewise.
4756 (rs6000_emit_prologue): Likewise.
4757 (rs6000_emit_epilogue): Likewise.
4758 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4759 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4760 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4762 (emit_fusion_p9_store): Likewise.
4763 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4764 resetting these macros if the assembler does not support ISA 3.0
4766 (TARGET_P9_DFORM_VECTOR): Likewise.
4767 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4768 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4770 2017-08-25 Alan Modra <amodra@gmail.com>
4772 PR rtl-optimization/81747
4773 * cse.c (cse_extended_basic_block): Don't attempt to record
4774 equivalences for degenerate conditional branches that branch
4775 to their fall-through.
4777 2017-08-24 Martin Sebor <msebor@redhat.com>
4780 * gimple-fold.c (size_must_be_zero_p): New function.
4781 (gimple_fold_builtin_memory_op): Call it.
4783 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4785 * config/rs6000/mm_malloc.h: New file.
4787 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4789 PR tree-optimization/81913
4790 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4791 analysis when either IVs in condition can wrap.
4793 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4795 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4796 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4798 2017-08-24 Richard Biener <rguenther@suse.de>
4801 * targhooks.c (default_target_can_inline_p): Properly
4802 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4803 is present and always compare.
4804 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4805 infer -mfpmath=sse from TARGET_SSE_P.
4806 (ix86_can_inline_p): Properly use target_option_default_node when
4807 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4809 2017-08-24 Richard Biener <rguenther@suse.de>
4812 * dwarf2out.c (output_die): Handle flag_generate_offload like
4814 (output_comp_unit): Likewise.
4815 (gen_array_type_die): Likewise.
4816 (dwarf2out_early_finish): Likewise.
4817 (note_variable_value_in_expr): Likewise.
4818 (dwarf2out_finish): Likewise. Adjust assert.
4819 * cgraphunit.c (symbol_table::compile): Move setting of
4820 flag_generate_offload earlier ...
4821 (symbol_table::finalize_compilation_unit): ... here, before
4822 early debug finalization.
4824 2017-08-24 Richard Biener <rguenther@suse.de>
4826 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4827 and ipa-fnsummary.h.
4828 (ix86_can_inline_p): When ix86_fpmath flags do not match
4829 check whether the callee uses FP math at all.
4831 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4834 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4835 instead of TYPE_PRECISION.
4837 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4838 Alan Hayward <alan.hayward@arm.com>
4839 David Sherwood <david.sherwood@arm.com>
4841 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4843 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4845 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4847 (ix86_frame::stack_realign_allocate): New field.
4848 (struct machine_frame_state): Modify comments.
4849 (machine_frame_state::sp_realigned_fp_end): New field.
4850 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4852 (sp_valid_at): Add assertion to assure no attempt to access invalid
4853 offset of a realigned stack.
4854 (fp_valid_at): Likewise.
4855 (choose_baseaddr): Modify comments.
4856 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4857 ix86_expand_prologue.
4858 (ix86_expand_prologue): Modify stack realignment and allocation.
4859 (ix86_expand_epilogue): Modify comments.
4860 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4861 avx2_runtime, avx512f, and avx512f_runtime.
4863 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4865 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4866 (mstackrealign): Do not init to -1.
4867 * config/i386/i386.c (ix86_option_override_internal):
4868 Check opts_set, not opts when setting default value of
4869 opts->x_ix86_force_align_arg_pointer.
4871 2017-08-23 Richard Biener <rguenther@suse.de>
4873 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4874 lang_hooks.decl_printable_name.
4875 * print-rtl-function.c (print_rtx_function): Likewise.
4876 * tree-pretty-print.c (dump_function_header): Likewise.
4878 2017-08-23 Richard Biener <rguenther@suse.de>
4881 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4882 -g0 at compile-time.
4884 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4887 * doc/sourcebuild.texi (Other hardware attributes):
4890 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4893 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4894 Fix single-use check.
4896 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4898 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4900 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4902 * doc/install.texi: Modify to add more details on running selected
4905 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4907 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4908 is combined with -mabi=ms.
4909 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4912 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4914 PR tree-optimization/81488
4915 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4916 and cached_basis fields.
4917 (MAX_SPREAD): New constant.
4918 (alloc_cand_and_find_basis): Initialize new fields.
4919 (clear_visited): New function.
4920 (create_phi_basis_1): Rename from create_phi_basis, set visited
4921 and cached_basis fields.
4922 (create_phi_basis): New wrapper function.
4923 (phi_add_costs_1): Rename from phi_add_costs, add spread
4924 parameter, set visited field, short-circuit when limits reached.
4925 (phi_add_costs): New wrapper function.
4926 (record_phi_increments_1): Rename from record_phi_increments, set
4928 (record_phi_increments): New wrapper function.
4929 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4930 (phi_incr_cost): New wrapper function.
4931 (all_phi_incrs_profitable_1): Rename from
4932 all_phi_incrs_profitable, set visited field.
4933 (all_phi_incrs_profitable): New wrapper function.
4935 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4936 Alan Hayward <alan.hayward@arm.com>
4937 David Sherwood <david.sherwood@arm.com>
4939 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4940 that takes the outer and inner modes.
4941 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4942 comparison as the canonical test for a paradoxical subreg.
4943 * combine.c (simplify_set): Use paradoxical_subreg_p.
4944 (make_extraction): Likewise.
4945 (force_to_mode): Likewise.
4946 (rtx_equal_for_field_assignment_p): Likewise.
4947 (gen_lowpart_for_combine): Likewise.
4948 (simplify_comparison): Likewise.
4949 * cse.c (equiv_constant): Likewise.
4950 * expmed.c (store_bit_field_1): Likewise.
4951 * final.c (alter_subreg): Likewise.
4952 * fwprop.c (propagate_rtx): Likewise.
4953 (forward_propagate_subreg): Likewise.
4954 * ira-conflicts.c (ira_build_conflicts): Likewise.
4955 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4956 * lra-constraints.c (curr_insn_transform): Likewise.
4957 (split_reg): Likewise.
4958 * lra-eliminations.c (move_plus_up): Likewise.
4959 (lra_eliminate_regs_1): Likewise.
4960 * recog.c (general_operand): Likewise.
4961 * ree.c (combine_reaching_defs): Likewise.
4962 * reload.c (push_reload): Likewise.
4963 (find_reloads): Likewise.
4964 * reload1.c (elimination_effects): Likewise.
4965 (compute_reload_subreg_offset): Likewise.
4966 (choose_reload_regs): Likewise.
4967 * rtlanal.c (subreg_lsb_1): Likewise.
4968 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4969 (simplify_subreg): Likewise.
4970 * var-tracking.c (track_loc_p): Likewise.
4971 * emit-rtl.c (byte_lowpart_offset): Likewise.
4972 (paradoxical_subreg_p): Delete out-of-line definition.
4974 2017-08-22 Jeff Law <law@redhat.com>
4976 PR tree-optimization/81741
4977 PR tree-optimization/71947
4978 * tree-ssa-dom.c: Include tree-inline.h.
4979 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4980 equivalences if one is more expensive to compute than the other.
4981 * tree-ssa-scopedtables.h (class const_or_copies): Make
4982 record_const_or_copy_raw method private.
4983 (class avail_exprs_stack): New method simplify_binary_operation.
4984 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4985 avail_exprs_stack::simplify_binary_operation as needed.
4986 (avail_exprs_stack::simplify_binary_operation): New function.
4988 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4990 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4991 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4992 (DOT_SYMBOLS): Likewise.
4993 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4994 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4995 (RS6000_ABI_NAME): Likewise.
4996 (TARGET_CMODEL): Likewise.
4997 (TOC_SECTION_ASM_OP): Likewise.
4998 (SET_CMODEL): New macro.
4999 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5001 2017-08-22 Richard Biener <rguenther@suse.de>
5003 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5004 to allow for free-lang-data replacements similar to verify_type_variant.
5006 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5007 Alan Hayward <alan.hayward@arm.com>
5008 David Sherwood <david.sherwood@arm.com>
5010 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5011 VOIDmode for the LABEL_REF.
5013 2017-08-22 Richard Biener <rguenther@suse.de>
5015 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5017 2017-08-22 Richard Biener <rguenther@suse.de>
5019 * common.opt (feliminate-dwarf2-dups): Ignore.
5020 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5021 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5022 same_die_p_wrap, compute_section_prefix,
5023 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5024 (comdat_symbol_id, comdat_symbol_number): Likewise.
5025 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5027 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5028 (output_die): Mark unreachable path unreachable.
5029 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5030 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5031 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5032 (dwarf2out_early_finish): Likewise.
5034 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5036 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5038 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5041 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5042 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5043 (avr_attribute_table) <io, io_low, address>: Initialize
5044 .decl_required with true.
5046 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5048 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5049 undocumented debugging options.
5050 (-mvsx-scalar-double): Likewise.
5051 (-mallow-df-permute): Likewise.
5052 (-mvectorize-builtins): Likewise.
5053 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5054 (rs6000_builtin_vectorized_function): Likewise.
5055 (rs6000_builtin_md_vectorized_function): Likewise.
5056 (rs6000_opt_vars): Likewise.
5058 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5061 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5062 (*btrq_imm): Rename from *btrq.
5063 (*btcq_imm): Rename from *btcq.
5064 (btsc): New code attribute.
5065 (*<btsc><mode>): New insn pattern.
5066 (*btr<mode>): Ditto.
5067 (*<btsc><mode>_mask): New insn_and_split pattern.
5068 (*btr<mode>_mask): Ditto.
5070 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5071 Alan Hayward <alan.hayward@arm.com>
5072 David Sherwood <david.sherwood@arm.com>
5074 * function.c (pad_below): Simplify padding calculation.
5076 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5077 Alan Hayward <alan.hayward@arm.com>
5078 David Sherwood <david.sherwood@arm.com>
5080 * target.def (function_prologue): Remove frame size argument.
5081 (function_epilogue): Likewise.
5082 * doc/tm.texi: Regenerate.
5083 * final.c (final_start_function): Update call to function_prologue.
5084 (final_end_function): Update call to function_epilogue.
5085 (default_function_pro_epilogue): Remove frame size argument.
5086 * output.h (default_function_pro_epilogue): Likewise.
5087 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5088 (arm_output_function_prologue): Likewise.
5089 * config/frv/frv.c (frv_function_prologue): Likewise.
5090 (frv_function_epilogue): Likewise.
5091 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5092 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5093 (ia64_output_function_epilogue): Likewise.
5094 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5095 (m32r_output_function_epilogue): Likewise.
5096 * config/microblaze/microblaze.c (microblaze_function_prologue)
5097 (microblaze_function_epilogue): Likewise.
5098 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5099 (mips_output_function_epilogue): Likewise.
5100 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5101 (mmix_target_asm_function_epilogue): Likewise.
5102 * config/msp430/msp430.c (msp430_start_function): Likewise.
5103 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5104 (nds32_asm_function_epilogue): Likewise.
5105 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5106 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5107 (pa_output_function_epilogue): Likewise.
5108 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5109 (rs6000_output_function_epilogue): Likewise.
5110 * config/rl78/rl78.c (rl78_start_function): Likewise.
5111 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5112 (rs6000_output_function_epilogue): Likewise.
5113 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5114 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5115 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5116 (sparc_asm_function_epilogue): Likewise.
5118 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5120 * tree.h (type_has_mode_precision_p): New function.
5121 * convert.c (convert_to_integer_1): Use it.
5122 * expr.c (expand_expr_real_2): Likewise.
5123 (expand_expr_real_1): Likewise.
5124 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5125 * match.pd: Likewise.
5126 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5127 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5128 * tree-tailcall.c (process_assignment): Likewise.
5129 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5130 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5131 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5132 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5133 (vectorizable_assignment): Likewise.
5134 (vectorizable_shift): Likewise.
5135 (vectorizable_operation): Likewise.
5136 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5138 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5140 * match.pd: Add pow (C, x) simplification.
5142 2017-08-21 Richard Biener <rguenther@suse.de>
5144 PR tree-optimization/81900
5145 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5146 for blocks with abnormal predecessors.
5147 (compute_antic): Do not set visited flag prematurely.
5149 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5152 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5154 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5156 * stor-layout.h (vector_type_mode): Move to...
5157 * tree.h (vector_type_mode): ...here.
5158 * stor-layout.c (vector_type_mode): Move to...
5159 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5161 2017-08-21 Richard Biener <rguenther@suse.de>
5163 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5164 register_external_die hooks.
5165 (debug_false_tree_charstarstar_uhwistar): Declare.
5166 (debug_nothing_tree_charstar_uhwi): Likewise.
5167 * debug.c (do_nothing_debug_hooks): Adjust.
5168 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5169 (debug_nothing_tree_charstar_uhwi): Likewise.
5170 * dbxout.c (dbx_debug_hooks): Adjust.
5171 (xcoff_debug_hooks): Likewise.
5172 * sdbout.c (sdb_debug_hooks): Likewise.
5173 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5174 * dwarf2out.c (macinfo_label_base): New global.
5175 (dwarf2out_register_external_die): New function for the
5176 register_external_die hook.
5177 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5178 (dwarf2_debug_hooks): Use them.
5179 (dwarf2_lineno_debug_hooks): Adjust.
5180 (struct die_struct): Add with_offset flag.
5181 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5182 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5183 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5184 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5185 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5186 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5187 defining section names for the early LTO debug variants.
5188 (reset_indirect_string): New helper.
5189 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5190 (print_dw_val): Add support for offsetted symbol references.
5191 (get_ultimate_context): Split out from is_cxx.
5192 (is_cxx): Use get_ultimate_context.
5193 (is_fortran): Add decl overload.
5194 (compute_comp_unit_symbol): Split out worker from
5195 compute_section_prefix.
5196 (compute_section_prefix): Call compute_comp_unit_symbol and
5197 set comdat_type_p here.
5198 (output_die): Skip DIE symbol output for the LTO added one.
5199 Handle DIE symbol references with offset.
5200 (output_comp_unit): Guard section name mangling properly.
5201 For LTO debug sections emit a symbol at the section beginning
5202 which we use to refer to its DIEs.
5203 (add_abstract_origin_attribute): For DIEs registered via
5204 dwarf2out_register_external_die directly refer to the early
5205 DIE rather than indirectly through the shadow one we created.
5206 Remove obsolete call to dwarf2out_abstract_function for
5207 non-function/block origins.
5208 (gen_array_type_die): When generating early LTO debug do
5209 not emit DW_AT_string_length.
5210 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5211 late when in LTO. As suggested place a gcc_unreachable for
5212 the DECL_ABSTRACT_P case.
5213 (gen_subprogram_die): Avoid another specification DIE
5214 for early built declarations/definitions for the late LTO case.
5215 (gen_variable_die): Add type references for late duplicated VLA dies
5217 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5218 we have the abstract instance already.
5219 (process_scope_var): Adjust decl DIE contexts in LTO which
5220 first puts them in limbo.
5221 (gen_decl_die): Do not generate type DIEs late apart from
5222 types for VLAs or for decls we do not yet have a DIE. Do not
5223 call dwarf2out_abstract_function late.
5224 (dwarf2out_early_global_decl): Make sure to create DIEs
5225 for abstract instances of a decl first.
5226 (dwarf2out_late_global_decl): Adjust comment.
5227 (output_macinfo_op): With multiple macro sections use
5228 macinfo_label_base to distinguish labels.
5229 (output_macinfo): Likewise. Update macinfo_label_base.
5230 Pass in the line info label.
5231 (note_variable_value_in_expr): When generating LTO resolve
5232 all variable values here by generating DIEs as needed.
5233 (init_sections_and_labels): Add early LTO debug flag parameter
5234 and generate different sections and names if set. Add generation
5235 counter for the labels so we can have multiple of them.
5236 (reset_dies): Helper to allow DIEs to be output multiple times.
5237 (dwarf2out_finish): When outputting DIEs to the fat part of an
5238 LTO object first reset DIEs.
5239 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5240 (modified_type_die): Check for decl_ultimate_origin being self
5242 (gen_type_die_with_usage): Likewise.
5243 (gen_typedef_die): Allow decl_ultimate_origin being self.
5244 (set_decl_abstract_flags): Remove.
5245 (set_block_abstract_flags): Likewise.
5246 (dwarf2out_abstract_function): Treat the early generated DIEs
5247 as the abstract copy and only add DW_AT_inline and
5248 DW_AT_artificial here and call set_decl_origin_self.
5249 If the DIE has an abstract origin don't do anything.
5250 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5251 if we have none yet (Go fails to build one, PR78628).
5252 (variably_modified_type_p): Prevent endless recursion for Ada
5253 cyclic pointer types.
5254 * lto-streamer-in.c: Include debug.h.
5255 (dref_queue): New global.
5256 (lto_read_tree_1): Stream in DIE references.
5257 (lto_input_tree): Register DIE references.
5258 (input_function): Stream DECL_DEBUG_ARGS.
5259 * lto-streamer-out.c: Include debug.h.
5260 (lto_write_tree_1): Output DIE references.
5261 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5262 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5263 (output_function): Stream DECL_DEBUG_ARGS.
5264 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5265 Stream DECL_ABSTRACT_ORIGIN.
5266 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5267 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5268 DECL_CONTEXT for file-scope decls.
5269 * lto-streamer.h (struct dref_entry): Declare.
5270 (dref_queue): Likewise.
5271 * cfgexpand.c (pass_expand::execute): Do not call the
5272 outlining_inline_function hook here.
5273 * lto-wrapper.c (debug_obj): New global.
5274 (tool_cleanup): Unlink it if required.
5275 (debug_objcopy): New function.
5276 (run_gcc): Handle early debug sections in the IL files by
5277 extracting them to separate files, partially linkin them and
5278 feeding the result back as result to the linker.
5279 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5280 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5281 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5282 sections into a separate segment.
5283 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5285 (darwin_asm_dwarf_section): Likewise.
5286 (darwin_asm_output_dwarf_offset): Likewise.
5287 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5289 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5290 Alan Hayward <alan.hayward@arm.com>
5291 David Sherwood <david.sherwood@arm.com>
5293 * read-md.h (md_reader::record_potential_iterator_use): Replace
5294 pointer argument with an rtx and an index.
5295 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5296 (apply_mode_iterator): Likewise.
5297 (apply_code_iterator): Likewise.
5298 (apply_int_iterator): Likewise.
5299 (apply_subst_iterator): Likewise.
5300 (record_iterator_use): Likewise.
5301 (record_attribute_use): Likewise.
5302 (md_reader::record_potential_iterator_use): Likewise. Update calls
5303 to record_iterator_use and apply_iterator.
5304 (iterator_use): Replace ptr with x and index.
5305 (attribute_use): Likewise.
5306 (apply_attribute_uses): Update calls to apply_iterator.
5307 (apply_iterators): Likewise. Update initialization of iterator_use.
5308 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5309 and record_potential_iterator_use.
5310 (rtx_reader::read_rtx_operand): Likewise.
5312 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5313 Alan Hayward <alan.hayward@arm.com>
5314 David Sherwood <david.sherwood@arm.com>
5316 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5319 2017-08-21 Richard Biener <rguenther@suse.de>
5322 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5323 at struct end conservatively when comparing common bases.
5325 2017-08-21 Richard Biener <rguenther@suse.de>
5327 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5328 (mem_ref_in_stmt): Remove.
5329 (determine_max_movement): Use ref index to get at the reference.
5330 (invariantness_dom_walker::before_dom_children): Deal with
5331 lim data already initialized.
5332 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5334 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5336 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5337 (TARGET_ISA_ROUND): Ditto.
5338 (TARGET_ROUND): Ditto.
5339 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5340 * config/i386/i386.md: Ditto.
5341 * config/i386/sse.md: Ditto.
5342 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5343 with OPTION_MASK_ISA_SSE4_1.
5345 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5348 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5349 __builtin_ia32_lzcnt_u16.
5351 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5354 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5355 (rs6000_set_current_function): Rewrite function to use it.
5357 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5360 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5361 and TYPE_WARN_IF_NOT_ALIGN.
5362 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5363 (handle_warn_if_not_align): New.
5364 (place_union_field): Call handle_warn_if_not_align.
5365 (place_field): Call handle_warn_if_not_align.
5366 Copy TYPE_WARN_IF_NOT_ALIGN.
5367 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5368 (layout_type): Likewise.
5369 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5371 (tree_decl_common): Add warn_if_not_align.
5372 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5373 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5374 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5375 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5376 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5377 * doc/extend.texi: Document warn_if_not_aligned attribute.
5378 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5380 2017-08-17 Martin Liska <mliska@suse.cz>
5383 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5384 (get_data_dependence): Use it as pointer type.
5385 (distribute_loop): Likewise.
5387 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5389 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5390 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5391 (p8_vmrgow_<mode>_direct): New define_insn.
5392 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5393 handle endianness for vmrgew and vmrgow permute patterns.
5395 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5397 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5398 * config/rs6000/rs6000-cpus.def: Remove comment.
5399 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5400 (POWERPC_MASKS): Likewise.
5401 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5402 use of TARGET_VSX_TIMODE.
5403 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5404 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5405 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5406 (rs6000_option_override_internal): Remove dead code.
5407 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5408 (rs6000_legitimize_reload_address): Likewise.
5409 (rs6000_legitimate_address_p): Likewise.
5410 (rs6000_opt_masks): Delete "vsx-timode".
5411 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5412 from function comment.
5413 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5414 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5415 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5417 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5418 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5419 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5422 (define_peephole2): Likewise.
5424 2017-08-17 Martin Sebor <msebor@redhat.com>
5427 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5428 past the end of an array.
5429 (test_pp_format): Add test cases.
5431 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5433 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5434 missing ECF_NOTHROW flags.
5436 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5439 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5440 operands residing in integer registers.
5441 (*vsx_le_perm_load_<mode>): Likewise.
5442 (*vsx_le_perm_store_<mode>): Likewise.
5443 (define_peephole2): Add peepholes to optimize the above.
5445 2017-08-17 Marek Polacek <polacek@redhat.com>
5448 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5449 to mimic what shorten_compare did. Change the return type to bool.
5450 (fold_cond_expr_with_comparison): Update call to
5451 operand_equal_for_comparison_p.
5452 (fold_ternary_loc): Likewise.
5454 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5456 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5458 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5460 2017-08-17 Richard Biener <rguenther@suse.de>
5462 * tree-ssa-structalias.c (solve_graph): When propagating
5463 to successors update the graphs succ edges and avoid duplicate work.
5465 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5468 * config/i386/i386.c (ix86_option_override_internal): Save target
5469 specific options after ix86_stack_protector_guard_reg was changed.
5471 2017-08-17 Richard Biener <rguenther@suse.de>
5473 PR tree-optimization/81827
5474 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5475 (new_var_info): Initialize it conservatively.
5476 (get_call_vi): Mark register vars.
5477 (new_scalar_tmp_constraint_exp): Likewise.
5478 (handle_rhs_call): Likewise.
5479 (handle_const_call): Likewise.
5480 (create_function_info_for): Likewise.
5481 (solve_constraints): Sort varinfos to separate register from
5482 non-register vars to pack points-to solution bitmaps during
5485 2017-08-17 Marek Polacek <polacek@redhat.com>
5487 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5489 2017-08-17 Richard Biener <rguenther@suse.de>
5491 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5492 to true when overflow is undefined and we saturated the result.
5494 2017-08-17 Alan Modra <amodra@gmail.com>
5497 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5498 Don't use store multiple if only one reg needs saving.
5499 (interesting_frame_related_regno): New function.
5500 (rs6000_frame_related): Don't emit frame info for regs that
5502 (rs6000_emit_epilogue): Likewise.
5504 2017-08-16 Nathan Sidwell <nathan@acm.org>
5506 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5507 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5508 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5509 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5510 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5511 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5512 (verify_type): Adjust for TYPE_BINFO move.
5513 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5514 process TYPE_BINFO directly.
5515 (hash_tree): Likewise.
5516 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5518 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5521 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5523 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5525 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5528 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5529 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5530 Add ix86_binary_operator_ok to insn constraint.
5531 (*iordi_1_bts): Ditto.
5532 (*xordi_1_btc): Ditto.
5533 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5534 Update corresponding peephole2 pattern.
5538 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5540 PR tree-optimization/81832
5541 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5542 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5544 2017-08-16 Marek Polacek <polacek@redhat.com>
5547 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5548 regarding size_zero_node.
5550 2017-08-16 Martin Liska <mliska@suse.cz>
5553 * config.gcc: Respect previously set extra_objs in case
5556 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5558 PR tree-optimization/81835
5559 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5560 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5561 not depend on the phi.
5563 2017-08-16 Alan Modra <amodra@gmail.com>
5565 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
5568 2017-08-16 Alan Modra <amodra@gmail.com>
5570 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
5571 (save_reg_p): ..into this. Update all callers.
5572 (first_reg_to_save): Simplify.
5574 2017-08-16 Alan Modra <amodra@gmail.com>
5576 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
5579 2017-08-15 Joseph Myers <joseph@codesourcery.com>
5583 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
5584 constant count if that count is less than 32.
5586 2017-08-15 Nathan Sidwell <nathan@acm.org>
5588 * gcc.c (execute): Emit friendlier message if inferior is killed
5589 by an external cause.
5591 2017-08-15 Richard Biener <rguenther@suse.de>
5593 PR tree-optimization/81790
5594 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
5595 CONSTRUCTORs from simplifying and VN.
5597 2017-08-14 Martin Sebor <msebor@redhat.com>
5599 * builtin-attrs.def: Add comments.
5601 2017-08-14 Martin Sebor <msebor@redhat.com>
5604 * doc/extend.texi (attribute nonstring): Document new attribute.
5606 2017-08-14 Martin Sebor <msebor@redhat.com>
5609 * tree-diagnostic.c (default_tree_printer): Handle %G.
5610 * gimple-pretty-print.h (percent_G_format): Declare new function.
5611 * gimple-pretty-print.c (percent_G_format): Define.
5612 * tree-pretty-print.c (percent_K_format): Add argument.
5614 2017-08-14 Martin Sebor <msebor@redhat.com>
5616 PR translation/79998
5617 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5618 Remove a stray space.
5620 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5623 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5624 (*iordi_1_bts): Ditto.
5625 (*xordi_1_btc): Ditto.
5627 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5630 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5631 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5633 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5634 quoted strings, and make more translator-friendly.
5635 (darwin_rs6000_override_options): Likewise.
5636 (rs6000_option_override_internal): Likewise.
5637 (rs6000_return_in_memory): Fix overlong line.
5638 (init_cmulative_args): Use quoted strings, and make more
5639 translator-friendly.
5640 (rs6000_pass_by_reference): Fix overlong line.
5641 (def_builtin): Use quoted strings.
5642 (altivec_expand_predicate_builtin): Use quoted strings, and make
5643 more translator-friendly.
5644 (htm_expand_builtin): Use quoted strings.
5645 (cpu_expand_builtin): Use quoted strings, and make more
5646 translator-friendly.
5647 (altivec_expand_builtin): Likewise.
5648 (paired_expand_predicate_builtin): Likewise.
5649 (rs6000_invalid_builtin): Likewise.
5650 (builtin_function_type): Use quoted strings.
5651 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5652 more translator-friendly.
5653 (rs6000_trampoline_init): Likewise.
5654 (rs6000_handle_altivec_attribute): Likewise.
5655 (rs6000_inner_target_options): Use quoted strings.
5656 (rs6000_disable_incompatible_switches): Likewise.
5657 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5658 strings, and make more translator-friendly.
5659 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5661 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5663 PR tree-optimization/81799
5664 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5665 cond_expr to simple gimple operand.
5667 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5670 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5672 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5676 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5677 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5678 TARGET_GASISR_PROLOGUES.
5679 * config/avr/avr.c (avr_option_override): Same.
5680 (avr_pass_pre_proep::execute): Same.
5682 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5685 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5686 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5688 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5690 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5691 define_insn_and_split. Split to a memory load from 0 in
5692 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5693 using PTR mode iterator.
5694 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5695 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5696 (*add_tp_<mode>): Redefine as define_insn_and_split.
5697 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5698 address space. Merge with *add_tp_x32 using PTR mode iterator.
5699 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5700 Split to an add with a memory load from 0 in
5701 DEFAULT_TLS_SEG_REG address space.
5703 2017-08-12 Andrew Pinski <apinski@cavium.com>
5705 * config/aarch64/aarch64-option-extensions.def (rdma):
5706 Fix feature string to what Linux prints out in /proc/cpuinfo.
5708 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5711 * dwarf2out.c (modified_type_die): For C typedef types that have
5712 an ultimate origin, process the ultimate origin instead of the
5714 (gen_typedef_die): Assert that input DECLs have no ultimate
5716 (gen_type_die_with_usage): For typedef variants that have an
5717 ultimate origin, just call gen_decl_die on the original DECL.
5718 (process_scope_var): Avoid creating DIEs for local typedefs and
5719 concrete static variables.
5721 2017-08-12 Alan Modra <amodra@gmail.com>
5725 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5726 match gnu-user.h startfile.
5727 (ENDFILE_LINUX_SPEC): Similarly.
5729 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5732 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5734 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5736 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5737 * config/aarch64/aarch64.md (mov<mode>): Change.
5738 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5739 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5740 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5742 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5744 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5745 for storage order barriers.
5747 2017-08-11 Martin Liska <mliska@suse.cz>
5750 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5751 variables of void type.
5753 2017-08-11 Martin Liska <mliska@suse.cz>
5755 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5756 TARGET_SUPPORTS_ALIASES.
5757 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5758 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5759 (optimize_weakref): Likewise.
5760 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5761 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5762 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5764 2017-08-11 Martin Liska <mliska@suse.cz>
5767 * config/i386/i386.c (make_resolver_func): Do complete
5768 refactoring of the function.
5770 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5773 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5774 * config/i386/i386.c (ix86_stack_protect_guard): Use
5775 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5776 * doc/invoke.texi (x86 Options): Document
5777 -mstack-protector-guard-symbol= option.
5779 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5781 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5782 * config/i386/i386.c (ix86_split_stack_guard): New function.
5783 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5784 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5785 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5786 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5787 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5788 (split_stack_space_check): Call ix86_split_stack_guard.
5790 2017-08-10 Martin Sebor <msebor@redhat.com>
5792 * print-tree.c (print_node): Print location using the established
5794 Replace spaces with colons.
5795 (debug_raw, debug): Ditto.
5797 2017-08-10 Martin Sebor <msebor@redhat.com>
5800 * pretty-print.c (pp_format): Correct the handling of %s precision.
5802 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5805 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5807 (ix86_finalize_stack_frame_flags): This. Also clear
5808 frame_pointer_needed if -fno-omit-frame-pointer is used without
5810 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5811 with ix86_finalize_stack_frame_flags.
5812 (ix86_expand_epilogue): Likewise.
5813 (ix86_expand_split_stack_prologue): Likewise.
5814 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5816 2017-08-10 Martin Liska <mliska@suse.cz>
5819 * c-attribs.c (handle_target_attribute):
5820 Report warning for an empty string argument of target attribute.
5822 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5825 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5827 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5828 or DECL_NONLOCAL labels.
5829 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5830 or DECL_NONLOCAL labels here.
5832 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5835 to indicate when early gimple folding has been disabled.
5836 (rs6000_gimple_fold_builtin): Add debug content.
5837 (rs6000_invalid_builtin): Fix whitespace.
5838 (rs6000_expand_builtin): Fix whitespace.
5839 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5841 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5844 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5845 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5847 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5849 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5850 (qdf24xx): Likewise.
5851 * config/aarch64/aarch64.md: Include falkor.md.
5852 * config/aarch64/falkor.md: New.
5854 2017-08-09 Marek Polacek <polacek@redhat.com>
5857 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5858 * diagnostic.c (emit_diagnostic): Add a comment.
5859 (emit_diagnostic_valist): New function.
5861 2017-08-09 Marek Polacek <polacek@redhat.com>
5864 * input.c (make_location): New overload.
5865 * input.h (make_location): Declare.
5867 2017-08-08 Alan Modra <amodra@gmail.com>
5868 H.J. Lu <hongjiu.lu@intel.com>
5871 * gcc.c (NO_PIE_SPEC): Delete.
5872 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5874 (LINK_PIE_SPEC): ..to here.
5875 (LINK_COMMAND_SPEC): Support -no-pie.
5876 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5877 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5879 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5880 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5881 (ENDFILE_CRTEND_SPEC): Similarly.
5883 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5886 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5887 (mstack-protector-guard-offset=): Ditto.
5888 * config/i386/i386.c (ix86_option_override): Handle
5889 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5891 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5892 ix86_stack_protect_guard_offset variables.
5893 (TARGET_STACK_PROTECT_GUARD): Always define.
5894 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5895 and -mstack-protector-guard-offset= options.
5897 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5899 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5900 boundary case for the last candidate.
5902 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5904 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5905 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5907 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5910 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5911 * config/aarch64/aarch64-builtins.c
5912 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5913 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5914 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5916 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5917 Andrew Pinski <pinskia@gmail.com>
5920 * internal-fn.def (XORSIGN): New.
5921 * optabs.def (xorsign_optab): New.
5922 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5923 (convert_expand_mult_copysign): New.
5924 (pass_optimize_widening_mul::execute): Call
5925 convert_expand_mult_copysign.
5927 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5929 PR tree-optimization/81354
5930 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5931 Insert on edges rather than explicitly creating landing pads.
5932 (analyze_candidates_and_replace): Commit edge inserts.
5934 2017-08-08 Richard Biener <rguenther@suse.de>
5937 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5938 (expand_simple_operations): Also look through ADDR_EXPRs with
5939 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5941 2017-08-08 Richard Biener <rguenther@suse.de>
5943 PR tree-optimization/81723
5944 * tree-vect-slp.c (struct bst_traits): New hash traits.
5945 (bst_fail): New global.
5946 (vect_build_slp_tree_2): New worker, split out from ...
5947 (vect_build_slp_tree): ... this now wrapping it with using
5948 bst_fail set to cache SLP tree build fails. Properly handle
5950 (vect_analyze_slp_instance): Allocate and free bst_fail.
5952 2017-08-08 Martin Liska <mliska@suse.cz>
5955 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5956 LABEL_DECLs that can be from a different function.
5958 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5960 PR tree-optimization/81744
5961 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5962 loop's number of iterations.
5964 2017-08-08 Martin Liska <mliska@suse.cz>
5966 * asan.c: Include header files.
5967 * attribs.c (build_decl_attribute_variant): New function moved
5969 (build_type_attribute_qual_variant): Likewise.
5970 (cmp_attrib_identifiers): Likewise.
5971 (simple_cst_list_equal): Likewise.
5972 (omp_declare_simd_clauses_equal): Likewise.
5973 (attribute_value_equal): Likewise.
5974 (comp_type_attributes): Likewise.
5975 (build_type_attribute_variant): Likewise.
5976 (lookup_ident_attribute): Likewise.
5977 (remove_attribute): Likewise.
5978 (merge_attributes): Likewise.
5979 (merge_type_attributes): Likewise.
5980 (merge_decl_attributes): Likewise.
5981 (merge_dllimport_decl_attributes): Likewise.
5982 (handle_dll_attribute): Likewise.
5983 (attribute_list_equal): Likewise.
5984 (attribute_list_contained): Likewise.
5985 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5986 (lookup_attribute_by_prefix): Likewise.
5987 * bb-reorder.c: Include header files.
5988 * builtins.c: Likewise.
5989 * calls.c: Likewise.
5990 * cfgexpand.c: Likewise.
5991 * cgraph.c: Likewise.
5992 * cgraphunit.c: Likewise.
5993 * convert.c: Likewise.
5994 * dwarf2out.c: Likewise.
5995 * final.c: Likewise.
5996 * fold-const.c: Likewise.
5997 * function.c: Likewise.
5998 * gimple-expr.c: Likewise.
5999 * gimple-fold.c: Likewise.
6000 * gimple-pretty-print.c: Likewise.
6001 * gimple.c: Likewise.
6002 * gimplify.c: Likewise.
6003 * hsa-common.c: Likewise.
6004 * hsa-gen.c: Likewise.
6005 * internal-fn.c: Likewise.
6006 * ipa-chkp.c: Likewise.
6007 * ipa-cp.c: Likewise.
6008 * ipa-devirt.c: Likewise.
6009 * ipa-fnsummary.c: Likewise.
6010 * ipa-inline.c: Likewise.
6011 * ipa-visibility.c: Likewise.
6013 * lto-cgraph.c: Likewise.
6014 * omp-expand.c: Likewise.
6015 * omp-general.c: Likewise.
6016 * omp-low.c: Likewise.
6017 * omp-offload.c: Likewise.
6018 * omp-simd-clone.c: Likewise.
6019 * opts-global.c: Likewise.
6020 * passes.c: Likewise.
6021 * predict.c: Likewise.
6022 * sancov.c: Likewise.
6023 * sanopt.c: Likewise.
6024 * symtab.c: Likewise.
6025 * toplev.c: Likewise.
6026 * trans-mem.c: Likewise.
6027 * tree-chkp.c: Likewise.
6028 * tree-eh.c: Likewise.
6029 * tree-into-ssa.c: Likewise.
6030 * tree-object-size.c: Likewise.
6031 * tree-parloops.c: Likewise.
6032 * tree-profile.c: Likewise.
6033 * tree-ssa-ccp.c: Likewise.
6034 * tree-ssa-live.c: Likewise.
6035 * tree-ssa-loop.c: Likewise.
6036 * tree-ssa-sccvn.c: Likewise.
6037 * tree-ssa-structalias.c: Likewise.
6038 * tree-ssa.c: Likewise.
6039 * tree-streamer-in.c: Likewise.
6040 * tree-vectorizer.c: Likewise.
6041 * tree-vrp.c: Likewise.
6043 * ubsan.c: Likewise.
6044 * varasm.c: Likewise.
6045 * varpool.c: Likewise.
6046 * tree.c: Remove functions moved to attribs.[ch].
6048 * config/aarch64/aarch64.c: Add attrs.h header file.
6049 * config/alpha/alpha.c: Likewise.
6050 * config/arc/arc.c: Likewise.
6051 * config/arm/arm.c: Likewise.
6052 * config/avr/avr.c: Likewise.
6053 * config/bfin/bfin.c: Likewise.
6054 * config/c6x/c6x.c: Likewise.
6055 * config/cr16/cr16.c: Likewise.
6056 * config/cris/cris.c: Likewise.
6057 * config/darwin.c: Likewise.
6058 * config/epiphany/epiphany.c: Likewise.
6059 * config/fr30/fr30.c: Likewise.
6060 * config/frv/frv.c: Likewise.
6061 * config/ft32/ft32.c: Likewise.
6062 * config/h8300/h8300.c: Likewise.
6063 * config/i386/winnt.c: Likewise.
6064 * config/ia64/ia64.c: Likewise.
6065 * config/iq2000/iq2000.c: Likewise.
6066 * config/lm32/lm32.c: Likewise.
6067 * config/m32c/m32c.c: Likewise.
6068 * config/m32r/m32r.c: Likewise.
6069 * config/m68k/m68k.c: Likewise.
6070 * config/mcore/mcore.c: Likewise.
6071 * config/microblaze/microblaze.c: Likewise.
6072 * config/mips/mips.c: Likewise.
6073 * config/mmix/mmix.c: Likewise.
6074 * config/mn10300/mn10300.c: Likewise.
6075 * config/moxie/moxie.c: Likewise.
6076 * config/msp430/msp430.c: Likewise.
6077 * config/nds32/nds32-isr.c: Likewise.
6078 * config/nds32/nds32.c: Likewise.
6079 * config/nios2/nios2.c: Likewise.
6080 * config/nvptx/nvptx.c: Likewise.
6081 * config/pa/pa.c: Likewise.
6082 * config/pdp11/pdp11.c: Likewise.
6083 * config/powerpcspe/powerpcspe.c: Likewise.
6084 * config/riscv/riscv.c: Likewise.
6085 * config/rl78/rl78.c: Likewise.
6086 * config/rx/rx.c: Likewise.
6087 * config/s390/s390.c: Likewise.
6088 * config/sh/sh.c: Likewise.
6089 * config/sol2.c: Likewise.
6090 * config/sparc/sparc.c: Likewise.
6091 * config/spu/spu.c: Likewise.
6092 * config/stormy16/stormy16.c: Likewise.
6093 * config/tilegx/tilegx.c: Likewise.
6094 * config/tilepro/tilepro.c: Likewise.
6095 * config/v850/v850.c: Likewise.
6096 * config/vax/vax.c: Likewise.
6097 * config/visium/visium.c: Likewise.
6098 * config/xtensa/xtensa.c: Likewise.
6100 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6103 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6104 constraints since the -mupper-regs-* switches have been
6106 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6107 into a vector from a double word element that was extracted from
6108 another vector, and eliminate extra XXPERMDI instructions.
6109 (vsx_concat_<mode>_2): Likewise.
6110 (vsx_concat_<mode>_3): Likewise.
6111 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6112 concat to allow optimizing inserts from previous extracts.
6114 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6116 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6117 memory reference to a SSP offset in TLS address space.
6118 (ix86_print_operand) <case '@'>: Remove.
6119 (ix86_print_operand_punct_valid_p): Remove '@' code.
6120 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6122 (stack_tls_protect_set_<mode>): Remove.
6123 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6124 (stack_tls_protect_test_<mode>): Remove.
6125 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6127 2017-08-07 Olivier Hainque <hainque@adacore.com>
6130 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6132 2017-08-07 Douglas Rupp <rupp@adacore.com>
6134 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6135 variable was referenced as multidir in command.
6137 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6140 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6143 2017-08-07 Martin Liska <mliska@suse.cz>
6145 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6146 * config/rl78/rl78.c: Add include of attribs.h.
6147 * config/sh/sh.c: Likewise.
6148 * config/v850/v850.c: Likewise.
6150 2017-08-07 Tom de Vries <tom@codesourcery.com>
6153 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6155 2017-08-07 Martin Liska <mliska@suse.cz>
6157 * config/mips/mips.c: Include attribs.h.
6159 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6162 * doc/invoke.texi: Document change in behvaior for -Ofast for
6165 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6167 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6169 (aarch64_pop_regs): Likewise.
6170 (aarch64_gen_load_pair): Likewise.
6171 (aarch64_save_callee_saves): Likewise.
6172 (aarch64_restore_callee_saves): Likewise.
6174 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6176 * config/i386/i386.c: Revert the last change.
6178 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6181 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6183 (ix86_finalize_stack_frame_flags): This. Also clear
6184 frame_pointer_needed if -fno-omit-frame-pointer is used without
6186 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6187 with ix86_finalize_stack_frame_flags.
6188 (ix86_expand_epilogue): Likewise.
6189 (ix86_expand_split_stack_prologue): Likewise.
6191 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6194 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6195 to P_AES for Westmere.
6197 2017-08-07 Jonathan Yong <10walls@gmail.com>
6199 * config/i386/mingw.opt (fset-stack-executable): Removed.
6200 * config/i386/cygming.opt (fset-stack-executable): Moved
6202 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6204 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6206 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6208 2017-08-07 Marek Polacek <polacek@redhat.com>
6211 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6213 2017-08-07 Martin Liska <mliska@suse.cz>
6215 * attribs.h (canonicalize_attr_name): New function.
6216 (cmp_attribs): Move from c-format.c and adjusted.
6217 (is_attribute_p): Moved from tree.h.
6218 * tree-inline.c: Add new includes.
6219 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6220 (private_is_attribute_p): Remove.
6221 (private_lookup_attribute): Likewise.
6222 (private_lookup_attribute_by_prefix): Simplify.
6223 (remove_attribute): Use is_attribute_p.
6224 * tree.h: Remove removed declarations.
6226 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6229 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6230 instead of computing it in the function. Formatting fix.
6231 (expand_case): Don't rely on default_edge being the first edge,
6232 clear it if removing it, pass default_edge to
6233 emit_case_dispatch_table.
6234 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6237 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6239 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6240 insn in the function, emit NOP after the insn.
6242 2017-08-06 Tom de Vries <tom@codesourcery.com>
6244 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6247 2017-08-06 Tom de Vries <tom@codesourcery.com>
6249 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6252 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6254 PR tree-optimization/57371
6255 * match.pd: New pattern.
6257 2017-08-04 Marek Polacek <polacek@redhat.com>
6260 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6261 perform the computation in offset_int.
6263 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6265 PR tree-optimization/81136
6266 * tree-vectorizer.h: Include tree-hash-traits.h.
6267 (vec_base_alignments): New typedef.
6268 (vec_info): Add a base_alignments field.
6269 (vect_record_base_alignments): Declare.
6270 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6272 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6273 (create_data_ref): Add an is_conditional_in_stmt argument.
6274 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6275 the is_conditional_in_stmt field.
6276 (data_ref_loc): Add an is_conditional_in_stmt field.
6277 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6278 (find_data_references_in_stmt): Update call to create_data_ref.
6279 (graphite_find_data_references_in_stmt): Likewise.
6280 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6281 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6282 (vect_record_base_alignment): New function.
6283 (vect_record_base_alignments): Likewise.
6284 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6285 for nested statements even if we fail to compute a misalignment.
6286 Use pooled base alignments for unconditional references.
6287 (vect_find_same_alignment_drs): Compare base addresses instead
6289 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6290 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6292 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6294 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6295 Add an explicit name for the enum. Use auto_vec for slp_instances
6297 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6299 (_bb_vec_info): Add a constructor and destructor.
6300 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6301 (destroy_loop_vec_info): Delete.
6302 (vect_destroy_datarefs): Likewise.
6303 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6304 (vec_info::vec_info): New function.
6305 (vec_info::~vec_info): Likewise.
6306 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6307 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6308 destroy_loop_vec_info.
6309 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6310 (_loop_vec_info::_loop_vec_info): ...this.
6311 (destroy_loop_vec_info): Replace with...
6312 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6313 the stmt_vec_infos. Leave handling of vec_info information to its
6314 destructor. Remove explicit vector releases.
6315 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6316 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6317 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6318 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6319 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6320 (destroy_bb_vec_info): Replace with...
6321 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6322 information to its destructor.
6323 (vect_slp_analyze_bb_1): Use new and delete instead of
6324 new_bb_vec_info and destroy_bb_vec_info.
6325 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6328 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6330 * tree-data-ref.h (subscript): Add access_fn field.
6331 (data_dependence_relation): Add could_be_independent_p.
6332 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6333 (same_access_functions): Move to tree-data-ref.c.
6334 * tree-data-ref.c (ref_contains_union_access_p): New function.
6335 (access_fn_component_p): Likewise.
6336 (access_fn_components_comparable_p): Likewise.
6337 (dr_analyze_indices): Add a reference to access_fn_component_p.
6338 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6340 (constant_access_functions): Likewise.
6341 (add_other_self_distances): Likewise.
6342 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6343 (initialize_data_dependence_relation): Use XCNEW and remove
6344 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6345 of access functions that have the same type. Allow the
6346 subsequence to end with different bases in some circumstances.
6347 Record the chosen access functions in SUB_ACCESS_FN.
6348 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6349 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6350 (subscript_dependence_tester_1): Likewise dra and drb.
6351 (build_classic_dist_vector): Update calls accordingly.
6352 (subscript_dependence_tester): Likewise.
6353 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6354 DDR_COULD_BE_INDEPENDENT_P.
6355 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6356 comp_alias_ddrs instead of may_alias_ddrs.
6357 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6359 (vect_analyze_data_ref_dependence): Use it if
6360 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6361 distance vectors if that fails.
6362 (dependence_distance_ge_vf): New function.
6363 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6364 LOOP_VINFO_MAY_ALIAS_DDRS.
6366 2017-08-04 Richard Biener <rguenther@suse.de>
6369 * fold-const.c (fold_binary_loc): Properly restrict
6370 minus_var0 && minus_var1 case when associating undefined overflow
6373 2017-08-04 Tom de Vries <tom@codesourcery.com>
6375 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6377 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6379 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6380 Don't start diagnostic messages with a capital letter.
6381 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6383 (rs6000_invalid_builtin): Likewise.
6384 (rs6000_trampoline_init): Likewise.
6386 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6389 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6390 after setting changeable df flags.
6392 2017-08-03 Richard Biener <rguenther@suse.de>
6394 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6395 up if the use is in USE - X.
6397 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6399 * toplev.c (dumpfile.h): New include.
6400 (internal_error_reentered): New static function. Use it...
6401 (internal_error_function): ...here to handle reentered internal_error.
6403 2017-08-03 Richard Biener <rguenther@suse.de>
6406 * fold-const.c (split_tree): Add minus_var and minus_con
6407 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6408 here but always use minus_*.
6409 (associate_trees): Assert we never associate with MINUS_EXPR
6410 and NULL first operand. Do not recurse for PLUS_EXPR operands
6411 when associating as MINUS_EXPR either.
6412 (fold_binary_loc): Track minus_var and minus_con.
6414 2017-08-03 Tom de Vries <tom@codesourcery.com>
6417 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6418 ACCEL_COMPILER, apply finish_options on
6419 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6421 2017-08-03 Tom de Vries <tom@codesourcery.com>
6424 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6425 function_entry_patch_area_size > 0.
6427 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6430 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6431 instead of 10??LU, perform unit multiplication in wide_int,
6432 don't change alloc_object_size_limit if the limit is larger
6435 PR tree-optimization/81655
6436 PR tree-optimization/81588
6437 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6438 the case when ranges[i].low and high are 1 for unsigned type with
6442 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6443 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6445 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6447 * tree-vrp.h: Add include guard.
6449 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6452 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6453 (ud2): New insn pattern.
6454 * config/i386/i386.c (ix86_expand_epilogue):
6455 For naked functions, generate ud2 instead of trap insn.
6457 2017-08-02 Marek Polacek <polacek@redhat.com>
6460 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6462 2017-08-02 Tom de Vries <tom@codesourcery.com>
6463 Cesar Philippidis <cesar@codesourcery.com>
6465 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6466 Add missing edge probabilities.
6468 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6470 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6473 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6476 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6477 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6478 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6479 prologue_seq sequences - if any.
6481 2017-08-02 Richard Biener <rguenther@suse.de>
6483 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6484 via vectors if supported, integer extracts via punning if supported
6485 or otherwise vector extracts.
6487 2017-08-02 Richard Biener <rguenther@suse.de>
6489 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6491 (bitmap_insert_into_set): ... this.
6493 2017-08-02 Richard Biener <rguenther@suse.de>
6495 PR tree-optimization/81633
6497 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6499 PR tree-optimization/71752
6500 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6502 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6504 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6505 (machine_function::call_ms2sysv_pad_out): Remove field.
6506 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6507 (ix86_compute_frame_layout): Likewise.
6509 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6512 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6513 attribute with interrupt attribute.
6515 2017-08-01 Andrew Pinski <apinski@cavium.com>
6517 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6518 BIT_INSERT_EXPR's operand 1
6519 to see if the types precision matches.
6521 2017-08-01 Martin Liska <mliska@suse.cz>
6524 * builtins.c (expand_builtin_memcpy_args): Remove.
6525 (expand_builtin_memcpy): Call newly added function
6526 expand_builtin_memory_copy_args.
6527 (expand_builtin_memcpy_with_bounds): Likewise.
6528 (expand_builtin_mempcpy): Remove last argument.
6529 (expand_builtin_mempcpy_with_bounds): Likewise.
6530 (expand_builtin_memory_copy_args): New function created from
6531 expand_builtin_mempcpy_args with small modifications.
6532 (expand_builtin_mempcpy_args): Remove.
6533 (expand_builtin_stpcpy): Remove unused argument.
6534 (expand_builtin): Likewise.
6535 (expand_builtin_with_bounds): Likewise.
6537 2017-08-01 Martin Liska <mliska@suse.cz>
6540 Make mempcpy more optimal (PR middle-end/70140).
6542 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6545 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6546 __builtin_vec_cmpne verify both arguments are compatible vectors
6547 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6548 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6549 move computation of aligned to after checking the argument types.
6553 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6554 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6557 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6558 Doug Rupp <rupp@adacore.com>
6559 Olivier Hainque <hainque@adacore.com>
6561 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6562 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6564 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6565 for TARGET_OS_CPP_BUILTIN.
6566 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
6567 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
6569 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
6570 passing required abi options to the assembler for EABI configurations.
6571 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
6572 of .text.hot and .text.unlikely sections for kernel modules when
6573 using ARM style exceptions.
6574 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
6575 options. Add EXTRA_CC1_SPEC.
6576 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
6578 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
6580 (ARM_TARGET2_DWARF_FORMAT): Define.
6581 * config/arm/t-vxworks: Adjust multilib control to removal of the
6582 Diab command line options.
6584 2017-08-01 Martin Liska <mliska@suse.cz>
6586 PR gcov-profile/81561
6587 * gcov.c (unblock): Make unblocking safe as we need to preserve
6588 index correspondence of blocks and block_lists.
6590 2017-08-01 Richard Biener <rguenther@suse.de>
6592 PR tree-optimization/81181
6593 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6594 (compute_antic): ... end of iteration here.
6596 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
6598 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
6599 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
6600 (ftree-slp-vectorize): Likewise.
6601 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
6602 can no longer be set independent of flag_tree_loop_vectorize.
6603 * omp-general.c (emp_max_vf): Likewise.
6604 * opts.c (enable_fdo_optimizations): Remove references to
6605 flag_tree_vectorize, these are now implicit.
6606 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6607 and leave it for the options machinery.
6609 2017-08-01 Martin Liska <mliska@suse.cz>
6612 * builtins.c (expand_builtin_memcpy_args): Remove.
6613 (expand_builtin_memcpy): Call newly added function
6614 expand_builtin_memory_copy_args.
6615 (expand_builtin_memcpy_with_bounds): Likewise.
6616 (expand_builtin_mempcpy): Remove last argument.
6617 (expand_builtin_mempcpy_with_bounds): Likewise.
6618 (expand_builtin_memory_copy_args): New function created from
6619 expand_builtin_mempcpy_args with small modifications.
6620 (expand_builtin_mempcpy_args): Remove.
6621 (expand_builtin_stpcpy): Remove unused argument.
6622 (expand_builtin): Likewise.
6623 (expand_builtin_with_bounds): Likewise.
6625 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6628 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6629 print "ds:" only for immediates in generic address space.
6631 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6634 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6635 (ix86_function_ok_for_sibcall): Return false for naked functions.
6637 2017-08-01 Richard Biener <rguenther@suse.de>
6639 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6640 (compute_antic): Seed worklist with exit block predecessors.
6641 * cfganal.c (dfs_find_deadend): For a cycle return the source
6642 of the edge closing it.
6644 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6646 * config/aarch64/aarch64.c
6647 (aarch64_can_const_movi_rtx_p): Move 0 check.
6649 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6651 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6652 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6654 * match.pd: Ditto in address comparison pattern.
6656 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6658 PR tree-optimization/81627
6659 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6660 closed ssa form for store-store chain.
6662 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6664 PR tree-optimization/81620
6665 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6666 for store-store chain.
6668 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6670 PR tree-optimization/81588
6671 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6672 ranges[i].in_p, invert comparison code ccode. For >/>=,
6673 swap rhs1 and rhs2 and comparison code unconditionally,
6674 for </<= don't do that. Don't swap rhs1/rhs2 again if
6675 ranges[i].in_p, instead invert comparison code ccode if
6676 opcode or oe->rank is BIT_IOR_EXPR.
6679 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6680 a direct optab to conversion optab.
6681 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6682 with GET_MODE_INNER as last argument instead of optab_handler.
6683 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6684 vector extraction if possible and optab is available.
6685 * expr.c (store_constructor): Use convert_optab_handler instead
6686 of optab_handler. Use vector initialization from smaller
6687 vectors if possible and optab is available.
6688 * tree-vect-stmts.c (vectorizable_load): Likewise.
6689 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6691 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6692 of vec_init from half-sized vectors with the same element mode.
6693 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6694 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6695 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6696 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6697 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6698 after mode in gen_vec_extract* calls.
6699 (vec_extract<mode>): Renamed to ...
6700 (vec_extract<mode><ssescalarmodelower>): ... this.
6701 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6702 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6703 element mode after mode in gen_vec_init* calls.
6704 (VEC_INIT_HALF_MODE): New mode iterator.
6705 (vec_init<mode>): Renamed to ...
6706 (vec_init<mode><ssescalarmodelower>): ... this.
6707 (vec_init<mode><ssehalfvecmodelower>): New expander.
6708 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6709 (vec_extractv2sfsf): ... this.
6710 (vec_initv2sf): Renamed to ...
6711 (vec_initv2sfsf): ... this.
6712 (vec_extractv2si): Renamed to ...
6713 (vec_extractv2sisi): ... this.
6714 (vec_initv2si): Renamed to ...
6715 (vec_initv2sisi): ... this.
6716 (vec_extractv4hi): Renamed to ...
6717 (vec_extractv4hihi): ... this.
6718 (vec_initv4hi): Renamed to ...
6719 (vec_initv4hihi): ... this.
6720 (vec_extractv8qi): Renamed to ...
6721 (vec_extractv8qiqi): ... this.
6722 (vec_initv8qi): Renamed to ...
6723 (vec_initv8qiqi): ... this.
6724 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6725 (vec_init<mode>): Renamed to ...
6726 (vec_init<mode><VEC_base_l>): ... this.
6727 (vec_extract<mode>): Renamed to ...
6728 (vec_extract<mode><VEC_base_l>): ... this.
6729 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6730 (vec_initv2sfsf): ... this.
6731 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6732 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6733 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6734 element mode after mode in gen_vec_init* calls.
6735 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6736 (vec_init<mode><Vel>): ... this.
6737 (vec_extract<mode>): Renamed to ...
6738 (vec_extract<mode><Vel>): ... this.
6739 * config/aarch64/iterators.md (Vel): New mode attribute.
6740 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6741 Add element mode after mode in gen_vec_extract* calls.
6742 * config/s390/vector.md (non_vec_l): New mode attribute.
6743 (vec_extract<mode>): Renamed to ...
6744 (vec_extract<mode><non_vec_l>): ... this.
6745 (vec_init<mode>): Renamed to ...
6746 (vec_init<mode><non_vec_l>): ... this.
6747 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6748 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6750 * config/arm/iterators.md (V_elem_l): New mode attribute.
6751 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6752 (vec_extract<mode><V_elem_l>): ... this.
6753 (vec_extractv2di): Renamed to ...
6754 (vec_extractv2didi): ... this.
6755 (vec_init<mode>): Renamed to ...
6756 (vec_init<mode><V_elem_l>): ... this.
6757 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6758 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6759 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6760 Add element mode after gen_vec_extract* calls.
6761 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6762 (vec_init<mode><unitmode>): ... this.
6763 (vec_extract<mode>): Renamed to ...
6764 (vec_extract<mode><unitmode>): ... this.
6765 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6766 (vec_init<mode><unitmode>): ... this.
6767 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6768 (vec_initv2sfsf): ... this.
6769 (vec_extractv2sf): Renamed to ...
6770 (vec_extractv2sfsf): ... this.
6771 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6772 Add element mode after gen_vec_extract* calls.
6773 * config/mips/mips.md (unitmode): New mode iterator.
6774 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6775 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6776 * config/spu/spu.md (inner_l): New mode attribute.
6777 (vec_init<mode>): Renamed to ...
6778 (vec_init<mode><inner_l>): ... this.
6779 (vec_extract<mode>): Renamed to ...
6780 (vec_extract<mode><inner_l>): ... this.
6781 * config/sparc/sparc.md (veltmode): New mode iterator.
6782 (vec_init<VMALL:mode>): Renamed to ...
6783 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6784 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6785 (vec_initv2sisi): ... this.
6786 (vec_initv2sf): Renamed to ...
6787 (vec_initv2sfsf): ... this.
6788 (vec_extractv2sf): Renamed to ...
6789 (vec_extractv2sfsf): ... this.
6790 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6791 (vec_init<mode>): Renamed to ...
6792 (vec_init<mode><VEC_base_l>): ... this.
6793 (vec_extract<mode>): Renamed to ...
6794 (vec_extract<mode><VEC_base_l>): ... this.
6795 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6796 (vec_initv2sfsf): ... this.
6797 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6798 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6799 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6800 gen_vec_init* calls.
6802 2017-08-01 Richard Biener <rguenther@suse.de>
6804 PR tree-optimization/81297
6805 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6806 TREE_OVERFLOW from INTEGER_CSTs.
6808 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6810 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6812 2017-07-31 Carl Love <cel@us.ibm.com>
6814 * config/rs6000/rs6000-c: Add support for built-in functions
6815 vector signed char vec_xl_be (signed long long, signed char *);
6816 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6817 vector signed int vec_xl_be (signed long long, signed int *);
6818 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6819 vector signed long long vec_xl_be (signed long long, signed long long *);
6820 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6821 vector signed short vec_xl_be (signed long long, signed short *);
6822 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6823 vector double vec_xl_be (signed long long, double *);
6824 vector float vec_xl_be (signed long long, float *);
6825 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6826 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6827 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6829 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6830 (altivec_expand_builtin): Add switch statement to call
6831 altivec_expand_xl_be for each builtin.
6832 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6833 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6834 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6835 __builtin_vsx_le_be_v16qi.
6836 * doc/extend.texi: Update the built-in documentation file for the
6837 new built-in functions.
6839 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6842 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6844 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6846 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6848 * config.gcc: Add z14.
6849 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6850 CPU model numbers for z13s and z14.
6851 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6853 * config/s390/s390-opts.h (enum processor_type): Rename
6854 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6855 * config/s390/s390.c (processor_table): Add field for CPU name to
6856 be passed to Binutils.
6857 (s390_asm_output_machine_for_arch): Use the new field in
6858 processor_table for Binutils.
6859 (s390_expand_builtin): Replace arch12 with z14.
6860 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6861 (s390_get_sched_attrmask): Likewise.
6862 (s390_get_unit_mask): Likewise.
6863 * config/s390/s390.opt: Add z14 to processor_type enum.
6865 2017-07-31 Martin Jambor <mjambor@suse.cz>
6868 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6869 regardless of optimization level.
6871 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6872 Martin Liska <mliska@suse.cz>
6874 * predict.def: Remove old comment and adjust probability.
6875 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6878 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6881 * config/i386/i386.c (ix86_function_naked): New function.
6882 (ix86_can_use_return_insn_p): Return false for naked functions.
6883 (ix86_expand_prologue): Skip prologue for naked functions.
6884 (ix86_expand_epilogue): Skip epilogue for naked functions
6885 and emit trap instruction.
6886 (ix86_warn_func_return): New function.
6887 (ix86_attribute_table): Add "naked" attribute specification.
6888 (TARGET_WARN_FUNC_RETURN): Define.
6889 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6891 2017-07-31 Martin Liska <mliska@suse.cz>
6893 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6894 (dump_gimple_bb_header): Always dump BB info.
6895 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6897 2017-07-31 Martin Liska <mliska@suse.cz>
6900 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6901 also with current_function_decl non-null equality.
6903 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6906 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6907 change type to the element type, instead add eltype variable and
6908 use it where we are interested in the element type.
6910 PR tree-optimization/81603
6911 * ipa-polymorphic-call.c
6912 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6913 offset arithmetic in offset_int, bail out if the resulting bit offset
6914 doesn't fit into shwi.
6916 2017-07-31 Martin Liska <mliska@suse.cz>
6918 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6919 (gimplify_save_expr): Fix comment.
6921 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6924 * config/i386/i386.c (ix86_function_arg): Update arguments for
6926 (ix86_compute_frame_layout): Set the initial stack offset to
6927 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6928 INCOMING_FRAME_SP_OFFSET.
6929 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6930 stack before exception handler returns.
6931 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6932 the 'ERROR_CODE' for exception handler.
6934 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6936 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6937 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6938 (ASM_OUTPUT_REG_POP): Ditto.
6939 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6940 instead of asm_fprintf to output pure string.
6942 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6944 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6945 to imported_module_or_decl hook.
6946 (debug_nothing_tree_tree_tree_bool): Remove.
6947 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6948 * debug.c (do_nothing_debug_hooks): Use
6949 debug_nothing_tree_tree_tree_bool_bool instead of
6950 debug_nothing_tree_tree_tree_bool.
6951 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6952 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6953 * sdbout.c (sdb_debug_hooks): Likewise.
6954 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6955 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6957 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6958 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6959 attribute, don't add anything.
6961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6963 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6964 (fold_build2_stat_loc): Likewise.
6965 (fold_build3_stat_loc): Likewise.
6966 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6967 (fold_build1_loc): Remove macro.
6968 (fold_build2_loc): Likewise.
6969 (fold_build3_loc): Likewise.
6971 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6973 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6974 (gimple_build_debug_bind_source_stat): Likewise.
6975 * gimple.h (gimple_build_debug_bind): Remove macro.
6976 (gimple_build_debug_bind_source): Likewise.
6978 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6980 * bitmap.c (bitmap_alloc): Adjust.
6981 (bitmap_gc_alloc): Likewise.
6982 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6984 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6986 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6987 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6988 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6989 (bitmap_gc_alloc_stat): Likewise.
6990 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6992 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6994 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6995 * rtl.h (shallow_copy_rtx): Remove macro.
6997 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6999 * emit-rtl.c (gen_raw_REG): Adjust.
7000 * gengenrtl.c (gendef): Likewise.
7001 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7002 * rtl.h (rtx_alloc): Remove macro.
7004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7006 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7007 (build_tree_list_stat): Likewise.
7008 * tree.h (build_tree_list): Remove macro.
7009 (build_tree_list_vec): Likewise.
7011 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7013 * tree.c (make_vector_stat): Remove _stat from name.
7014 (build_vector_stat): Likewise.
7015 * tree.h (make_vector_stat): Remove macro.
7016 (build_vector_stat): Likewise.
7018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7020 * tree.h (build_var_debug_value): Remove prototype.
7022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7024 * tree.c (tree_cons_stat): Remove _stat from name.
7025 * tree.h (tree_cons): Remove macro.
7027 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7029 * tree.c (build_vl_exp_stat): Remove _stat from name.
7030 * tree.h (build_vl_exp): Remove macro.
7032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7034 * tree.c (build_decl_stat): Remove _stat from name.
7035 * tree.h (build_decl): Remove macro.
7037 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7039 * gimple.c (gimple_build_with_ops_stat): Adjust.
7040 (gimple_alloc_stat): Remove _stat from name.
7041 * gimple.h (gimple_alloc): Remove macro.
7043 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7045 * tree.c (make_tree_vec_stat): Remove _stat from name.
7046 (grow_tree_vec_stat): Likewise.
7047 * tree.h (make_tree_vec_stat): Adjust prototype.
7048 (grow_tree_vec_stat): Likewise.
7049 (make_tree_vec): Remove macro.
7050 (grow_tree_vec): Likewise.
7052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7054 * fold-const.c (fold_build1_stat_loc): Adjust.
7055 (fold_build2_stat_loc): Likewise.
7056 (fold_build3_stat_loc): Likewise.
7057 * tree.c (build0_stat): Remove _stat from name.
7058 (build1_stat): Likewise.
7059 (build2_stat): Likewise.
7060 (build3_stat): Likewise.
7061 (build4_stat): Likewise.
7062 (build5_stat): Likewise.
7063 * tree.h (build1_loc): Remove macro, and rename _stat function
7065 (build2_loc): Likewise.
7066 (build3_loc): Likewise.
7067 (build4_loc): Likewise.
7068 (build5_loc): Likewise.
7070 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7072 * tree.c (make_int_cst_stat): Remove _stat from name.
7073 * tree.h (make_int_cst_stat): Adjust prototype.
7074 (make_int_cst): Remove macro.
7076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7078 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7079 * tree.h (make_tree_binfo_stat): Adjust prototype.
7080 (make_tree_binfo): Remove.
7082 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7084 * tree.c (copy_node_stat): Rename to copy_node.
7085 (build_distinct_type_copy): Adjust.
7086 * tree.h (copy_node_stat): Adjust prototype.
7087 (copy_node): Remove macro.
7089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7091 * tree.c (make_node_stat): rename to make_node.
7092 (build_tree_list_stat): Adjust.
7093 (build0_stat): Likewise.
7094 (build2_stat): Likewise.
7095 (build3_stat): Likewise.
7096 (build4_stat): Likewise.
7097 (build5_stat): Likewise.
7098 (build_decl_stat): Likewise.
7099 * tree.h (make_node_stat): Adjust prototype.
7100 (make_node): remove macro.
7102 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7104 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7105 (PPC_FEATURE2_SCV): Likewise.
7106 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7108 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7110 * config/aarch64/aarch64.c
7111 (aarch64_internal_mov_immediate): Add new special pattern.
7112 * config/aarch64/aarch64.md (*movdi_aarch64):
7113 Add reg/32bit const mov case.
7115 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7116 Richard Sandiford <richard.sandiford@linaro.org>
7118 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7119 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7120 Add integer and movi cases.
7121 (movi-split-hf-df-sf split, fp16): New.
7122 (enabled): Added TARGET_FP_F16INST.
7123 * config/aarch64/iterators.md (GPF_HF): New.
7124 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7126 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7128 * config/aarch64/aarch64.c
7129 (aarch64_simd_container_mode): Add prototype.
7130 (aarch64_expand_mov_immediate): Add HI support.
7131 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7132 (aarch64_can_const_movi_rtx_p): New.
7133 (aarch64_preferred_reload_class):
7134 Remove restrictions of using FP registers for certain SIMD operations.
7135 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7136 (aarch64_valid_floating_const): Add integer move validation.
7137 (aarch64_simd_imm_scalar_p): Remove.
7138 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7139 (aarch64_legitimate_constant_p): Expand list of supported cases.
7140 * config/aarch64/aarch64-protos.h
7141 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7142 (aarch64_reinterpret_float_as_int): New.
7143 (aarch64_simd_imm_scalar_p): Remove.
7144 * config/aarch64/constraints.md (Uvi): New.
7145 (Dd): Split into Ds and new Dd.
7146 * config/aarch64/aarch64.md (*movsi_aarch64):
7148 (*movdi_aarch64): Add SIMD mov case.
7150 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7152 * tree-predcom.c: (struct chain): Handle store-store chain in which
7153 stores for elimination only store loop invariant values.
7154 (execute_pred_commoning_chain): Ditto.
7155 (prepare_initializers_chain_store_elim): Ditto.
7156 (prepare_finalizers): Ditto.
7157 (is_inv_store_elimination_chain): New function.
7158 (initialize_root_vars_store_elim_1): New function.
7160 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7162 * tree-predcom.c: Revise general description of the pass.
7163 (enum chain_type): New enum type for store elimination.
7164 (struct chain): New field supporting store elimination.
7165 (struct component): Ditto.
7166 (dump_chain): Dump store-stores chain.
7167 (release_chain): Release resources.
7168 (split_data_refs_to_components): Compute and create component
7169 contains only stores for elimination.
7170 (get_chain_last_ref_at): New function.
7171 (make_invariant_chain): Initialization.
7172 (make_rooted_chain): Specify chain type in parameter and record it.
7173 (add_looparound_copies): Skip for store-stores chain.
7174 (determine_roots_comp): Compute type of chain and pass it to
7176 (initialize_root_vars_store_elim_2): New function.
7177 (finalize_eliminated_stores): New function.
7178 (remove_stmt): Handle store for elimination.
7179 (execute_pred_commoning_chain): Execute predictive commoning on
7181 (determine_unroll_factor): Skip unroll for store-stores chain.
7182 (prepare_initializers_chain_store_elim): New function.
7183 (prepare_initializers_chain): Hanlde store-store chain.
7184 (prepare_finalizers_chain, prepare_finalizers): New function.
7185 (tree_predictive_commoning_loop): Return integer value indicating
7186 if loop is unrolled or lcssa form is corrupted.
7187 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7189 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7191 * tree-predcom.c (initialize_root): Delete.
7192 (execute_pred_commoning_chain): Initialize root vars and replace
7193 reference of non-combined chain directly, rather than call above
7196 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7198 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7199 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7201 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7203 * tree-predcom.c (struct chain): New field init_seq.
7204 (release_chain): Release init_seq.
7205 (prepare_initializers_chain): Record intialization stmts in above
7207 (insert_init_seqs): New function.
7208 (tree_predictive_commoning_loop): Call insert_init_seqs.
7210 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7212 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7214 2017-07-28 Richard Biener <rguenther@suse.de>
7216 * match.pd: Remove superfluous :c.
7217 * genmatch.c (simplify::id): Add member.
7218 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7220 (current_id): New global.
7221 (dt_node::parent): Move from ...
7222 (dt_operand::parent): ... here. Add for_id member.
7223 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7224 (decision_tree::find_node): Relax order requirement when
7225 merging DT_TRUE nodes to ones inbetween the current simplify
7226 and the one we try to merge with. Add diagnostic whenever
7227 we need to enforce pattern order by not merging.
7228 (decision_tree::insert): Set current_id.
7229 (decision_tree::print_node): Dump parent node and for_id.
7230 (parser::last_id): Add member.
7231 (parser::push_simplify): Assign unique id.
7232 (parser::parser): Initialize last_id.
7234 2017-07-28 Martin Liska <mliska@suse.cz>
7237 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7238 gimple_build_debug_bind.
7240 2017-07-28 Richard Biener <rguenther@suse.de>
7242 PR tree-optimization/81502
7243 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7245 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7247 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7248 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7249 for BIT_FIELD_REF args.
7250 * fold-const.c (make_bit_field_ref): Likewise.
7251 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7253 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7256 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7257 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7258 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7259 Or it into SANITIZER_UNDEFINED.
7260 * ubsan.c: Include gimple-fold.h and varasm.h.
7261 (ubsan_expand_ptr_ifn): New function.
7262 (instrument_pointer_overflow): New function.
7263 (maybe_instrument_pointer_overflow): New function.
7264 (instrument_object_size): Formatting fix.
7265 (pass_ubsan::execute): Call instrument_pointer_overflow
7266 and maybe_instrument_pointer_overflow.
7267 * internal-fn.c (expand_UBSAN_PTR): New function.
7268 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7269 * sanitizer.def (__ubsan_handle_pointer_overflow,
7270 __ubsan_handle_pointer_overflow_abort): New builtins.
7271 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7272 * internal-fn.def (UBSAN_PTR): New internal function.
7273 * opts.c (sanitizer_opts): Add pointer-overflow.
7274 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7275 * fold-const.c (build_range_check): Compute pointer range check in
7276 integral type if pointer arithmetics would be needed. Formatting
7279 2017-07-28 Martin Liska <mliska@suse.cz>
7282 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7283 parameters that are of a variable-length.
7285 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7287 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7289 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7290 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7291 (CRT_CALL_STATIC_FUNCTION): Likewise.
7292 (ASM_DEFAULT_SPEC): New define.
7293 (ASM_SPEC32): Likewise.
7294 (ASM_SPEC64): Likewise.
7295 (ASM_SPEC_COMMON): Likewise.
7296 (ASM_SPEC): Likewise.
7297 (INVALID_64BIT): Likewise.
7298 (LINK_OS_DEFAULT_SPEC): Likewise.
7299 (LINK_OS_SPEC32): Likewise.
7300 (LINK_OS_SPEC64): Likewise.
7301 (POWERPC_LINUX): Likewise.
7302 (PTRDIFF_TYPE): Likewise.
7303 (RESTORE_FP_PREFIX): Likewise.
7304 (RESTORE_FP_SUFFIX): Likewise.
7305 (SAVE_FP_PREFIX): Likewise.
7306 (SAVE_FP_SUFFIX): Likewise.
7307 (SIZE_TYPE): Likewise.
7308 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7309 (TARGET_64BIT): Likewise.
7310 (TARGET_64BIT): Likewise.
7311 (TARGET_AIX): Likewise.
7312 (WCHAR_TYPE_SIZE): Likewise.
7313 (WCHAR_TYPE): Undefine.
7314 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7315 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7316 (CPP_OS_RTEMS_SPEC): Delete.
7317 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7318 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7320 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7322 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7324 PR tree-optimization/81578
7325 * tree-parloops.c (build_new_reduction): Bail out if
7326 reduction_code isn't one of the standard OpenMP reductions.
7327 Move the details printing after that decision.
7329 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7331 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7332 related to reload_in_progress.
7333 (splat_input_operand): Likewise.
7334 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7336 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7338 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7339 (TARGET_INSTANTIATE_DECLS): Likewise.
7340 (legitimate_indexed_address_p): Delete reload_in_progress code.
7341 (rs6000_debug_legitimate_address_p): Likewise.
7342 (rs6000_eliminate_indexed_memrefs): Likewise.
7343 (rs6000_emit_le_vsx_store): Likewise.
7344 (rs6000_emit_move_si_sf_subreg): Likewise.
7345 (rs6000_emit_move): Likewise.
7346 (register_to_reg_type): Likewise.
7347 (rs6000_pre_atomic_barrier): Likewise.
7348 (rs6000_machopic_legitimize_pic_address): Likewise.
7349 (rs6000_allocate_stack_temp): Likewise.
7350 (rs6000_address_for_fpconvert): Likewise.
7351 (rs6000_address_for_altivec): Likewise.
7352 (rs6000_secondary_memory_needed_rtx): Delete function.
7353 (rs6000_check_sdmode): Likewise.
7354 (rs6000_alloc_sdmode_stack_slot): Likewise.
7355 (rs6000_instantiate_decls): Likewise.
7356 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7357 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7358 Delete reload_in_progress.
7359 (*vec_reload_and_plus_<mptrsize>): Likewise.
7360 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7361 (vsx_div_v2di): Likewise.
7362 (vsx_udiv_v2di): Likewise.
7364 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7366 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7367 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7368 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7369 (rs6000_debug_reg_global): Delete print of LRA status.
7370 (rs6000_option_override_internal): Delete dead LRA related code.
7371 (rs6000_lra_p): Delete function.
7372 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7374 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7376 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7377 * config/riscv/rtems.h: New file.
7379 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7380 Sudakshina Das <sudi.das@arm.com>
7382 * config/aarch64/aarch64.md
7383 (define_split for and<mode>3nr_compare): Move
7384 non aarch64_logical_operand to a register.
7385 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7386 register immediate operand to a register.
7387 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7389 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7392 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7394 2017-07-27 Richard Biener <rguenther@suse.de>
7396 PR tree-optimization/81573
7397 PR tree-optimization/81494
7398 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7399 multi defuse cycle case.
7401 2017-07-27 Richard Biener <rguenther@suse.de>
7403 PR tree-optimization/81571
7404 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7407 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7409 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7410 earlier and only if MASK_FPU is set. Adjust formatting.
7412 2017-07-27 Martin Liska <mliska@suse.cz>
7414 * opt-functions.awk: Add validation of value of Init.
7415 * optc-gen.awk: Pass new argument.
7417 2017-07-27 Martin Liska <mliska@suse.cz>
7419 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7420 Fix wrong condition.
7422 2017-07-27 Martin Liska <mliska@suse.cz>
7424 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7425 BBs and edges seen by autoFDO.
7427 2017-07-27 Richard Biener <rguenther@suse.de>
7429 PR tree-optimization/81502
7430 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7431 with incompatible but same sized type.
7432 (execute_update_addresses_taken): Likewise.
7434 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7436 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7437 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7439 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7442 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7443 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7444 Change s_operand to memory_operand.
7446 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7448 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7449 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7450 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7451 Emit instructions rather than returning an expression. Handle TFmode
7452 and KFmode by casting to TImode.
7453 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7454 (rs6000_emit_le_vsx_store): Likewise.
7455 * config/rs6000/vsx.md (VSX_TI): New iterator.
7456 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7457 (*vsx_le_undo_permute_<mode>): Likewise.
7458 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7459 emit the split sequence.
7460 (*vsx_le_perm_store_<mode>): Likewise.
7462 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7464 PR tree-optimization/81555
7465 PR tree-optimization/81556
7466 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7467 if true, force CHANGED for the recursive invocation.
7468 (reassociate_bb): Remember original length of ops array, pass
7469 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7471 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7472 attributes for noipa attribute. For naked attribute use
7473 lookup_attribute first before lookup_attribute_spec.
7474 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7476 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7477 for functions with noipa attribute.
7478 (cgraph_externally_visible_p): Return true for functions with noipa
7480 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7481 for functions with noipa attribute.
7482 * doc/extend.texi: Document noipa function attribute.
7483 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7484 also for functions with noipa attribute.
7485 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7487 2017-07-26 Andrew Pinski <apinski@cavium.com>
7489 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7490 vec_unalign_load_cost and vec_unalign_store_cost.
7492 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7494 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7495 -mvsx-small-integer option.
7496 (ISA_3_0_MASKS_IEEE): Likewise.
7497 (OTHER_VSX_VECTOR_MASKS): Likewise.
7498 (POWERPC_MASKS): Likewise.
7499 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7500 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7501 code, only testing for DImode being allowed in non-VSX floating
7503 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7504 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7506 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7507 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7508 TARGET_P8_VECTOR test.
7509 (rs6000_secondary_reload_simple_move): Likewise.
7510 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7511 since TARGET_P9_VECTOR was already tested.
7512 (rs6000_opt_masks): Remove -mvsx-small-integer.
7513 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7514 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7516 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7517 test for TARGET_VEXTRACTUB was used, and that uses
7519 (p9 extract splitter): Likewise.
7520 (vsx_extract_<mode>_di_p9): Likewise.
7521 (vsx_extract_<mode>_store_p9): Likewise.
7522 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7523 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7524 the elimination of TARGET_VSX_SMALL_INTEGER.
7525 (vsx_extract_<mode>_p8): Likewise.
7526 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7527 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7528 (vsx_set_<mode>_p9): Likewise.
7529 (vsx_set_v4sf_p9): Likewise.
7530 (vsx_set_v4sf_p9_zero): Likewise.
7531 (vsx_insert_extract_v4sf_p9): Likewise.
7532 (vsx_insert_extract_v4sf_p9_2): Likewise.
7533 * config/rs6000/rs6000.md (sign extend splitter): Change
7534 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7535 (floatsi<mode>2_lfiwax_mem): Likewise.
7536 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7537 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7538 since a test for TARGET_P9_VECTOR was used.
7539 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7540 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7541 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7542 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7543 TARGET_P8_VECTOR test.
7544 (fix_trunc<mode>si2_stfiwx): Likewise.
7545 (fix_trunc<mode>si2_internal): Likewise.
7546 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7547 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7549 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7550 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7551 TARGET_P8_VECTOR test.
7552 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7553 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7554 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7556 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7557 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7558 since a test for TARGET_P9_VECTOR was used.
7559 (splitter for loading small constants): Likewise.
7561 2017-07-26 Andrew Pinski <apinski@cavium.com>
7563 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7566 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
7569 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
7570 (fp_valid_at): Likewise.
7572 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7574 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
7575 (qdf24xx_addrcost_table): Likewise.
7576 (cortexa57_tunings): Update to use generic_branch_cost.
7577 (cortexa72_tunings): Likewise.
7578 (cortexa73_tunings): Likewise.
7579 (qdf24xx_tunings): Likewise.
7581 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
7583 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
7584 (thunderx2t99_branch_cost): Likewise.
7585 (cortexa35_tunings): Update to use generic_branch_cost.
7586 (cortexa53_tunings): Likewise.
7587 (cortexa57_tunings): Likewise.
7588 (cortexa72_tunings): Likewise.
7589 (cortexa73_tunings): Likewise.
7590 (thunderx2t99_tunings): Likewise.
7592 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7594 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
7595 (sparc_option_override): Honour MASK_FSMULD.
7596 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
7597 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
7598 * config/sparc/sparc.opt (mfsmuld): New option.
7599 * doc/invoke.texi (mfsmuld): Document option.
7601 2017-07-26 Marek Polacek <polacek@redhat.com>
7604 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7606 2017-07-26 Richard Biener <rguenther@suse.de>
7608 * gimple-match-head.c (do_valueize): Return OP if valueize
7610 (get_def): New helper to get at the def stmt of a SSA name
7612 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7613 do_valueize to get at the def stmt.
7614 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7616 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7619 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7621 2017-07-26 Martin Liska <mliska@suse.cz>
7624 * function.c (expand_function_start): Make expansion of
7625 nonlocal_goto_save_area after parm_birth_insn.
7627 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7629 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7630 from all CPU target flags enable members.
7632 2017-07-26 Richard Biener <rguenther@suse.de>
7634 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7635 (decision_tree::gen): Make 'type' const.
7636 (write_predicate): Likewise.
7638 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7640 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7641 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7642 (rs6000_option_override_internal): Likewise.
7643 (rs6000_expand_vector_set): Likewise.
7644 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7645 (TARGET_UPPER_REGS_SF): Likewise.
7646 (TARGET_UPPER_REGS_DI): Likewise.
7647 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7648 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7649 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7650 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7651 (Splitters for DI constants in Altivec registers): Likewise.
7652 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7653 (vsx_set_v4sf_p9): Likewise.
7654 (vsx_set_v4sf_p9_zero): Likewise.
7655 (vsx_insert_extract_v4sf_p9): Likewise.
7656 (vsx_insert_extract_v4sf_p9_2): Likewise.
7658 2017-07-25 Carl Love <cel@us.ibm.com>
7660 * doc/extend.texi: Update the built-in documentation file for the
7661 existing built-in functions
7662 vector signed char vec_cnttz (vector signed char);
7663 vector unsigned char vec_cnttz (vector unsigned char);
7664 vector signed short vec_cnttz (vector signed short);
7665 vector unsigned short vec_cnttz (vector unsigned short);
7666 vector signed int vec_cnttz (vector signed int);
7667 vector unsigned int vec_cnttz (vector unsigned int);
7668 vector signed long long vec_cnttz (vector signed long long);
7669 vector unsigned long long vec_cnttz (vector unsigned long long);
7671 2017-07-25 Andrew Pinski <apinski@cavium.com>
7673 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7674 accesses where the use is for the first operand of a BIT_INSERT.
7676 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7679 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7680 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7682 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7684 * config/i386/gstabs.h: Delete.
7685 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7687 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7689 * config/i386/i386.c (ix86_decompose_address): Do not check for
7690 register RTX when looking at index_reg or base_reg.
7691 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7693 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7695 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7696 to update EH info here.
7698 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7700 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7702 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7704 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7706 2017-07-25 Torsten Duwe <duwe@suse.de>
7708 * common.opt: Introduce -fpatchable-function-entry
7709 command line option, and its variables function_entry_patch_area_size
7710 and function_entry_patch_area_start.
7711 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7712 including a two-value parser.
7713 * target.def (print_patchable_function_entry): New target hook.
7714 * targhooks.h (default_print_patchable_function_entry): New function.
7715 * targhooks.c (default_print_patchable_function_entry): Likewise.
7716 * toplev.c (process_options): Switch off IPA-RA if
7717 patchable function entries are being generated.
7718 * varasm.c (assemble_start_function): Look at the
7719 patchable-function-entry command line switch and current
7720 function attributes and maybe generate NOP instructions by
7721 calling the print_patchable_function_entry hook.
7722 * doc/extend.texi: Document patchable_function_entry attribute.
7723 * doc/invoke.texi: Document -fpatchable_function_entry
7724 command line option.
7725 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7727 * doc/tm.texi: Re-generate.
7729 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7732 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7733 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7735 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7737 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7739 2017-07-25 Richard Biener <rguenther@suse.de>
7741 PR tree-optimization/81455
7742 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7743 not walk in cycles when looking for guards.
7745 2017-07-25 Richard Biener <rguenther@suse.de>
7747 PR tree-optimization/81529
7748 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7749 when optimizing backedge uses.
7751 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7753 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7755 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7756 to dl_section_ref. On AIX, append an expression to subtract
7757 the size of the section length to dl_section_ref.
7759 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7761 * configure.ac: If any of the config.* scripts fail, exit 1.
7762 * configure: Regenerate.
7764 2017-07-25 Richard Biener <rguenther@suse.de>
7767 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7768 of immediate uses, be more verbose on errors.
7770 2017-07-25 Richard Biener <rguenther@suse.de>
7772 PR tree-optimization/81510
7773 * tree-vect-loop.c (vect_is_simple_reduction): When the
7774 reduction stmt is not inside the loop bail out.
7776 2017-07-25 Richard Biener <rguenther@suse.de>
7778 PR tree-optimization/81303
7779 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7780 profitability check against LOOP_VINFO_NITERSM1.
7782 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7784 * domwalk.c (cmp_bb_postorder): Simplify.
7785 (sort_bbs_postorder): New function. Use it...
7786 (dom_walker::walk): ...here to optimize common cases.
7788 2017-07-25 Martin Liska <mliska@suse.cz>
7791 * ipa-visibility.c (function_and_variable_visibility): Make the
7792 redirection just on target that supports aliasing.
7793 Fix GNU coding style.
7795 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7798 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7800 * config/arm/rtems.h: Likewise.
7801 * config/bfin/rtems.h: Likewise.
7802 * config/i386/rtemself.h: Likewise.
7803 * config/lm32/rtems.h: Likewise.
7804 * config/m32c/rtems.h: Likewise.
7805 * config/m68k/rtemself.h: Likewise.
7806 * config/microblaze/rtems.h: Likewise.
7807 * config/mips/rtems.h: Likewise.
7808 * config/moxie/rtems.h: Likewise.
7809 * config/nios2/rtems.h: Likewise.
7810 * config/powerpcspe/rtems.h: Likewise.
7811 * config/rs6000/rtems.h: Likewise.
7812 * config/rtems.h: Likewise.
7813 * config/sh/rtems.h: Likewise.
7814 * config/sh/rtemself.h: Likewise.
7815 * config/sparc/rtemself.h: Likewise.
7817 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7820 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7821 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7822 * tree-ssa-structalias.c (alias_get_name): Same.
7824 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7827 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7828 instructions if no du chain is found.
7830 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7832 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7834 2017-07-25 Richard Biener <rguenther@suse.de>
7837 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7840 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7842 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7844 (ISA_2_7_MASKS_SERVER): Likewise.
7845 (ISA_3_0_MASKS_IEEE): Likewise.
7846 (OTHER_P8_VECTOR_MASKS): Likewise.
7847 (OTHER_VSX_VECTOR_MASKS): Likewise.
7848 (POWERPC_MASKS): Likewise.
7849 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7850 duplicate list of options.
7851 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7852 explicit -mupper-regs options.
7853 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7854 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7855 alias for -mupper-regs-df.
7856 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7857 (rs6000_init_hard_regno_mode_ok): Likewise.
7858 (rs6000_option_override_internal): Likewise.
7859 (rs6000_opt_masks): Likewise.
7860 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7861 options in terms of whether -mvsx or -mpower8-vector was used.
7862 (TARGET_UPPER_REGS_DI): Likewise.
7863 (TARGET_UPPER_REGS_SF): Likewise.
7864 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7865 -mupper-regs-* options.
7867 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7869 * passes.c (emergency_dump_function): Print some empty lines and a
7870 header before the RTL dump.
7872 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7874 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7876 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7879 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7880 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7882 2017-07-24 Carl Love <cel@us.ibm.com>
7884 * config/rs6000/rs6000-c.c: Add support for built-in functions
7885 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7886 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7887 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7888 vec_extract_fp_from_shortl): Add defines for the two builtins.
7889 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7890 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7892 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7893 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7894 * doc/extend.texi: Update the built-in documentation file for the
7895 new built-in function.
7897 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7900 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7902 * doc/generic.texi: Likewise.
7903 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7904 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7906 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7908 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7909 (aarch64_mls_elt_merge<mode>): Likewise.
7911 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7913 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7914 having __cxa_atexit.
7916 2017-07-23 Michael Collison <michael.collison@arm.com>
7918 * config/arm/arm.c (arm_option_override): Deprecate
7919 use of -mstructure-size-boundary.
7920 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7921 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7923 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7926 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7927 Reduce cost estimate for direct moves.
7929 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7932 * config/i386/i386.c (ix86_option_override_internal): Disable
7933 BMI, BMI2 and TBM instructions for -m16.
7935 2017-07-21 Carl Love <cel@us.ibm.com>
7937 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7938 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7939 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7940 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7941 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7942 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7943 VMULOSW): New enum "unspec" values.
7944 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7945 altivec_vmulosw): New patterns.
7946 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7947 VMULOSW): Add definitions.
7949 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7951 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7952 (qdf24xx): Likewise.
7953 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7954 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7955 (AARCH64_FL_V8_1): Renumber.
7956 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7957 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7958 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7959 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7960 rdma to feature modifiers list.
7962 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7965 * ipa-visibility (function_and_variable_visibility): Convert
7966 recursive PLT call to direct call if appropriate.
7968 2017-07-21 Andrew Pinski <apinski@cavium.com>
7970 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7971 operand 1 to see if the types precision matches.
7972 * fold-const.c (operand_equal_p): Likewise.
7974 2017-07-21 Richard Biener <rguenther@suse.de>
7976 PR tree-optimization/81303
7977 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7978 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7979 (vect_peeling_hash_get_lowest_cost): Adjust.
7980 (vect_enhance_data_refs_alignment): Likewise. Use
7981 vect_get_peeling_costs_all_drs to compute the penalty for no
7982 peeling to match up costs.
7984 2017-07-21 Richard Biener <rguenther@suse.de>
7986 PR tree-optimization/81500
7987 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7988 we didn't identify a reduction path.
7990 2017-07-21 Tom de Vries <tom@codesourcery.com>
7991 Cesar Philippidis <cesar@codesourcery.com>
7993 PR gcov-profile/81442
7994 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7997 2017-07-21 Tom de Vries <tom@codesourcery.com>
8000 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8002 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8003 nvptx_override_options_after_change.
8005 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8007 * dwarf2out.c (output_file_names): Avoid double testing for
8010 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8012 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8014 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8016 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8018 (try_crossjump_to_edge): Do not punt on partitioned functions.
8020 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8022 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8023 Put all BBs reachable only via paths crossing cold region to cold
8025 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8027 2016-07-21 Richard Biener <rguenther@suse.de>
8029 PR tree-optimization/81303
8030 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8031 into account prologue and epilogue iterations when raising
8032 min_profitable_iters to sth at least covering one vector iteration.
8034 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8036 * config/arm/arm.c (arm_test_cpu_arch_dat):
8039 2017-07-20 Nathan Sidwell <nathan@acm.org>
8041 Remove TYPE_METHODS.
8042 * tree.h (TYPE_METHODS): Delete.
8043 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8044 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8045 (dbxout_type_methods): Scan TYPE_FIELDS.
8046 (dbxout_type): Don't check TYPE_METHODS here.
8047 * function.c (use_register_for_decl): Always ignore register for
8048 class types when not optimizing.
8049 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8050 * tree.c (free_lang_data_in_type): Stitch out member functions and
8051 templates from TYPE_FIELDS.
8052 (build_distinct_type_copy, verify_type_variant,
8053 verify_type): Member fns are on TYPE_FIELDS.
8054 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8055 * tree-pretty-print.c (dump_generic_node): Likewise.
8057 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8060 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8061 V2TImode and V4TImode.
8062 (ix86_expand_vector_extract): Likewise.
8063 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8064 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8065 (ssescalarmode): Handle V4TImode and V2TImode.
8066 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8067 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8068 (VEXTRACTI128_MODE): New mode iterator.
8069 (splitter for *vec_extractv?ti first element): New.
8070 (VEC_INIT_MODE): New mode iterator.
8071 (vec_init<mode>): Consolidate 3 expanders into one using
8072 VEC_INIT_MODE mode iterator.
8074 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8076 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8077 non_spilled_static_chain_regno_p.
8079 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8081 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8083 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8085 * bb-reorder.c (connect_traces): Allow copying of blocks within
8088 2017-07-20 Richard Biener <rguenther@suse.de>
8090 * gimple.h (gimple_phi_result): Add gphi * overload.
8091 (gimple_phi_result_ptr): Likewise.
8092 (gimple_phi_arg): Likewise. Adjust index assert to only
8093 allow actual argument accesses rather than all slots available
8095 (gimple_phi_arg_def): Add gphi * overload.
8096 * tree-phinodes.c (make_phi_node): Initialize only actual
8098 (resize_phi_node): Clear memory not covered by old node,
8099 do not initialize excess argument slots.
8100 (reserve_phi_args_for_new_edge): Initialize new argument slot
8103 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8105 PR tree-optimization/81388
8107 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8109 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8110 by removing computation of may_be_zero.
8112 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8113 Tom de Vries <tom@codesourcery.com>
8116 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8117 when gimple level profile disagrees with what RTL expander did.
8119 2017-07-20 Richard Biener <rguenther@suse.de>
8121 PR tree-optimization/61171
8122 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8123 (vect_analyze_stmt): Add slp instance parameter.
8124 (vectorizable_reduction): Likewise.
8125 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8126 (vect_is_simple_reduction): Deal with chains not detected
8127 as SLP reduction chain, specifically not properly associated
8128 chains containing a mix of plus/minus.
8129 (get_reduction_op): Remove.
8130 (get_initial_defs_for_reduction): Simplify, pass in whether
8131 this is a reduction chain, pass in the SLP node for the PHIs.
8132 (vect_create_epilog_for_reduction): Get the SLP instance as
8134 (vectorizable_reduction): Get the SLP instance as arg.
8135 During analysis remember the SLP node with the PHIs in the
8136 instance. Simplify getting at the vectorized reduction PHIs.
8137 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8138 through SLP instance.
8139 (vect_slp_analyze_operations): Likewise.
8140 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8141 (vect_transform_stmt): Likewise.
8143 2017-07-20 Tom de Vries <tom@codesourcery.com>
8145 PR tree-optimization/81489
8146 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8147 read of phi arg location to before loop that modifies phi.
8149 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8151 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8157 * except.c (execute): Fix ordering issue.
8159 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8161 PR rtl-optimization/81423
8162 * combine.c (make_compound_operation_int): Don't try to optimize
8163 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8165 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8167 PR rtl-optimization/81423
8168 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8169 with a constant that is -1 in the truncated to mode.
8171 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8173 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8174 (determine_unlikely_bbs): ... here.
8175 * predict.h (propagate_unlikely_bbs_forward): Declare.
8176 * cfgexpand.c (pass_expand::execute): Use it.
8177 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8179 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8180 propagate_unlikely_bbs_forward.
8182 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8185 * except.c (maybe_add_nop_after_section_switch): New function.
8188 2017-07-19 Tom de Vries <tom@codesourcery.com>
8190 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8192 2017-07-19 Tom de Vries <tom@codesourcery.com>
8194 * gimple.h (gimple_phi_arg): Make assert more strict.
8196 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8198 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8199 * config/rs6000/mmintrin.h: New file.
8200 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8202 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8204 PR tree-optimization/81346
8205 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8207 2017-07-19 Tom de Vries <tom@codesourcery.com>
8209 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8211 2017-07-19 Tom de Vries <tom@codesourcery.com>
8213 * config/nvptx/nvptx-modes.def: Add V2DImode.
8214 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8215 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8216 (nvptx_output_mov_insn): Handle lack of mov.b128.
8217 (nvptx_print_operand): Handle 'H' and 'L' codes.
8218 (nvptx_vector_mode_supported): Allow V2DImode.
8219 (nvptx_preferred_simd_mode): New function.
8220 (nvptx_data_alignment): New function.
8221 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8222 nvptx_preferred_simd_mode.
8223 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8225 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8227 2017-07-19 Tom de Vries <tom@codesourcery.com>
8229 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8230 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8231 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8232 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8233 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8234 (mov<VECIM>_insn): New define_insn.
8235 (define_expand "mov<VECIM>): New define_expand.
8237 2017-07-19 Tom de Vries <tom@codesourcery.com>
8239 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8241 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8243 PR tree-optimization/81346
8244 * fold-const.h (fold_div_compare, range_check_type): Declare.
8245 * fold-const.c (range_check_type): New function.
8246 (build_range_check): Use range_check_type.
8247 (fold_div_compare): No longer static, rewritten into
8248 a match.pd helper function.
8249 (fold_comparison): Don't call fold_div_compare here.
8250 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8253 2017-07-19 Nathan Sidwell <nathan@acm.org>
8255 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8256 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8257 * tree.c (find_decls_types_r, verify_type): Use
8258 TYPE_{MIN,MAX}_VALUE_RAW.
8259 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8260 (hash_tree): Likewise.
8261 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8263 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8266 2017-07-18 Tom de Vries <tom@codesourcery.com>
8269 * omp-expand.c (expand_omp_for_static_chunk): Handle
8270 equal-argument loop exit phi.
8272 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8275 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8276 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8277 operand 2 predicate.
8278 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8279 operand 2 predicate.
8280 (ror,rol -> rorx splitters): Use const_int_operand as
8281 operand 2 predicate.
8283 2017-06-18 Richard Biener <rguenther@suse.de>
8285 PR tree-optimization/81410
8286 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8287 the gap in the ! slp_perm SLP case after each group.
8289 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8292 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8295 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8298 * predict.c (set_even_probabilities): Cleanup; do not affect
8299 probabilities that are already known.
8300 (combine_predictions_for_bb): Call even when count is set.
8302 2017-07-18 Nathan Sidwell <nathan@acm.org>
8304 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8307 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8310 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8311 optimization for loop niter analysis.
8313 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8316 * config/avr/avr.c (avr_optimize_casesi): Don't use
8317 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8319 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8321 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8322 body_cost_vec from _vect_peel_extended_info.
8323 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8324 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8327 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8329 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8331 2017-07-18 Richard Biener <rguenther@suse.de>
8333 PR tree-optimization/80620
8334 PR tree-optimization/81403
8335 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8336 info when re-using a VN table entry.
8338 2017-07-18 Richard Biener <rguenther@suse.de>
8340 PR tree-optimization/81418
8341 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8342 vectype_in. Verify that with lane-reducing reduction operations
8343 we have a single def-use cycle.
8345 2017-07-17 Carl Love <cel@us.ibm.com>
8347 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8350 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8351 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8352 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8353 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8354 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8355 VMULOSW): New enum "unspec" values.
8356 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8357 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8358 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8359 altivec_vmulosw): New patterns.
8360 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8361 VMULOSW): Add definitions.
8362 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8364 * config/alpha/alpha.c: Include predict.h.
8366 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8368 * tree-vrp.c (compare_assert_loc): Fix comparison function
8369 to return predictable results.
8371 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8373 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8376 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8377 * doc/invoke.texi (mexpand-adddi): Update text.
8379 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8381 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8382 that also clobbers the CC register. The old expand code is moved
8384 (*arc_clzsi2): ... here.
8385 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8386 the CC register. The old expand code is moved to ...
8387 (arc_ctzsi2): ... here.
8389 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8391 * config/arc/arc.opt (mindexed-loads): Use initial value
8392 TARGET_INDEXED_LOADS_DEFAULT.
8393 (mauto-modify-reg): Use initial value
8394 TARGET_AUTO_MODIFY_REG_DEFAULT.
8395 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8396 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8397 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8398 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8400 2017-07-17 Martin Liska <mliska@suse.cz>
8403 * opts.c (finish_options): Do not allow -fgnu-tm
8404 w/ -fsanitize={kernel-,}address. Say sorry.
8406 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8409 * tree-loop-distribution.c (classify_partition): Only assert on
8410 numer of iterations.
8411 (merge_dep_scc_partitions): Delete prameter. Update function call.
8412 (distribute_loop): Remove code handling loop with unknown niters.
8413 (pass_loop_distribution::execute): Skip loop with unknown niters.
8415 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8418 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8419 function sort_partitions_by_post_order.
8421 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8423 PR tree-optimization/81374
8424 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8425 the max index of basic blocks, rather than number of basic blocks.
8427 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8429 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8431 (arc_legitimate_pic_operand_p): Likewise.
8432 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8434 (arc_needs_pcl_p): Likewise.
8435 (arc_legitimate_pc_offset_p): Likewise.
8436 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8437 function is also used in constrains.md.
8438 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8439 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8440 PLUS. Only return true/false in known cases, otherwise assert.
8441 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8442 is already called in arc_legitimate_constant_p.
8443 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8445 (LEGITIMATE_PIC_OPERAND_P): Use
8446 arc_raw_symbolic_reference_mentioned_p function.
8447 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8452 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8453 Andrew Burgess <andrew.burgess@embecosm.com>
8455 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8456 (arc_return_address_register): New function.
8457 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8458 (arc_handle_fndecl_attribute): Add naked attribute.
8459 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8460 (TARGET_WARN_FUNC_RETURN): Likewise.
8461 (arc_allocate_stack_slots_for_args): New function.
8462 (arc_warn_func_return): Likewise.
8463 (machine_function): Change type fn_type.
8464 (arc_compute_function_type): Consider new naked function type,
8465 change function return type.
8466 (arc_must_save_register): Adapt to handle new
8467 arc_compute_function_type's return type.
8468 (arc_expand_prologue): Likewise.
8469 (arc_expand_epilogue): Likewise.
8470 (arc_return_address_regs): Delete.
8471 (arc_return_address_register): New function.
8472 (arc_epilogue_uses): Use above function.
8473 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8474 (arc_function_type): Change encoding, add naked type.
8475 (ARC_INTERRUPT_P): Change to handle the new encoding.
8476 (ARC_FAST_INTERRUPT_P): Likewise.
8477 (ARC_NORMAL_P): Define.
8478 (ARC_NAKED_P): Likewise.
8479 (arc_compute_function_type): Delete prototype.
8480 * config/arc/arc.md (in_ret_delay_slot): Use
8481 arc_return_address_register function.
8482 (simple_return): Likewise.
8483 (p_return_i): Likewise.
8485 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8487 PR tree-optimization/81428
8488 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8489 can't be built for those types.
8491 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8493 Remove stuff dead since r239246.
8495 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8496 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8497 (avr_inform_devices): Remove dead stuff.
8499 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8501 * config/arm/arm_neon.h: Fix softp typo.
8503 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8505 PR tree-optimization/81365
8506 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8507 aggregate moves onto bb predecessor edges, make sure there are no
8508 loads that could alias the lhs in between the start of bb and the
8511 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8514 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8515 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8516 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8518 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8520 PR tree-optimization/81396
8521 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8522 (init_symbolic_number): Initialize it to 1.
8523 (perform_symbolic_merge): Add n_ops from both operands into the new
8525 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8526 without base_addr as useless if they need more than one operation.
8527 (bswap_replace): Handle !bswap case for NULL base_addr.
8529 2017-07-17 Tom de Vries <tom@codesourcery.com>
8532 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8535 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8537 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8538 conditional builtin define __FIX_LEON3FT_B2BST.
8540 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8542 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8543 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8546 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8548 PR rtl-optimization/81424
8549 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8550 to remove potential trapping from operands if -fnon-call-exceptions.
8552 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8554 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8555 profile_proability for scalling.
8556 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8558 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8560 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8562 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8564 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8565 fixpoint arithmetics.
8567 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8569 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
8570 fixpoint arithmetics.
8572 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8574 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
8575 fixpoint arithmetics.
8577 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8579 * profile-count.h (profile_probability::from_reg_br_prob_note,
8580 profile_probability::to_reg_br_prob_note): New functions.
8581 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
8582 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
8583 * predict.c (probability_reliable_p): Update.
8584 (edge_probability_reliable_p): Update.
8585 (br_prob_note_reliable_p): Update.
8586 (invert_br_probabilities): Update.
8587 (add_reg_br_prob_note): New function.
8588 (combine_predictions_for_insn): Update.
8589 * asan.c (asan_clear_shadow): Update.
8590 * cfgbuild.c (compute_outgoing_frequencies): Update.
8591 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
8592 (update_br_prob_note): Update.
8593 (rtl_verify_edges): Update.
8594 (purge_dead_edges): Update.
8595 (fixup_reorder_chain): Update.
8596 * emit-rtl.c (try_split): Update.
8597 * ifcvt.c (cond_exec_process_insns): Update.
8598 (cond_exec_process_if_block): Update.
8599 (dead_or_predicable): Update.
8600 * internal-fn.c (expand_addsub_overflow): Update.
8601 (expand_neg_overflow): Update.
8602 (expand_mul_overflow): Update.
8603 * loop-doloop.c (doloop_modify): Update.
8604 * loop-unroll.c (compare_and_jump_seq): Update.
8605 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8606 * predict.h: Update.
8607 * reorg.c (mostly_true_jump): Update.
8609 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8610 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8611 * config/arc/arc.c: (emit_unlikely_jump): Update.
8612 * config/arm/arm.c: (emit_unlikely_jump): Update.
8613 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8614 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8615 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8616 (ix86_print_operand): Update.
8617 (ix86_split_fp_branch): Update.
8618 (predict_jump): Update.
8619 * config/ia64/ia64.c (ia64_print_operand): Update.
8620 * config/mmix/mmix.c (mmix_print_operand): Update.
8621 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8622 (rs6000_expand_split_stack_prologue): Update.
8623 * config/rs6000/rs6000.c: Update.
8624 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8625 (s390_expand_vec_movstr): Update.
8626 (s390_expand_cs_tdsi): Update.
8627 (s390_expand_split_stack_prologue): Update.
8628 * config/sh/sh.c (sh_print_operand): Update.
8629 (expand_cbranchsi4): Update.
8630 (expand_cbranchdi4): Update.
8631 * config/sparc/sparc.c (output_v9branch): Update.
8632 * config/spu/spu.c (get_branch_target): Update.
8633 (ea_load_store_inline): Update.
8634 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8635 * config/tilepro/tilepro.c: Update.
8637 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8639 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8640 (gimplify_save_expr): Likewise.
8642 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8644 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8646 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8648 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8650 * timevar.def (TV_IPA_FNSUMMARY): Define.
8652 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8654 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8655 to back store errata sensitive sequence from being generated.
8656 (sqrtdf2_fix): Likewise.
8658 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8660 * tree-ssa-threadupdate.c (compute_path_counts,
8661 update_joiner_offpath_counts): Use profile_probability.
8663 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8666 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8668 * config/arm/arm-c.c (arm_cpu_builtins): Define
8669 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8671 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8673 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8674 array entries to represent __ieee128 versions of the
8675 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8676 scalar_extract_sig, and scalar_insert_exp built-in functions.
8677 (altivec_resolve_overloaded_builtin): Add special case handling
8678 for the __builtin_scalar_insert_exp function, as represented by
8679 the P9V_BUILTIN_VEC_VSIEDP constant.
8680 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8681 exponent support for __ieee128 argument.
8682 (VSESQP): Add scalar extract signature support for __ieee128
8684 (VSTDCNQP): Add scalar test negative support for __ieee128
8686 (VSIEQP): Add scalar insert exponent support for __int128 argument
8687 with __ieee128 result.
8688 (VSIEQPF): Add scalar insert exponent support for __ieee128
8689 argument with __ieee128 result.
8690 (VSTDCQP): Add scalar test data class support for __ieee128
8692 (VSTDCNQP): Add overload support for scalar test negative with
8694 (VSTDCQP): Add overload support for scalar test data class
8696 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8698 (UNSPEC_VSX_SIEXPQP): New constant.
8699 (xsxexpqp): New insn for VSX scalar extract exponent quad
8701 (xsxsigqp): New insn for VSX scalar extract significand quad
8703 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8704 precision with floating point argument.
8705 (xststdcqp): New expand for VSX scalar test data class quad
8707 (xststdcnegqp): New expand for VSX scalar test negative quad
8709 (xststdcqp): New insn to match expansions for VSX scalar test data
8710 class quad precision and VSX scalar test negative quad precision.
8711 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8712 special case operand checking to enforce that second operand of
8713 VSX scalar test data class with quad precision argument is a 7-bit
8715 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8716 prototypes and descriptions of __ieee128 versions of
8717 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8718 scalar_test_data_class, and scalar_test_neg built-in functions.
8720 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8722 PR tree-optimization/81162
8723 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8724 replace a negate with an add.
8726 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8728 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8730 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8732 * config/arm/arm-c.c (arm_cpu_builtins): Define
8733 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8735 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8737 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8738 (armv8-r): Set ARM Cortex-R52 as default CPU.
8739 * config/arm/arm-tables.opt: Regenerate.
8740 * config/arm/arm-tune.md: Regenerate.
8741 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8743 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8744 extension for -mcpu=cortex-r52.
8746 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8748 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8749 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8750 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8751 (fp-armv8): Define it as FP_ARMv8 only.
8752 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8753 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8755 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8756 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8757 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8758 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8759 than TARGET_FPU_ARMV8.
8760 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8761 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8762 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8764 * config/arm/neon.md (neon_vrint): Likewise.
8765 (neon_vcvt): Likewise.
8766 (neon_<fmaxmin_op><mode>): Likewise.
8767 (<fmaxmin><mode>3): Likewise.
8768 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8769 * config/arm/predicates.md (arm_cond_move_operator): Check against
8770 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8772 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8774 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8777 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8779 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8780 loop in comment with memset.
8782 2017-07-14 Martin Liska <mliska@suse.cz>
8784 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8785 * dwarf2out.c (is_java): Remove the function.
8786 (output_pubname): Remove usage of the function.
8787 (lower_bound_default): Remove usage of DW_LANG_Java.
8788 (gen_compile_unit_die): Likewise.
8789 * gcc.c: Remove compiler defaults for .java and .zip files.
8790 * gimple-expr.c (remove_suffix): Change as there's no longer
8791 extension than 4-letter one.
8792 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8793 (gimplify_save_expr): Likewise.
8794 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8795 as it's possible even for other languages than Java.
8796 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8797 * lto-opts.c (lto_write_options): Remove reference to Java.
8798 * opts.c (strip_off_ending): Update file extension handling.
8799 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8800 * tree-eh.c (lower_resx): Likewise.
8801 * tree.c (free_lang_data_in_type): Remove dead code.
8802 (find_decls_types_r): Likewise.
8803 (build_common_builtin_nodes): Remove Java from a comment.
8804 (verify_type): Remove dead code.
8805 * varasm.c (assemble_external): Remove Java from a comment.
8807 2017-07-14 Martin Liska <mliska@suse.cz>
8809 * opts.c (finish_options): Add quotes.
8810 (common_handle_option): Likewise.
8812 2017-07-14 Martin Liska <mliska@suse.cz>
8814 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8815 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8817 * dwarf2out.c (lower_bound_default): Do not handle
8819 (gen_compile_unit_die): Likewise.
8820 * gcc.c: Remove default extension binding for GNU Pascal.
8821 * stmt.c: Remove Pascal language from a comment.
8822 * xcoffout.c: Likewise.
8824 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8827 * diagnostic-show-locus.c (fixit_cmp): New function.
8828 (layout::layout): Sort m_fixit_hints.
8829 (column_range::column_range): Assert that the values are valid.
8830 (struct char_span): New struct.
8831 (correction::overwrite): New method.
8832 (struct source_line): New struct.
8833 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8834 calls in terms of classes source_line and char_span, and
8835 correction::overwrite.
8836 (selftest::test_overlapped_fixit_printing_2): New function.
8837 (selftest::diagnostic_show_locus_c_tests): Call it.
8839 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8841 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8842 early if there is no lhs.
8844 2017-07-13 Martin Liska <mliska@suse.cz>
8846 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8847 (gen_reference_type_die): Likewise.
8848 * stor-layout.c: Remove Pascal-related comment.
8850 2017-07-13 Martin Liska <mliska@suse.cz>
8852 * opts.c (finish_options): Add quotes to error messages.
8853 (parse_sanitizer_options): Likewise.
8855 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8857 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8859 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8861 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8863 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8865 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8867 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8869 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8872 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8873 provides the hardware capability bits, define the macro
8874 __BUILTIN_CPU_SUPPORTS__.
8875 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8876 if GLIBC does not provide the hardware capability bits. Add a
8877 gcc_unreachable call if the built-in cpu function is neither
8878 __builtin_cpu_is nor __builtin_cpu_supports.
8879 (rs6000_get_function_versions_dispatcher): Change the warning
8880 that an old GLIBC is used which does not export the capability
8881 bits to be an error.
8882 * doc/extend.texi (target_clones attribute): Document the
8883 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8884 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8885 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8886 the macros defined by GCC if the newer GLIBC is available.
8888 2017-07-12 Jeff Law <law@redhat.com>
8890 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8891 remaining includes slightly.
8892 * config/riscv/riscv-builtins.c: Include profile-count.h.
8894 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8897 * config/avr/avr.c (avr_set_current_function): In diagnostic
8898 messages: Quote keywords and (parts of) identifiers.
8899 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8902 2017-07-12 Carl Love <cel@us.ibm.com>
8904 * config/rs6000/rs6000-c.c: Add support for built-in functions
8905 vector bool char vec_revb (vector bool char);
8906 vector bool short vec_revb (vector short char);
8907 vector bool int vec_revb (vector bool int);
8908 vector bool long long vec_revb (vector bool long long);
8909 * doc/extend.texi: Update the built-in documentation file for the
8910 new built-in functions.
8912 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8914 * config/s390/s390.md: Remove movcc splitter.
8916 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8918 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8919 load/store on condition.
8921 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8924 * config/avr/avr.c (avr_encode_section_info)
8925 [progmem && !TREE_READONLY]: Error if progmem object needs
8928 2017-07-11 Michael Collison <michael.collison@arm.com>
8930 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8933 2017-07-11 Carl Love <cel@us.ibm.com>
8935 * config/rs6000/rs6000-c.c: Add support for builtins
8936 vector unsigned int vec_parity_lsbb (vector signed int);
8937 vector unsigned int vec_parity_lsbb (vector unsigned int);
8938 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8939 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8940 vector unsigned long long vec_parity_lsbb (vector signed long long);
8941 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8942 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8943 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8944 * doc/extend.texi: Update the built-in documentation file for the
8945 new built-in functions.
8947 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8949 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8950 (layout::m_primary_loc): New field.
8951 (layout::layout): Initialize new field. Move location filtering
8952 logic from here to...
8953 (layout::maybe_add_location_range): ...this new method. Add
8954 support for filtering to just the lines already specified by other
8956 (layout::will_show_line_p): New method.
8957 (gcc_rich_location::add_location_if_nearby): New method.
8958 (selftest::test_add_location_if_nearby): New test function.
8959 (selftest::diagnostic_show_locus_c_tests): Call it.
8960 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8963 2017-07-11 Tom de Vries <tom@codesourcery.com>
8965 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8966 (bb_first_real_insn): New function.
8967 (nvptx_single): Add extra initialization of broadcasted condition
8970 2017-07-11 Nathan Sidwell <nathan@acm.org>
8972 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8974 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8976 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8977 Binutils doc as TEXI will mess them up.
8978 * doc/invoke.texi (AVR Options): Same here.
8980 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8982 * config/sparc/sparc.opt (mfix-ut700): New option.
8983 (mfix-gr712rc): Likewise.
8984 (sparc_fix_b2bst): New variable.
8985 * doc/invoke.texi (SPARC options): Document them.
8986 (ARM options): Fix warnings.
8987 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8988 instructions to prevent sequences that can trigger the store-store
8989 errata for certain LEON3FT processors.
8990 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8991 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8992 * config/sparc/sparc.md (fix_b2bst): New attribute.
8993 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8995 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8998 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9000 (*rsqrtsf2_sse): Ditto.
9002 (div<mode>3): Macroize insn from divdf3 and divsf3
9003 using MODEF mode iterator.
9005 2017-07-10 Martin Sebor <msebor@redhat.com>
9007 PR tree-optimization/80397
9008 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9009 instead of testing for equality to INTEGER_TYPE.
9011 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9013 * config.gcc: Remove uclibc from arc target spec.
9015 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9017 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9019 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9022 * lto-wrapper.c (remove_option): New function.
9023 (merge_and_complain): Merge PIC/PIE options more realistically.
9025 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9027 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9031 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9032 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9033 * config.in: Regenerate.
9034 * configure: Regenerate.
9035 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9036 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9037 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9038 (TARGET_GASISR_PROLOGUES): ...target mask.
9039 * common/config/avr/avr-common.c
9040 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9041 Set -mgas-isr-prologues.
9042 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9043 INSERT_PASS_BEFORE for it.
9044 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9045 * config/avr/avr.c (avr_option_override)
9046 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9047 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9048 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9049 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9050 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9052 (make_avr_pass_pre_proep): New function.
9053 (emit_push_sfr) <treg>: Add argument to function and use it
9055 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9056 and set machine->gasisr.yes.
9057 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9058 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9059 __gcc_isr.n_pushed to .L__stack_usage.
9060 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9061 (avr_asm_final_postscan_insn): ...this new static function.
9062 * config/avr/avr.h (machine_function)
9063 <is_no_gccisr, use_L__stack_usage>: New fields.
9064 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9065 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9066 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9067 (gasisr, *gasisr): New expander and insn.
9068 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9069 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9070 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9072 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9074 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9077 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9079 Move jump-tables out of .text again.
9082 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9083 (ASM_OUTPUT_ADDR_VEC): New function.
9084 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9085 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9086 INSN_ADDRESSes as asm comment.
9087 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9088 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9089 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9090 * config/avr/avr.md (*tablejump): Adjust comment.
9091 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9092 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9094 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9095 (avr_output_addr_vec): New proto.
9096 (avr_log_t) <insn_addresses>: New field.
9098 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9101 * config/i386/i386.c (ix86_function_arg_advance): Set
9102 outgoing_args_on_stack to true if there are outgoing arguments
9104 (ix86_function_arg): Likewise.
9105 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9106 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9107 * config/i386/i386.h (machine_function): Add
9108 outgoing_args_on_stack.
9110 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9112 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9114 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9116 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9118 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9119 (REAL_H): Remove $(MACHMODE_H).
9120 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9122 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9123 $(MACHMODE_H) and double-int.h.
9124 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9126 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9129 2017-07-07 Andrew Pinski <apinski@cavium.com>
9131 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9132 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9134 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9136 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9137 Add warning if GCC was not configured to link against a GLIBC that
9138 exports the hardware capability bits.
9139 (make_resolver_func): Make resolver function private and not a
9140 COMDAT function. Create the name with clone_function_name instead
9141 of make_unique_name.
9144 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9145 correct operand in doing the split.
9147 2017-07-07 Carl Love <cel@us.ibm.com>
9149 * config/rs6000/rs6000-c: Add support for built-in function
9150 vector unsigned short vec_pack_to_short_fp32 (vector float,
9152 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9153 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9154 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9155 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9156 (convert_4f32_8i16): Add define_expand.
9157 * doc/extend.texi: Update the built-in documentation file for the
9158 new built-in function.
9160 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9162 * config/sparc/m8.md: New file.
9163 * config/sparc/sparc.md: Include m8.md.
9165 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9167 * config/sparc/sparc.opt: New option -mvis4b.
9168 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9169 (sparc_option_override): Handle VIS4B.
9170 (enum sparc_builtins): Define
9171 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9172 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9173 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9174 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9175 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9176 (check_constant_argument): New function.
9177 (sparc_vis_init_builtins): Define builtins
9178 __builtin_vis_dictunpack{8,16,32},
9179 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9180 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9181 __builtin_vis_fpcmpde{8,16,32}shl and
9182 __builtin_vis_fpcmpur{8,16,32}shl.
9183 (sparc_expand_builtin): Check that the constant operands to
9184 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9185 constant and in range.
9186 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9188 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9189 (SPARC_IMM5_P): Likewise.
9190 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9191 (enabled): Handle vis4b.
9192 (UNSPEC_DICTUNPACK): New unspec.
9193 (UNSPEC_FPCMPSHL): Likewise.
9194 (UNSPEC_FPUCMPSHL): Likewise.
9195 (UNSPEC_FPCMPDESHL): Likewise.
9196 (UNSPEC_FPCMPURSHL): Likewise.
9197 (cpu_feature): New CPU feature `vis4b'.
9198 (dictunpack{8,16,32}): New insns.
9199 (FPCSMODE): New mode iterator.
9200 (fpcscond): New code iterator.
9201 (fpcsucond): Likewise.
9202 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9203 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9204 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9205 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9206 * config/sparc/constraints.md: Define constraints `q' for unsigned
9207 2-bit integer constants and `t' for unsigned 5-bit integer
9209 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9211 (imm5_operand_dictunpack16): Likewise.
9212 (imm5_operand_dictunpack32): Likewise.
9213 (imm2_operand): Likewise.
9214 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9215 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9216 ditunpack* and fpcmp*shl builtins.
9218 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9220 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9221 * config.in: Add HAVE_AS_SPARC6 define.
9222 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9224 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9226 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9227 (CPP_CPU_SPEC): Handle m8.
9228 (ASM_CPU_SPEC): Likewise.
9229 * config/sparc/sparc-opts.h (enum processor_type): Add
9231 * config/sparc/sparc.c (m8_costs): New struct.
9232 (sparc_option_override): Handle TARGET_CPU_m8.
9233 (sparc32_initialize_trampoline): Likewise.
9234 (sparc64_initialize_trampoline): Likewise.
9235 (sparc_issue_rate): Likewise.
9236 (sparc_register_move_cost): Likewise.
9237 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9238 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9239 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9240 (CPP_CPU_SPEC): Handle M8.
9241 (ASM_CPU_SPEC): Likewise.
9242 (AS_M8_FLAG): Define.
9243 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9244 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9245 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9247 * configure: Regenerate.
9248 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9251 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9253 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9255 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9256 ("*movdi_insn_sp32"): Do not set v3pipe.
9257 ("*movsi_insn"): Likewise.
9258 ("*movdi_insn_sp64"): Likewise.
9259 ("*movsf_insn"): Likewise.
9260 ("*movdf_insn_sp32"): Likewise.
9261 ("*movdf_insn_sp64"): Likewise.
9262 ("*zero_extendsidi2_insn_sp64"): Likewise.
9263 ("*sign_extendsidi2_insn"): Likewise.
9264 ("*mov<VM32:mode>_insn"): Likewise.
9265 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9266 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9267 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9268 ("<vlop:code><VL:mode>3"): Likewise.
9269 ("*not_<vlop:code><VL:mode>3"): Likewise.
9270 ("*nand<VL:mode>_vis"): Likewise.
9271 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9272 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9273 ("one_cmpl<VL:mode>2"): Likewise.
9274 ("faligndata<VM64:mode>_vis"): Likewise.
9275 ("alignaddrsi_vis"): Likewise.
9276 ("alignaddrdi_vis"): Likweise.
9277 ("alignaddrlsi_vis"): Likewise.
9278 ("alignaddrldi_vis"): Likewise.
9279 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9280 ("bmaskdi_vis"): Likewise.
9281 ("bmasksi_vis"): Likewise.
9282 ("bshuffle<VM64:mode>_vis"): Likewise.
9283 ("cmask8<P:mode>_vis"): Likewise.
9284 ("cmask16<P:mode>_vis"): Likewise.
9285 ("cmask32<P:mode>_vis"): Likewise.
9286 ("pdistn<P:mode>_vis"): Likewise.
9287 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9289 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9291 * config/sparc/sparc.md ("subtype"): New insn attribute.
9292 ("*wrgsr_sp64"): Set insn subtype.
9293 ("*rdgsr_sp64"): Likewise.
9294 ("alignaddrsi_vis"): Likewise.
9295 ("alignaddrdi_vis"): Likewise.
9296 ("alignaddrlsi_vis"): Likewise.
9297 ("alignaddrldi_vis"): Likewise.
9298 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9299 ("fexpand_vis"): Likewise.
9300 ("fpmerge_vis"): Likewise.
9301 ("faligndata<VM64:mode>_vis"): Likewise.
9302 ("bshuffle<VM64:mode>_vis"): Likewise.
9303 ("cmask8<P:mode>_vis"): Likewise.
9304 ("cmask16<P:mode>_vis"): Likewise.
9305 ("cmask32<P:mode>_vis"): Likewise.
9306 ("fchksm16_vis"): Likewise.
9307 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9308 ("fmean16_vis"): Likewise.
9309 ("fp<plusminus_insn>64_vis"): Likewise.
9310 ("<plusminus_insn>v8qi3"): Likewise.
9311 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9312 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9313 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9314 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9315 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9316 ("*movqi_insn"): Likewise.
9317 ("*movhi_insn"): Likewise.
9318 ("*movsi_insn"): Likewise.
9319 ("movsi_pic_gotdata_op"): Likewise.
9320 ("*movdi_insn_sp32"): Likewise.
9321 ("*movdi_insn_sp64"): Likewise.
9322 ("movdi_pic_gotdata_op"): Likewise.
9323 ("*movsf_insn"): Likewise.
9324 ("*movdf_insn_sp32"): Likewise.
9325 ("*movdf_insn_sp64"): Likewise.
9326 ("*zero_extendhisi2_insn"): Likewise.
9327 ("*zero_extendqihi2_insn"): Likewise.
9328 ("*zero_extendqisi2_insn"): Likewise.
9329 ("*zero_extendqidi2_insn"): Likewise.
9330 ("*zero_extendhidi2_insn"): Likewise.
9331 ("*zero_extendsidi2_insn_sp64"): Likewise.
9332 ("ldfsr"): Likewise.
9333 ("prefetch_64"): Likewise.
9334 ("prefetch_32"): Likewise.
9335 ("tie_ld32"): Likewise.
9336 ("tie_ld64"): Likewise.
9337 ("*tldo_ldub_sp32"): Likewise.
9338 ("*tldo_ldub1_sp32"): Likewise.
9339 ("*tldo_ldub2_sp32"): Likewise.
9340 ("*tldo_ldub_sp64"): Likewise.
9341 ("*tldo_ldub1_sp64"): Likewise.
9342 ("*tldo_ldub2_sp64"): Likewise.
9343 ("*tldo_ldub3_sp64"): Likewise.
9344 ("*tldo_lduh_sp32"): Likewise.
9345 ("*tldo_lduh1_sp32"): Likewise.
9346 ("*tldo_lduh_sp64"): Likewise.
9347 ("*tldo_lduh1_sp64"): Likewise.
9348 ("*tldo_lduh2_sp64"): Likewise.
9349 ("*tldo_lduw_sp32"): Likewise.
9350 ("*tldo_lduw_sp64"): Likewise.
9351 ("*tldo_lduw1_sp64"): Likewise.
9352 ("*tldo_ldx_sp64"): Likewise.
9353 ("*mov<VM32:mode>_insn"): Likewise.
9354 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9355 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9359 * config/sparc/sparc.md ("type"): New insn type viscmp.
9360 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9362 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9363 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9364 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9365 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9367 ("n7_vis_logical_11cycle"): Likewise.
9368 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9369 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9370 * config/sparc/niagara.md ("niag_vis"): Likewise.
9371 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9372 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9374 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9376 * config/sparc/sparc.md: New instruction type `bmask'.
9377 (bmaskdi_vis): Use the `bmask' type.
9378 (bmasksi_vis): Likewise.
9379 * config/sparc/ultra3.md (us3_array): Likewise.
9380 * config/sparc/niagara7.md (n7_array): Likewise.
9381 * config/sparc/niagara4.md (n4_array): Likewise.
9382 * config/sparc/niagara2.md (niag2_vis): Likewise.
9383 (niag3_vis): Likewise.
9384 * config/sparc/niagara.md (niag_vis): Likewise.
9386 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9388 * ipa-comdats.c: Remove optimize check from gate.
9389 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9390 for functions not optimized.
9391 (ipa_fn_summary_read): Skip optimize check.
9392 (ipa_fn_summary_write): Likewise.
9393 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9395 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9397 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9398 (check_callers): Give up on caller which is not optimized.
9399 (inline_small_functions): Likewise.
9400 (ipa_inline): Do not give up when not optimizing.
9401 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9402 away unoptimizes cdtors.
9403 (whole_program_function_and_variable_visibility): Do
9404 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9405 * ipa.c (process_references): Do not check optimize.
9406 (symbol_table::remove_unreachable_nodes): Update optimize check.
9407 (set_writeonly_bit): Update optimize check.
9408 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9409 (pass_ipa_single_use::gate): Remove.
9411 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9413 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9414 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9415 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9416 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9417 permute_load, permute_store, adjust_extract, adjust_splat,
9418 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9419 replace_swap_with_copy, dump_swap_insn_table,
9420 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9421 recombine_lvx_pattern, recombine_stvx_pattern,
9422 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9423 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9424 to file rs6000-p8swap.c.
9425 * config/rs6000/rs6000-p8swap.c: New file.
9426 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9427 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9428 and rs6000*-*-* targets.
9430 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9432 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9434 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9436 * lto-wrapper.c (merge_and_complain): Do not merge
9437 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9438 fsigned_zeros, ftrapping_math, fwrapv.
9439 (append_compiler_options): Do not track these options.
9440 (append_linker_options): Likewie
9442 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9444 * cgraphunit.c (cgraph_node::finalize_function): When
9445 !flag_toplevel_reorde set no_reorder flag.
9446 (varpool_node::finalize_decl): Likewise.
9447 (symbol_table::compile): Drop no toplevel reorder path.
9449 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9451 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9452 edges; zero probability is not better than uninitialized.
9454 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9456 * asan.h (asan_sanitize_allocas_p): Declare.
9457 * asan.c (asan_sanitize_allocas_p): New function.
9458 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9459 (handle_builtin_alloca): Likewise.
9460 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9461 if !asan_sanitize_allocas_p.
9462 * params.def (asan-instrument-allocas): Add new option.
9463 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9464 * opts.c (common_handle_option): Disable allocas sanitization for
9467 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9469 * asan.c: Include gimple-fold.h.
9470 (get_last_alloca_addr): New function.
9471 (handle_builtin_stackrestore): Likewise.
9472 (handle_builtin_alloca): Likewise.
9473 (asan_emit_allocas_unpoison): Likewise.
9474 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9475 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9476 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9477 (instrument_builtin_call): Pass gimple iterator to
9478 get_mem_refs_of_builtin_call.
9479 (last_alloca_addr): New global.
9480 * asan.h (asan_emit_allocas_unpoison): Declare.
9481 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9482 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9483 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9484 if function calls alloca.
9485 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9486 * gimple-fold.h (replace_call_with_value): Declare.
9487 * internal-fn.c: Include asan.h.
9488 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9489 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9491 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9493 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9494 (C_SELFTEST_FLAGS): New.
9495 (CPP_SELFTEST_FLAGS): New.
9496 (SELFTEST_DEPS): New, from deps of s-selftest.
9497 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9498 (CPP_SELFTEST_DEPS): New.
9499 (selftest): Add dependency on s-selftest-c++.
9500 (s-selftest): Rename to...
9501 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9502 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9503 than SELFTEST_FLAGS.
9504 (selftest-gdb): Rename to...
9505 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9507 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9508 (selftest-valgrind): Rename to...
9509 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9511 (selftest-valgrind): Reintroduce as an alias for
9512 selftest-c-valgrind.
9513 (s-selftest-c++): New.
9514 (selftest-c++-gdb): New.
9515 (selftest-c++-valgrind): New.
9517 2017-07-06 Olivier Hainque <hainque@adacore.com>
9519 * gcc.c (process_command): When deciding if undefined variables
9520 should be ignored when processing specs, accept "gcc -v" as well.
9522 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9524 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9525 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9527 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9529 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9530 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9531 * config/arm/arm-tables.opt: Regenerate.
9532 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9534 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9536 2017-07-06 Carl Love <cel@us.ibm.com>
9538 * ChangeLog: Clean up from mid air collision
9540 2017-07-06 Carl Love <cel@us.ibm.com>
9542 * config/rs6000/rs6000-c.c: Add support for built-in functions
9543 vector signed int vec_subc (vector signed int, vector signed int);
9544 vector signed __int128 vec_subc (vector signed __int128,
9545 vector signed __int128);
9546 vector unsigned __int128 vec_subc (vector unsigned __int128,
9547 vector unsigned __int128);
9548 vector signed int vec_sube (vector signed int, vector signed int,
9550 vector unsigned int vec_sube (vector unsigned int,
9551 vector unsigned int,
9552 vector unsigned int);
9553 vector signed __int128 vec_sube (vector signed __int128,
9554 vector signed __int128,
9555 vector signed__int128);
9556 vector unsigned __int128 vec_sube (vector unsigned __int128,
9557 vector unsigned __int128,
9558 vector unsigned __int128);
9559 vector signed int vec_subec (vector signed int, vector signed int,
9561 vector unsigned int vec_subec (vector unsigned int,
9562 vector unsigned int,
9563 vector unsigned int);
9564 vector signed __int128 vec_subec (vector signed __int128,
9565 vector signed __int128,
9566 vector signed__int128);
9567 vector unsigned __int128 vec_subec (vector unsigned __int128,
9568 vector unsigned __int128,
9569 vector unsigned __int128);
9570 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
9571 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
9572 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
9573 BU_ALTIVEC_OVERLOAD_X definitions.
9574 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
9575 * doc/extend.texi: Update the built-in documentation file for the new
9578 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9581 * diagnostic-show-locus.c (layout::layout): Use start and finish
9582 spelling location for the start and finish of each range.
9583 * genmatch.c (linemap_client_expand_location_to_spelling_point):
9584 Add unused aspect param.
9585 * input.c (expand_location_1): Add "aspect" param, and use it
9586 to access the correct part of the location.
9587 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
9589 (expand_location_to_spelling_point): Likewise.
9590 (linemap_client_expand_location_to_spelling_point): Add "aspect"
9591 param, and pass it to expand_location_1.
9593 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
9595 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
9596 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
9597 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
9598 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
9599 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
9600 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
9601 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
9602 _mm_maskz_getmant_ss): New intrinsics.
9603 (__builtin_ia32_getexpss128_mask): Changed to ...
9604 __builtin_ia32_getexpss128_round ... this.
9605 (__builtin_ia32_getexpsd128_mask): Changed to ...
9606 __builtin_ia32_getexpsd128_round ... this.
9607 * config/i386/i386-builtin-types.def
9608 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9609 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9610 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9611 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9612 __builtin_ia32_getmantss_mask_round): New builtins.
9613 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9614 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9615 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9616 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9617 * config/i386/sse.md
9618 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9619 avx512f_sgetexp<mode><mask_scalar_name>
9620 <round_saeonly_scalar_name> ... this.
9621 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9622 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9623 vgetexp<ssescalarmodesuffix>
9624 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9625 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9626 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9627 avx512f_vgetmant<mode><mask_scalar_name>
9628 <round_saeonly_scalar_name> ... this.
9629 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9630 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9631 vgetmant<ssescalarmodesuffix>
9632 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9633 %0<mask_scalar_operand4>, %1, %2
9634 <round_saeonly_scalar_mask_op4>, %3} ... this.
9635 * config/i386/subst.md (mask_scalar_operand4,
9636 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9637 round_saeonly_scalar_nimm_predicate): New subst attributes.
9639 2017-07-06 Julia Koval <julia.koval@intel.com>
9641 * config/i386/i386.c (ix86_erase_embedded_rounding):
9642 Remove code for old rounding pattern.
9644 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9646 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9648 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9650 * doc/sourcebuild.texi (Test Directives, Variants of
9651 dg-require-support): Add documentation for dg-require-stack-check.
9653 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9655 * config/i386/subst.md (mask_scalar, round_scalar,
9656 round_saeonly_scalar): New meta-templates.
9657 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9658 round_scalar_mask_operand3, round_scalar_mask_op3,
9659 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9660 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9661 round_saeonly_scalar_constraint,
9662 round_saeonly_scalar_prefix): New subst attribute.
9663 * config/i386/sse.md
9664 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9665 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9666 <round_scalar_name> ... this.
9667 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9668 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9669 <round_scalar_name> ... this.
9670 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9671 <sse>_vm<code><mode>3<mask_scalar_name>
9672 <round_saeonly_scalar_name> ... this.
9673 (v<plusminus_mnemonic><ssescalarmodesuffix>
9674 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9675 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9676 v<plusminus_mnemonic><ssescalarmodesuffix>
9677 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9678 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9679 (v<multdiv_mnemonic><ssescalarmodesuffix>
9680 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9681 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9682 v<multdiv_mnemonic><ssescalarmodesuffix>
9683 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9684 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9685 (v<maxmin_float><ssescalarmodesuffix>
9686 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9687 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9688 v<maxmin_float><ssescalarmodesuffix>
9689 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9690 %0<mask_scalar_operand3>, %1, %<iptr>2
9691 <round_saeonly_scalar_mask_op3>} ... this.
9693 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9695 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9696 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9698 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9699 Alan Hayward <alan.hayward@arm.com>
9700 David Sherwood <david.sherwood@arm.com>
9702 * combine.c (simplify_if_then_else): Remove "enum" before
9704 * compare-elim.c (can_eliminate_compare): Likewise.
9705 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9707 (aarch64_lookup_simd_builtin_type): Likewise.
9708 (aarch64_simd_builtin_type): Likewise.
9709 (aarch64_init_simd_builtin_types): Likewise.
9710 (aarch64_simd_expand_args): Likewise.
9711 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9713 (aarch64_reverse_mask): Likewise.
9714 (aarch64_simd_emit_reg_reg_move): Likewise.
9715 (aarch64_gen_adjusted_ldpstp): Likewise.
9716 (aarch64_ccmp_mode_to_code): Likewise.
9717 (aarch64_operands_ok_for_ldpstp): Likewise.
9718 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9719 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9721 (aarch64_min_divisions_for_recip_mul): Likewise.
9722 (aarch64_reassociation_width): Likewise.
9723 (aarch64_get_condition_code_1): Likewise.
9724 (aarch64_simd_emit_reg_reg_move): Likewise.
9725 (aarch64_simd_attr_length_rglist): Likewise.
9726 (aarch64_reverse_mask): Likewise.
9727 (aarch64_operands_ok_for_ldpstp): Likewise.
9728 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9729 (aarch64_gen_adjusted_ldpstp): Likewise.
9730 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9732 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9733 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9734 (arm_lookup_simd_builtin_type): Likewise.
9735 (arm_simd_builtin_type): Likewise.
9736 (arm_init_simd_builtin_types): Likewise.
9737 (arm_expand_builtin_args): Likewise.
9738 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9739 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9740 (ft32_setup_incoming_varargs): Likewise.
9741 (ft32_function_arg): Likewise.
9742 (ft32_function_arg_advance): Likewise.
9743 (ft32_pass_by_reference): Likewise.
9744 (ft32_arg_partial_bytes): Likewise.
9745 (ft32_valid_pointer_mode): Likewise.
9746 (ft32_addr_space_pointer_mode): Likewise.
9747 (ft32_addr_space_legitimate_address_p): Likewise.
9748 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9750 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9751 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9752 (iamcu_alignment): Likewise.
9753 (canonicalize_vector_int_perm): Likewise.
9754 (ix86_noce_conversion_profitable_p): Likewise.
9755 (ix86_mpx_bound_mode): Likewise.
9756 (ix86_operands_ok_for_move_multiple): Likewise.
9757 * config/microblaze/microblaze-protos.h
9758 (microblaze_expand_conditional_branch_reg): Likewise.
9759 * config/microblaze/microblaze.c
9760 (microblaze_expand_conditional_branch_reg): Likewise.
9761 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9763 (rs6000_reassociation_width): Likewise.
9764 (rs6000_invalid_binary_op): Likewise.
9765 (fusion_p9_p): Likewise.
9766 (emit_fusion_p9_load): Likewise.
9767 (emit_fusion_p9_store): Likewise.
9768 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9770 (riscv_hard_regno_mode_ok_p): Likewise.
9771 (riscv_address_insns): Likewise.
9772 (riscv_split_symbol): Likewise.
9773 (riscv_legitimize_move): Likewise.
9774 (riscv_function_value): Likewise.
9775 (riscv_hard_regno_nregs): Likewise.
9776 (riscv_expand_builtin): Likewise.
9777 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9778 (riscv_build_integer): Likewise.
9779 (riscv_split_integer): Likewise.
9780 (riscv_legitimate_constant_p): Likewise.
9781 (riscv_cannot_force_const_mem): Likewise.
9782 (riscv_regno_mode_ok_for_base_p): Likewise.
9783 (riscv_valid_base_register_p): Likewise.
9784 (riscv_valid_offset_p): Likewise.
9785 (riscv_valid_lo_sum_p): Likewise.
9786 (riscv_classify_address): Likewise.
9787 (riscv_legitimate_address_p): Likewise.
9788 (riscv_address_insns): Likewise.
9789 (riscv_load_store_insns): Likewise.
9790 (riscv_force_binary): Likewise.
9791 (riscv_split_symbol): Likewise.
9792 (riscv_force_address): Likewise.
9793 (riscv_legitimize_address): Likewise.
9794 (riscv_move_integer): Likewise.
9795 (riscv_legitimize_const_move): Likewise.
9796 (riscv_legitimize_move): Likewise.
9797 (riscv_address_cost): Likewise.
9798 (riscv_subword): Likewise.
9799 (riscv_output_move): Likewise.
9800 (riscv_canonicalize_int_order_test): Likewise.
9801 (riscv_emit_int_order_test): Likewise.
9802 (riscv_function_arg_boundary): Likewise.
9803 (riscv_pass_mode_in_fpr_p): Likewise.
9804 (riscv_pass_fpr_single): Likewise.
9805 (riscv_pass_fpr_pair): Likewise.
9806 (riscv_get_arg_info): Likewise.
9807 (riscv_function_arg): Likewise.
9808 (riscv_function_arg_advance): Likewise.
9809 (riscv_arg_partial_bytes): Likewise.
9810 (riscv_function_value): Likewise.
9811 (riscv_pass_by_reference): Likewise.
9812 (riscv_setup_incoming_varargs): Likewise.
9813 (riscv_print_operand): Likewise.
9814 (riscv_elf_select_rtx_section): Likewise.
9815 (riscv_save_restore_reg): Likewise.
9816 (riscv_for_each_saved_reg): Likewise.
9817 (riscv_register_move_cost): Likewise.
9818 (riscv_hard_regno_mode_ok_p): Likewise.
9819 (riscv_hard_regno_nregs): Likewise.
9820 (riscv_class_max_nregs): Likewise.
9821 (riscv_memory_move_cost): Likewise.
9822 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9823 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9824 (rl78_addr_space_address_mode): Likewise.
9825 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9827 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9828 (rs6000_reassociation_width): Likewise.
9829 (rs6000_invalid_binary_op): Likewise.
9830 (fusion_p9_p): Likewise.
9831 (emit_fusion_p9_load): Likewise.
9832 (emit_fusion_p9_store): Likewise.
9833 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9834 (ok_for_simple_move_operands): Likewise.
9835 (ok_for_simple_move_strict_operands): Likewise.
9836 (ok_for_simple_arith_logic_operands): Likewise.
9837 (visium_legitimize_reload_address): Likewise.
9838 (visium_select_cc_mode): Likewise.
9839 (output_cbranch): Likewise.
9840 (visium_split_double_move): Likewise.
9841 (visium_expand_copysign): Likewise.
9842 (visium_expand_int_cstore): Likewise.
9843 (visium_expand_fp_cstore): Likewise.
9844 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9845 (visium_function_arg): Likewise.
9846 (visium_function_arg_advance): Likewise.
9847 (visium_libcall_value): Likewise.
9848 (visium_setup_incoming_varargs): Likewise.
9849 (visium_legitimate_constant_p): Likewise.
9850 (visium_legitimate_address_p): Likewise.
9851 (visium_legitimize_address): Likewise.
9852 (visium_secondary_reload): Likewise.
9853 (visium_register_move_cost): Likewise.
9854 (visium_memory_move_cost): Likewise.
9855 (prepare_move_operands): Likewise.
9856 (ok_for_simple_move_operands): Likewise.
9857 (ok_for_simple_move_strict_operands): Likewise.
9858 (ok_for_simple_arith_logic_operands): Likewise.
9859 (visium_function_value_1): Likewise.
9860 (rtx_ok_for_offset_p): Likewise.
9861 (visium_legitimize_reload_address): Likewise.
9862 (visium_split_double_move): Likewise.
9863 (visium_expand_copysign): Likewise.
9864 (visium_expand_int_cstore): Likewise.
9865 (visium_expand_fp_cstore): Likewise.
9866 (visium_split_cstore): Likewise.
9867 (visium_select_cc_mode): Likewise.
9868 (visium_split_cbranch): Likewise.
9869 (output_cbranch): Likewise.
9870 (visium_print_operand_address): Likewise.
9871 * expmed.c (flip_storage_order): Likewise.
9872 * expmed.h (emit_cstore): Likewise.
9873 (flip_storage_order): Likewise.
9874 * genrecog.c (validate_pattern): Likewise.
9875 * hsa-gen.c (gen_hsa_addr): Likewise.
9876 * internal-fn.c (expand_arith_overflow): Likewise.
9877 * ira-color.c (allocno_copy_cost_saving): Likewise.
9878 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9879 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9880 (process_invariant_for_inheritance): Likewise.
9881 * lra-eliminations.c (move_plus_up): Likewise.
9882 * omp-low.c (lower_oacc_reductions): Likewise.
9883 * simplify-rtx.c (simplify_subreg): Likewise.
9884 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9885 (TARGET_CHKP_BOUND_MODE): Likewise..
9886 * targhooks.c (default_chkp_bound_mode): Likewise.
9887 (default_setup_incoming_vararg_bounds): Likewise.
9888 * targhooks.h (default_chkp_bound_mode): Likewise.
9889 (default_setup_incoming_vararg_bounds): Likewise.
9890 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9891 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9892 (have_whole_vector_shift): Likewise.
9893 * tree-vect-stmts.c (vectorizable_load): Likewise.
9894 * doc/tm.texi: Regenerate.
9896 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9898 Graceful degrade if Binutils PR21472 is not available.
9901 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9902 .rodata in flash test fails.
9903 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9904 * confgure: Regenerate.
9905 * config.in: Regenerate.
9906 * config/avr/avr.c (avr_asm_named_section)
9907 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9908 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9909 (avr_asm_init_sections): Same.
9911 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9913 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9914 (fma<VH:mode>4_intrinsic): Likewise.
9915 (*fmsub<VCVTF:mode>4): Likewise.
9916 (*fmsub<VH:mode>4_intrinsic): Likewise.
9918 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9921 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9922 Don't depend on "optimize > 0".
9923 (out_movhi_r_mr, out_movqi_mr_r): Same.
9924 (out_movhi_mr_r, out_movqi_r_mr): Same.
9925 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9926 io_address_operand on "optimize > 0".
9928 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9930 * tree-loop-distribution.c: Add general explanantion on the pass.
9931 (generate_loops_for_partition): Mark distributed loop.
9932 (pg_add_dependence_edges): New parameter. Handle alias data
9933 dependence specially and record it in the parameter if asked.
9934 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9935 (init_partition_graph_vertices, add_partition_graph_edge): New.
9936 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9937 (free_partition_graph_vdata, build_partition_graph): New.
9938 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9939 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9940 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9941 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9942 (version_for_distribution_p, finalize_partitions): New.
9943 (distribute_loop): Handle alias data dependence specially. Factor
9944 out loop fusion code as functions and call these functions.
9946 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9948 * tree-loop-distribution.c (classify_partition): New parameter and
9949 better handle reduction statement.
9950 (rdg_build_partitions): Revise comment.
9951 (distribute_loop): Compute statements in all partitions and pass it
9952 to classify_partition.
9954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9956 * tree-loop-distribution.c (enum partition_type): New.
9957 (struct partition): New field type.
9958 (partition_merge_into): Add parameter. Update partition type.
9959 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9960 (build_rdg_partition_for_vertex): Compute partition type.
9961 (rdg_build_partitions): Dump partition type.
9962 (distribute_loop): Update calls to partition_merge_into.
9964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9966 * tree-loop-distribution.c (struct ddr_hasher): New.
9967 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9969 (classify_partition): Call get_data_dependence.
9970 (pg_add_dependence_edges): Ditto.
9971 (distribute_loop): Release data dependence hash table.
9973 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9975 * tree-loop-distribution.c (ref_base_address): Delete.
9976 (similar_memory_accesses): Rename ...
9977 (share_memory_accesses): ... to this. Check if partitions access
9978 the same memory reference.
9979 (distribute_loop): Call share_memory_accesses.
9981 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9983 * tree-loop-distribution.c (struct partition): New field recording
9985 (partition_alloc, partition_free): Init and release data refs.
9986 (partition_merge_into): Merge data refs.
9987 (build_rdg_partition_for_vertex): Collect data refs for partition.
9988 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9990 (distribute_loop): Remve data refs from vertice data of partition
9993 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9995 * tree-loop-distribution.c (params.h): Include header file.
9996 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9997 (datarefs_vec): New global var.
9998 (create_rdg_vertices): Use datarefs_vec directly.
9999 (free_rdg): Don't free data references.
10000 (build_rdg): Update use. Don't free data references.
10001 (distribute_loop): Compute global variable for data references.
10002 Bail out if there are too many data references.
10004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10006 * tree-loop-distribution.c (loop_nest): New global var.
10007 (build_rdg): Use loop directly, rather than loop nest.
10008 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10010 (distribute_loop): Compute global variable loop nest. Update use.
10012 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10014 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10015 (partition_merge_into): New parameter. Dump reason for fusion.
10016 (distribute_loop): Update use of partition_merge_into.
10018 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10020 * tree-loop-distribution.c (bb_top_order_index): New.
10021 (bb_top_order_index_size, bb_top_order_cmp): New.
10022 (stmts_from_loop): Use topological order.
10023 (pass_loop_distribution::execute): Compute and release topological
10024 order for basic blocks.
10026 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10028 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10031 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10033 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10034 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10035 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10036 * internal-fn.def (LOOP_DIST_ALIAS): New.
10037 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10038 (fold_loop_internal_call): ... this.
10039 (vect_loop_dist_alias_call): New function.
10040 (set_uid_loop_bbs): Call fold_loop_internal_call.
10041 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10044 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10047 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10048 Require dead FLAGS_REG at the beginning of a peephole.
10050 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10053 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10054 arguments in the call to __builtin_ia32_sbb_u32.
10055 (_subborrow_u64): Swap _X and _Y arguments in the call to
10056 __builtin_ia32_sbb_u64.
10058 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10061 * tree-vrp.c (compare_assert_loc): Turn into a function template
10062 with stable template parameter. Only test if a->e is NULL,
10063 !a->e == !b->e has been verified already. Use e == NULL or
10064 e != NULL instead of e or ! e tests. If stable is true, don't use
10065 iterative_hash_expr, on the other side allow a or b or both NULL
10066 and sort the NULLs last.
10067 (process_assert_insertions): Sort using compare_assert_loc<false>
10068 instead of compare_assert_loc, later sort using
10069 compare_assert_loc<true> before calling process_assert_insertions_for
10070 in a loop. Use break instead of continue once seen NULL pointer.
10072 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10074 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10075 Cortex-R7 and Cortex-R8 processors.
10077 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10079 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10080 uninitialized while src is not.
10082 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10084 * common/config/arm/arm-common.c: Adjust include path for
10086 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10087 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10088 (arm-cpu-data.h): Likewise.
10089 (arm-cpu-cdata.h): Likewise.
10090 * config/arm/arm-cpu.h: Delete.
10091 * config/arm/arm-cpu-cdata.h: Delete.
10092 * config/arm/arm-cpu-data.h: Delete.
10094 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10096 * config/arm/arm-cpus.in (cortex-a55): New.
10097 (cortex-a75): Likewise.
10098 (cortex-a75.cortex-a55): Likewise.
10099 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10101 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10102 * config/arm/arm-cpu-cdata.h: Regenerate.
10103 * config/arm/arm-cpu-data.h: Regenerate.
10104 * config/arm/arm-cpu.h: Regenerate.
10105 * config/arm/arm-tables.opt: Regenerate.
10106 * config/arm/arm-tune.md: Regenerate.
10108 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10110 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10112 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10114 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10117 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10119 PR tree-optimization/81292
10120 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10121 full_string_p, also call adjust_related_strinfos if the adjustment
10122 is simple, otherwise invalidate related strinfos.
10124 2017-07-04 Martin Liska <mliska@suse.cz>
10127 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10128 newly created variable as DECL_IGNORED_P.
10130 2017-07-04 Martin Liska <mliska@suse.cz>
10133 * ipa-inline.c (inline_small_functions):
10134 Use xstrdup_for_dump.
10136 2017-07-04 Tom de Vries <tom@codesourcery.com>
10138 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10140 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10143 * config/darwin.c (darwin_function_switched_text_sections):
10146 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10148 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10150 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10152 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10154 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10156 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10157 min_profitable_iters, and th as inclusive lower bounds.
10158 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10159 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10160 for min_profitable_iters and min_profitable_estimate.
10161 (vect_transform_loop): Treat th as an inclusive lower bound.
10162 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10164 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10167 * config/darwin.c (darwin_function_switched_text_sections):
10168 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10169 in two pieces, and suppress the use of buf.
10171 2017-07-03 Nathan Sidwell <nathan@acm.org>
10173 * hash-table.h (hash_table_mod1): Fix indentation.
10175 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10177 PR middle-end/81290
10178 * predict.c (force_edge_cold): Be more careful about propagation
10180 * profile-count.h (profile_probability::guessed,
10181 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10183 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10185 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10187 * doc/invoke.texi (rcpc architecture extension): Document it.
10189 2017-07-03 Richard Biener <rguenther@suse.de>
10191 PR tree-optimization/60510
10192 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10193 the scalar reduction PHI and use it.
10194 (vectorizable_reduction): Properly guard the single_defuse_cycle
10195 path for non-SLP reduction chains where we cannot use it.
10196 Rework reduc_def/index and vector type deduction. Rework
10197 vector operand gathering during reduction op code-gen.
10198 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10199 chains dissolve the chain and leave it to non-SLP reduction
10202 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10204 * tree-data-ref.h (dr_alignment): Declare.
10205 * tree-data-ref.c (dr_alignment): New function.
10206 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10207 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10209 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10211 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10213 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10214 and base_misalignment fields.
10215 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10216 * tree-data-ref.c: Include builtins.h.
10217 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10218 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10219 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10220 * tree-vect-data-refs.c: Include tree-cfg.h.
10221 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10222 fields instead of calculating an alignment here.
10223 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10224 innermost_loop_behavior fields.
10226 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10228 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10230 (DR_STEP_ALIGNMENT): New macro.
10231 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10232 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10233 (create_data_ref): Print it.
10234 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10235 to tell whether the step preserves vector (mis)alignment.
10236 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10237 Move the check for an integer step and generalise to all INTEGER_CST.
10238 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10239 Print the outer step alignment.
10241 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10243 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10244 with offset_alignment.
10245 (DR_ALIGNED_TO): Delete.
10246 (DR_OFFSET_ALIGNMENT): New macro.
10247 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10248 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10249 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10250 (create_data_ref): Likewise.
10251 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10252 (vect_analyze_data_refs): Likewise.
10253 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10254 creating dummy innermost behavior.
10256 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10258 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10259 with a "innermost_loop_behavior *" and refeence tree.
10260 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10261 (create_data_ref): Update call accordingly.
10262 * tree-predcom.c (find_looparound_phi): Likewise.
10264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10266 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10267 fields with dr_wrt_vec_loop.
10268 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10269 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10270 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10271 (vect_dr_behavior): New function.
10272 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10273 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10274 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10275 track whether the step preserves the misalignment.
10276 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10277 Use vect_dr_behavior.
10278 (vect_setup_realignment): Update call accordingly.
10279 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10280 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10281 call to vect_create_addr_base_for_vector_ref.
10282 (vect_create_cond_for_align_checks): Likewise.
10283 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10284 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10285 (vect_recog_mask_conversion_pattern): Likewise.
10286 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10287 (new_stmt_vec_info): Remove redundant zeroing.
10289 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10291 * common/config/arm/arm-common.c (arm_be8_option): New function.
10292 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10293 (ISA_ARMv6): Add isa_bit_be8.
10294 * config/arm/arm.h (arm_be8_option): Add prototype.
10295 (BE8_SPEC_FUNCTION): New define.
10296 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10297 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10298 (mlittle-endian): Similarly.
10299 (mbe8, mbe32): New options.
10300 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10301 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10303 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10305 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10307 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10309 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10310 (cleanup_tree_cfg_bb): Use it.
10311 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10313 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10315 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10318 * loop-doloop.c (add_test): Update profile.
10320 2017-07-03 Martin Liska <mliska@suse.cz>
10323 * sanopt.c (rewrite_usage_of_param): New function.
10324 (sanitize_rewrite_addressable_params): Likewise.
10325 (pass_sanopt::execute): Call rewrite_usage_of_param.
10327 2017-07-03 Richard Biener <rguenther@suse.de>
10329 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10330 back to using VIEW_CONVERT_EXPR.
10332 2017-07-03 Martin Liska <mliska@suse.cz>
10335 * doc/extend.texi: Document when a resolver function is
10336 generated for target_clones.
10338 2017-07-03 Martin Liska <mliska@suse.cz>
10340 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10341 and shadow memory of auto variables which are subject of
10342 use-after-scope sanitization.
10343 (asan_expand_mark_ifn): Add do set only when is_poison.
10345 2016-07-03 Richard Biener <rguenther@suse.de>
10347 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10349 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10350 (vectorizable_reduction): Perform reduction PHI creation when
10351 visiting a reduction PHI and adjust and simplify code generation
10352 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10353 (vect_transform_loop): Visit reduction PHIs.
10354 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10355 defs into the SLP tree.
10356 (vect_build_slp_tree): Reduction defs terminate the recursion.
10357 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10359 (vect_get_vec_defs_for_stmt_copy): Export.
10360 (vect_get_vec_defs): Likewise.
10361 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10363 (vect_get_vec_defs_for_stmt_copy): Declare.
10364 (vect_get_vec_defs): Likewise.
10366 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10368 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10369 parameter with a "loop" parameter and use it instead of the
10370 loop containing DR_STMT. Don't check simple_iv when doing
10371 BB analysis. Describe the two analysis modes in the comment.
10373 2017-07-03 Tom de Vries <tom@codesourcery.com>
10375 PR tree-optimization/69468
10376 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10377 (find_same_succ_bb): Handle ignore_edge_flags.
10379 2017-07-03 Tom de Vries <tom@codesourcery.com>
10381 PR tree-optimization/81192
10382 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10384 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10386 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10388 2017-07-03 Tom de Vries <tom@codesourcery.com>
10390 PR tree-optimization/81192
10391 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10392 BB_SAME_SUCC (bb) == NULL.
10394 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10396 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10399 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10401 * dumpfile.c: Include profile-count.h
10402 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10404 (insert_cond_bb): Update profile.
10405 * tree-cfg.h (insert_cond_bb): Update prototype.
10406 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10407 * tree-dump.c: Do not include tree-cfg.
10409 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10411 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10413 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10415 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10418 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10420 * tree-complex.c (expand_complex_div_wide): update profile.
10422 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10423 Alan Hayward <alan.hayward@arm.com>
10424 David Sherwood <david.sherwood@arm.com>
10426 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10427 (CORETYPES_H): New define.
10428 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10429 (insn-modes-inline.h, s-modes-inline-h): New rules.
10430 (generated_files): Add insn-modes-inline.h.
10431 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10432 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10433 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10434 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10435 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10436 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10437 (build/gencodes.o, build/genconditions.o): Likewise.
10438 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10439 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10440 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10441 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10442 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10443 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10444 * coretypes.h: Include everything up to real.h for generators.
10445 Include insn-modes.h first. Include wide-int-print.h after
10446 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10447 * machmode.h: Don't include insn-modes.h here.
10448 * function-tests.c: Remove includes of signop.h, machmode.h,
10449 double-int.h and wide-int.h.
10451 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10453 * optc-save-gen.awk: Likewise.
10454 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10455 * godump.c: Remove include of wide-int-print.h.
10456 * pretty-print.h: Likewise.
10457 * wide-int-print.cc: Likewise.
10458 * wide-int.cc: Likewise.
10459 * hash-map-tests.c: Remove include of signop.h.
10460 * hash-set-tests.c: Likewise.
10461 * rtl-tests.c: Likewise.
10462 * mkconfig.sh: Remove include of machmode.h.
10463 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10465 (emit_insn_modes_inline_h): ...this new function. Emit the code
10466 into an insn-modes-inline.h header file, adding appropriate
10467 include guards and end comments.
10468 (emit_insn_modes_c_header): Remove include of machmode.h.
10469 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10471 (main): Handle -i flag and call emit_insn_modes_inline_h when
10474 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10476 * tree-ssa-strlen.c (strinfo): Rename the length field to
10477 nonzero_chars. Add a full_string_p field.
10478 (compare_nonzero_chars, zero_length_string_p): New functions.
10479 (get_addr_stridx): Add an offset_out parameter.
10480 Use compare_nonzero_chars.
10481 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10482 (new_strinfo): Update after above changes to strinfo.
10483 (set_endptr_and_length): Set full_string_p.
10484 (get_string_length): Update after above changes to strinfo.
10485 (unshare_strinfo): Update call to new_strinfo.
10486 (maybe_invalidate): Likewise.
10487 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10488 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10489 as a uhwi instead of an shwi. Update after above changes to
10490 strinfo and new_strinfo.
10491 (zero_length_string): Assert that chainsi contains full strings.
10492 Use zero_length_string_p. Update call to new_strinfo.
10493 (adjust_related_strinfos): Update after above changes to strinfo.
10494 Copy full_string_p from origsi.
10495 (adjust_last_stmt): Use zero_length_string_p.
10496 (handle_builtin_strlen): Update after above changes to strinfo and
10497 new_strinfo. Install the lhs as the string length if the previous
10498 entry didn't describe a full string.
10499 (handle_builtin_strchr): Update after above changes to strinfo
10501 (handle_builtin_strcpy): Likewise.
10502 (handle_builtin_strcat): Likewise.
10503 (handle_builtin_malloc): Likewise.
10504 (handle_pointer_plus): Likewise.
10505 (handle_builtin_memcpy): Likewise. Track nonzero characters
10506 that aren't necessarily followed by a nul terminator.
10507 (handle_char_store): Likewise.
10509 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10511 PR tree-optimization/80769
10512 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10513 for malloc and calloc. Document the new invariant that all related
10514 strinfos have delayed lengths or none do.
10515 (verify_related_strinfos): Move earlier in file.
10516 (set_endptr_and_length): New function, split out from...
10517 (get_string_length): ...here. Also set the lengths of related
10519 (zero_length_string): Assert that chainsi has known (rather than
10521 (adjust_related_strinfos): Likewise.
10523 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10525 PR tree-optimization/81136
10526 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10527 assert that two references with the same misalignment have the same
10528 compile-time misalignment if those compile-time misalignments
10531 2017-07-01 Andi Kleen <ak@linux.intel.com>
10533 * print-tree.c (print_node): Print all attributes.
10535 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10537 * cfg.c (scale_bbs_frequencies): New function.
10538 * cfg.h (scale_bbs_frequencies): Declare it.
10539 * cfgloopanal.c (single_likely_exit): Cleanup.
10540 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10542 (scale_loop_profile): Likewise.
10543 (loop_version): Likewise.
10544 (create_empty_loop_on_edge): Update.
10545 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10546 scale_loop_frequencies, scale_loop_profile, loopify,
10547 loop_version): Update prototypes.
10548 * modulo-sched.c (sms_schedule): Update.
10549 * predict.c (unlikely_executed_edge_p): Also check probability.
10550 (probably_never_executed_edge_p): Fix typo.
10551 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10552 * tree-parloops.c (gen_parallel_loop): Update.
10553 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10554 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10555 * tree-ssa-loop-split.c (split_loop): Update.
10556 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10557 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10558 (vect_loop_versioning): Update.
10559 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10561 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10563 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10565 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10567 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
10568 to keep profile consistent.
10570 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10572 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
10573 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
10574 * profile-count.h (max_safe_multiplier): Make unsigned.
10575 (profile_count::guessed_zero): New.
10577 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10579 * bb-reorder.c (fix_up_crossing_landing_pad,
10580 fix_crossing_conditional_branches): Use make_single_succ_edge
10581 to keep profile consistent.
10583 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10585 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
10588 2017-07-01 Jakub Jelinek <jakub@redhat.com>
10591 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
10592 the right scopes, make sure cond_jump isn't preserved between multiple
10593 iterations. Search for fallthru edge whenever there are 3+ edges and
10594 use find_fallthru_edge for it.
10596 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10598 Patch by Alexander Monakov <amonakov@ispras.ru>
10599 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
10600 probabilities consistently.
10602 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10604 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10605 emit_cmp_and_jump_insns.
10607 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10610 * tree-inline.c (expand_call_inline): Combine profile statuses.
10612 2017-06-30 Andrew Pinski <apinski@cavium.com>
10614 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10615 fold_stmt returned true.
10617 2017-06-30 Nathan Sidwell <nathan@acm.org>
10619 * ggc.h (empty_string): Delete.
10620 * cfgexpand.c (expand_asm_stmt): Use plain "".
10621 * optabs.c (expand_asm_memory_barrier): Likewise.
10622 * stringpool.c (empty_string): Delete.
10623 (digit_vector, digit_string): Delete.
10624 (ggc_alloc_string): Use plain "", don't optimize single digit
10625 strings. Use ggc_alloc_atomic.
10627 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10629 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10630 comparison set and one other set, use the cost of the non-comparison
10633 2017-06-30 Nathan Sidwell <nathan@acm.org>
10635 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10638 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10640 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10641 loops. Remove now unneeded calls to gimple_switch_set_label() that
10642 just set removed labels to NULL_TREE.
10644 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10646 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10647 (set_range_info): ...here. Only call set_range_info_raw if domain
10649 (set_nonzero_bits): Call set_range_info_raw.
10650 * tree-ssanames.h (set_range_info_raw): New.
10652 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10655 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10656 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10657 of nonimmediate_operand and <store_mask_constraint> instead of m
10658 for the input operand. For V8FI iterator, always split if input
10659 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10660 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10661 <store_mask_predicate> instead of register_operand and
10662 <store_mask_constraint> instead of v for the input operand. Make
10663 sure both operands aren't MEMs for if not <mask_applied>.
10665 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10667 * lto-wrapper.c (copy_file) Close both file descriptors before
10670 2017-06-30 Martin Liska <mliska@suse.cz>
10673 * multiple_target.c (create_dispatcher_calls): Make ifunc
10674 also for function that don't have calls or are not referenced.
10676 2017-06-30 Richard Biener <rguenther@suse.de>
10678 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10679 analyze the first scalar stmt. Move vector type computation
10680 for the BB case here from ...
10681 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10682 live operation processing in the SLP case properly.
10684 2017-06-30 Richard Biener <rguenther@suse.de>
10686 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10688 2017-06-30 Martin Liska <mliska@suse.cz>
10691 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10692 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10694 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10696 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10699 2017-06-30 Richard Biener <rguenther@suse.de>
10701 PR tree-optimization/81249
10702 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10703 condition reduction result to original scalar type.
10705 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10707 * profile-count.h (enum profile_quality): Fix typos and whitespace
10710 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10712 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10713 type for branch probabilities.
10715 2017-06-29 Julian Brown <julian@codesourcery.com>
10716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10718 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10719 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10720 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10721 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10723 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10725 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10726 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10727 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10728 CC usage from generic code to here.
10729 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10730 CC usage into the target macros.
10732 2017-06-29 Maya Rashish <coypu@sdf.org>
10734 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10737 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10739 * arm/arm-builtins.c: Include profile-count.h
10740 * except.c (sjlj_emit_function_enter): Use
10741 profile_probability::unlikely.
10743 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10745 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10746 and tocrel_offset be pointer args rather than implicitly using
10748 (legitimate_constant_pool_address_p, rs6000_emit_move,
10749 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10750 tocrel_offset and use in toc_relative_expr_p call.
10751 (print_operand, print_operand_address): Use static tocrel_base_oac
10752 and tocrel_offset_oac.
10753 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10756 2017-06-29 Maya Rashish <coypu@sdf.org>
10758 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10760 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10762 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10763 objects, take into account only the alignment of 'op0' and 'mode1' if
10766 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10768 * ccmp.c (ccmp_tree_comparison_p): New function.
10769 (ccmp_candidate_p): Update to use above function.
10770 (get_compare_parts): New function.
10771 (expand_ccmp_next): Update to use new functions.
10772 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10774 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10775 take mode as argument.
10776 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10777 * expr.c (expand_expr_real_1): Pass mode as argument.
10779 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10781 * combine.c (combine_instructions): Print insns to dump_file, together
10784 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10786 * asan.c (asan_emit_stack_protection): Update.
10787 (create_cond_insert_point): Update.
10788 * auto-profile.c (afdo_propagate_circuit): Update.
10789 * basic-block.h (struct edge_def): Turn probability to
10790 profile_probability.
10791 (EDGE_FREQUENCY): Update.
10792 * bb-reorder.c (find_traces_1_round): Update.
10793 (better_edge_p): Update.
10794 (sanitize_hot_paths): Update.
10795 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10796 (make_single_succ_edge): Update.
10797 (check_bb_profile): Update.
10798 (dump_edge_info): Update.
10799 (update_bb_profile_for_threading): Update.
10800 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10801 probabilitycount to 0.
10802 * cfgbuild.c (compute_outgoing_frequencies): Update.
10803 * cfgcleanup.c (try_forward_edges): Update.
10804 (outgoing_edges_match): Update.
10805 (try_crossjump_to_edge): Update.
10806 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10807 (expand_gimple_tailcall): Update.
10808 (construct_init_block): Use make_single_succ_edge.
10809 (construct_exit_block): Use make_single_succ_edge.
10810 * cfghooks.c (verify_flow_info): Update.
10811 (redirect_edge_succ_nodup): Update.
10812 (split_edge): Update.
10813 (account_profile_record): Update.
10814 * cfgloopanal.c (single_likely_exit): Update.
10815 * cfgloopmanip.c (scale_loop_profile): Update.
10816 (set_zero_probability): Remove.
10817 (duplicate_loop_to_header_edge): Update.
10818 * cfgloopmanip.h (loop_version): Update prototype.
10819 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10820 (force_nonfallthru_and_redirect): Update.
10821 (update_br_prob_note): Update.
10822 (rtl_verify_edges): Update.
10823 (purge_dead_edges): Update.
10824 (rtl_lv_add_condition_to_bb): Update.
10825 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10826 * cgraphunit.c (init_lowered_empty_function): Update.
10827 (cgraph_node::expand_thunk): Update.
10828 * cilk-common.c: Include profile-count.h
10829 * dojump.c (inv): Remove.
10830 (jumpifnot): Update.
10831 (jumpifnot_1): Update.
10832 (do_jump_1): Update.
10834 (do_jump_by_parts_greater_rtx): Update.
10835 (do_compare_rtx_and_jump): Update.
10836 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10837 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10838 * dwarf2cfi.c: Include profile-count.h
10839 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10840 (sjlj_emit_dispatch_table): Likewise.
10841 * explow.c: Include profile-count.h
10842 * expmed.c (emit_store_flag_force): Update.
10843 (do_cmp_and_jump): Update.
10844 * expr.c (compare_by_pieces_d::generate): Update.
10845 (compare_by_pieces_d::finish_mode): Update.
10846 (emit_block_move_via_loop): Update.
10847 (store_expr_with_bounds): Update.
10848 (store_constructor): Update.
10849 (expand_expr_real_2): Update.
10850 (expand_expr_real_1): Update.
10851 * expr.h (try_casesi, try_tablejump): Update prototypes.
10852 * gimple-pretty-print.c (dump_probability): Update.
10853 (dump_profile): New.
10854 (dump_gimple_label): Update.
10855 (dump_gimple_bb_header): Update.
10856 * graph.c (draw_cfg_node_succ_edges): Update.
10857 * hsa-gen.c (convert_switch_statements): Update.
10858 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10859 (find_if_case_1): Update.
10860 (find_if_case_2): Update.
10861 * internal-fn.c (expand_arith_overflow_result_store): Update.
10862 (expand_addsub_overflow): Update.
10863 (expand_neg_overflow): Update.
10864 (expand_mul_overflow): Update.
10865 (expand_vector_ubsan_overflow): Update.
10866 * ipa-cp.c (good_cloning_opportunity_p): Update.
10867 * ipa-split.c (split_function): Use make_single_succ_edge.
10868 * ipa-utils.c (ipa_merge_profiles): Update.
10869 * loop-doloop.c (add_test): Update.
10870 (doloop_modify): Update.
10871 * loop-unroll.c (compare_and_jump_seq): Update.
10872 (unroll_loop_runtime_iterations): Update.
10873 * lra-constraints.c (lra_inheritance): Update.
10874 * lto-streamer-in.c (input_cfg): Update.
10875 * lto-streamer-out.c (output_cfg): Update.
10876 * mcf.c (adjust_cfg_counts): Update.
10877 * modulo-sched.c (sms_schedule): Update.
10878 * omp-expand.c (expand_omp_for_init_counts): Update.
10879 (extract_omp_for_update_vars): Update.
10880 (expand_omp_ordered_sink): Update.
10881 (expand_omp_for_ordered_loops): Update.
10882 (expand_omp_for_generic): Update.
10883 (expand_omp_for_static_nochunk): Update.
10884 (expand_omp_for_static_chunk): Update.
10885 (expand_cilk_for): Update.
10886 (expand_omp_simd): Update.
10887 (expand_omp_taskloop_for_outer): Update.
10888 (expand_omp_taskloop_for_inner): Update.
10889 * omp-simd-clone.c (simd_clone_adjust): Update.
10890 * optabs.c (expand_doubleword_shift): Update.
10891 (expand_abs): Update.
10892 (emit_cmp_and_jump_insn_1): Update.
10893 (expand_compare_and_swap_loop): Update.
10894 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10895 * predict.c (predictable_edge_p): Update.
10896 (edge_probability_reliable_p): Update.
10897 (set_even_probabilities): Update.
10898 (combine_predictions_for_insn): Update.
10899 (combine_predictions_for_bb): Update.
10900 (propagate_freq): Update.
10901 (estimate_bb_frequencies): Update.
10902 (force_edge_cold): Update.
10903 * profile-count.c (profile_count::dump): Add missing space into dump.
10904 (profile_count::debug): Add newline.
10905 (profile_count::differs_from_p): Explicitly convert to unsigned.
10906 (profile_count::stream_in): Update.
10907 (profile_probability::dump): New member function.
10908 (profile_probability::debug): New member function.
10909 (profile_probability::differs_from_p): New member function.
10910 (profile_probability::differs_lot_from_p): New member function.
10911 (profile_probability::stream_in): New member function.
10912 (profile_probability::stream_out): New member function.
10913 * profile-count.h (profile_count_quality): Rename to ...
10914 (profile_quality): ... this one.
10915 (profile_probability): New.
10916 (profile_count): Update.
10917 * profile.c (compute_branch_probabilities): Update.
10918 * recog.c (peep2_attempt): Update.
10919 * sched-ebb.c (schedule_ebbs): Update.
10920 * sched-rgn.c (find_single_block_region): Update.
10921 (compute_dom_prob_ps): Update.
10922 (schedule_region): Update.
10923 * sel-sched-ir.c (compute_succs_info): Update.
10924 * stmt.c (struct case_node): Update.
10925 (do_jump_if_equal): Update.
10926 (get_outgoing_edge_probs): Update.
10927 (conditional_probability): Update.
10928 (emit_case_dispatch_table): Update.
10929 (expand_case): Update.
10930 (expand_sjlj_dispatch_table): Update.
10931 (emit_case_nodes): Update.
10932 * targhooks.c: Update.
10933 * tracer.c (better_p): Update.
10934 (find_best_successor): Update.
10935 * trans-mem.c (expand_transaction): Update.
10936 * tree-call-cdce.c: Update.
10937 * tree-cfg.c (gimple_split_edge): Upate.
10938 (move_sese_region_to_fn): Upate.
10939 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10940 * tree-eh.c (lower_resx): Upate.
10941 (cleanup_empty_eh_move_lp): Upate.
10942 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10943 * tree-inline.c (copy_edges_for_bb): Update.
10944 (copy_cfg_body): Update.
10945 * tree-parloops.c (gen_parallel_loop): Update.
10946 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10947 (gimple_gen_time_profiler): Update.
10948 * tree-ssa-dce.c (remove_dead_stmt): Update.
10949 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10950 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10951 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10952 (unloop_loops): Update.
10953 (try_peel_loop): Update.
10954 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10955 * tree-ssa-loop-split.c (connect_loops): Update.
10956 (split_loop): Update.
10957 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10958 (hoist_guard): Update.
10959 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10960 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10961 (value_replacement): Update.
10962 * tree-ssa-reassoc.c (branch_fixup): Update.
10963 * tree-ssa-tail-merge.c (replace_block_by): Update.
10964 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10965 (create_edge_and_update_destination_phis): Update.
10966 (compute_path_counts): Update.
10967 (recompute_probabilities): Update.
10968 (update_joiner_offpath_counts): Update.
10969 (freqs_to_counts_path): Update.
10970 (duplicate_thread_path): Update.
10971 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10972 (struct switch_conv_info): Update.
10973 (gen_inbound_check): Update.
10974 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10975 (vect_do_peeling): Update.
10976 (vect_loop_versioning): Update.
10977 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10978 (optimize_mask_stores): Update.
10979 * ubsan.c (ubsan_expand_null_ifn): Update.
10980 * value-prof.c (gimple_divmod_fixed_value): Update.
10981 (gimple_divmod_fixed_value_transform): Update.
10982 (gimple_mod_pow2): Update.
10983 (gimple_mod_pow2_value_transform): Update.
10984 (gimple_mod_subtract): Update.
10985 (gimple_mod_subtract_transform): Update.
10986 (gimple_ic): Update.
10987 (gimple_stringop_fixed_value): Update.
10988 (gimple_stringops_transform): Update.
10989 * value-prof.h: Update.
10991 2017-06-29 Carl Love <cel@us.ibm.com>
10993 * config/rs6000/rs6000-c.c: Add support for built-in functions
10994 vector signed int vec_signed (vector float);
10995 vector signed long long vec_signed (vector double);
10996 vector signed int vec_signed2 (vector double, vector double);
10997 vector signed int vec_signede (vector double);
10998 vector signed int vec_signedo (vector double);
10999 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11000 instruction generator.
11001 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11002 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11003 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11005 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11006 vunsignede_v2df): Add define_expands.
11007 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11008 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11009 VEC_UNSIGNEDO): Add definitions.
11010 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11011 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11012 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11013 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11014 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11015 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11016 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11017 vec_unsignede, vec_unsignedo): Add builtin defines.
11018 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11020 * doc/extend.texi: Update the built-in documentation file for the
11021 new built-in functions.
11023 2017-06-29 Richard Biener <rguenther@suse.de>
11025 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11026 reduction chains to LOOP_VINFO_REDUCTIONS.
11027 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11028 SLP reductions after processing reduction chains.
11030 2017-06-29 Nathan Sidwell <nathan@acm.org>
11032 * builtins.c (fold_builtin_FUNCTION): Use
11033 lang_hooks.decl_printable_name.
11035 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11037 PR middle-end/81194
11038 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11039 with only one label.
11040 * stmt.c (expand_case): Assert NCASES is greater than one.
11042 2017-06-29 Richard Biener <rguenther@suse.de>
11044 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11046 (group_case_labels): Likewise.
11047 (find_taken_edge): Push sanity checking on val to workers...
11048 (find_taken_edge_cond_expr): ... here
11049 (find_taken_edge_switch_expr): ... and here, handle cases
11050 with just a default label.
11051 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11052 (group_case_labels): Likewise.
11053 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11054 group_case_labels does anything cleanup the CFG again.
11056 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11058 PR tree-optimization/81196
11059 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11060 exit condition comparing two IVs.
11062 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11064 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11065 profile to the dummy entry at the end of the list of architectures.
11066 * config/arm/arm-cpu-cdata.h: Regenerated.
11068 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11069 Michael Collison <michael.collison@arm.com>
11072 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11074 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11075 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11076 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11077 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11078 of shift when the shift amount is masked with constant equal to
11079 the size of the mode.
11080 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11083 2017-06-29 Martin Liska <mliska@suse.cz>
11085 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11087 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11090 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11093 (edge_growth_cache_entry::edge_growth_cache_entry): New
11095 (reset_edge_growth_cache): Update to use constructor.
11097 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11099 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11100 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11101 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11103 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11105 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11106 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11108 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11110 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11111 (*-linux-uclibc*): Add t-uclibc tmake_file.
11112 * config/t-musl: New.
11113 * config/t-uclibc: New.
11115 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11117 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11119 (gen_comm_data): Emit architectural setting of arch_prof.
11120 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11122 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11123 (armv8-m.base, armv8-m.main): Likewise.
11124 * arm-protos.h (arm_build_target): Add profile field.
11125 (arch_option): Likewise.
11126 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11128 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11129 arm_active_target.profile.
11131 2017-06-28 Richard Biener <rguenther@suse.de>
11133 PR middle-end/81227
11134 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11135 TYPE_OVERFLOW_WRAPS.
11136 * match.pd (negate_expr_p): Likewise.
11137 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11138 fold_build2, not fold_binary.
11140 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11142 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11143 Convert memory address to Pmode.
11144 (aarch64_print_operand): Assert MEM operands are always Pmode.
11146 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11149 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11150 Remove redundant if.
11151 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11152 * config/arm/aarch-common-protos.h
11153 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11154 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11156 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11159 * multiple_target.c (create_dispatcher_calls): Set the default
11160 clone to be static, not public.
11162 2017-06-28 Richard Biener <rguenther@suse.de>
11164 * tree-vect-loop.c (vectorizable_reduction): Move special
11165 cond reduction IV var creation ...
11166 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11167 parameter. Use STMT_VINFO_VECTYPE.
11168 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11171 2017-06-28 Martin Liska <mliska@suse.cz>
11174 * ipa-visibility.c (non_local_p): Handle visibility.
11176 2017-06-28 Martin Liska <mliska@suse.cz>
11179 * common.opt: Add IntegerRange to various options.
11180 * opt-functions.awk (integer_range_info): New function.
11181 * optc-gen.awk: Add integer_range_info to cl_options struct.
11182 * opts-common.c (decode_cmdline_option): Handle
11183 CL_ERR_INT_RANGE_ARG.
11184 (cmdline_handle_error): Likewise.
11185 * opts.c (print_filtered_help): Show valid interval in
11186 when --help is provided.
11187 * opts.h (struct cl_option): Add range_min and range_max fields.
11188 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11190 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11192 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11193 (x * C EQ/NE y * C): New transformation.
11195 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11197 * genmultilib (combination_space): Accept '+' in option names.
11199 2017-06-28 Martin Liska <mliska@suse.cz>
11202 * asan.c (instrument_derefs): Bail out inner references
11203 that are hard register variables.
11205 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11208 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11209 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11211 2017-06-28 Richard Biener <rguenther@suse.de>
11213 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11214 (vect_get_slp_defs): Adjust.
11215 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11217 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11219 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11220 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11221 (vectorizable_reduction): Adjust.
11222 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11224 (vect_get_slp_defs): Likewise.
11225 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11226 (vectorizable_bswap): Adjust.
11227 (vectorizable_call): Likewise.
11228 (vectorizable_conversion): Likewise.
11229 (vectorizable_assignment): Likewise.
11230 (vectorizable_shift): Likewise.
11231 (vectorizable_operation): Likewise.
11232 (vectorizable_store): Likewise.
11233 (vectorizable_condition): Likewise.
11234 (vectorizable_comparison): Likewise.
11236 2017-06-28 Michael Collison <michael.collison@arm.com>
11239 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11241 (arm_gen_movmemqi): Removed unused variable 'i'.
11242 Convert 'for' loop into 'while' loop.
11243 (arm_expand_prologue): Remove last unnecessary set of insn.
11244 (thumb_pop): Remove unused variable 'pushed_words'.
11245 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11247 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11249 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11250 * config/s390/s390-protos.h: Add prototype of
11251 s390_rel_address_ok_p.
11252 * config/s390/s390.c (s390_got_symbol): New function.
11253 (s390_rel_address_ok_p): New function.
11254 (legitimize_pic_address): Use s390_rel_address_ok_p.
11255 (s390_load_got): Use s390_got_symbol.
11256 (s390_option_override): Issue error if
11257 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11258 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11260 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11262 2017-06-27 Andrew Pinski <apinski@cavium.com>
11264 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11265 (X * copysign (1.0, X)): New pattern.
11266 (X * copysign (1.0, -X)): New pattern.
11267 (copysign (-1.0, CST)): New pattern.
11269 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11271 * genmultilib (combination_space): Remove variable.
11272 Validate reuse rules against regular expression for any sequence
11273 of multilib options in any order.
11275 2017-06-27 Michael Collison <michael.collison@arm.com>
11277 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11278 call aarch64_split_simd_combine.
11279 * (aarch64_combine_internal<mode>): Delete pattern.
11280 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11281 Allow register and subreg operands.
11283 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11285 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11286 specific need, just fallback on defaults.
11287 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11289 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11290 Olivier Hainque <hainque@adacore.com>
11292 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11294 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11295 targets. Pick a default if no particular attempt applied.
11296 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11299 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11301 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11302 (x86_64-wrs-vxworks7): Likewise.
11304 2017-06-27 Marek Polacek <polacek@redhat.com>
11307 * ubsan.c (instrument_null): Check get_base_address's result for null.
11309 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11311 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11313 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11315 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11316 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11317 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11318 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11319 New function types.
11320 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11321 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11322 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11323 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11324 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11325 BUILT_IN_FEUPDATEENV): New builtins.
11326 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11327 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11328 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11329 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11331 (builtin_structptr_types): Adjust size.
11332 * tree.c (builtin_structptr_types): Add four entries.
11334 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11335 Olivier Hainque <hainque@adacore.com>
11337 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11338 (TLS_SYM): New local macro, forcing reference to __tls__ on
11339 link command lines for VxWorks 7 RTPs, triggering initialization
11341 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11342 OS features TLS support, true for RTPs on VxWorks 7.
11343 * config/vxworks.c (vxworks_override_options): Setup emutls
11346 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11348 * predict.c (test_prediction_value_range): Use -1U instead of -1
11349 to avoid narrowing conversion warning.
11350 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11351 to avoid narrowing conversion warning.
11352 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11354 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11356 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11358 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11359 64bit configurations.
11360 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11361 (SIZE_TYPE): Likewise.
11362 * config/vxworks.c (vxworks_emutls_var_fields): Use
11363 long_unsigned_type_node instead of unsigned_type_node as the offset
11364 field type, which is "pointer" mode in emutls.c.
11366 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11369 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11371 PR middle-end/81207
11372 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11373 gimple_vuse copying separately from gimple_vdef copying.
11375 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11377 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11380 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11381 Olivier Hainque <hainque@adacore.com>
11383 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11384 for all vxworks7 targets.
11385 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11386 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11387 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11388 variations for VX6/VX7 and 32/64bits later on in ...
11389 (VXWORKS_LIB_SPEC): Leverage new macros.
11390 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11391 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11393 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11395 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11396 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11398 2017-06-26 Carl Love <cel@us.ibm.com>
11400 * config/rs6000/rs6000-c.c: Add support for built-in functions
11401 vector bool char vec_reve (vector bool char);
11402 vector signed char vec_reve (vector signed char);
11403 vector unsigned char vec_reve (vector unsigned char);
11404 vector bool int vec_reve (vector bool int);
11405 vector signed int vec_reve (vector signed int);
11406 vector unsigned int vec_reve (vector unsigned int);
11407 vector bool long long vec_reve (vector bool long long);
11408 vector signed long long vec_reve (vector signed long long);
11409 vector unsigned long long vec_reve (vector unsigned long long);
11410 vector bool short vec_reve (vector bool short);
11411 vector signed short vec_reve (vector signed short);
11412 vector double vec_reve (vector double);
11413 vector float vec_reve (vector float);
11414 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11415 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11416 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11417 (altivec_vreve): New pattern.
11418 * config/rs6000/altivec.h (vec_reve): New define.
11419 * doc/extend.texi (vec_rev): Update the built-in documentation file
11420 for the new built-in functions.
11422 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11424 PR tree-optimization/71815
11425 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11427 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11429 (slsr_process_phi): Likewise.
11430 (replace_uncond_cands_and_profitable_phis): Don't replace a
11431 multiply candidate with a stride of 1 (copy or cast).
11432 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11434 (lowest_cost_path): Likewise.
11435 (total_savings): Likewise.
11437 2017-06-26 Richard Biener <rguenther@suse.de>
11440 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11441 Use def_builtin_pure for all gather builtins.
11443 2017-06-26 Richard Biener <rguenther@suse.de>
11445 PR tree-optimization/81203
11446 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11447 non-dominating BBs.
11449 2017-06-26 Marek Polacek <polacek@redhat.com>
11452 * doc/invoke.texi: Document -Wmultistatement-macros.
11454 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11456 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11457 arm_neon_ok_no_float_abi effective target.
11459 2017-06-26 Richard Biener <rguenther@suse.de>
11461 PR tree-optimization/80928
11462 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11463 (copy_bbs): Set BB_DUPLICATED flag early.
11464 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11466 (execute_on_shrinking_pred): Likewise.
11467 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11468 BB_DUPLICATED blocks.
11469 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11470 iterate over all PHIs considering removal of *gsi.
11472 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11474 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11477 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11479 * config/rs6000/rs6000-string.c: (expand_block_clear,
11480 do_load_for_compare, select_block_compare_mode,
11481 compute_current_alignment, expand_block_compare,
11482 expand_strncmp_align_check, expand_strn_compare,
11483 expand_block_move, rs6000_output_load_multiple)
11484 Move functions related to string/block move/compare
11485 to a separate file.
11486 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11487 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11488 for this function which is now used in two files.
11489 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11490 * config.gcc: Add rs6000-string.o to extra_objs for
11491 targets powerpc*-*-* and rs6000*-*-*.
11493 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11496 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11497 32-bit, since indexed is not valid for DImode.
11498 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11499 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11500 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11501 (define_peephole2 for Altivec d-form store): Likewise.
11504 * multiple_target.c (create_dispatcher_calls): Only create the
11505 dispatcher call if the function is the default clone of a
11506 versioned function.
11508 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11510 PR middle-end/80902
11511 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11512 a call, force the call to not be a tail call.
11514 2017-06-23 Jeff Law <law@redhat.com>
11516 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11519 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11521 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11522 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11523 handling for early expansion of vector shifts (sl,sr,sra,rl).
11524 (builtin_function_type): Add vector shift right instructions
11525 to the unsigned argument list.
11527 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11529 rtl-optimizatoin/79286
11530 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11531 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11532 trap. PIC register plus a const unspec without offset can never trap.
11534 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11536 * tree.h (builtin_structptr_type): New type.
11537 (builtin_structptr_types): Declare new array.
11538 * tree.c (builtin_structptr_types): New array.
11539 (free_lang_data, build_common_tree_nodes): Use it.
11541 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11544 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11547 2017-06-22 Matt Turner <mattst88@gmail.com>
11549 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11550 Lake models to skylake case. Assume skylake for unknown
11551 models with clflushopt.
11553 2017-06-22 Jeff Law <law@redhat.com>
11555 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11556 frame sizes that do not satisfy aarch64_uimm12_shift.
11558 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11560 * profile-count.h (apply_probability,
11561 apply_scale, probability_in): Fix checks for zero.
11563 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11565 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11566 * doc/cppdiropts.texi (-I @var{dir}): Document it.
11568 2016-06-22 Richard Biener <rguenther@suse.de>
11570 * tree-vect-loop.c (vect_model_reduction_cost): Handle
11571 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
11572 REDUC_MAX_EXPR support.
11573 (vectorizable_reduction): Likewise.
11574 (vect_create_epilog_for_reduction): Likewise.
11576 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
11578 * match.pd (A / (1 << B) -> A >> B): New.
11579 * generic-match-head.c: Include optabs-tree.h.
11580 * gimple-match-head.c: Likewise.
11581 * optabs-tree.h (target_supports_op_p): New.
11582 * optabs-tree.c (target_supports_op_p): New.
11584 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11586 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
11587 $gcc_cv_ld --help output.
11588 (gcc_cv_ld_demangle): Likewise.
11589 (gcc_cv_ld_eh_frame_hdr): Likewise.
11590 (gcc_cv_ld_pie): Likewise.
11591 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
11592 (gcc_cv_ld_buildid): Likewise.
11593 (gcc_cv_ld_sysroot): Likewise.
11594 (ld_bndplt_support): Likewise.
11595 (ld_pushpopstate_support): Likewise.
11596 * configure: Regenerate.
11597 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
11599 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11602 * config/i386/sse.md (round<mode>2): Renumber match_dup and
11603 operands indexes to avoid gap between operands and match_dups.
11605 2017-06-21 Andrew Pinski <apinski@cavium.com>
11607 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11608 Increment Arith_shift and Arith_shift_reg by 1.
11609 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11611 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11612 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11613 (aarch64_strip_extend): Add new argument and test for it.
11614 (aarch64_cheap_mult_shift_p): New function.
11615 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11616 add a cost if it is true.
11617 Update calls to aarch64_strip_extend.
11618 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11620 2017-06-21 Andrew Pinski <apinski@cavium.com>
11622 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11624 (thunderxt88): Likewise.
11625 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11626 (thunderx_prefetch_tune): New variable.
11627 (thunderx2t99_prefetch_tune): Update for the correct values.
11628 (thunderxt88_tunings): New variable.
11629 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11630 generic_prefetch_tune.
11631 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11633 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11635 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11636 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11637 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11638 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11639 (aarch64_atomic_cas<mode>, GPI): Likewise.
11641 2017-06-21 Martin Liska <mliska@suse.cz>
11643 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11644 statements on cold and hot labels.
11645 * predict.c (tree_estimate_probability_bb): Remove the
11646 prediction from this place.
11648 2017-06-21 Martin Liska <mliska@suse.cz>
11650 PR tree-optimization/79489
11651 * gimplify.c (maybe_add_early_return_predict_stmt): New
11653 (gimplify_return_expr): Call the function.
11654 * predict.c (tree_estimate_probability_bb): Remove handling
11656 * predict.def: Update comment about early return predictor.
11657 * gimple-predict.h (is_gimple_predict): New function.
11658 * predict.def: Change default value of early return to 66.
11659 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11661 * passes.def: Put pass_strip_predict_hints to the beginning of
11664 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11666 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11667 FUNCTION_DECL declarations.
11668 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11670 (dwaf2out_decl): Likewise.
11671 * godump.c (go_early_global_decl): Skip call to the real debug hook
11672 for FUNCTION_DECL declarations.
11673 * passes.c (rest_of_decl_compilation): Skip call to the
11674 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11675 -fdump-go-spec is passed.
11677 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11679 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11680 Reorder for compactness.
11681 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11682 (def_builtin_pure, def_builtin_pure2): New functions.
11683 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11685 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11687 * match.pd (nop_convert): New predicate.
11688 ((A +- CST1) +- CST2): Allow some NOP conversions.
11690 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11693 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11694 with ctors/dtors if GOVD_SHARED is set.
11696 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11698 * config/aarch64/aarch64.md (movti_aarch64):
11699 Emit mov rather than orr.
11700 (movtf_aarch64): Likewise.
11701 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11702 Emit mov rather than orr.
11704 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11706 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11707 Swap alternatives, make integer dup more expensive.
11709 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11711 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11712 Return true for non-tls symbols.
11714 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11716 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11717 (cortex-a75): Likewise.
11718 (cortex-a75.cortex-a55): Likewise.
11719 * config/aarch64/aarch64-tune.md: Regenerate.
11720 * doc/invoke.texi (-mtune): Document new values for -mtune.
11722 2017-06-21 Tom de Vries <tom@codesourcery.com>
11724 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11725 stack_size feature.
11726 (Effective-Target Keywords, Other attributes): Suggest using
11727 dg-add-options stack_size feature to get stack limit in stack_size
11728 effective target documentation.
11730 2017-06-21 Julian Brown <julian@codesourcery.com>
11731 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11733 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11734 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11735 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11737 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11738 attribute type list for neon_multiply.
11739 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11740 attribute type list for neon_multiply.
11741 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11742 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11743 attribute type list for neon_multiply.
11744 * config/arm/types.md (crypto_pmull): Add.
11745 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11746 attribute type list.
11748 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11750 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11753 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11755 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11756 to make sure not to dereference a NULL cost_classes_ptr pointer.
11758 2017-06-20 Carl Love <cel@us.ibm.com>
11760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11761 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11762 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11763 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11764 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11765 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11766 VMULOSW): New enum "unspec" values.
11767 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11768 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11769 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11770 altivec_vmulosw): New patterns.
11771 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11772 VMULOSW): Add definitions.
11774 2017-06-20 Julia Koval <julia.koval@intel.com>
11776 * config/i386/i386.c: Fix rounding expand for new pattern.
11777 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11779 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11781 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11782 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11784 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11786 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11789 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11791 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11792 architecture, then by implementer ID.
11793 * config/aarch64/aarch64-tune.md: Regenerate.
11795 2017-06-20 Richard Biener <rguenther@suse.de>
11797 PR middle-end/81097
11798 * fold-const.c (split_tree): Fold to type before negating.
11800 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11802 * diagnostic-show-locus.c
11803 (selftest::test_fixit_deletion_affecting_newline): New function.
11804 (selftest::diagnostic_show_locus_c_tests): Call it.
11806 2017-06-20 Andreas Schwab <schwab@suse.de>
11809 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11812 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11814 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11815 __ARM_FEATURE_COPROC according to support.
11817 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11819 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11820 Rewritten to avoid overflow for > 32-bit pointers.
11823 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11824 by removing enum keyword.
11825 (ubsan_type_descriptor): Likewise. Formatting fix.
11828 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11829 splitter): Require TARGET_SSE2 in the condition.
11831 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11834 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11835 for doing vector set of SFmode on ISA 3.0.
11836 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11837 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11839 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11840 SFmode value into a V4SF variable that was extracted from another
11841 V4SF variable without converting the element to double precision
11842 and back to single precision vector format.
11843 (vsx_insert_extract_v4sf_p9_2): Likewise.
11845 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11847 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11848 in UWHI to avoid undefined overflow.
11851 * ubsan.h (enum ubsan_encode_value_phase): New.
11852 (ubsan_encode_value): Change second argument to
11853 enum ubsan_encode_value_phase with default value of
11854 UBSAN_ENCODE_VALUE_GENERIC.
11855 * ubsan.c (ubsan_encode_value): Change second argument to
11856 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11857 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11858 create_tmp_var_raw instead of create_tmp_var and use a
11860 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11861 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11862 ubsan_encode_value callers.
11865 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11866 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11867 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11869 2017-06-19 Richard Biener <rguenther@suse.de>
11871 PR middle-end/81118
11872 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11873 estimates if we changed anything.
11875 2017-06-19 Richard Biener <rguenther@suse.de>
11877 PR tree-optimization/80887
11878 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11879 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11880 simplified lookups, then reset mprts_hook.
11881 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11883 (try_to_simplify): Likewise.
11885 2017-06-19 Martin Liska <mliska@suse.cz>
11888 * gimplify.c (gimplify_switch_expr):
11889 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11891 2017-06-19 Martin Liska <mliska@suse.cz>
11893 * doc/install.texi: Document that PGO runs in 4 stages.
11895 2017-06-19 Martin Liska <mliska@suse.cz>
11898 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11899 to dispatcher function name.
11900 * multiple_target.c (replace_function_decl): New function.
11901 (create_dispatcher_calls): Redirect both edges and references.
11903 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11905 * profile-count.c (profile_count::dump): Dump quality.
11906 (profile_count::differs_from_p): Update for unsigned val.
11907 * profile-count.h (profile_count_quality): New enum.
11908 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11910 2017-06-19 Richard Biener <rguenther@suse.de>
11912 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11913 struct function as arg.
11914 (estimate_numbers_of_iterations): Export overload with loop arg.
11915 (free_numbers_of_iterations_estimates_loop): Use an overload of
11916 free_numbers_of_iterations_estimates instead.
11917 * tree-cfg.c (remove_bb): Adjust.
11918 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11919 * tree-parloops.c (gen_parallel_loop): Likewise.
11920 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11922 (tree_unroll_loops_completely): Likewise.
11923 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11924 Use an overload instead and export.
11925 (estimated_loop_iterations): Adjust.
11926 (max_loop_iterations): Likewise.
11927 (likely_max_loop_iterations): Likewise.
11928 (estimate_numbers_of_iterations): Take struct function as arg
11930 (loop_exits_before_overflow): Adjust.
11931 (free_numbers_of_iterations_estimates_loop): Use an overload.
11932 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11933 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11935 2017-06-19 Richard Biener <rguenther@suse.de>
11938 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11939 RANGE_EXPR conservatively.
11941 2017-06-16 Carl Love <cel@us.ibm.com>
11943 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11944 definitions for vec_float, vec_float2, vec_floato,
11945 vec_floate built-ins.
11946 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11947 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11949 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11950 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11951 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11952 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11953 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11954 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11955 vec_floato): Add builtin defines.
11956 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11957 Update the built-in documentation file for the new built-in
11960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11962 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11963 (mthumb): Mark as the negative of -marm.
11965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11967 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11969 (ARM Options, -mtune): Document that this accepts the same
11970 extension options as -mcpu.
11971 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11975 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11976 permitted extensions.
11978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11980 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11981 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11982 (armv8-m.main): Add option +nodsp.
11983 * config/arm/arm-cpu-cdata.h: Regenerated.
11985 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11987 * config/arm/t-fuchsia: New file.
11988 * config.gcc (arm*-*-fuchsia*): Use it.
11990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11992 * config/arm/t-symbian: Rewrite for new option infrastructure.
11994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11996 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11997 (MULTILIB_REQUIRED): Likewise.
11999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12001 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12002 (MULTILIB_RESUE): Likewise.
12003 (MULTILIB_MATCHES): Likewise.
12004 (MULTLIB_REQUIRED): Likewise.
12006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12008 * config/arm/t-rtems: Rewrite for new option framework.
12010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12012 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12013 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12014 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12015 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12016 * config/arm/t-multilib: ... here.
12017 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12018 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12019 armv7-a and armv8*-a when A-profile libraries have not been built.
12020 * config/arm/t-rmprofile: Rewrite.
12022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12024 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12025 with a backslash. Remove the backslash after substituting unescaped
12027 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12031 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12032 the driver to pass through the default setting of -mfloat-abi.
12033 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12035 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12036 (all_feat_combs): New rule.
12037 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12039 * config/arm/t-aprofile: Rewrite.
12041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12043 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12044 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12045 fpu is not specified by the user/command-line.
12046 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12047 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12048 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12049 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12050 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12051 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12055 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12056 * config/arm/t-arm-elf: Rewritten.
12058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12060 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12061 have some floating-point instructions.
12062 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12063 (TARGET_MAYBE_HARD_FLOAT): New macro.
12064 * config/arm/arm-builtins.c (arm_init_builtins): Use
12065 TARGET_MAYBE_HARD_FLOAT.
12066 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12070 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12071 (configargs.h): Include it.
12072 (arm_print_hint_for_fpu_option): New function.
12073 (arm_parse_fpu_option): New function.
12074 (candidate_extension): New class.
12075 (arm_canon_for_multilib): New function.
12076 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12077 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12078 (ARCH_CANONICAL_SPECS): New macro.
12079 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12081 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12083 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12084 are set after handling multilib fragments. Set target_cpu_default2
12087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12089 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12091 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12092 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12094 (check_arch): Likewise.
12095 * config/arm/arm.c (arm_configure_build_target): Handle
12096 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12097 options in the default.
12099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12101 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12102 when an option is an alias of another.
12103 * config/arm/parsecpu.awk (optalias): New parser token.
12104 (gen_comm_data): Mark non-alias options as such. Emit entries
12105 for extension aliases.
12106 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12107 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12108 (armv6kz, armv6zk, armv6t2): Likewise.
12109 (armv7): Make vfpv3-d16 an alias.
12110 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12112 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12113 Sort in canonical order.
12114 (armv8-a): Sort in canonical order.
12115 (armv8.1-a, armv8.2-a): Likewise.
12116 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12118 (cortex-a9): Sort in canonical order.
12119 * config/arm/arm.c (selftests.h): Include it.
12120 (arm_test_cpu_arch_data): New function.
12121 (arm_run_self_tests): New function.
12122 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12123 (targetm): Move declaration to the end of the file.
12124 * arm-cpu-cdata.h: Regenerated.
12126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12128 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12129 call to target_mode_check describing the type of option passed.
12130 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12131 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12132 arm_parse_cpu_option_name to match parameters against list of
12134 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12135 arm_arch_core_flags data structure.
12136 * config/arm/arm-cpu_cdata.h: Regenerated.
12138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12140 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12142 (arm_print_hint_for_cpu_option): Likewise.
12143 (arm_print_hint_for_arch_option): Likewise.
12144 (arm_parse_cpu_option_name): Likewise.
12145 (arm_parse_arch_option_name): Likewise.
12146 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12147 of entries in the all_fpus list.
12148 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12149 (arm_parse_cpu_option_name): Declare.
12150 (arm_parse_arch_option_name): Declare.
12151 (arm_parse_option_features): Declare.
12152 (arm_intialize_isa): Declare.
12153 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12155 (gen_comm_data): ... here. Make definitions non-static.
12156 * config/arm/arm-cpu-data.h: Regenerated.
12157 * config/arm/arm-cpu-cdata.h: Regenerated.
12159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12161 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12162 (cpu_arch_extension): New structure.
12163 (cpu_arch_option, arch_option, cpu_option): New structures.
12164 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12165 architecture types.
12166 (gen_data): Generate new format data tables.
12167 * config/arm/arm.c (cpu_tune): New structure.
12168 (cpu_option, processors): Delete.
12169 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12170 (arm_print_hint_for_cpu_option): ... this and ...
12171 (arm_print_hint_for_arch_option): ... this.
12172 (arm_parse_arch_cpu_name): Delete. Replace with ...
12173 (arm_parse_cpu_option_name): ... this and ...
12174 (arm_parse_arch_option_name): ... this.
12175 (arm_unrecognized_feature): Change type of target parameter to
12177 (arm_parse_arch_cpu_features): Delete. Replace with ...
12178 (arm_parse_option_features): ... this.
12179 (arm_configure_build_target): Rework to use new configuration data
12181 (arm_print_tune_info): Rework for new configuration data tables.
12182 * config/arm/arm-cpu-data.h: Regenerated.
12183 * config/arm/arm-cpu.h: Regenerated.
12185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12187 * Makefile.in (OBJS): Move sbitmap.o from here ...
12188 (OBJS-libcommon): ... to here.
12190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12192 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12193 (ISA_ALL_CRYPTO): New macro.
12194 (ISA_ALL_SIMD): New macro
12195 (ISA_ALL_FP): New macro.
12196 * config/arm/arm.c (fpu_bitlist): Update initializer.
12197 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12199 (arm9e): Add fpu. Add option for nofp
12200 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12201 (arm926ej-s, arm1026ej-s): Likewise.
12202 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12203 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12204 neon-fp16, neon-vfpv4, nofp and nosimd.
12205 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12206 (cortex-a8): Add fpu. Add option for nofp.
12207 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12208 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12209 (cortex-r4f): Add fpu.
12210 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12211 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12213 (cortex-r8): Likewise.
12214 (cortex-m4): Add fpu. Add option for nofp.
12215 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12216 (cortex-a17.cortex-a7): Likewise.
12217 (cortex-a32): Add fpu. Add options for crypto and nofp.
12218 (cortex-a35, cortex-a53): Likewise.
12219 (cortex-a57): Add fpu. Add option for crypto.
12220 (cortex-a72, cortex-a73): Likewise.
12221 (exynos-m1): Likewise.
12222 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12223 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12224 (cortex-m33): Add fpu. Add option for nofp.
12225 * config/arm/arm-cpu-cdata.h: Regenerated
12226 * config/arm/arm-cpu-data.h: Regenerated.
12228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12230 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12231 (armv5te, armv5tej): Likewise.
12232 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12233 (armv7): Add options fp and vfpv3-d16.
12234 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12235 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12237 (armv7ve): Likewise.
12238 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12239 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12240 (armv8-a): Add nocrypto option.
12241 (armv8.1-a, armv8.2-a): Likewise.
12242 (armv8-m.main): add options fp, fp.dp and nofp.
12244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12246 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12248 (armv8-a+crc): Delete.
12249 (armv8.1-a): Add options simd, crypto and nofp.
12250 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12251 (armv8.2-a+fp16): Delete.
12252 (armv8-m.main): Add option dsp.
12253 (armv8-m.main+dsp): Delete.
12254 (cortex-a8): Add fpu. Add nofp option.
12255 (cortex-a9): Add fpu. Add nofp and nosimd options.
12256 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12257 link to main cpu and architecture data structures.
12258 (gen_comm_data): Only put isa attributes from the main architecture
12260 (option): New statement for architecture and CPU entries.
12261 * arm.c (struct cpu_option): New structure.
12262 (struct processors): Add entry for options.
12263 (arm_unrecognized_feature): New function.
12264 (arm_parse_arch_cpu_name): Ignore any characters after the first
12266 (arm_parse_arch_cpu_feature): New function.
12267 (arm_configure_build_target): Separate out any CPU and architecture
12268 features and parse separately. Don't error out if -mfpu=auto is
12269 used with only an architecture string.
12270 (arm_print_asm_arch_directives): New function.
12271 (arm_file_start): Call it.
12272 * config/arm/arm-cpu-cdata.h: Regenerated.
12273 * config/arm/arm-cpu-data.h: Likewise.
12274 * config/arm/arm-tables.opt: Likewise.
12276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12278 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12279 assembler when it is not -mfpu=auto.
12281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12283 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12284 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12285 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12286 (ASM_CPU_SPEC): Rewrite.
12287 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12288 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12289 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12290 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12291 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12292 copied string is NUL-terminated. Also strip any characters prefixed
12294 (arm_rewrite_selected_arch): New function.
12295 (arm_rewrite_march): New function.
12297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12299 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12300 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12301 (march, mcpu, mtune): Convert to string-based options.
12302 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12303 (arm_parse_arch_cpu_name): New function.
12304 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12305 identify selected architecture or CPU.
12306 (arm_option_save): New function.
12307 (TARGET_OPTION_SAVE): Redefine.
12308 (arm_option_restore): Restore string options.
12309 (arm_option_print): Print string options.
12311 2017-06-16 Martin Sebor <msebor@redhat.com>
12313 PR tree-optimization/80933
12314 PR tree-optimization/80934
12315 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12316 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12317 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12318 (gimple_fold_builtin): Call them.
12320 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12322 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12323 as unlikely; update profile.
12325 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12327 * predict.c (force_edge_cold): Handle declaring edges impossible
12330 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12332 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12334 (try_unroll_loop_completely): Fix reporting.
12336 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12338 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12340 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12343 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12344 if given a non-constant argument for an intrinsic which requires a
12347 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12349 * profile.c (compare_freqs): New function.
12350 (branch_prob): Sort edge list.
12351 (find_spanning_tree): Assume that the list is priority sorted.
12353 2017-06-16 Richard Biener <rguenther@suse.de>
12355 PR tree-optimization/81090
12356 * passes.def (pass_record_bounds): Remove.
12357 * tree-pass.h (make_pass_record_bounds): Likewise.
12358 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12359 make_pass_record_bounds): Likewise.
12360 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12361 not free niter estimates at the beginning but at the end.
12362 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12364 2017-06-16 Richard Biener <rguenther@suse.de>
12366 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12367 initializer to workaround ICE in host GCC 4.8.
12369 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12371 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12373 (clone_inlined_nodes): Update.
12375 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12377 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12378 prefetch settings, and enable prefetching by default at -O3.
12380 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12382 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12383 Set flag_prefetch_loop_arrays according to tuning data.
12385 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12387 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12388 New tune structure.
12389 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12390 [Unrelated to main purpose of the patch] Place the pointer field last
12391 to enable type checking errors when tune structure are wrongly merged.
12392 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12393 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12394 (thunderx2t99_prefetch_tune): New tune constants.
12395 (tune_params *_tunings): Update all tunings (no functional change).
12396 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12397 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12398 from tunings structures.
12400 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12403 * ubsan.c (instrument_null): Add T argument, use it instead
12404 of computing it based on IS_LHS.
12405 (instrument_object_size): Likewise.
12406 (pass_ubsan::execute): Adjust instrument_null and
12407 instrument_object_size callers to pass gimple_get_lhs or
12408 gimple_assign_rhs1 result to it. Use instrument_null instead of
12409 calling get_base_address and instrument_mem_ref. Handle
12410 aggregate call arguments for object-size sanitization.
12412 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12414 PR tree-optimization/81089
12415 * tree-vrp.c (is_masked_range_test): Validate operands of
12418 2017-06-15 Martin Sebor <msebor@redhat.com>
12421 * dumpfile.c (dump_register): Avoid calling memset to initialize
12422 a class with a default ctor.
12423 * gcc.c (struct compiler): Remove const qualification.
12424 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12425 * hash-table.h: Ditto.
12426 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12428 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12429 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12431 * params.h (struct param_info): Make struct members non-const.
12432 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12433 with default initialization.
12434 * vec.h (vec_copy_construct, vec_default_construct): New helper
12436 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12437 with vec_copy_construct.
12438 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12439 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12440 * doc/invoke.texi (-Wclass-memaccess): Document.
12442 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12444 * emit-rtl.h (is_leaf): Update comment about local
12445 register allocator.
12447 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12450 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12451 for a variable to have a section before checking if the section has a
12453 Set section to.persistent if persistent attribute is set.
12454 Warn if .persistent attribute is used on an automatic variable.
12456 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12458 PR rtl-optimization/80474
12459 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12461 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12463 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12466 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12468 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12469 (casesi): Emit bounds checking as RTL.
12470 (casesi_internal_mips16_<mode>): Remove bounds checking.
12472 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12474 * config/xtensa/xtensa.c (xtensa_option_override): Append
12475 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12476 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12477 xtensa_doloop_hooks): Define unconditionally.
12478 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12480 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12481 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12482 for it in xtensa_option_override.
12483 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12484 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12486 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12488 * doc/cppopts.texi: Document '-' special value to -MF.
12490 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12492 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12493 (cortex_a53_fconst): Likewise.
12494 (cortex_a53_fpmul): Likewise.
12495 (cortex_a53_f_load_64): Likewise.
12496 (cortex_a53_f_load_many): Likewise.
12497 (cortex_a53_advsimd_alu): Likewise.
12498 (cortex_a53_advsimd_alu_q): Likewise.
12499 (cortex_a53_advsimd_mul): Likewise.
12500 (cortex_a53_advsimd_mul_q): Likewise.
12501 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12502 Add missing fmul, r2f_cvt and fconst cases.
12504 2017-06-14 Richard Biener <rguenther@suse.de>
12506 PR middle-end/81088
12507 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12509 (fold_binary_loc): When associating do not treat pre-existing
12510 TREE_OVERFLOW on literal constants as a reason to allow
12511 TREE_OVERFLOW on associated literal constants.
12513 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12515 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12516 (MASK_FEATURES): New macro.
12517 * config/sparc/sparc.c (sparc_option_override): Remove the special
12518 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12520 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12522 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12523 a division of 0 if non-call exceptions are enabled.
12525 2017-06-14 Andrew Pinski <apinski@cavium.com>
12526 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12529 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12530 Improve vector initialization code gen for only variable case.
12532 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12534 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12536 2017-06-14 Richard Biener <rguenther@suse.de>
12538 PR tree-optimization/81083
12539 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12542 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12544 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12545 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12546 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12547 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12548 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12549 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12553 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12554 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12556 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12558 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12560 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12562 * config/rs6000/t-rtems: Don't handle SPE.
12564 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12566 * config/rs6000/t-linux: Don't handle SPE.
12568 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12570 * config/rs6000/eabispe.h: Delete file.
12572 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12574 * config/rs6000/t-spe: Delete file.
12576 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12578 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
12579 (rs6000_legitimate_offset_address_p): Return false for anything in
12580 V2SImode or V2SFmode.
12582 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12584 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
12585 except V2SF and V2SI. Rearrange the vector modes, and add comments.
12586 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
12588 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
12589 (rs6000_legitimate_offset_address_p): Ditto.
12590 (rs6000_emit_move): Ditto.
12591 (rs6000_init_builtins): Remove V4HI_type_node.
12593 2017-06-13 Martin Liska <mliska@suse.cz>
12596 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
12597 (gate_asan): Likewise.
12598 * asan.h (asan_no_sanitize_address_p): Remove the function.
12599 (sanitize_flags_p): New function.
12600 * builtins.def: Fix coding style.
12601 * common.opt: Use renamed enum value.
12602 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
12603 * doc/extend.texi: Document no_sanitize attribute.
12604 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12605 to SANITIZE_UNDEFINED_NONDEFAULT.
12606 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12607 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12608 Use sanitize_flags_p.
12609 * gimplify.c (gimplify_function_tree): Likewise.
12610 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12611 * opts.c (parse_no_sanitize_attribute): New function.
12612 (common_handle_option): Use renamed enum value.
12613 * opts.h (parse_no_sanitize_attribute): Declare.
12614 * tree.c (sanitize_flags_p): New function.
12615 * tree.h: Declared here.
12616 * tsan.c: Use sanitize_flags_p.
12617 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12618 (instrument_mem_ref): Likewise.
12619 (instrument_bool_enum_load): Likewise.
12620 (do_ubsan_in_current_function): Remove the function.
12621 (pass_ubsan::execute): Use sanitize_flags_p.
12622 * ubsan.h: Remove do_ubsan_in_current_function
12623 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12624 (dump_function_to_file): Use it here.
12626 2017-06-13 Martin Jambor <mjambor@suse.cz>
12628 PR tree-optimization/80803
12629 PR tree-optimization/81063
12630 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12631 (propagate_subaccesses_across_link): Enqueue subtree whenever
12632 necessary instead of relying on the caller.
12634 2017-06-13 Martin Jambor <mjambor@suse.cz>
12636 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12637 that have a first_link.
12638 (sort_and_splice_var_accesses): Do not check first_link before
12640 (subtree_mark_written_and_enqueue): Likewise.
12641 (propagate_all_subaccesses): Likewise and do not stop at first
12642 parent with a first_link.
12644 2017-06-13 Martin Jambor <mjambor@suse.cz>
12646 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12649 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12651 * match.pd: New pattern.
12653 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12655 * tree-vrp.c (is_masked_range_test): New function.
12656 (register_edge_assert_for): Determine ranges for
12659 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12661 PR tree-optimization/67328
12662 * fold-const.c (maskable_range_p): New function.
12663 (build_range_check): Generate bittests if possible.
12665 2017-06-13 Martin Liska <mliska@suse.cz>
12667 * gimple-pretty-print.c (dump_probability): Add new argument.
12668 (dump_edge_probability): Dump both probability and count.
12669 (dump_gimple_label): Likewise.
12670 (dump_gimple_bb_header): Likewise.
12672 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12675 * config/avr/avr-devices.c: Fix indentation.
12676 * config/avr/gen-avr-mmcu-specs.c: Dito.
12678 2017-06-13 Richard Biener <rguenther@suse.de>
12680 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12681 instead get vector type from stmt_info.
12682 (vectorizable_reduction): Adjust. Remove dead code.
12684 2017-06-13 Richard Biener <rguenther@suse.de>
12686 PR middle-end/81065
12687 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12688 case of C * (x * C2 + C3).
12689 (fold_addr_of_array_ref_difference): Properly fold index difference.
12691 2017-06-12 David S. Miller <davem@davemloft.net>
12694 * config/sparc/sparc.md (return expander): Emit frame blockage if
12695 function uses alloca.
12697 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12699 * combine.c (make_field_assignment): Check len rather than the mode
12700 precision when calling force_to_mode.
12702 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12704 Support multilibs and devices that see flash in RAM address range.
12707 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12708 (avr_mcu_t) <flash_pm_offset>: New field.
12709 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12710 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12711 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12712 (AVR_TINY_PM_OFFSET): Remove macro.
12713 * config/avr/avr.opt (-mshort-calls): New option.
12714 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12715 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12716 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12717 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12718 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12719 instead of avr_arch->have_jmp_call.
12720 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12721 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12722 avr_arch->flash_pm_offset to define.
12723 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12724 new field flash_pm_offset. Add entry for avrxmega3.
12725 (avr_texinfo): Add entry for avrxmega3.
12726 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12727 attiny212, attiny214,
12728 attiny412, attiny414, attiny416, attiny417,
12729 attiny814, attiny816, attiny817,
12730 attiny1614, attiny1616, attiny1617,
12731 attiny3214, attiny3216, attiny3217.
12732 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12733 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12734 (avr_print_operand_address) [AVR_TINY]: Same.
12735 (avr_asm_init_sections) <readonly_data_section>: Only patch
12736 callback if avr_arch->flash_pm_offset = 0.
12737 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12738 for rodata if avr_arch->flash_pm_offset != 0.
12739 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12740 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12741 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12742 (m_options): Append opt_rcall.
12743 (m_dirnames): Append dir_rcall.
12744 * config/avr/t-multilib: Regenerate.
12746 * configure.ac [target=avr]: Check whether avrxmega3 default
12747 linker description file works as needed.
12748 * configure: Regenerate.
12749 * doc/avr-mmcu.texi: Regenerate.
12750 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12751 <__AVR_ARCH__>: Document avrxmega3 and 103.
12752 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12753 <__AVR_SHORT_CALLS__>: Document it.
12754 <__AVR_PM_BASE_ADDRESS__>: Document it.
12755 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12756 (AVR Variable Attributes) <progmem>: Document this is
12757 not needed for avrxmega3.
12758 (AVR Named Address Spaces) <__flash>: Dito.
12760 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12762 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12764 2017-06-12 Doug Rupp <rupp@adacore.com>
12766 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12767 Append vxworks-stdint.h to the tm_file list.
12768 * config/vxworks-stdint.h: New file.
12770 2017-06-12 Martin Liska <mliska@suse.cz>
12772 PR tree-optimization/81041
12773 * tree-profile.c (gimple_gen_ic_func_profiler):
12774 Create an extra BB in profile-generate
12775 (gimple_gen_time_profiler): Likewise.
12777 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12779 PR tree-optimization/81003
12780 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12781 (update_range_test): Use it instead of force_gimple_operand_gsi.
12783 2017-06-12 Richard Biener <rguenther@suse.de>
12785 PR tree-optimization/81053
12786 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12787 with backedge value not defined in loop. Simplify def stmt
12790 2017-06-11 Tom de Vries <tom@codesourcery.com>
12793 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12795 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12796 nvptx_cannot_force_const_mem.
12798 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12800 * opts.c (finish_options): Move test for flag_split_stack after
12801 it has been initialized.
12803 2017-06-11 Jason Merrill <jason@redhat.com>
12805 * tree.h (id_equal): New.
12806 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12807 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12808 instead of strcmp of IDENTIFIER_POINTER.
12810 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12812 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12813 (mark_all_inlined_calls_cdtor): Fix formating.
12814 (inline_transform): Rescale profile before inlining.
12816 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12818 * cgraph.h (cgraph_edge::clone): Update prototype.
12819 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12820 (cgraph_node::create_clone): Update.
12821 (cgraph_node::create_version_clone): Update.
12822 * tree-inline.c (copy_bb): Update.
12823 (expand_call_inline): Update.
12825 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12827 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12828 factored out from ...
12829 (rs6000_emit_prologue): ... here.
12831 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12833 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12834 factored out from ...
12835 (rs6000_emit_prologue): ... here.
12837 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12839 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12841 (handle_missing_profiles): Fix computation of tp_first_run.
12842 (counts_to_freqs): Do not touch freqs when count is 0.
12844 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12846 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12849 2017-06-10 Tom de Vries <tom@codesourcery.com>
12851 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12852 attributes): Document signal effective target.
12854 2017-06-10 Tom de Vries <tom@codesourcery.com>
12856 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12857 Document effective target stack_size.
12859 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12861 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12862 to the edit_context if they can be auto-applied.
12864 2017-06-9 Ian Lance Taylor <iant@golang.org>
12866 * opts.c (finish_options): If -fsplit-stack, disable implicit
12867 -forder-blocks-and-partition.
12868 * doc/invoke.texi (Optimize Options): Document that when using
12869 -fsplit-stack -forder-blocks-and-partition is not implicitly
12872 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12874 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12875 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12876 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12877 * builtins.def (abort, trap, unreachable): Declare cold.
12878 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12879 * tree-core.h (ECF_COLD): New.
12880 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12881 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12883 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12885 * predict.c (unlikely_executed_stmt_p): Cleanup.
12887 2017-06-09 Richard Biener <rguenther@suse.de>
12889 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12890 model if the ref is always written to.
12892 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12894 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12896 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12898 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12901 2017-06-09 Tom de Vries <tom@codesourcery.com>
12904 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12905 "target cannot support label values" when encountering LABEL_REF.
12907 2017-06-09 Martin Liska <mliska@suse.cz>
12909 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12910 (gimple_gen_ic_func_profiler): Emit direct comparison
12911 of __gcov_indirect_call_callee with NULL.
12912 (gimple_gen_time_profiler): Change probability from
12913 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12915 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12917 * profile.c (edge_gcov_counts): Turn to pointer.
12918 (compute_branch_probabilities, compute_branch_probabilities): Update.
12919 (branch_prob): Do not clear edge_gcov_count.
12920 * profile.h (edge_gcov_counts): Turn to pointer.
12921 (edge_gcov_count): Update.
12923 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12925 * gimple.h (gimple_check_failed): Mark cold.
12927 2017-06-09 Richard Biener <rguenther@suse.de>
12929 PR tree-optimization/66623
12930 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12931 refactor check_reduction into two parts, properly computing
12932 whether we have to check reduction validity for outer loop
12935 2017-06-09 Richard Biener <rguenther@suse.de>
12937 PR tree-optimization/79483
12938 * graphite-scop-detection.c (order): New global.
12939 (get_order): Compute bb to order mapping that satisfies code
12940 generation constraints.
12941 (cmp_pbbs): New helper.
12942 (build_scops): Start domwalk at entry block, sort generated
12945 2017-06-09 Richard Biener <rguenther@suse.de>
12947 PR middle-end/81007
12948 * ipa-polymorphic-call.c
12949 (ipa_polymorphic_call_context::restrict_to_inner_class):
12950 Skip FIELD_DECLs with error_mark_node type.
12951 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12954 2017-06-09 Martin Liska <mliska@suse.cz>
12956 * predict.c (struct branch_predictor): New struct.
12957 (test_prediction_value_range): New test.
12958 (predict_c_tests): New function.
12959 * selftest-run-tests.c (selftest::run_tests): Run the function.
12960 * selftest.h: Declare new tests.
12962 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12965 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12966 gen_add3_insn did not fail.
12967 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12968 r0, construct that number in a temporary reg and add that reg to r0.
12969 If asked to put the result in r0 as well, fail.
12971 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12973 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12974 for early expansion of vec_eqv.
12976 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12978 PR middle-end/81005
12979 * ubsan.c (instrument_null): Avoid pointless code temporary.
12980 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12982 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12986 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12989 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12992 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12994 * predict.c (unlikely_executed_edge_p): Move ahead.
12995 (probably_never_executed_edge_p): Use it.
12997 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12999 PR middle-end/79988
13000 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13001 gimple_call_builtin_p call.
13003 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13005 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13006 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13007 rtl_check_failed_type2, rtl_check_failed_code1,
13008 rtl_check_failed_code2, rtl_check_failed_code_mode,
13009 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13010 rtvec_check_failed_bounds, rtl_check_failed_flag,
13011 _fatal_insn_not_found, _fatal_insn): Likewise.
13012 * tree.h (tree_contains_struct_check_failed,
13013 tree_check_failed, tree_not_check_failed,
13014 tree_class_check_failed, tree_range_check_failed,
13015 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13016 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13017 tree_operand_check_failed, omp_clause_check_failed,
13018 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13021 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13023 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13024 flag_branch_probabilities.
13025 * ipa-inline.c (edge_badness): Likewise.
13026 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13027 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13028 * predict.c (maybe_hot_frequency_p): Likewise.
13029 (probably_never_executed): Likewise.
13030 * sched-ebb.c (schedule_ebbs): Likewise.
13031 * sched-rgn.c (find_single_block_region): Likewise.
13032 * tracer.c (tail_duplicate): Likewise.
13034 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13036 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13037 longer requires x_flag_profile_use.
13039 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13041 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13042 instead of flag_reorder_blocks_and_partition.
13043 * dbxout.c (dbxout_function_end): Likewise.
13044 * dwarf2out.c (gen_subprogram_die): Likewise.
13045 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13046 * hw-doloop.c (reorg_loops): Likewise.
13047 * varasm.c (assemble_start_function,
13048 assemble_end_function): Likewise.
13049 (decide_function_section): Do not check for
13050 flag_reorder_blocks_and_partition.
13052 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13054 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13056 (chkp_get_hard_register_fake_addr_expr): Ditto.
13057 (chkp_build_addr_expr): Add check for hard reg case.
13058 (chkp_parse_array_and_component_ref): Ditto.
13059 (chkp_find_bounds_1): Ditto.
13060 (chkp_process_stmt): Don't generate bounds store for
13063 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13065 * predict.c (maybe_hot_bb_p): Do not check profile status.
13066 (maybe_hot_edge_p): Likewise.
13067 (probably_never_executed): Check for zero counts even if profile
13069 (unlikely_executed_edge_p): New function.
13070 (unlikely_executed_stmt_p): New function.
13071 (unlikely_executed_bb_p): New function.
13072 (set_even_probabilities): Use unlikely predicates.
13073 (combine_predictions_for_bb): Likewise.
13074 (predict_paths_for_bb): Likewise.
13075 (predict_paths_leading_to_edge): Likewise.
13076 (determine_unlikely_bbs): New function.
13077 (estimate_bb_frequencies): Use it.
13078 (compute_function_frequency): Use zero counts even if profile is
13080 * profile-count.h: Fix typo.
13082 2017-08-08 Julia Koval <julia.koval@intel.com>
13084 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13085 _mm512_mask_cvtsepi16_storeu_epi8,
13086 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13087 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13088 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13089 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13090 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13091 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13092 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13093 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13094 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13095 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13096 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13097 __builtin_ia32_pmovuswb256mem_mask,
13098 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13099 __builtin_ia32_pmovwb512mem_mask): New builtins.
13101 2017-08-08 Julia Koval <julia.koval@intel.com>
13103 PR target/73350,80862
13104 * config/i386/subst.md (round): Fix round pattern.
13105 * config/i386/i386.c (ix86_erase_embedded_rounding):
13106 Fix erasing rounding for the fixed pattern.
13108 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13110 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13112 2017-06-08 Martin Liska <mliska@suse.cz>
13114 PR gcov-profile/80911
13115 * gcov.c (block_info::block_info): New constructor.
13117 2017-06-07 Carl Love <cel@us.ibm.com>
13119 * config/rs6000/rs6000-c: The return type of the following
13120 built-in functions was implemented as int not long long. Fix sign
13121 of return value for the unsigned version of vec_mulo and vec_mule.
13122 vector unsigned long long vec_bperm (vector unsigned long long,
13123 vector unsigned char)
13124 vector signed long long vec_mule (vector signed int,
13126 vector unsigned long long vec_mule (vector unsigned int,
13127 vector unsigned int)
13128 vector signed long long vec_mulo (vector signed int,
13130 vector unsigned long long vec_mulo (vector unsigned int,
13131 vector unsigned int)
13132 * doc/extend.texi: Fix the documentation for the built-in
13135 2017-06-07 Carl Love <cel@us.ibm.com>
13138 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13141 2017-06-07 Carl Love <cel@us.ibm.com>
13143 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13144 support, Generate doublehv for signed int/float for BE case only.
13146 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13148 * doc/invoke.texi (mcx16): Rewrite.
13150 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13152 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13153 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13154 *mov<mode>_softfloat, and an anonymous splitter): Use
13155 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13159 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13161 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13162 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13163 (rs6000_debug_reg_global): Adjust.
13164 (rs6000_init_hard_regno_mode_ok): Adjust.
13165 (rs6000_dbx_register_number): Adjust.
13166 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13167 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13168 Remove SPE_ACC and SPEFSCR.
13169 (REG_ALLOC_ORDER): Ditto.
13170 (FRAME_POINTER_REGNUM): Change to 111.
13171 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13172 (REG_CLASS_NAMES): Ditto.
13173 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13174 (REGISTER_NAMES): Ditto.
13175 (ADDITIONAL_REG_NAMES): Ditto.
13176 (rs6000_reg_names): Ditto.
13177 * config/rs6000/rs6000.md: Renumber some register number
13180 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13182 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13184 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13185 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13187 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13188 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13189 Delete the SPE high registers.
13190 (REG_ALLOC_ORDER): Ditto.
13191 (enum reg_class): Remove SPE_HIGH_REGS.
13192 (REG_CLASS_NAMES): Ditto.
13193 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13194 (REGISTER_NAMES): Ditto.
13195 (rs6000_reg_names): Ditto.
13196 * doc/tm.texi.in: Remove SPE as example.
13197 * doc/tm.texi: Regenerate.
13199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13201 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13202 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13203 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13204 * config/rs6000/rs6000.md (type): Remove "brinc".
13206 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13208 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13209 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13210 * config/rs6000/linuxspe.h: Delete file.
13211 * config/rs6000/rs6000.md: Don't include spe.md.
13212 * config/rs6000/spe.h: Delete file.
13213 * config/rs6000/spe.md: Delete file.
13214 * config/rs6000/t-rs6000: Remove spe.md.
13216 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13218 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13219 (reg_or_none500mem_operand): Delete.
13220 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13221 instead of reg_or_none500mem_operand.
13223 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13225 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13226 handling of SPE flags.
13227 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13229 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13231 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13233 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13234 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13235 paired_divv2sf3): Similar.
13236 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13237 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13238 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13240 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13241 Rename the paired_* instruction patterns.
13242 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13244 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13245 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13246 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13247 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13248 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13249 PAIRED_VECTOR_MODE.
13250 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13251 (spe_func_has_64bit_regs_p): Delete.
13252 (spe_expand_predicate_builtin): Delete.
13253 (spe_expand_evsel_builtin): Delete.
13254 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13255 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13256 (invalid_e500_subreg): Delete.
13257 (rs6000_legitimize_address): Always force_reg op2 as well, for
13258 paired single memory accesses.
13259 (rs6000_member_type_forces_blk): Delete.
13260 (rs6000_spe_function_arg): Delete.
13261 (rs6000_expand_unop_builtin): Delete SPE handling.
13262 (rs6000_expand_binop_builtin): Ditto.
13263 (spe_expand_stv_builtin): Delete.
13264 (bdesc_2arg_spe): Delete.
13265 (spe_expand_builtin): Delete.
13266 (spe_expand_predicate_builtin): Delete.
13267 (spe_expand_evsel_builtin): Delete.
13268 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13269 (spe_init_builtins): Delete.
13270 (spe_func_has_64bit_regs_p): Delete.
13271 (savres_routine_name): Delete "info" parameter. Adjust callers.
13272 (rs6000_emit_stack_reset): Ditto.
13273 (rs6000_dwarf_register_span): Delete.
13274 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13275 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13276 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13278 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13280 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13281 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13282 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13283 mulv2sf3, divv2sf3): Delete expanders.
13285 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13287 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13291 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13292 * config/rs6000/rs6000.c: Ditto.
13294 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13296 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13297 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13298 comparison_operator.
13300 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13302 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13303 * config/rs6000/rs6000.opt: Ditto.
13304 * config/rs6000/t-rtems: Ditto.
13306 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13308 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13309 TARGET_E500_SINGLE by 0, simplify.
13310 * config/rs6000/rs6000.c: Ditto.
13311 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13312 (spe_build_register_parallel): Delete.
13313 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13314 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13315 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13316 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13317 (E500_CONVERT): Delete.
13318 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13320 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13322 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13323 * config/rs6000/dfp.md: Ditto.
13324 (negdd2, *negdd2_fpr): Merge.
13325 (absdd2, *absdd2_fpr): Merge.
13326 (negtd2, *negtd2_fpr): Merge.
13327 (abstd2, *abstd2_fpr): Merge.
13328 * config/rs6000/e500.h: Delete file.
13329 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13330 TARGET_FPRS by 1 and simplify.
13331 * config/rs6000/rs6000-c.c: Ditto.
13332 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13333 TARGET_DF_SPE by 0.
13334 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13336 * config/rs6000/rs6000.md: Ditto.
13337 (floatdidf2, *floatdidf2_fpr): Merge.
13338 (move_from_CR_gt_bit): Delete.
13339 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13340 (E500_CR_IOR_COMPARE): Delete.
13341 (All patterns that require !TARGET_FPRS): Delete.
13342 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13344 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13346 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13348 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13350 * graphds.c (add_edge): Intitialize edge's attached data.
13351 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13352 pointer parameter. Call pointed function on each edge during
13353 graph traversing. Skip traversing the edge when the function
13355 (graphds_dfs, graphds_scc): Ditto.
13356 (for_each_edge): New parameter. Pass the new parameter to callback
13358 * graphds.h (skip_edge_callback): New function pointer type.
13359 (graphds_dfs, graphds_scc): New function pointer parameter.
13360 (graphds_edge_callback, for_each_edge): New parameter.
13362 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13364 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13365 out code checking if runtime alias check is possible to below ...
13366 Call the new function.
13367 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13368 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13370 2017-06-07 Marek Polacek <polacek@redhat.com>
13373 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13374 TYPE_OVERFLOW_WRAPS check.
13376 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13378 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13379 if versioning is required.
13380 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13381 peeling with the check for versioning.
13383 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13385 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13386 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13387 Set true to new parameter if new ssa variable is defined.
13388 (vect_gen_vector_loop_niters): Refactor. Set range information
13389 for the new vector loop bound variable.
13390 (vect_do_peeling): Ditto.
13392 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13394 * tree-affine.c (ssa.h): Include header file.
13395 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13396 has wrapping overflow behavior.
13398 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13400 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13402 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13404 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13405 (tree_to_aff_combination): ... here.
13407 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13409 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13410 reg_pressure model function.
13411 (ivopts_global_cost_for_size): Delete.
13412 (determine_set_costs, iv_ca_recount_cost): Call new model function
13413 ivopts_estimate_reg_pressure.
13415 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13417 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13418 expensive than udiv. Remove floating point cases from mod.
13420 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13422 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13423 Increase idiv cost.
13425 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13427 * config/aarch64/aarch64.md
13428 (copysignsf3): Fix mask generation.
13430 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13432 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13434 (class dump_manager): Add register_dumps method.
13435 * dumpfile.c: Include langhooks.h.
13436 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13437 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13438 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13439 (dump_manager::dump_register): Start with 512 entries instead of 32.
13440 (dump_manager::register_dumps): New method.
13441 * toplev.c (general_init): Instead of invoking register_dumps
13442 langhook, invoke register_dumps method on the dump manager.
13443 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13446 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13448 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13449 Say that # requires an associated define_split to exist, and that
13450 the define_split must be suitable for use after register allocation.
13452 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13454 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13455 (compute_outgoing_frequencies): Also initialize zero counts.
13456 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13457 around loops; preserve more of profile when nothing changes.
13459 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13461 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13463 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13464 * config/arm/arm-cpu-cdata.h: Regenerate.
13465 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13466 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13467 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13468 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13469 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13471 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13473 * config/arm/t-rmprofile: Likewise.
13474 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13476 2017-06-06 David S. Miller <davem@davemloft.net>
13479 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13480 blockage if function uses alloca.
13482 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13484 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13485 New "uid" fields to hold pretty-print IDs of group and ref.
13486 Memory references are now identified as <group_id>:<ref_id>
13487 instead of using [random] addresses.
13488 (dump_mem_details): Simplify, no functional change.
13489 (dump_mem_ref): Simplify and make output more concise.
13490 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13491 (find_or_create_group): Initialize group uid.
13492 (record_ref): Initialize ref uid. Improve debug output.
13493 (prune_group_by_reuse, should_issue_prefetch_p,)
13494 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13495 (mark_nontemporal_store, determine_loop_nest_reuse):
13496 Improve debug output.
13498 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13500 * dbgcnt.def (prefetch): New debug counter.
13501 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13502 (schedule_prefetches): Stop issueing prefetches if debug counter
13505 2017-06-06 Tom de Vries <tom@codesourcery.com>
13507 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13508 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13510 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13512 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13513 Use aarch64_reg_or_zero predicate for operand 4.
13514 (aarch64_compare_and_swap<mode> define_insn_and_split):
13515 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13516 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13518 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13520 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13521 (arm_compute_save_core_reg_mask): This.
13522 (thumb1_compute_save_reg_mask): Rename into ...
13523 (thumb1_compute_save_core_reg_mask): This.
13524 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13525 (arm_compute_frame_layout): Likewise.
13527 2017-06-06 Richard Biener <rguenther@suse.de>
13529 PR tree-optimization/80974
13530 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13531 keep or clear leaders SSA info.
13533 2017-06-06 Tom de Vries <tom@codesourcery.com>
13535 * config/nvptx/nvptx.c (split_mode_p): New function.
13536 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13538 2017-06-06 Tom de Vries <tom@codesourcery.com>
13540 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13542 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13545 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13548 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13550 * shrink-wrap.c (handle_simple_exit): Update profile.
13551 (try_shrink_wrapping): Upate profile.
13553 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13555 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13556 (tree_guess_outgoing_edge_probabilities): New.
13557 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13558 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13560 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13562 * ipa-split.c (split_function): Initialize return bb profile.
13564 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13566 * profile.c (compute_branch_probabilities): Also initialize
13567 EXIT_BLOCK profile.
13569 2017-06-06 Richard Biener <rguenther@suse.de>
13571 PR tree-optimization/80928
13572 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
13573 (vect_analyze_loop_operations): Properly guard analysis for
13575 (vect_transform_loop): Likewise.
13576 (vect_analyze_loop_2): Also reset SLP type on PHIs.
13577 (vect_model_induction_cost): Do not cost for pure SLP.
13578 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
13579 of induction in inner loop vectorization.
13580 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
13581 (vect_get_and_check_slp_defs): Handle vect_induction_def.
13582 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
13584 (vect_analyze_slp_cost_1): Cost induction.
13585 (vect_detect_hybrid_slp_stmts): Handle PHIs.
13586 (vect_get_slp_vect_defs): Likewise.
13587 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
13588 (vect_transform_stmt): Handle SLP reductions.
13589 * tree-vectorizer.h (vectorizable_induction): Adjust.
13591 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13593 * config/rs6000/rs6000.c (make_resolver_func): Update
13594 init_lowered_empty_function call.
13596 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
13598 * doc/invoke.texi: Document the -fprofile-abs-path option.
13599 * common.opt (fprofile-abs-path): New option.
13600 * gcov-io.h (gcov_write_filename): Declare.
13601 * gcov-io.c (gcov_write_filename): New function.
13602 * coverage.c (coverage_begin_function): Use gcov_write_filename.
13603 * profile.c (output_location): Likewise.
13605 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13607 * shring-wrap.c: Revert accidental commit.
13609 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13611 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13613 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13615 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13617 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13618 profile in callgraph edge.
13619 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13620 (apply_scale): Likewise.
13621 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13622 Also scale profile when inlining function with zero profile.
13623 (initialize_cfun): Update exit block profile even when it is zero.
13624 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13625 when profile is read.
13627 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13629 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13630 (CLONE_*): New constants to define the processors we can generate
13631 code for with the target_clone attribute.
13632 (rs6000_clone_map): New array to identify which clone processors
13633 the current program is running on.
13634 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13635 target_clone attribute.
13636 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13637 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13638 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13639 (cpu_expand_builtin): Add support for target_clone attribute.
13640 (rs6000_valid_attribute_p): Allow "default" attribute.
13641 (get_decl_name): New debug function to simplify printing the
13642 current function name in debugging statements.
13643 (rs6000_clone_priority): New functions to support the target_clone
13644 attribute, and be able to generate code to switch between ISA 2.05
13645 through ISA 3.0 (power6 through power9).
13646 (rs6000_compare_version_priority): Likewise.
13647 (rs6000_get_function_versions_dispatcher): Likewise.
13648 (make_resolver_func): Likewise.
13649 (add_condition_to_bb): Likewise.
13650 (dispatch_function_versions): Likewise.
13651 (rs6000_generate_version_dispatcher_body): Likewise.
13652 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13653 (fusion_gpr_load_p): Fix a spacing issue.
13654 * doc/extend.texi (Common Function Attributes): Document that the
13655 PowerPC supports the target_clone attribute.
13657 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13659 * config/arm/arm.h: explain F symbol found in description of ARM
13660 register allocation in its legend.
13662 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13664 * config/mips/frame-header-opt.c: Include profile-count.h.
13665 * config/riscv/riscv.c: Include profile-count.h
13667 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13669 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13671 (sm_set_flag_if_changed): Add bbs field.
13672 (execute_sm_if_changed_flag_set): Pass BBS.
13673 (execute_sm): Update.
13675 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13677 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13680 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13682 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13683 (peephole2): New peephole2 to emit the above.
13684 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13686 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13688 * config/aarch64/aarch64.c (define_peephole2 above
13689 *sub_<shift>_<mode>): New peephole.
13691 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13693 * config/i386/i386.c (make_resolver_func): Update.
13694 * Makefile.in: Add profile-count.h and profile-count.o
13695 * auto-profile.c (afdo_indirect_call): Update to new API.
13696 (afdo_set_bb_count): Update.
13697 (afdo_propagate_edge): Update.
13698 (afdo_propagate_circuit): Update.
13699 (afdo_calculate_branch_prob): Update.
13700 (afdo_annotate_cfg): Update.
13701 * basic-block.h: Include profile-count.h
13702 (struct edge_def): Turn count to profile_count.
13703 (struct basic_block_def): Likewie.
13704 (REG_BR_PROB_BASE): Move to profile-count.h
13705 (RDIV): Move to profile-count.h
13706 * bb-reorder.c (max_entry_count): Turn to profile_count.
13707 (find_traces): Update.
13708 (rotate_loop):Update.
13709 (connect_traces):Update.
13710 (sanitize_hot_paths):Update.
13711 * bt-load.c (migrate_btr_defs): Update.
13712 * cfg.c (RDIV): Remove.
13713 (init_flow): Use alloc_block.
13714 (alloc_block): Uninitialize count.
13715 (unchecked_make_edge): Uninitialize count.
13716 (check_bb_profile): Update.
13717 (dump_edge_info): Update.
13718 (dump_bb_info): Update.
13719 (update_bb_profile_for_threading): Update.
13720 (scale_bbs_frequencies_int): Update.
13721 (scale_bbs_frequencies_gcov_type): Update.
13722 (scale_bbs_frequencies_profile_count): New.
13723 * cfg.h (update_bb_profile_for_threading): Update.
13724 (scale_bbs_frequencies_profile_count): Declare.
13725 * cfgbuild.c (compute_outgoing_frequencies): Update.
13726 (find_many_sub_basic_blocks): Update.
13727 * cfgcleanup.c (try_forward_edges): Update.
13728 (try_crossjump_to_edge): Update.
13729 * cfgexpand.c (expand_gimple_tailcall): Update.
13730 (construct_exit_block): Update.
13731 * cfghooks.c (verify_flow_info): Update.
13732 (dump_bb_for_graph): Update.
13733 (split_edge): Update.
13734 (make_forwarder_block): Update.
13735 (duplicate_block): Update.
13736 (account_profile_record): Update.
13737 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13738 (get_estimated_loop_iterations): Update.
13739 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13740 (single_likely_exit): Update.
13741 * cfgloopmanip.c (scale_loop_profile): Update.
13743 (set_zero_probability): Update.
13744 (lv_adjust_loop_entry_edge): Update.
13745 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13746 (purge_dead_edges): Update.
13747 (rtl_account_profile_record): Update.
13748 * cgraph.c (cgraph_node::create): Uninitialize count.
13749 (symbol_table::create_edge): Uninitialize count.
13750 (cgraph_update_edges_for_call_stmt_node): Update.
13751 (cgraph_edge::dump_edge_flags): Update.
13752 (cgraph_node::dump): Update.
13753 (cgraph_edge::maybe_hot_p): Update.
13754 * cgraph.h: Include profile-count.h
13755 (create_clone), create_edge, create_indirect_edge): Update.
13756 (cgraph_node): Turn count to profile_count.
13757 (cgraph_edge0: Likewise.
13758 (make_speculative, clone): Update.
13759 (create_edge): Update.
13760 (init_lowered_empty_function): Update.
13761 * cgraphclones.c (cgraph_edge::clone): Update.
13762 (duplicate_thunk_for_node): Update.
13763 (cgraph_node::create_clone): Update.
13764 * cgraphunit.c (cgraph_node::analyze): Update.
13765 (cgraph_node::expand_thunk): Update.
13766 * final.c (dump_basic_block_info): Update.
13767 * gimple-streamer-in.c (input_bb): Update.
13768 * gimple-streamer-out.c (output_bb): Update.
13769 * graphite.c (print_global_statistics): Update.
13770 (print_graphite_scop_statistics): Update.
13771 * hsa-brig.c: Include basic-block.h.
13772 * hsa-dump.c: Include basic-block.h.
13773 * hsa-gen.c (T sum_slice): Update.
13774 (convert_switch_statements):Update.
13775 * hsa-regalloc.c: Include basic-block.h.
13776 * ipa-chkp.c (chkp_produce_thunks): Update.
13777 * ipa-cp.c (struct caller_statistics): Update.
13778 (init_caller_stats): Update.
13779 (gather_caller_stats): Update.
13780 (ipcp_cloning_candidate_p): Update.
13781 (good_cloning_opportunity_p): Update.
13782 (get_info_about_necessary_edges): Update.
13783 (dump_profile_updates): Update.
13784 (update_profiling_info): Update.
13785 (update_specialized_profile): Update.
13786 (perhaps_add_new_callers): Update.
13787 (decide_about_value): Update.
13788 (ipa_cp_c_finalize): Update.
13789 * ipa-devirt.c (struct odr_type_warn_count): Update.
13790 (struct decl_warn_count): Update.
13791 (struct final_warning_record): Update.
13792 (possible_polymorphic_call_targets): Update.
13793 (ipa_devirt): Update.
13794 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13795 * ipa-icf.c (sem_function::merge): Update.
13796 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13797 * ipa-inline.c (compute_uninlined_call_time): Update.
13798 (compute_inlined_call_time): Update.
13799 (want_inline_small_function_p): Update.
13800 (want_inline_self_recursive_call_p): Update.
13801 (edge_badness): Update.
13802 (lookup_recursive_calls): Update.
13803 (recursive_inlining): Update.
13804 (inline_small_functions): Update.
13805 (dump_overall_stats): Update.
13806 (dump_inline_stats): Update.
13807 * ipa-profile.c (ipa_profile_generate_summary): Update.
13808 (ipa_propagate_frequency): Update.
13809 (ipa_profile): Update.
13810 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13811 * ipa-utils.c (ipa_merge_profiles): Update.
13812 * loop-doloop.c (doloop_modify): Update.
13813 * loop-unroll.c (report_unroll): Update.
13814 (unroll_loop_runtime_iterations): Update.
13815 * lto-cgraph.c (lto_output_edge): Update.
13816 (lto_output_node): Update.
13817 (input_node): Update.
13818 (input_edge): Update.
13819 (merge_profile_summaries): Update.
13820 * lto-streamer-in.c (input_cfg): Update.
13821 * lto-streamer-out.c (output_cfg): Update.
13822 * mcf.c (create_fixup_graph): Update.
13823 (adjust_cfg_counts): Update.
13824 (sum_edge_counts): Update.
13825 * modulo-sched.c (sms_schedule): Update.
13826 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13827 * predict.c (maybe_hot_count_p): Update.
13828 (probably_never_executed): Update.
13829 (dump_prediction): Update.
13830 (combine_predictions_for_bb): Update.
13831 (propagate_freq): Update.
13832 (handle_missing_profiles): Update.
13833 (counts_to_freqs): Update.
13834 (rebuild_frequencies): Update.
13835 (force_edge_cold): Update.
13836 * predict.h: Include profile-count.h
13837 (maybe_hot_count_p, counts_to_freqs): UPdate.
13838 * print-rtl-function.c: Do not include cfg.h
13839 * print-rtl.c: Include basic-block.h
13840 * profile-count.c: New file.
13841 * profile-count.h: New file.
13842 * profile.c (is_edge_inconsistent): Update.
13843 (correct_negative_edge_counts): Update.
13844 (is_inconsistent): Update.
13845 (set_bb_counts): Update.
13846 (read_profile_edge_counts): Update.
13847 (compute_frequency_overlap): Update.
13848 (compute_branch_probabilities): Update; Initialize and deinitialize
13850 (branch_prob): Update.
13851 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13852 (edge_gcov_count): New.
13853 (bb_gcov_count): New.
13854 * shrink-wrap.c (try_shrink_wrapping): Update.
13855 * tracer.c (better_p): Update.
13856 * trans-mem.c (expand_transaction): Update.
13857 (ipa_tm_insert_irr_call): Update.
13858 (ipa_tm_insert_gettmclone_call): Update.
13859 * tree-call-cdce.c: Update.
13860 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13861 (gimple_duplicate_sese_tail): Update.
13862 (gimple_account_profile_record): Update.
13863 (execute_fixup_cfg): Update.
13864 * tree-inline.c (copy_bb): Update.
13865 (copy_edges_for_bb): Update.
13866 (initialize_cfun): Update.
13867 (freqs_to_counts): Update.
13868 (copy_cfg_body): Update.
13869 (expand_call_inline): Update.
13870 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13871 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13872 (try_unroll_loop_completely): Update.
13873 (try_peel_loop): Update.
13874 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13875 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13876 * tree-ssa-loop-split.c (connect_loops): Update.
13877 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13878 * tree-ssa-reassoc.c (branch_fixup): Update.
13879 * tree-ssa-tail-merge.c (replace_block_by): Update.
13880 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13881 (compute_path_counts): Update.
13882 (update_profile): Update.
13883 (recompute_probabilities): Update.
13884 (update_joiner_offpath_counts): Update.
13885 (estimated_freqs_path): Update.
13886 (freqs_to_counts_path): Update.
13887 (clear_counts_path): Update.
13888 (ssa_fix_duplicate_block_edges): Update.
13889 (duplicate_thread_path): Update.
13890 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13891 (struct switch_conv_info): Update.
13892 * tree-tailcall.c (decrease_profile): Update.
13893 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13894 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13895 * value-prof.c (check_counter): Update.
13896 (gimple_divmod_fixed_value): Update.
13897 (gimple_mod_pow2): Update.
13898 (gimple_mod_subtract): Update.
13899 (gimple_ic_transform): Update.
13900 (gimple_stringop_fixed_value): Update.
13901 * value-prof.h (gimple_ic): Update.
13903 2017-06-02 Carl Love <cel@us.ibm.com>
13905 * config/rs6000/rs6000-c: Add support for built-in functions
13906 vector double vec_doublee (vector signed int);
13907 vector double vec_doublee (vector unsigned int);
13908 vector double vec_doublee (vector float);
13909 vector double vec_doubleh (vector signed int);
13910 vector double vec_doubleh (vector unsigned int);
13911 vector double vec_doubleh (vector float);
13912 vector double vec_doublel (vector signed int);
13913 vector double vec_doublel (vector unsigned int);
13914 vector double vec_doublel (vector float);
13915 vector double vec_doubleo (vector signed int);
13916 vector double vec_doubleo (vector unsigned int);
13917 vector double vec_doubleo (vector float);.
13918 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13919 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13921 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13922 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13923 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13925 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13926 vec_doublel, vec_doubleh.
13927 * doc/extend.texi: Update the built-in documentation file for the
13928 new built-in functions.
13930 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13933 * ipa-inline-analysis.c (free_growth_caches): Set
13934 edge_removal_hook_holder to NULL after removing it.
13936 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13938 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13939 comparision with zero.
13941 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13942 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13943 for early expansion of vec_min and vec_max builtins.
13944 (builtin_function_type): Add min/max unsigned variants to those
13945 identified as having unsigned arguments.
13947 2017-06-02 Olivier Hainque <hainque@adacore.com>
13949 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13951 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13953 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13954 Use VALL_F16 iterator rather than VALL.
13956 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13958 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13959 Emit CBNZ inside loop when doing a strong exchange and comparing
13960 against zero. Generate the CC flags after the loop.
13962 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13964 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13965 (dl_section_ref): New.
13966 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13967 On AIX, append an expression to subtract the size of the
13968 section length to dl_section_ref.
13970 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13972 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13973 for early expansion of vector absolute builtins.
13975 2017-06-02 Richard Biener <rguenther@suse.de>
13977 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13978 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13980 2017-06-02 Richard Biener <rguenther@suse.de>
13982 PR tree-optimization/80948
13983 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13984 stmt order as well.
13986 2017-06-02 Richard Biener <rguenther@suse.de>
13988 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13990 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13991 for inductions as relevant.
13993 2017-06-02 Richard Biener <rguenther@suse.de>
13995 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13996 (vectorizable_induction): ... this. Remove dead code.
13998 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14000 * builtins. (expand_builtin_alloca): Remove second parameter and
14001 infer its value from the first parameter instead.
14002 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14004 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14006 PR rtl-optimization/80903
14007 * loop-doloop.c (add_test): Unshare sequence.
14009 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14011 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14013 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14015 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14017 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14018 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14019 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14020 (xlogue_layout::get_stub_rtx): Make static.
14021 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14022 (xlogue_layout::compute_stub_managed_regs): Rename to...
14023 (xlogue_layout::count_stub_managed_regs): ...this.
14024 (xlogue_layout::is_stub_managed_reg): New function.
14025 (xlogue_layout::m_stub_names): Rename to...
14026 (xlogue_layout::s_stub_names): ...this, make static.
14027 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14028 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14029 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14030 xlogue_layout::s_stub_names): Instantiate statics.
14031 (stub_managed_regs): Remove.
14032 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14033 (disable_call_ms2sysv_xlogues): Rename to...
14034 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14035 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14037 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14038 change after reload_completed.
14039 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14041 (ix86_expand_prologue): Likewise.
14042 (ix86_expand_epilogue): Likewise.
14043 (ix86_expand_split_stack_prologue): Likewise.
14044 (ix86_compute_frame_layout): Remove frame parameter ...
14045 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14046 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14048 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14049 (ix86_frame): Move from here ...
14050 * config/i386/i386.h (ix86_frame): ... to here.
14051 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14052 complete ix86_frame data structure instead. Remove some_ld_name.
14054 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14056 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14057 symbols that hold a DECL_VALUE_EXPR.
14059 2017-06-01 Martin Jambor <mjambor@suse.cz>
14061 PR tree-optimization/80898
14062 * tree-sra.c (process_subtree_disqualification): Removed.
14063 (disqualify_candidate): Do not acll
14064 process_subtree_disqualification.
14065 (subtree_mark_written_and_enqueue): New function.
14066 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14067 RHS has been disqualified and re-queue LHS if necessary. Apart
14068 from that, ignore disqualified RHS.
14070 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14072 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14073 address fetch for z10 or later.
14075 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14077 * config/arc/arc.md (tst_movb): Add guard when splitting.
14079 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14081 * config/arc/arc.c (arc_can_eliminate): Test against
14082 arc_frame_pointer_needed.
14084 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14086 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14087 to prevent store reordering.
14088 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14089 (type): Add block type.
14090 (stack_tie): Define special instruction to be used in
14093 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14095 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14096 constraint. It is not valid for the pattern.
14097 (noncommutative_binary_comparison): Likewise.
14099 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14101 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14104 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14106 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14107 be used by the reg-alloc.
14109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14111 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14112 reg-alloc when having mul64 or mul32x16 instructions.
14113 (mulsidi3): Likewise.
14114 (umulsidi3): Likewise.
14115 (mulsi32x16): New pattern.
14116 (mulsi64): Likewise.
14117 (mulsidi64): Likewise.
14118 (umulsidi64): Likewise.
14119 (MUL32x16_REG): Define.
14120 (mul64_600): Use MUL32x16_REG.
14121 (mac64_600): Likewise.
14122 (umul64_600): Likewise.
14123 (umac64_600): Likewise.
14125 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14127 * config/arc/arc.md (mulsi3_700): Make it commutative.
14129 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14131 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14133 (*sign_extendsidi2_insn): Likewise for movstosw.
14135 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14137 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14138 the type of the input discriminant value. Convert the
14139 discriminant value of signedness vary.
14141 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14143 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14144 Add to -Wall section.
14146 2017-06-01 Richard Biener <rguenther@suse.de>
14148 PR middle-end/66313
14149 * fold-const.c (fold_plusminus_mult_expr): If the factored
14150 factor may be zero use a wrapping type for the inner operation.
14151 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14152 and handle moved defs.
14153 (process_assignment): Properly guard the unary op case. Return a
14154 tri-state indicating that moving the stmt before the call may allow
14155 to continue. Pass through to_move.
14156 (find_tail_calls): Handle moving unrelated defs before
14159 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14162 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14163 splitter result in the canonical way.
14165 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14167 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14168 also for 32bit target. Update insn attributes.
14169 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14171 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14173 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14174 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14175 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14176 (_mm_maskz_min_ss): New intrinsics.
14178 2017-05-31 Martin Liska <mliska@suse.cz>
14180 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14181 Change comment style to one we normally use.
14182 (vectorizable_reduction): Likewise.
14183 (vectorizable_induction): Likewise.
14184 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14185 (vectorizable_call): Likewise.
14186 (vectorizable_simd_clone_call): Likewise.
14187 (vectorizable_conversion): Likewise.
14188 (vectorizable_assignment): Likewise.
14189 (vectorizable_shift): Likewise.
14190 (vectorizable_operation): Likewise.
14191 (vectorizable_store): Likewise.
14192 (vectorizable_load): Likewise.
14193 * tree-vectorizer.h: Likewise.
14195 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14197 * passes.c (emergency_dump_function): New.
14198 * tree-pass.h (emergency_dump_function): Declare.
14199 * plugin.c (plugins_internal_error_function): Remove.
14200 * plugin.h (plugins_internal_error_function): Remove declaration.
14201 * toplev.c (internal_error_function): New static function. Use it...
14202 (general_init): ...here.
14204 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14206 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14207 (arc_rtx_costs): Add costs for new patterns.
14208 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14209 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14211 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14213 * tree-ssa-strlen.c (get_next_strinfo): New function.
14214 (get_stridx_plus_constant): Use it.
14215 (zero_length_string): Likewise.
14216 (adjust_related_strinfos): Likewise.
14217 (adjust_last_stmt): Likewise.
14219 2017-05-31 Richard Biener <rguenther@suse.de>
14222 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14223 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14225 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14227 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14228 loop_vinfo argument and use of dependence distance vectors.
14229 Check instead whether the two references differ only in their
14230 initial value and assume that they have the same alignment if the
14231 difference is a multiple of the vector alignment.
14232 (vect_analyze_data_refs_alignment): Update call accordingly.
14234 2017-05-31 Martin Liska <mliska@suse.cz>
14237 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14239 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14241 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14242 (create_intersect_range_checks): Move from ...
14243 * tree-data-ref.c (create_intersect_range_checks_index)
14244 (create_intersect_range_checks): ... to here.
14245 (create_runtime_alias_checks): New function factored from ...
14246 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14247 here. Call above function.
14248 * tree-data-ref.h (create_runtime_alias_checks): New function.
14250 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14252 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14253 segment length for dr_b and compute it in wide_int.
14255 2017-05-31 Richard Biener <rguenther@suse.de>
14257 PR tree-optimization/80906
14258 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14259 and pass through iv_map.
14260 (copy_bb_and_scalar_dependences): Adjust.
14261 (translate_pending_phi_nodes): Likewise.
14262 (copy_loop_close_phi_args): Handle code-generating IVs instead
14265 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14267 * diagnostic-color.c (color_dict): Add "type-diff".
14268 (parse_gcc_colors): Update comment.
14269 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14270 -fdiagnostics-show-template-tree and -fno-elide-type.
14271 (GCC_COLORS): Add type-diff to example.
14273 (-fdiagnostics-show-template-tree): New.
14274 (-fno-elide-type): New.
14275 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14276 the pp_format_decoder callback. Call any m_format_postprocessor's
14278 (pretty_printer::pretty_printer): Initialize
14279 m_format_postprocessor.
14280 (pretty_printer::~pretty_printer): Delete any
14281 m_format_postprocessor.
14282 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14283 (class format_postprocessor): New class.
14284 (struct pretty_printer::format_decoder): Document the new parameters.
14285 (struct pretty_printer::m_format_postprocessor): New field.
14286 * tree-diagnostic.c (default_tree_printer): Update for new
14287 bool and const char ** params.
14288 * tree-diagnostic.h (default_tree_printer): Likewise.
14290 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14292 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14293 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14294 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14295 rs6000_gen_cell_microcode code.
14296 (rs6000_final_prescan_insn): Delete.
14297 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14298 "warn-cell-microcode" entries.
14299 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14300 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14301 throughout. Change cc_reg_not_micro_cr0_operand to
14302 cc_reg_not_cr0_operand throughout.
14303 (*extendhi<mode>2_noload): Delete.
14304 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14305 (mwarn-cell-microcode): Delete.
14306 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14307 -mgen-cell-microcode and -mwarn-cell-microcode.
14309 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14312 * config/i386/constraints.md (Yd): New constraint.
14314 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14315 and (?Yd, r) alternatives. Update insn attributes.
14316 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14317 and (?*Yd, r) alternatives. Update insn attributes.
14318 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14320 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14322 * gimplify.c (gimplify_modify_expr): Don't create a
14323 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14326 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14328 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14330 2017-05-30 Richard Biener <rguenther@suse.de>
14332 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14333 and reduc_def fields.
14334 (STMT_VINFO_REDUC_TYPE): New define.
14335 (STMT_VINFO_REDUC_DEF): Likewise.
14336 (vect_force_simple_reduction): Adjust prototype.
14337 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14338 (vect_is_simple_reduction): Remove check_reduction argument.
14339 (vect_force_simple_reduction): Adjust and set
14340 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14341 (vectorizable_reduction): Do not re-do reduction analysis
14342 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14343 * tree-parloops.c (gather_scalar_reductions): Adjust.
14345 2017-05-30 Richard Biener <rguenther@suse.de>
14347 PR middle-end/80901
14348 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14351 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14353 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14354 Introduce unknown_misalignment parameter and remove vf.
14355 (vect_peeling_hash_get_lowest_cost):
14356 Pass unknown_misalignment parameter.
14357 (vect_enhance_data_refs_alignment):
14358 Fix unsupportable data ref treatment.
14360 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14362 * tree-vect-data-refs.c (vect_get_data_access_cost):
14363 Workaround for SLP handling.
14364 (vect_enhance_data_refs_alignment):
14365 Compute costs for doing no peeling at all, compare to the best
14366 peeling costs so far and avoid peeling if cheaper.
14368 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14370 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14371 Return peeling info and set costs to zero for unlimited cost
14373 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14374 with unknown misalignment. Compute and costs for unknown
14375 misalignment, compare them to the costs for known misalignment
14376 and choose the cheapest for peeling.
14378 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14380 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14381 (vect_get_peeling_costs_all_drs): Create function.
14382 (vect_peeling_hash_get_lowest_cost):
14383 Use vect_get_peeling_costs_all_drs.
14384 (vect_peeling_supportable): Create function.
14385 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14387 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14389 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14390 DR_HAS_NEGATIVE_STEP.
14391 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14392 (vect_enhance_data_refs_alignment): Use.
14393 (vect_duplicate_ssa_name_ptr_info): Use.
14394 * tree-vectorizer.h (dr_misalignment): Use.
14395 (known_alignment_for_access_p): Use.
14397 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14400 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14402 (has_section_name): New function.
14404 2017-05-30 Martin Liska <mliska@suse.cz>
14407 * auto-profile.c (get_function_decl_from_block): Fix
14410 2017-05-30 Richard Biener <rguenther@suse.de>
14412 PR middle-end/80876
14413 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14415 2017-05-30 Martin Liska <mliska@suse.cz>
14417 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14418 * dumpfile.h (struct dump_file_info): Remove ctors.
14420 2017-05-30 Martin Liska <mliska@suse.cz>
14422 * predict.def: Fix GNU coding style.
14424 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14426 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14427 Mark 'to' argument with ATTRIBUTE_UNUSED.
14429 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14431 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14432 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14433 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14436 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14438 * doc/install.texi (Options specification): Restore entry of
14439 --enable-sjlj-exceptions.
14441 2017-05-27 Michael Eager <eager@eagercon.com>
14444 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14446 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14448 * config/microblaze/microblaze.h
14449 (FIXED_REGISTERS): Update in macro.
14450 (CALL_USED_REGISTERS): Update in macro.
14452 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14454 * doc/install.texi: Add links to macOS binary distributions.
14456 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14460 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14462 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14464 2017-05-26 Martin Liska <mliska@suse.cz>
14466 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14468 2017-05-26 Martin Liska <mliska@suse.cz>
14470 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14471 always leading ';; '.
14472 (dump_bb_info): Likewise.
14473 (brief_dump_cfg): Likewise.
14474 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14475 * dumpfile.c: Remove usage of TDF_VERBOSE.
14476 * dumpfile.h (enum dump_kind): Likewise.
14477 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14478 * print-tree.c (debug_verbose): Remove.
14479 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14480 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14481 * tree-diagnostic.c (default_tree_printer): Replace
14482 TDF_DIAGNOSTIC with TDF_SLIM.
14484 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14486 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14487 in parameter loop, rather than loop_vinfo.
14488 (create_intersect_range_checks): Ditto.
14489 (vect_create_cond_for_alias_checks): Update call to above functions.
14491 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14493 PR tree-optimization/80815
14494 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14495 for merging runtime alias checks. Handle negative DR_STEPs.
14497 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14499 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14501 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14502 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14503 out code pruning runtime alias checks.
14504 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14505 factored out from above.
14506 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14508 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14510 (prune_runtime_alias_test_list): New decalaration.
14512 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14514 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14515 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14516 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14517 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14518 (operator==, comp_dr_with_seg_len_pair): Ditto.
14519 (vect_prune_runtime_alias_test_list): Ditto.
14521 2017-05-26 Martin Liska <mliska@suse.cz>
14524 * params.def: Bound partial-inlining-entry-probability param.
14526 2017-05-26 Marek Polacek <polacek@redhat.com>
14529 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14532 2017-05-26 Richard Biener <rguenther@suse.de>
14534 PR tree-optimization/80842
14535 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14538 2017-05-26 Richard Biener <rguenther@suse.de>
14540 PR tree-optimization/80844
14541 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14543 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14545 * doc/md.texi (Machine Constraints): Update x86 family
14546 machine constraints section to match 'config/i386/constraints.md'.
14548 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14550 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14552 2017-05-25 Nathan Sidwell <nathan@acm.org>
14554 * doc/invoke.texi (--enable-languages): Update documentation.
14556 2017-05-25 Martin Liska <mliska@suse.cz>
14558 * dumpfile.c: Add TDF_FOLDING.
14559 * dumpfile.h (enum dump_kind): Likewise.
14560 * genmatch.c (dt_simplify::gen_1): Use it.
14562 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14564 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14566 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14568 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14569 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
14571 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14573 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
14574 * match.pd (X == C): Rewrite it here.
14575 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
14576 with_certain_nonzero_bits2): New predicates.
14577 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
14579 2017-05-24 Nathan Sidwell <nathan@acm.org>
14581 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
14584 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
14587 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14589 * config/powerpcspe: New port. Files are copied from the rs6000
14590 port, with "rs6000" in filenames replaced by "powerpcspe".
14592 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
14594 PR rtl-optimization/80754
14595 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
14597 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
14599 * config/sparc/sparc.md (length): Return the correct value for -mflat
14600 sibcalls to match output_sibcall.
14602 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14606 * config/rs6000/rs6000.c (struct machine_function): Add new field
14608 (rs6000_get_separate_components): Init that field, use it.
14609 (rs6000_components_for_bb): Use the field.
14611 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14613 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14615 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14617 PR middle-end/80823
14618 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14620 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14623 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14624 address against address_operand predicate.
14625 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14627 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14629 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14630 parameters passed indirectly.
14632 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14634 * config/i386/i386.md (*movdi_internal): Remove SSE4
14635 alternative 18 (?r, *v). Update insn attributes.
14636 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14637 Update insn attributes.
14638 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14639 Update insn attributes.
14640 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14641 alternative 1 (r, v). Remove isa attribute.
14642 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14643 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14644 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14646 2017-05-23 Tom de Vries <tom@codesourcery.com>
14648 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14651 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14653 * cgraphunit.c (symbol_table::process_new_functions): Update.
14654 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14655 (inline_generate_summary): Rename to ...
14656 (ipa_fn_summary_generate): ... this one.
14657 (inline_read_summary): Rename to ...
14658 (ipa_fn_summary_read): ... this one.
14659 (inline_write_summary): Rename to ...
14660 (ipa_fn_summary_write): ... this one.
14661 (inline_free_summary): Rename to ...
14662 (ipa_free_fn_summary): ... this one.
14663 (pass_data_local_fn_summary, pass_local_fn_summary,
14664 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14665 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14666 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14667 make_pass_ipa_fn_summary): New.
14668 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14669 inline_write_summary, inline_free_summary): Remove.
14670 (ipa_free_fn_summary) : New.
14671 * ipa-inline.c (ipa_inline): Update.
14672 (pass_ipa_inline): Do not generate summaries.
14673 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14675 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14676 and add pass_ipa_fn_summary.
14677 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14679 (make_pass_inline_parameters): Remove.
14681 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14683 * omp-low.c (struct omp_context): Remove "default_kind" member.
14686 * omp-offload.c (execute_oacc_device_lower): Remove the
14687 parallelism dimensions function attributes for unparallelized
14688 OpenACC kernels constructs.
14690 2017-05-23 Martin Liska <mliska@suse.cz>
14692 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14694 (cgraph_edge::make_speculative): Likewise.
14695 (cgraph_edge::resolve_speculation): Likewise.
14696 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14697 (cgraph_node::dump): Likewise.
14698 * cgraph.h: Likewise.
14699 * cgraphunit.c (analyze_functions): Likewise.
14700 (symbol_table::compile): Likewise.
14701 * ipa-cp.c (print_all_lattices): Likewise.
14702 (determine_versionability): Likewise.
14703 (initialize_node_lattices): Likewise.
14704 (ipcp_verify_propagated_values): Likewise.
14705 (estimate_local_effects): Likewise.
14706 (update_profiling_info): Likewise.
14707 (create_specialized_node): Likewise.
14708 (perhaps_add_new_callers): Likewise.
14709 (decide_about_value): Likewise.
14710 (decide_whether_version_node): Likewise.
14711 (identify_dead_nodes): Likewise.
14712 (ipcp_store_bits_results): Likewise.
14713 * ipa-devirt.c (dump_targets): Likewise.
14714 (ipa_devirt): Likewise.
14715 * ipa-icf.c (sem_item::dump): Likewise.
14716 (sem_function::equals): Likewise.
14717 (sem_variable::equals): Likewise.
14718 (sem_item_optimizer::read_section): Likewise.
14719 (sem_item_optimizer::execute): Likewise.
14720 (congruence_class::dump): Likewise.
14721 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14722 (dump_inline_summary): Likewise.
14723 (estimate_node_size_and_time): Likewise.
14724 (inline_analyze_function): Likewise.
14725 * ipa-inline-transform.c (inline_call): Likewise.
14726 * ipa-inline.c (report_inline_failed_reason): Likewise.
14727 (want_early_inline_function_p): Likewise.
14728 (edge_badness): Likewise.
14729 (update_edge_key): Likewise.
14730 (inline_small_functions): Likewise.
14731 * ipa-profile.c (ipa_profile): Likewise.
14732 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14733 (ipa_make_edge_direct_to_target): Likewise.
14734 (remove_described_reference): Likewise.
14735 (ipa_impossible_devirt_target): Likewise.
14736 (propagate_controlled_uses): Likewise.
14737 (ipa_print_node_params): Likewise.
14738 (ipcp_transform_function): Likewise.
14739 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14740 (propagate_pure_const): Likewise.
14741 * ipa-reference.c (generate_summary): Likewise.
14742 (read_write_all_from_decl): Likewise.
14743 (propagate): Likewise.
14744 (ipa_reference_read_optimization_summary): Likewise.
14745 * ipa-utils.c (ipa_merge_profiles): Likewise.
14746 * ipa.c (walk_polymorphic_call_targets): Likewise.
14747 (symbol_table::remove_unreachable_nodes): Likewise.
14748 (ipa_single_use): Likewise.
14749 * passes.c (execute_todo): Likewise.
14750 * predict.c (drop_profile): Likewise.
14751 * symtab.c (symtab_node::get_dump_name): New function.
14752 (symtab_node::dump_name): Likewise.
14753 (symtab_node::dump_asm_name): Likewise.
14754 (symtab_node::dump_references): Likewise.
14755 (symtab_node::dump_referring): Likewise.
14756 (symtab_node::dump_base): Likewise.
14757 (symtab_node::debug_symtab): Likewise.
14758 * tree-sra.c (convert_callers_for_node): Likewise.
14759 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14760 * value-prof.c (init_node_map): Likewise.
14762 2017-05-23 Martin Liska <mliska@suse.cz>
14764 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14765 and symtab_node::debug_symtab to symbol_table::debug.
14766 * cgraphunit.c (analyze_functions): Use the renamed function.
14767 (symbol_table::compile): Likewise.
14768 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14769 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14770 * passes.c (execute_todo): Likewise.
14771 * symtab.c (symbol_table::dump): New function.
14772 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14774 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14776 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14777 that nonconst implies exec.
14779 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14781 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14782 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14783 (inline_edge_summary_vec): Turn into ...
14784 (ipa_call_summaries): ... this one.
14785 (redirect_to_unreachable, edge_set_predicate,
14786 evaluate_properties_for_edge, inline_summary_alloc,
14787 reset_ipa_call_summary, reset_inline_summary,
14788 inline_summary_t::duplicate): Update.
14789 (inline_edge_duplication_hook): Turn to ...
14790 (ipa_call_summary_t::duplicate): ... this one.
14791 (inline_edge_removal_hook): Turn to ...
14792 (ipa_call_summary_t::remove): ... this one.
14793 (dump_inline_edge_summary): Turn to ...
14794 (dump_ipa_call_summary): ... this one.
14795 (estimate_function_body_sizes): Update.
14796 (inline_update_callee_summaries): Update.
14797 (remap_edge_change_prob): Update.
14798 (remap_edge_summaries): Update.
14799 (inline_merge_summary): Update.
14800 (do_estimate_edge_time): Update.
14801 (inline_generate_summary): Update.
14802 (inline_read_section): Update.
14803 (inline_read_summary): Update.
14804 (inline_free_summary): Update.
14805 * ipa-inline.c (can_inline_edge_p): Update.
14806 (compute_inlined_call_time): Update.
14807 (want_inline_small_function_p): Update.
14808 (edge_badness): Update.
14809 (early_inliner): Update.
14810 * ipa-inline.h (inline_edge_summary): Turn to ...
14811 (ipa_call_summary): ... this one.
14812 (ipa_call_summary_t): New class.
14813 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14814 (ipa_call_summaries): New.
14815 (inline_edge_summary): Remove.
14816 (estimate_edge_growth): Update.
14817 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14818 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14819 * ipa-split.c (execute_split_functions): Update.
14820 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14822 2017-05-23 Tom de Vries <tom@codesourcery.com>
14824 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14825 attributes): Document rdrand effective target.
14827 2017-05-23 Tom de Vries <tom@codesourcery.com>
14829 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14830 attributes): Sort alphabetically.
14832 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14834 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14836 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14839 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14840 V2DF/V2DI splat into two separate patterns, one that handles
14841 registers, and the other that only handles memory. Drop support
14842 for splatting from a GPR on ISA 2.07 and then splitting the
14843 splat into direct move and splat.
14844 (vsx_splat_<mode>_reg): Likewise.
14845 (vsx_splat_<mode>_mem): Likewise.
14847 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14849 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14851 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14853 PR middle-end/80809
14854 * omp-low.c (finish_taskreg_remap): New function.
14855 (finish_taskreg_scan): If unit size of ctx->record_type
14856 is non-constant, unshare the size expression and replace
14857 decls in it with possible outer var refs.
14859 PR middle-end/80809
14860 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14861 GOVD_SHARED rather than GOVD_PRIVATE with it.
14862 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14863 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14865 PR middle-end/80853
14866 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14867 as last argument to build_outer_var_ref for pointer bases of array
14868 section reductions.
14870 2017-05-19 Martin Sebor <msebor@redhat.com>
14872 * print-tree.c (print_node): Print DECL_READ_P flag.
14874 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14876 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14877 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14878 * cgraph.c: Likewise.
14879 * cgraphunit.c: Likewise.
14880 * gengtype.c: Likewise.
14881 * ipa-cp.c: Likewise.
14882 * ipa-devirt.c: Likewise.
14883 * ipa-icf.c: Likewise.
14884 * ipa-predicate.c: Likewise.
14885 * ipa-profile.c: Likewise.
14886 * ipa-prop.c: Likewise.
14887 * ipa-split.c: Likewise.
14889 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14890 edge_predicate_pool, dump_inline_hints,
14891 inline_summary::account_size_time, redirect_to_unreachable,
14892 edge_set_predicate, set_hint_predicate,
14893 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14894 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14895 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14896 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14897 ipa_call_summary_t::remove, initialize_growth_caches,
14898 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14899 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14900 mark_modified, unmodified_parm_1, unmodified_parm,
14901 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14902 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14903 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14904 will_be_nonconstant_predicate, record_modified_bb_info,
14905 get_minimal_bb, record_modified, param_change_prob,
14906 phi_result_unknown_predicate, predicate_for_phi_result,
14907 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14908 estimate_function_body_sizes, compute_inline_parameters,
14909 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14910 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14911 inline_update_callee_summaries, remap_edge_change_prob,
14912 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14913 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14914 inline_analyze_function, inline_summary_t::insert,
14915 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14916 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14917 inline_free_summary): Move to ipa-fnsummary.h
14918 (predicate_t): Remove.
14919 * ipa-fnsummary.c: New file.
14920 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14921 (enum inline_hints_vals, inline_hints, agg_position_info,
14922 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14923 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14924 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14925 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14926 inline_read_summary, inline_write_summary, inline_free_summary,
14927 inline_analyze_function, initialize_inline_failed,
14928 inline_merge_summary, inline_update_overall_summary,
14929 compute_inline_parameters): Move to ipa-fnsummary.h
14930 * ipa-fnsummary.h: New file.
14931 * ipa-inline-transform.h: Include ipa-inline.h.
14932 * ipa-inline.c: LIkewise.
14934 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14936 * ipa-inline.c (edge_badness): Use inlined_time instead of
14937 inline_summaries->get.
14939 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14941 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14943 2017-05-22 Nathan Sidwell <nathan@acm.org>
14945 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14946 (fdump-lang): Document 'raw' option.
14947 * dumpfile.h (TDI_tu): Delete.
14948 * dumpfile.c (dump_files): Remove translation-unit.
14949 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14951 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14953 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14954 command option from $(AWK) call.
14955 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14956 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14957 [FORMAT]: Remove handling of variable.
14958 * config/avr/t-multilib: Regenerate.
14960 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14962 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14964 (dump_inline_summary): Do not print self_time.
14965 (estimate_function_body_sizes): Do not set self_time.
14966 (compute_inline_parameters): Likewise.
14967 (inline_read_section, inline_write_summary): Do not stream self_time.
14968 * ipa-inline.h (inline_summary): Drop self_time.
14970 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14972 * ipa-inline-analysis.c (account_size_time): Rename to ...
14973 (inline_summary::account_size_time): ... this one.
14974 (reset_ipa_call_summary): Turn to ...
14975 (ipa_call_summary::reset): ... this one.
14976 (reset_inline_summary): Turn to ...
14977 (inline_summary::reset): ... this one.
14978 (inline_summary_t::remove): Update.
14979 (inline_summary_t::duplicate): Update.
14980 (ipa_call_summary_t::remove): Update.
14981 (dump_inline_summary): Update.
14982 (estimate_function_body_sizes): Update.
14983 (compute_inline_parameters): Update.
14984 (estimate_node_size_and_time): Update.
14985 (inline_merge_summary): Update.
14986 (inline_update_overall_summary): Update.
14987 (inline_read_section): Update.
14988 (inline_write_summary): Update.
14989 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14990 add account_size_time and reset member functions.
14991 (ipa_call_summary): Add reset function.
14992 * ipa-predicate.h (predicate::operator &): Constify.
14994 2017-05-22 Richard Biener <rguenther@suse.de>
14996 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14998 2017-05-19 Jason Merrill <jason@redhat.com>
15000 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15002 2017-05-19 Marek Polacek <polacek@redhat.com>
15005 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15006 TYPE_OVERFLOW_WRAPS checks.
15008 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15010 * tree-core.h (enum omp_clause_default_kind): Add
15011 "OMP_CLAUSE_DEFAULT_PRESENT".
15012 * tree-pretty-print.c (dump_omp_clause): Handle it.
15013 * gimplify.c (enum gimplify_omp_var_data): Add
15014 "GOVD_MAP_FORCE_PRESENT".
15015 (gimplify_adjust_omp_clauses_1): Map it to
15016 "GOMP_MAP_FORCE_PRESENT".
15017 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15019 * gimplify.c (oacc_default_clause): Clarify.
15021 2017-05-19 Nathan Sidwell <nathan@acm.org>
15023 LANG_HOOK_REGISTER_DUMPS
15024 * toplev.c (general_init): Call register dump lang hook.
15025 * doc/invoke.texi: Document -fdump-lang option family.
15026 * dumpfile.c (dump_files): Remove class dump here.
15027 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15028 * dumpfile.h (tree_dump_index): Remove TDI_class.
15029 * langhooks-def.h (lhd_register_dumps): Declare.
15030 (LANG_HOOKS_REGISTER_DUMPS): Define.
15031 (LANG_HOOKS_INITIALIZER): Add it.
15032 * langhooks.c (lhd_register_dumps): Define.
15033 * langhooks.h (struct lang_hooks): Add register_dumps.
15035 2017-05-19 Nathan Sidwell <nathan@acm.org>
15037 * context.h (context::set_passes): New.
15038 * context.c (context::context): Do not create pass manager.
15039 * toplev.c (general_init): Create pass manager here.
15041 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15043 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15044 use this splitter if two add or or instructions would also work for
15045 the constant we want to generate.
15047 2017-05-19 Richard Biener <rguenther@suse.de>
15050 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15051 predicate evaluation.
15053 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15055 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15057 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15060 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15062 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15063 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15064 (inline_edge_summary_vec): Turn into ...
15065 (ipa_call_summaries): ... this one.
15066 (redirect_to_unreachable, edge_set_predicate,
15067 evaluate_properties_for_edge, inline_summary_alloc,
15068 reset_ipa_call_summary, reset_inline_summary,
15069 inline_summary_t::duplicate): Update.
15070 (inline_edge_duplication_hook): Turn to ...
15071 (ipa_call_summary_t::duplicate): ... this one.
15072 (inline_edge_removal_hook): Turn to ...
15073 (ipa_call_summary_t::remove): ... this one.
15074 (dump_inline_edge_summary): Turn to ...
15075 (dump_ipa_call_summary): ... this one.
15076 (estimate_function_body_sizes): Update.
15077 (inline_update_callee_summaries): Update.
15078 (remap_edge_change_prob): Update.
15079 (remap_edge_summaries): Update.
15080 (inline_merge_summary): Update.
15081 (do_estimate_edge_time): Update.
15082 (inline_generate_summary): Update.
15083 (inline_read_section): Update.
15084 (inline_read_summary): Update.
15085 (inline_free_summary): Update.
15086 * ipa-inline.c (can_inline_edge_p): Update.
15087 (compute_inlined_call_time): Update.
15088 (want_inline_small_function_p): Update.
15089 (edge_badness): Update.
15090 (early_inliner): Update.
15091 * ipa-inline.h (inline_edge_summary): Turn to ...
15092 (ipa_call_summary): ... this one.
15093 (ipa_call_summary_t): New class.
15094 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15095 (ipa_call_summaries): New.
15096 (inline_edge_summary): Remove.
15097 (estimate_edge_growth): Update.
15098 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15099 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15100 * ipa-split.c (execute_split_functions): Update.
15101 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15103 2017-05-19 Richard Biener <rguenther@suse.de>
15105 PR middle-end/80764
15106 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15108 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15110 * config/rs6000/rs6000.c (struct machine_function): Add field
15111 fpr_is_wrapped_separately.
15112 (rs6000_get_separate_components): Use 64 components. Handle the
15113 new FPR components.
15114 (rs6000_components_for_bb): Handle the FPR components.
15115 (rs6000_emit_prologue_components): Handle the FPR components.
15116 (rs6000_emit_epilogue_components): Handle the FPR components.
15117 (rs6000_set_handled_components): Handle the FPR components.
15118 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15119 that are already separately shrink-wrapped.
15120 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15121 that are already separately shrink-wrapped.
15123 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15126 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15129 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15130 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15131 cases where the register allocator uses a move and an offsettable
15132 memory operation to/from a FPR register on ISA 2.06/2.07.
15133 (define_peephole2 for Altivec d-form store): Likewise.
15135 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15138 * config/i386/mmx.md (*mov<mode>_internal): Enable
15139 alternatives 11, 12, 13 and 14 also for 32bit targets.
15140 Remove alternatives 15, 16, 17 and 18.
15141 * config/i386/sse.md (vec_concatv2di): Change
15142 alternative (!x, *y) to (x, ?!*Yn).
15144 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15146 * dumpfile.h (enum dump_kind): Remove stray comma.
15148 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15150 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15151 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15152 predicate::num_conditions
15153 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15154 (CHANGED): turn into predicate::changed.
15155 (agg_position_info): Move to ipa-predicate.h
15156 (add_condition, predicate::add_clause, predicate::operator &=,
15157 predicate::or_with, predicate::evaluate, predicate::probability,
15158 dump_condition, dump_clause, predicate::dump,
15159 predicate::remap_after_duplication, predicate::remap_after_inlining,
15160 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15161 (evaluate_conditions_for_known_args): Update.
15162 (set_cond_stmt_execution_predicate): Update.
15163 * ipa-inline.h: Include ipa-predicate.h
15164 (condition, inline_param_summary, conditions, agg_position_info,
15165 predicate): Move to ipa-predicate.h
15166 * ipa-predicate.c: New file.
15167 * ipa-predicate.h: New file.
15169 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15171 * final.c (leaf_function_p): Check we are not in a sequence.
15173 2017-05-18 Martin Liska <mliska@suse.cz>
15175 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15176 * dumpfile.c (dump_register): Use new enum dump_kind.
15177 (get_dump_file_name): Likewise.
15178 (dump_enable_all): Likewise.
15179 (dump_switch_p_1): Likewise.
15180 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15181 * dumpfile.h (enum dump_kind): New enum type.
15182 (struct dump_file_info): Create constructor and
15183 format fields and comments.
15184 * passes.c (pass_manager::register_one_dump_file):
15186 * statistics.c (statistics_early_init): Likewise.
15187 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15188 TDF_TREE with TDF_SLIM.
15189 (gather_memory_references_ref): Likewise.
15191 2017-05-18 Martin Liska <mliska@suse.cz>
15193 * vec.h (struct vnull): Use it.
15195 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15197 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15198 (true_predicate, false_predicate, true_predicate_p,
15199 false_predicate_p): Remove.
15200 (single_cond_predicate, not_inlined_predicate): Turn to member function
15202 (add_condition): Update.
15203 (add_clause): Turn to...
15204 (predicate::add_clause): ... this one; update; allow passing NULL
15206 (and_predicates): Turn to ...
15207 (predicate::operator &=): ... this one.
15208 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15209 (or_predicates): Turn to ...
15210 (predicate::or_with): ... this one.
15211 (evaluate_predicate): Turn to ...
15212 (predicate::evaluate): ... this one.
15213 (predicate_probability): Turn to ...
15214 (predicate::probability): ... this one.
15215 (dump_condition): Update.
15216 (dump_predicate): Turn to ...
15217 (predicate::dump): ... this one.
15218 (account_size_time): Update.
15219 (edge_set_predicate): Update.
15220 (set_hint_predicate): UPdate.
15221 (evaluate_conditions_for_known_args): Update.
15222 (evaluate_properties_for_edge): Update.
15223 (remap_predicate_after_duplication): Turn to...
15224 (predicate::remap_after_duplication): ... this one.
15225 (remap_hint_predicate_after_duplication): Update.
15226 (inline_summary_t::duplicate): UPdate.
15227 (dump_inline_edge_summary): Update.
15228 (dump_inline_summary): Update.
15229 (set_cond_stmt_execution_predicate): Update.
15230 (set_switch_stmt_execution_predicate): Update.
15231 (compute_bb_predicates): Update.
15232 (will_be_nonconstant_expr_predicate): Update.
15233 (will_be_nonconstant_predicate): Update.
15234 (phi_result_unknown_predicate): Update.
15235 (predicate_for_phi_result): Update.
15236 (array_index_predicate): Update.
15237 (estimate_function_body_sizes): Update.
15238 (estimate_node_size_and_time): Update.
15239 (estimate_ipcp_clone_size_and_time): Update.
15240 (remap_predicate): Rename to ...
15241 (predicate::remap_after_inlining): ... this one.
15242 (remap_hint_predicate): Update.
15243 (inline_merge_summary): Update.
15244 (inline_update_overall_summary): Update.
15245 (estimate_size_after_inlining): Update.
15246 (read_predicate): Rename to ...
15247 (predicate::stream_in): ... this one.
15248 (read_inline_edge_summary): Update.
15249 (write_predicate): Rename to ...
15250 (predicate::stream_out): ... this one.
15251 (write_inline_edge_summary): Update.
15252 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15253 (clause_t): Turn to uint32_t
15254 (predicate): Turn to class; implement constructor and operators
15256 (size_time_entry): Update.
15257 (inline_summary): Update.
15258 (inline_edge_summary): Update.
15260 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15262 * fold-const.c (fold_binary_loc): Move transformation...
15263 * match.pd (C - X CMP X): ... here.
15265 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15267 * config/sparc/sparc.c (sparc_option_override): Set function
15268 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15269 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15271 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15273 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15275 2017-05-18 Marek Polacek <polacek@redhat.com>
15278 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15279 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15280 gimple_assign_load_p.
15282 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15284 PR middle-end/80692
15285 * real.c (do_compare): Give decimal_do_compare preference over
15286 comparing just the signs.
15288 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15290 * doc/md.texi (Canonicalization of Instructions): Describe the
15291 canonical form of instructions that inherently set a condition
15294 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15296 PR middle-end/80775
15297 * tree-cfg.c: Move deletion of unreachable case statements to after
15298 the merging of consecutive case labels.
15300 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15302 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15303 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15304 restoring of callee-saved registers.
15306 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15308 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15309 * config/visium/visium.c (single_set_and_flags): Likewise.
15310 * config/visium/visium.md (Substitutions): Likewise.
15312 2017-05-17 Martin Liska <mliska@suse.cz>
15314 * cfg.c: Introduce dump_flags_t type and
15315 use it instead of int type.
15317 * cfghooks.c: Likewise.
15318 * cfghooks.h (struct cfg_hooks): Likewise.
15319 * cfgrtl.c: Likewise.
15320 * cfgrtl.h: Likewise.
15321 * cgraph.c (cgraph_node::get_body): Likewise.
15322 * coretypes.h: Likewise.
15323 * domwalk.c: Likewise.
15324 * domwalk.h: Likewise.
15325 * dumpfile.c (struct dump_option_value_info): Likewise.
15326 (dump_enable_all): Likewise.
15327 (dump_switch_p_1): Likewise.
15328 (opt_info_switch_p): Likewise.
15329 * dumpfile.h (enum tree_dump_index): Likewise.
15330 (struct dump_file_info): Likewise.
15331 * genemit.c: Likewise.
15332 * generic-match-head.c: Likewise.
15333 * gengtype.c (open_base_files): Likewise.
15334 * gimple-pretty-print.c: Likewise.
15335 * gimple-pretty-print.h: Likewise.
15336 * graph.c (print_graph_cfg): Likewise.
15337 * graphite-scop-detection.c (dot_all_sese): Likewise.
15338 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15339 * loop-unroll.c (report_unroll): Likewise.
15340 * passes.c (pass_manager::register_one_dump_file): Likewise.
15341 * print-tree.c: Likewise.
15342 * statistics.c: Likewise.
15343 * tree-cfg.c: Likewise.
15344 * tree-cfg.h: Likewise.
15345 * tree-dfa.c: Likewise.
15346 * tree-dfa.h: Likewise.
15347 * tree-dump.c (dump_function): Likewise.
15348 * tree-dump.h (struct dump_info): Likewise.
15349 * tree-pretty-print.c: Likewise.
15350 * tree-pretty-print.h: Likewise.
15351 * tree-ssa-live.c: Likewise.
15352 * tree-ssa-live.h: Likewise.
15353 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15354 * tree-vect-loop.c: Likewise.
15355 * tree-vect-slp.c: Likewise.
15357 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15358 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15360 PR tree-optimization/80457
15361 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15362 of all arguments to a statement as scalar_to_vec operations.
15363 (vectorizable_call): Adjust call to vect_model_simple_cost for
15365 (vectorizable_conversion): Likewise.
15366 (vectorizable_assignment): Likewise.
15367 (vectorizable_shift): Likewise.
15368 (vectorizable_operation): Likewise.
15369 (vectorizable_comparison): Likewise.
15370 (vect_is_simple_cond): Record the def types for operands.
15371 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15372 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15373 for statement argument count.
15375 2017-05-16 Carl Love <cel@us.ibm.com>
15377 * config/rs6000/rs6000-c: Add support for built-in functions
15378 vector unsigned long long vec_bperm (vector unsigned long long,
15379 vector unsigned char)
15380 vector signed long long vec_mule (vector signed int,
15382 vector unsigned long long vec_mule (vector unsigned int,
15383 vector unsigned int)
15384 vector signed long long vec_mulo (vector signed int,
15386 vector unsigned long long vec_mulo (vector unsigned int,
15387 vector unsigned int)
15388 vector signed char vec_sldw (vector signed char,
15389 vector signed char,
15391 vector unsigned char vec_sldw (vector unsigned char,
15392 vector unsigned char,
15394 vector signed short vec_sldw (vector signed short,
15395 vector signed short,
15397 vector unsigned short vec_sldw (vector unsigned short,
15398 vector unsigned short,
15400 vector signed int vec_sldw (vector signed int,
15403 vector unsigned int vec_sldw (vector unsigned int,
15404 vector unsigned int,
15406 vector signed long long vec_sldw (vector signed long long,
15407 vector signed long long,
15409 vector unsigned long long vec_sldw (vector unsigned long long,
15410 vector unsigned long long,
15412 * config/rs6000/rs6000-c: Add support for built-in functions
15413 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15414 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15415 * doc/extend.texi: Update the built-in documentation for the
15416 new built-in functions.
15418 2017-05-16 Marek Polacek <polacek@redhat.com>
15422 * tree.c (save_expr): Don't fold the expression.
15424 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15426 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15427 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15428 and (?*y,m). Update insn attributes.
15430 2017-05-16 Martin Liska <mliska@suse.cz>
15432 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15433 flags argument of print_gimple_stmt, print_gimple_expr,
15434 print_generic_stmt and print_generic_expr.
15435 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15436 * coretypes.h: Likewise.
15437 * except.c (dump_eh_tree): Likewise.
15438 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15439 * gimple-pretty-print.h: Likewise.
15440 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15441 (backprop::push_to_worklist): Likewise.
15442 (backprop::pop_from_worklist): Likewise.
15443 (backprop::process_use): Likewise.
15444 (backprop::intersect_uses): Likewise.
15445 (note_replacement): Likewise.
15446 * gimple-ssa-store-merging.c
15447 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15448 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15449 (pass_store_merging::execute): Likewise.
15450 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15451 (ssa_base_cand_dump_callback): Likewise.
15452 (dump_incr_vec): Likewise.
15453 (replace_refs): Likewise.
15454 (replace_mult_candidate): Likewise.
15455 (create_add_on_incoming_edge): Likewise.
15456 (create_phi_basis): Likewise.
15457 (insert_initializers): Likewise.
15458 (all_phi_incrs_profitable): Likewise.
15459 (introduce_cast_before_cand): Likewise.
15460 (replace_one_candidate): Likewise.
15461 * gimplify.c (gimplify_expr): Likewise.
15462 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15463 (set_rename): Likewise.
15464 (rename_uses): Likewise.
15465 (copy_loop_phi_nodes): Likewise.
15466 (add_close_phis_to_merge_points): Likewise.
15467 (copy_loop_close_phi_args): Likewise.
15468 (copy_cond_phi_args): Likewise.
15469 (graphite_copy_stmts_from_block): Likewise.
15470 (translate_pending_phi_nodes): Likewise.
15471 * graphite-poly.c (print_pdr): Likewise.
15472 (dump_gbb_cases): Likewise.
15473 (dump_gbb_conditions): Likewise.
15474 (print_scop_params): Likewise.
15475 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15476 (build_cross_bb_scalars_use): Likewise.
15477 (gather_bbs::before_dom_children): Likewise.
15478 * hsa-dump.c (dump_hsa_immed): Likewise.
15479 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15480 (get_replacement_map): Likewise.
15481 * ipa-inline-analysis.c (dump_condition): Likewise.
15482 (estimate_function_body_sizes): Likewise.
15483 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15484 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15485 * ipa-prop.c (ipa_dump_param): Likewise.
15486 (ipa_print_node_jump_functions_for_edge): Likewise.
15487 (ipa_modify_call_arguments): Likewise.
15488 (ipa_modify_expr): Likewise.
15489 (ipa_dump_param_adjustments): Likewise.
15490 (ipa_dump_agg_replacement_values): Likewise.
15491 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15492 * ipa-pure-const.c (check_stmt): Likewise.
15493 (pass_nothrow::execute): Likewise.
15494 * ipa-split.c (execute_split_functions): Likewise.
15495 * omp-offload.c (dump_oacc_loop_part): Likewise.
15496 (dump_oacc_loop): Likewise.
15497 * trans-mem.c (tm_log_emit): Likewise.
15498 (tm_memopt_accumulate_memops): Likewise.
15499 (dump_tm_memopt_set): Likewise.
15500 (dump_tm_memopt_transform): Likewise.
15501 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15502 (print_loop): Likewise.
15503 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15504 (chkp_gather_checks_info): Likewise.
15505 (chkp_get_check_result): Likewise.
15506 (chkp_remove_check_if_pass): Likewise.
15507 (chkp_use_outer_bounds_if_possible): Likewise.
15508 (chkp_reduce_bounds_lifetime): Likewise.
15509 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15510 (chkp_mark_completed_bounds): Likewise.
15511 (chkp_register_incomplete_bounds): Likewise.
15512 (chkp_mark_invalid_bounds): Likewise.
15513 (chkp_maybe_copy_and_register_bounds): Likewise.
15514 (chkp_build_returned_bound): Likewise.
15515 (chkp_get_bound_for_parm): Likewise.
15516 (chkp_build_bndldx): Likewise.
15517 (chkp_get_bounds_by_definition): Likewise.
15518 (chkp_generate_extern_var_bounds): Likewise.
15519 (chkp_get_bounds_for_decl_addr): Likewise.
15520 * tree-chrec.c (chrec_apply): Likewise.
15521 * tree-data-ref.c (dump_data_reference): Likewise.
15522 (dump_subscript): Likewise.
15523 (dump_data_dependence_relation): Likewise.
15524 (analyze_overlapping_iterations): Likewise.
15525 * tree-inline.c (expand_call_inline): Likewise.
15526 (tree_function_versioning): Likewise.
15527 * tree-into-ssa.c (dump_defs_stack): Likewise.
15528 (dump_currdefs): Likewise.
15529 (dump_names_replaced_by): Likewise.
15530 (dump_update_ssa): Likewise.
15531 (update_ssa): Likewise.
15532 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15533 * tree-parloops.c (build_new_reduction): Likewise.
15534 (try_create_reduction_list): Likewise.
15535 (ref_conflicts_with_region): Likewise.
15536 (oacc_entry_exit_ok_1): Likewise.
15537 (oacc_entry_exit_single_gang): Likewise.
15538 * tree-pretty-print.h: Likewise.
15539 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15540 (get_scalar_evolution): Likewise.
15541 (add_to_evolution): Likewise.
15542 (get_loop_exit_condition): Likewise.
15543 (analyze_evolution_in_loop): Likewise.
15544 (analyze_initial_condition): Likewise.
15545 (analyze_scalar_evolution): Likewise.
15546 (instantiate_scev): Likewise.
15547 (number_of_latch_executions): Likewise.
15548 (gather_chrec_stats): Likewise.
15549 (final_value_replacement_loop): Likewise.
15550 (scev_const_prop): Likewise.
15551 * tree-sra.c (dump_access): Likewise.
15552 (disqualify_candidate): Likewise.
15553 (create_access): Likewise.
15554 (reject): Likewise.
15555 (maybe_add_sra_candidate): Likewise.
15556 (create_access_replacement): Likewise.
15557 (analyze_access_subtree): Likewise.
15558 (analyze_all_variable_accesses): Likewise.
15559 (sra_modify_assign): Likewise.
15560 (initialize_constant_pool_replacements): Likewise.
15561 (find_param_candidates): Likewise.
15562 (decide_one_param_reduction): Likewise.
15563 (replace_removed_params_ssa_names): Likewise.
15564 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15565 * tree-ssa-copy.c (dump_copy_of): Likewise.
15566 (copy_prop_visit_cond_stmt): Likewise.
15567 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
15568 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
15569 (record_equivalences_from_stmt): Likewise.
15570 * tree-ssa-dse.c (compute_trims): Likewise.
15571 (delete_dead_call): Likewise.
15572 (delete_dead_assignment): Likewise.
15573 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
15574 (forward_propagate_into_cond): Likewise.
15575 (pass_forwprop::execute): Likewise.
15576 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15577 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
15579 (move_computations_worker): Likewise.
15580 (execute_sm): Likewise.
15581 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
15582 (remove_exits_and_undefined_stmts): Likewise.
15583 (remove_redundant_iv_tests): Likewise.
15584 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
15585 (adjust_iv_update_pos): Likewise.
15586 * tree-ssa-math-opts.c (bswap_replace): Likewise.
15587 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
15588 (value_replacement): Likewise.
15589 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
15590 * tree-ssa-pre.c (print_pre_expr): Likewise.
15591 (get_representative_for): Likewise.
15592 (create_expression_by_pieces): Likewise.
15593 (insert_into_preds_of_block): Likewise.
15594 (eliminate_insert): Likewise.
15595 (eliminate_dom_walker::before_dom_children): Likewise.
15596 (eliminate): Likewise.
15597 (remove_dead_inserted_code): Likewise.
15598 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
15599 * tree-ssa-reassoc.c (get_rank): Likewise.
15600 (eliminate_duplicate_pair): Likewise.
15601 (eliminate_plus_minus_pair): Likewise.
15602 (eliminate_not_pairs): Likewise.
15603 (undistribute_ops_list): Likewise.
15604 (eliminate_redundant_comparison): Likewise.
15605 (update_range_test): Likewise.
15606 (optimize_range_tests_var_bound): Likewise.
15607 (optimize_vec_cond_expr): Likewise.
15608 (rewrite_expr_tree): Likewise.
15609 (rewrite_expr_tree_parallel): Likewise.
15610 (linearize_expr): Likewise.
15611 (break_up_subtract): Likewise.
15612 (linearize_expr_tree): Likewise.
15613 (attempt_builtin_powi): Likewise.
15614 (attempt_builtin_copysign): Likewise.
15615 (transform_stmt_to_copy): Likewise.
15616 (transform_stmt_to_multiply): Likewise.
15617 (dump_ops_vector): Likewise.
15618 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15619 (print_scc): Likewise.
15620 (set_ssa_val_to): Likewise.
15621 (visit_reference_op_store): Likewise.
15622 (visit_use): Likewise.
15623 (sccvn_dom_walker::before_dom_children): Likewise.
15624 (run_scc_vn): Likewise.
15625 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15627 (expr_hash_elt::print): Likewise.
15628 (const_and_copies::pop_to_marker): Likewise.
15629 (const_and_copies::record_const_or_copy_raw): Likewise.
15630 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15631 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15632 (dump_predicates): Likewise.
15633 (find_uninit_use): Likewise.
15634 (warn_uninitialized_phi): Likewise.
15635 (pass_late_warn_uninitialized::execute): Likewise.
15636 * tree-ssa.c (verify_vssa): Likewise.
15637 (verify_ssa): Likewise.
15638 (maybe_optimize_var): Likewise.
15639 * tree-vrp.c (dump_value_range): Likewise.
15640 (dump_all_value_ranges): Likewise.
15641 (dump_asserts_for): Likewise.
15642 (register_edge_assert_for_2): Likewise.
15643 (vrp_visit_cond_stmt): Likewise.
15644 (vrp_visit_switch_stmt): Likewise.
15645 (vrp_visit_stmt): Likewise.
15646 (vrp_visit_phi_node): Likewise.
15647 (simplify_cond_using_ranges_1): Likewise.
15648 (fold_predicate_in): Likewise.
15649 (evrp_dom_walker::before_dom_children): Likewise.
15650 (evrp_dom_walker::push_value_range): Likewise.
15651 (evrp_dom_walker::pop_value_range): Likewise.
15652 (execute_early_vrp): Likewise.
15654 2017-05-16 Richard Biener <rguenther@suse.de>
15656 * dwarf2out.c (loc_list_from_tree_1): Do not create
15657 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15659 2017-05-16 Richard Biener <rguenther@suse.de>
15661 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15663 (note_variable_value_in_expr): If we resolved the decl ref
15664 do not push to the stack.
15666 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15668 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15669 operations in fast-math mode.
15670 (vaddq_f16): Likewise.
15671 (vmul_f16): Likewise.
15672 (vmulq_f16): Likewise.
15673 (vsub_f16): Likewise.
15674 (vsubq_f16): Likewise.
15675 * config/arm/neon.md (add<mode>3): New.
15677 (fma:<VH:mode>3): New. Also remove outdated comment.
15680 2017-05-16 Martin Liska <mliska@suse.cz>
15684 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15685 (odr_types_equivalent_p): Likewise.
15687 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15689 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15691 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15694 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15695 non-interunit SSE move alternatives with '?'.
15696 (zero-extendsidi peephole2): New peephole to skip intermediate
15697 general register in SSE zero-extend sequence.
15699 2017-05-15 Jeff Law <law@redhat.com>
15701 * reorg.c (relax_delay_slots): Create a new variable to hold
15702 the temporary target rather than clobbering TARGET_LABEL.
15704 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15705 missing argument to extract_bit_field call.
15706 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15708 2017-05-15 Martin Liska <mliska@suse.cz>
15711 * gcc.c (process_command): Do not allow empty argument of -o option.
15713 2017-05-15 Renlin Li <renlin.li@arm.com>
15715 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15716 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15717 * config/aarch64/constraints.md (Usf): Add long call check.
15718 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15719 (call_value): Likewise.
15720 (sibcall): Likewise.
15721 (sibcall_value): Likewise.
15723 (call_value_insn): New.
15724 (sibcall_insn): Update rtx pattern.
15725 (sibcall_value_insn): Likewise.
15726 (call_internal): Remove.
15727 (call_value_internal): Likewise.
15728 (sibcall_internal): Likewise.
15729 (sibcall_value_internal): Likewise.
15730 (call_reg): Likewise.
15731 (call_symbol): Likewise.
15732 (call_value_reg): Likewise.
15733 (call_value_symbol): Likewise.
15735 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15738 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15740 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15742 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15743 compatible with CCGOCmode and with CCZmode.
15745 2017-05-14 Martin Sebor <msebor@redhat.com>
15747 PR middle-end/77671
15748 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15749 (gimple_fold_builtin_snprintf): Same.
15750 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15751 (gimple_fold_builtin_snprintf): Same.
15752 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15753 of character types.
15754 (is_call_safe): New function.
15755 (try_substitute_return_value): Call it.
15756 (try_simplify_call): New function.
15757 (pass_sprintf_length::handle_gimple_call): Call it.
15759 2017-05-14 Martin Sebor <msebor@redhat.com>
15761 PR middle-end/80669
15762 * builtins.c (expand_builtin_stpncpy): Simplify.
15764 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15766 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15767 * config/i386/i386.h
15768 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15769 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15770 (struct machine_function): Add new members call_ms2sysv,
15771 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15772 (struct machine_frame_state): New fields sp_realigned and
15773 sp_realigned_offset.
15774 * config/i386/i386.c
15775 (enum xlogue_stub): New enum.
15776 (enum xlogue_stub_sets): New enum.
15777 (class xlogue_layout): New class.
15778 (struct ix86_frame): New fields stack_realign_allocate_offset,
15779 stack_realign_offset and outlined_save_offset. Modify comments to
15780 detail stack layout when using out-of-line stubs.
15781 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15782 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15783 -mcall-ms2sysv-xlogues.
15784 (stub_managed_regs): New static variable.
15785 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15786 registers managed by out-of-line stub.
15787 (disable_call_ms2sysv_xlogues): New function.
15788 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15789 m->call_ms2sysv when appropriate and compute frame layout for
15791 (sp_valid_at, fp_valid_at): New inline functions.
15792 (choose_basereg): New function.
15793 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15795 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15796 Use align parameter of choose_baseaddr to generated aligned SSE movs
15798 (pro_epilogue_adjust_stack): Modify to track
15799 machine_frame_state::sp_realigned.
15800 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15801 (ix86_nsaved_sseregs): Likewise.
15802 (ix86_emit_save_regs): Likewise.
15803 (ix86_emit_save_regs_using_mov): Likewise.
15804 (ix86_emit_save_sse_regs_using_mov): Likewise.
15805 (get_scratch_register_on_entry): Likewise.
15806 (gen_frame_set): New function.
15807 (gen_frame_load): Likewise.
15808 (gen_frame_store): Likewise.
15809 (emit_outlined_ms2sysv_save): Likewise.
15810 (emit_outlined_ms2sysv_restore): Likewise.
15811 (ix86_expand_prologue): Modify stack re-alignment code and call
15812 emit_outlined_ms2sysv_save when appropriate.
15813 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15814 parameter rtx_insn *insn, which allows the function to be used to only
15815 generate the notes.
15816 (ix86_expand_epilogue): Modify validity checks of frame and stack
15817 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15818 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15819 * config/i386/predicates.md
15820 (save_multiple): New predicate.
15821 (restore_multiple): Likewise.
15822 * config/i386/sse.md
15823 (save_multiple<mode>): New pattern.
15824 (save_multiple_realign<mode>): Likewise.
15825 (restore_multiple<mode>): Likewise.
15826 (restore_multiple_and_return<mode>): Likewise.
15827 (restore_multiple_leave_return<mode>): Likewise.
15828 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15830 2017-05-14 Julia Koval <julia.koval@intel.com>
15832 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15833 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15834 (__builtin_ia32_xsetbv): New builtins.
15835 * config/i386/i386.c (ix86_expand_special_args_builtin):
15837 (ix86_expand_builtin): Special expand for new intrinsics.
15838 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15839 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15840 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15842 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15844 * cfganal.c (inverted_post_order_compute): Change argument type
15846 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15847 * df-core.c (rest_of_handle_df_initialize): Adjust.
15848 (rest_of_handle_df_finish): Likewise.
15849 (df_analyze_1): Likewise.
15850 (df_analyze): Likewise.
15851 (loop_inverted_post_order_compute): Change argument to be a vec *.
15852 (df_analyze_loop): Adjust.
15853 (df_get_n_blocks): Likewise.
15854 (df_get_postorder): Likewise.
15855 * df.h (struct df_d): Change field to be a vec.
15856 * lcm.c (compute_laterin): Adjust.
15857 (compute_available): Likewise.
15858 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15859 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15860 * tree-ssa-pre.c (compute_antic): Likewise.
15862 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15864 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15865 (depth_first_search::depth_first_search): Change structure init
15866 function to this constructor.
15867 (depth_first_search::add_bb): Rename function to this member.
15868 (depth_first_search::execute): Likewise.
15869 (flow_dfs_compute_reverse_finish): Adjust.
15871 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15873 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15874 (longest_simple_path): Likewise.
15875 * shrink-wrap.c (spread_components): Likewise.
15876 (disqualify_problematic_components): Likewise.
15877 (emit_common_heads_for_components): Likewise.
15878 (emit_common_tails_for_components): Likewise.
15879 (insert_prologue_epilogue_for_components): Likewise.
15881 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15883 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15886 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15888 * df-core.c (df_set_blocks): Start using auto_bitmap.
15889 (df_compact_blocks): Likewise.
15890 * df-problems.c (df_rd_confluence_n): Likewise.
15891 * df-scan.c (df_insn_rescan_all): Likewise.
15892 (df_process_deferred_rescans): Likewise.
15893 (df_update_entry_block_defs): Likewise.
15894 (df_update_exit_block_uses): Likewise.
15895 (df_entry_block_bitmap_verify): Likewise.
15896 (df_exit_block_bitmap_verify): Likewise.
15897 (df_scan_verify): Likewise.
15898 * lra-constraints.c (lra_constraints): Likewise.
15899 (undo_optional_reloads): Likewise.
15900 (lra_undo_inheritance): Likewise.
15901 * lra-remat.c (calculate_gen_cands): Likewise.
15902 (do_remat): Likewise.
15903 * lra-spills.c (assign_spill_hard_regs): Likewise.
15904 (spill_pseudos): Likewise.
15905 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15906 (bitmap_set_subtract_values): Likewise.
15908 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15910 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15911 management with auto_bitmap.
15912 (fix_inter_tick): Likewise.
15913 (fix_recovery_deps): Likewise.
15914 * ira.c (add_store_equivs): Likewise.
15915 (find_moveable_pseudos): Likewise.
15916 (split_live_ranges_for_shrink_wrap): Likewise.
15917 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15918 (rtx_reuse_manager::seen_def_p): Likewise.
15919 (rtx_reuse_manager::set_seen_def): Likewise.
15920 * print-rtl.h (class rtx_reuse_manager): Likewise.
15922 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15924 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15926 (migrate_btr_def): Likewise.
15927 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15928 * df-core.c (loop_post_order_compute): Likewise.
15929 (loop_inverted_post_order_compute): Likewise.
15930 * hsa-common.h: Likewise.
15931 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15932 * init-regs.c (initialize_uninitialized_regs): Likewise.
15933 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15934 (inline_small_functions): Likewise.
15935 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15936 * ira.c (combine_and_move_insns): Likewise.
15937 (build_insn_chain): Likewise.
15938 * loop-invariant.c (find_invariants): Likewise.
15939 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15940 * predict.c (tree_predict_by_opcode): Likewise.
15941 (predict_paths_leading_to): Likewise.
15942 (predict_paths_leading_to_edge): Likewise.
15943 (estimate_loops_at_level): Likewise.
15944 (estimate_loops): Likewise.
15945 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15946 (spread_components): Likewise.
15947 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15948 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15949 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15950 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15951 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15952 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15953 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15954 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15955 (mark_threaded_blocks): Likewise.
15956 (thread_through_all_blocks): Likewise.
15957 * tree-ssa.c (verify_ssa): Likewise.
15958 (execute_update_addresses_taken): Likewise.
15959 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15961 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15963 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15965 (post_order_compute): Likewise.
15966 (inverted_post_order_compute): Likewise.
15967 (pre_and_rev_post_order_compute_fn): Likewise.
15969 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15971 * genrecog.c (int_set::int_set): Explicitly construct our
15972 auto_vec base class.
15973 * vec.h (auto_vec::auto_vec): New constructor.
15975 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15977 * bitmap.h (class auto_bitmap): New constructor taking
15978 bitmap_obstack * argument.
15980 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15982 * bitmap.h (class auto_bitmap): Change type of m_bits to
15983 bitmap_head, and adjust ctor / dtor and member operators.
15985 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15987 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15988 when returned register mode doesn't match original mode.
15990 2017-05-12 Jeff Law <law@redhat.com>
15991 Jakub Jelinek <jakub@redhat.com>
15993 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15994 we look for cc setter after the compare-elim changes.
15995 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15996 within the vector to match what compare-elim now expects.
15997 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15998 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16000 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16001 after the compare-elim changes.
16002 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16003 the vector to match what compare-elim now expects.
16004 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16005 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16006 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16007 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16008 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16010 * config/visium/visium.c (single_set_and_flags): Fix where
16011 we look for cc setter after the compare-elim changes.
16012 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16013 with the vector to match what compare-elim now expects.
16014 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16015 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16016 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16017 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16018 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16019 (neg<mode>2_insn_set_overflow): Likewise.
16021 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16023 PR middle-end/79794
16024 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16025 maybe_expand_insn call, set ops[0].target. If still set after call,
16026 set alt_rtl. Add extra arg to recursive calls.
16027 (extract_bit_field): Add alt_rtl argument. Pass to
16029 * expmed.h (extract_bit_field): Fix prototype.
16030 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16031 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16032 to extract_bit_field_calls.
16033 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16034 Pass alt_rtl to extract_bit_field calls.
16035 * calls.c (store_unaligned_arguments_into_psuedos)
16036 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16037 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16039 * optabs.h (struct expand_operand): Add target bitfield.
16041 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16043 * compare-elim.c (try_eliminate_compare): Canonicalize
16044 operation with embedded compare to
16045 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16046 (set (reg) (operation)].
16048 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16050 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16053 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16054 cost of adding a carry flag for ADC instruction.
16055 [case MINUS]: Ignore the cost of subtracting a carry flag
16056 for SBB instruction.
16058 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16060 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16062 * config/rs6000/bmiintrin.h: New file.
16063 * config/rs6000/bmi2intrin.h: New file.
16064 * config/rs6000/x86intrin.h: New file.
16066 2017-05-12 Jeff Law <law@redhat.com>
16068 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16071 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16073 PR middle-end/80707
16074 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16076 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16079 early expansion of vector divide builtins.
16080 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16081 builtins identified as having unsigned arguments.
16083 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16085 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16086 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16087 expansion of vector logical operations (and, andc, or, xor,
16090 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16092 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16093 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16095 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16097 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16098 early GIMPLE expansion of vector multiplies.
16100 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16102 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16103 TARGET_HAVE_MOVT conditional.
16104 (movt splitter): Likewise.
16106 2017-05-12 Richard Biener <rguenther@suse.de>
16108 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16109 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16110 Fold all stmts not inplace.
16112 2017-05-12 Richard Biener <rguenther@suse.de>
16114 PR tree-optimization/80713
16115 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16116 inserted_exprs bit for not removed stmts.
16118 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16120 PR middle-end/69921
16121 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16122 parallelized" attribute for parallelized OpenACC kernels.
16123 * omp-offload.c (execute_oacc_device_lower): Use it.
16125 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16126 Set "oacc kernels" attribute.
16127 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16128 parameter. Adjust all users.
16129 (oacc_fn_attrib_kernels_p): Remove function.
16130 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16131 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16132 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16133 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16134 assert "oacc kernels" attribute is set.
16136 2017-05-11 Carl Love <cel@us.ibm.com>
16138 * config/rs6000/rs6000-c: Add support for built-in functions
16139 vector unsigned char vec_popcnt (vector signed char)
16140 vector unsigned char vec_popcnt (vector unsigned char)
16141 vector unsigned short vec_popcnt (vector signed short)
16142 vector unsigned short vec_popcnt (vector unsigned short)
16143 vector unsigned int vec_popcnt (vector signed int)
16144 vector unsigned int vec_popcnt (vector unsigned int)
16145 vector unsigned long long vec_popcnt (vector signed long long)
16146 vector unsigned long long vec_popcnt (vector unsigned long long)
16147 vector signed long long vec_slo (vector signed long long,
16148 vector signed char)
16149 vector signed long long vec_slo (vector signed long long,
16150 vector unsigned char)
16151 vector unsigned long long vec_slo (vector unsigned long long,
16152 vector signed char)
16153 vector unsigned long long vec_slo (vector unsigned long long,
16154 vector unsigned char)
16155 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16156 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16157 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16158 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16159 * doc/extend.texi: Update the built-in documentation file for the
16160 new built-in functions.
16162 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16164 * attribs.h (sorted_attr_string): Move machine independent
16165 functions for target clone support from the i386 port to common
16166 code. Rename ix86_function_versions to common_function_versions.
16167 Rename make_name to make_unique_name.
16168 (common_function_versions): Likewise.
16169 (make_unique_name): Likewise.
16170 (make_dispatcher_decl): Likewise.
16171 (is_function_default_version): Likewise.
16172 * attribs.c (attr_strcmp): Likewise.
16173 (sorted_attr_string): Likewise.
16174 (common_function_versions): Likewise.
16175 (make_unique_name): Likewise.
16176 (make_dispatcher_decl): Likewise.
16177 (is_function_default_version): Likewise.
16178 * config/i386/i386.c (attr_strcmp): Likewise.
16179 (sorted_attr_string): Likewise.
16180 (ix86_function_versions): Likewise.
16181 (make_name): Likewise.
16182 (make_dispatcher_decl): Likewise.
16183 (is_function_default_version): Likewise.
16184 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16186 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16189 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16190 Account for direct move costs for vec_construct of integer
16193 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16196 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16197 (UNSPEC_STX_ATOMIC): Ditto.
16198 (loaddi_via_sse): New insn.
16199 (storedi_via_sse): Ditto.
16200 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16201 Update corresponding peephole2 patterns.
16202 (atomic_storedi_fpu): Ditto.
16204 2017-05-11 Julia Koval <julia.koval@intel.com>
16206 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16207 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16209 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16210 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16211 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16213 2017-05-11 Nathan Sidwell <nathan@acm.org>
16215 * graphite-poly.c: Include dumpfile.h.
16217 * dumpfle.h (dump_function): Declare here ...
16218 * tree-dump.h (dump_function): ... not here.
16219 * dumpfile.c: #include tree-cfg.h.
16220 (dump_function): Move here from ...
16221 * tree-dump.c (dump_function): ... here.
16222 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16223 * graphite-poly.c: Don't include tree-dump.h.
16224 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16225 * print-tree.c: Likewise.
16226 * stor-layout.c: Likewise.
16227 * tree-nested.c: Likewise.
16229 * dumpfile.c (dump_start): Use TDF_FLAGS.
16230 (dump_enable_all): Fix TDF_KIND check thinko.
16232 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16234 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16235 array entries to represent two legal parameterizations of the
16236 overloaded __builtin_cmpb function, as represented by the
16237 P6_OV_BUILTIN_CMPB constant.
16238 (altivec_resolve_overloaded_builtin): Add special case handling
16239 for the __builtin_cmpb function, as represented by the
16240 P6_OV_BUILTIN_CMPB constant.
16241 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16242 (BU_P6_64BIT_2): New macro.
16243 (BU_P6_OVERLOAD_2): New macro
16244 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16245 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16246 (CMPB): Add overload support to represent both 32-bit and 64-bit
16247 compare-bytes function.
16248 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16249 support for TARGET_CMPB.
16250 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16251 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16252 documentation of the __builtin_cmpb overloaded built-in function.
16254 2017-05-11 Richard Biener <rguenther@suse.de>
16256 PR tree-optimization/80705
16257 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16258 bases are not vectorizable.
16260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16262 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16263 when counting register pressure.
16265 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16267 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16268 (iv_ca_more_deps): Renamed to ...
16269 (iv_ca_compare_deps): ... this.
16270 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16272 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16274 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16276 (determine_group_iv_costs): ... here.
16277 (find_inv_vars_cb): Record inv var if it's not recorded before.
16279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16281 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16282 (get_shiftadd_cost): Ditto.
16284 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16286 * tree-ssa-address.c: Include header file.
16287 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16289 (add_to_parts): Refactor.
16290 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16291 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16294 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16296 PR tree-optimization/53090
16297 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16299 (extract_cond_operands): Detect condition with IV on both sides
16300 and return COMP_IV_EXPR_2.
16301 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16302 (rewrite_use_compare): Simplify by removing call to function
16303 extract_cond_operands.
16305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16307 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16308 (extract_cond_operands): Detect condition comparing against non-
16309 invariant bound and return appropriate enum value.
16310 (find_interesting_uses_cond): Update use of extract_cond_operands.
16311 Handle its return value accordingly.
16312 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16316 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16317 nonlinear iv_use computation in loop invariant sensitive way.
16319 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16321 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16322 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16326 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16327 (dump_cand): Support iv_cand.inv_exprs.
16328 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16330 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16333 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16335 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16337 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16338 as local function. Include necessary header files.
16339 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16341 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16343 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16345 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16347 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16348 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16349 RSHIFT_EXPR and BIT_NOT_EXPR.
16351 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16353 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16354 (adjust_setup_cost): New parameter supporting round up adjustment.
16355 (struct address_cost_data): Delete.
16356 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16357 (split_address_cost, ptr_difference_cost): Delete.
16358 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16359 (struct ainc_cost_data): New struct.
16360 (get_address_cost_ainc): New function.
16361 (get_address_cost, get_computation_cost): Reimplement.
16362 (determine_group_iv_cost_address): Record inv_expr for all uses of
16364 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16365 (iv_ca_has_deps): Reimplemented to ...
16366 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16368 (iv_ca_extend): Call iv_ca_more_deps.
16370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16372 * tree-ssa-address.c (struct mem_address): Move to header file.
16373 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16374 * tree-ssa-address.h (struct mem_address): Move from C file.
16375 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16379 * tree-affine.h (aff_combination_type): New interface.
16380 (aff_combination_zero_p): Remove static.
16381 (aff_combination_const_p): New interface.
16382 (aff_combination_singleton_var_p): New interfaces.
16384 2017-05-11 Richard Biener <rguenther@suse.de>
16386 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16387 Skip unreachable blocks and destinations.
16388 (eliminate): Move stmt removal and fixup ...
16389 (fini_eliminate): ... here. Skip inserted exprs.
16390 (pass_pre::execute): Move fini_pre after fini_eliminate.
16391 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16392 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16393 PRE to get rid of dead code that has invalid SSA form and
16394 split critical edges again.
16396 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16398 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16400 2017-05-11 Richard Biener <rguenther@suse.de>
16402 * passes.c (execute_function_todo): Verify loops if they are
16403 said to be up-to-date.
16404 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16405 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16407 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16410 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16411 handle calling assemble_external ourself.
16414 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16415 modes with zero size. Enhance comment.
16417 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16419 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16420 built-ins for vec_xl and vec_xst with short and char pointer
16423 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16425 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16426 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16427 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16428 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16429 (_mm_maskz_min_round_ss): New intrinsics.
16430 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16431 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16432 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16433 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16434 (__builtin_ia32_minss_mask_round): New builtins.
16435 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16436 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16437 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16439 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16440 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16442 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16445 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16447 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16448 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16449 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16450 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16451 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16452 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16453 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16454 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16455 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16456 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16457 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16458 (__builtin_ia32_mulss_mask_round): New builtins.
16459 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16460 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16461 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16463 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16464 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16466 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16469 2017-05-10 Julia Koval <julia.koval@intel.com>
16471 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16472 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16473 (_mm256_setr_m128i): New intrinsics.
16475 2017-05-10 Julia Koval <julia.koval@intel.com>
16477 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16478 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16479 (_mm_maskz_rcp14_ss): New intrinsics.
16480 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16481 (__builtin_ia32_rcp14ss_mask): New builtins.
16482 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16484 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16486 PR tree-optimization/51513
16487 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16488 (assert_unreachable_fallthru_edge_p): Use it.
16489 (group_case_labels_stmt): Likewise.
16490 * tree-cfg.h: Prototype it.
16491 * stmt.c: Include cfghooks.h and tree-cfg.h.
16492 (emit_case_dispatch_table) <gap_label>: New local variable.
16493 Use it to fill dispatch table gaps.
16494 Test for default_label before updating probabilities.
16495 (expand_case) <default_label>: Remove unneeded initialization.
16496 Test for unreachable default case statement and remove its edge.
16497 Set default_label accordingly.
16498 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16500 2017-05-10 Carl Love <cel@us.ibm.com>
16502 * config/rs6000/rs6000-c: Add support for built-in functions
16503 vector signed char vec_neg (vector signed char)
16504 vector signed short int vec_neg (vector short int)
16505 vector signed int vec_neg (vector signed int)
16506 vector signed long long vec_neg (vector signed long long)
16507 vector float vec_neg (vector float)
16508 vector double vec_neg (vector double)
16509 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16511 * config/rs6000/altivec.h: Add define for vec_neg
16512 * doc/extend.texi: Update the built-in documentation for the
16513 new built-in functions.
16515 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16517 PR tree-optimization/77644
16518 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16520 2017-05-10 Nathan Sidwell <nathan@acm.org>
16522 * dumpfile.h (TDI_lang_all): New.
16523 (TDF_KIND): New. Renumber others
16524 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16526 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16528 (get_dump_file_name): Adjust suffix generation.
16529 (dump_enable_all): Use TDF_KIND.
16530 * doc/invoke.texi (-fdump-lang-all): Document.
16532 * dumpfile.h: Tabify.
16534 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16537 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16538 Move member access before delete.
16540 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16542 * tree-inline.c (expand_call_inline): Split block at stmt
16545 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16548 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16549 are now unused after splitting mov{sf,sd}_hardfloat.
16550 (f32_lr2): Likewise.
16551 (f32_lm): Likewise.
16552 (f32_lm2): Likewise.
16553 (f32_li): Likewise.
16554 (f32_li2): Likewise.
16555 (f32_lv): Likewise.
16556 (f32_sr): Likewise.
16557 (f32_sr2): Likewise.
16558 (f32_sm): Likewise.
16559 (f32_sm2): Likewise.
16560 (f32_si): Likewise.
16561 (f32_si2): Likewise.
16562 (f32_sv): Likewise.
16563 (f32_dm): Likewise.
16564 (f32_vsx): Likewise.
16565 (f32_av): Likewise.
16566 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
16567 For movsf, order stores so the VSX stores occur before the GPR
16568 store which encourages the register allocator to use a traditional
16569 FPR instead of a GPR. For movsd, order the stores so that the GPR
16570 store comes before the VSX stores to allow the power6 to work.
16571 This is due to the power6 not having a 32-bit integer store
16572 instruction from a FPR.
16573 (movsf_hardfloat): Likewise.
16574 (movsd_hardfloat): Likewise.
16576 2017-05-09 Martin Sebor <msebor@redhat.com>
16578 PR translation/80280
16579 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
16582 PR translation/80280
16583 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
16584 data member added in r247778.
16585 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
16587 2017-05-09 Nathan Sidwell <nathan@acm.org>
16589 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
16591 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
16594 2017-05-09 Marek Polacek <polacek@redhat.com>
16596 * doc/invoke.texi: Fix typo.
16598 2017-05-09 Richard Biener <rguenther@suse.de>
16600 * tree-vrp.c (vrp_val_is_max): Adjust comment.
16601 (vrp_val_is_min): Likewise.
16602 (set_value_range_to_value): Likewise.
16603 (set_value_range_to_nonnegative): Likewise.
16604 (gimple_assign_nonzero_p): Likewise.
16605 (gimple_stmt_nonzero_p): Likewise.
16606 (vrp_int_const_binop): Likewise. Remove unreachable case.
16607 (adjust_range_with_scev): Adjust comments.
16608 (compare_range_with_value): Likewise.
16609 (extract_range_from_phi_node): Likewise.
16610 (test_for_singularity): Likewise.
16612 2017-05-09 Richard Biener <rguenther@suse.de>
16614 * tree-vrp.c (get_single_symbol): Add assert that we don't
16615 get overflowed constants as invariant part.
16616 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16617 checks. Use wi::cmp instead of recursing for integer constants.
16618 (compare_values): Just ignore whether we assumed undefined
16619 overflow instead of failing the compare.
16620 (extract_range_for_var_from_comparison_expr): Add comment before the
16621 TREE_NO_WARNING sets.
16622 (test_for_singularity): Likewise.
16623 (extract_range_from_comparison): Do not disable optimization
16624 when we assumed undefined overflow.
16625 (extract_range_basic): Remove init of unused var.
16627 2017-05-09 Richard Biener <rguenther@suse.de>
16629 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16630 (extract_range_from_multiplicative_op_1): Adjust.
16631 (extract_range_from_binary_expr_1): Use int_const_binop.
16633 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16636 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16637 rs6000_store_data_bypass_p in seven define_bypass directives and
16638 in several comments.
16639 * config/rs6000/rs6000-protos.h: Add prototype for
16640 rs6000_store_data_bypass_p function.
16641 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16642 function implements slightly different (rs6000-specific) semantics
16643 than store_data_bypass_p, returning false rather than aborting
16644 with assertion error when arguments do not satisfy the
16645 requirements of store data bypass.
16646 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16647 rs6000_store_data_bypass_p.
16649 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16651 * config/xtensa/xtensa-protos.h
16652 (xtensa_initial_elimination_offset): New declaration.
16653 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16654 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16655 macro definition, add case for FRAME_POINTER_REGNUM when
16656 FRAME_GROWS_DOWNWARD.
16657 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16658 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16659 xtensa_initial_elimination_offset.
16661 2017-05-08 Nathan Sidwell <nathan@acm.org>
16663 * doc/invoke.texi: Alphabetize -fdump options.
16665 2017-05-08 Martin Sebor <msebor@redhat.com>
16667 PR translation/80280
16668 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16670 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16672 * target.def (compute_frame_layout): New optional target hook.
16673 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16674 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16675 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16677 * reload1.c (verify_initial_elim_offsets): Likewise.
16678 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16679 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16680 (arm_get_frame_offsets): Split up into this ...
16681 (arm_compute_frame_layout): ... and this function.
16683 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16685 * config/aarch64/constraints.md (Usa): New constraint.
16686 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16688 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16690 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16691 with_multilib_list after it has been checked.
16693 2017-05-08 Richard Biener <rguenther@suse.de>
16695 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16696 (bitmap_set_subtract_values): Likewise.
16698 2017-05-08 Richard Biener <rguenther@suse.de>
16700 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16701 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16703 (gimple_stmt_nonzero_warnv_p): Rename to ...
16704 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16706 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16707 (extract_range_basic): Adjust, do not disable propagation on
16708 strict overflow sensitive simplification.
16709 (vrp_visit_cond_stmt): Likewise.
16711 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16713 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16714 body size unconditionally.
16716 2017-05-07 Jeff Law <law@redhat.com>
16719 2017-05-06 Jeff Law <law@redhat.com>
16720 PR tree-optimization/78496
16721 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16724 PR tree-optimization/78496
16725 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16726 (simplify_stmt_using_ranges): Call it.
16727 (vrp_dom_walker::before_dom_children): Extract equivalences
16728 from an ASSERT_EXPR with an equality comparison against a
16731 2017-05-06 Jeff Law <law@redhat.com>
16733 PR tree-optimization/78496
16734 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16737 PR tree-optimization/78496
16738 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16739 (simplify_stmt_using_ranges): Call it.
16740 (vrp_dom_walker::before_dom_children): Extract equivalences
16741 from an ASSERT_EXPR with an equality comparison against a
16744 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16746 * lra-constraints.c (lra_copy_reg_equiv): New function.
16747 (split_reg): Use it to copy equivalence information from the
16748 original register to the spill register.
16750 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16752 PR rtl-optimization/75964
16753 * simplify-rtx.c (simplify_const_relational_operation): Remove
16754 invalid handling of comparisons of integer ABS.
16756 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16758 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16759 initialize to zero.
16760 (init_regs): Remove declaration.
16761 (function_arg_advance_32): Initialize error_p as boolean variable.
16763 2017-05-05 Nathan Sidwell <nathan@acm.org>
16765 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16766 lines. Use for (;;).
16768 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16770 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16771 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16772 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16773 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16774 VF=2 that require versioning.
16776 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16778 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16783 * diagnostic.h (diagnostic_override_option_index): Convert from
16784 macro to inline function.
16786 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16788 * diagnostic.c (last_module_changed_p): New function.
16789 (set_last_module): New function.
16790 (diagnostic_report_current_module): Convert macro usage to
16791 the above functions.
16792 * diagnostic.h (diagnostic_context::last_module): Strengthen
16793 from const line_map * to const line_map_ordinary *.
16794 (diagnostic_last_module_changed): Delete macro.
16795 (diagnostic_set_last_module): Delete macro.
16797 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16799 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16800 with diagnostic_report_diagnostic.
16801 (diagnostic_n_impl_richloc): Likewise.
16802 * diagnostic.h (report_diagnostic): Delete macro.
16803 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16804 with diagnostic_report_diagnostic.
16805 * substring-locations.c (format_warning_va): Likewise.
16807 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16809 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16810 save/restor of format_spec. Move option-printing code to...
16811 (print_option_information): ...this new function, and
16812 reimplement by simply printing to the pretty_printer,
16813 rather than appending to the format string.
16815 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16817 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16818 handling logic into...
16819 (update_effective_level_from_pragmas): ...this new function.
16821 2017-05-04 Andrew Waterman <andrew@sifive.com>
16823 * config/riscv/riscv.opt (mstrict-align): New option.
16824 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16825 (SLOW_UNALIGNED_ACCESS): Define.
16826 (riscv_slow_unaligned_access): Declare.
16827 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16829 (riscv_slow_unaligned_access): New variable.
16830 (rocket_tune_info): Set slow_unaligned_access to true.
16831 (optimize_size_tune_info): Set slow_unaligned_access to false.
16832 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16833 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16834 (riscv_option_override): Set riscv_slow_unaligned_access.
16835 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16837 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16839 * config/riscv/riscv.md: Unify indentation.
16841 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16846 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16848 (extendsi<mode>2): Add support for doing sign extension via
16849 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16850 don't have ISA 3.0 instructions.
16851 (extendsi<mode>2 splitter): Likewise.
16852 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16853 generate the normal insns since SImode can now go in vector
16854 registers. Disallow the special UNSPECs needed for previous
16855 machines to hide SImode being used. Add new insns
16856 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16857 (fix_trunc<mode>si2_stfiwx): Likewise.
16858 (fix_trunc<mode>si2_internal): Likewise.
16859 (fixuns_trunc<mode>si2): Likewise.
16860 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16861 (fctiw<u>z_<mode>_smallint): Likewise.
16862 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16863 of floating point to 32-bit integer from doing a direct move to
16864 the GPR registers to do a store.
16865 (fctiwz_<mode>): Break long line.
16867 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16869 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16870 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16871 (addr_list, addr_offset_valid_p): New.
16872 (split_address_groups): Check offset validity with above function.
16873 (gt-tree-ssa-loop-ivopts.h): Include header file.
16875 2017-05-05 Nathan Sidwell <nathan@acm.org>
16877 * config.gcc (arm*-*-*): Add missing 'fi'.
16879 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16881 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16882 included in -fopt-info does not matter.
16883 * doc/optinfo.texi (-fopt-info): Fix description of default
16884 behavour. Explicitly say order of options included in -fopt-info
16887 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16889 * config.gcc: Allow combinations of aprofile and rmprofile values for
16890 --with-multilib-list.
16891 * config/arm/t-multilib: New file.
16892 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16893 variables. Remove setting of ISA and floating-point ABI in
16894 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16895 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16896 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16897 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16899 * config/arm/t-rmprofile: Likewise except for the matches changes.
16900 * doc/install.texi (--with-multilib-list): Document the combination of
16901 aprofile and rmprofile values and warn about pitfalls in doing that.
16903 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16905 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16906 (movdi_aarch64): Likewise.
16908 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16910 PR tree-optimization/80632
16911 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16913 (build_arrays): Initialize it for virtual phis.
16914 (fix_phi_nodes): Use it for virtual phis.
16916 PR tree-optimization/80558
16917 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16918 [x, y] op z into [x op, y op z] for op & or | if conditions
16921 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16922 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16925 * config/arm/arm.md (use_literal_pool): Remove.
16926 (64-bit immediate split): No longer takes cost into consideration
16927 if arm_disable_literal_pool is enabled.
16928 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16929 used when arm_disable_literal_pool is enabled.
16930 (arm_max_const_double_inline_cost): Remove use of
16931 arm_disable_literal_pool.
16932 (push_minipool_fix): Add assert.
16933 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16934 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16935 (no_literal_pool_sf_immediate): New.
16937 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16939 PR tree-optimization/80613
16940 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16941 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16943 2017-05-05 Richard Biener <rguenther@suse.de>
16945 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16947 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16949 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16950 of this flag from insn conditions due to removal from r247495.
16952 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16954 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16956 (arm_early_store_addr_dep_ptr): Likewise.
16957 * config/arm/aarch-common-protos.h
16958 (arm_early_load_addr_dep_ptr): Add prototype.
16959 (arm_early_store_addr_dep_ptr): Likewise.
16960 * config/arm/cortex-a53.md: Add new bypasses.
16962 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16964 * tree.c (next_type_uid): Change type to unsigned.
16965 (type_hash_canon): Decrement back next_type_uid if
16966 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16967 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16970 2017-05-04 Martin Sebor <msebor@redhat.com>
16972 * builtins.c: Fix a trivial typo in a comment.
16974 PR middle-end/79234
16975 * builtins.c (check_sizes): Adjust to handle reading past the end.
16976 Avoid printing excessive upper bound of ranges. Use %E to print
16977 tree nodes instead of converting them to %wu.
16978 (expand_builtin_memchr): New function.
16979 (compute_dest_size): Rename...
16980 (compute_objsize): ...to this.
16981 (expand_builtin_memcpy): Adjust.
16982 (expand_builtin_mempcpy): Adjust.
16983 (expand_builtin_strcat): Adjust.
16984 (expand_builtin_strcpy): Adjust.
16985 (check_strncat_sizes): Adjust.
16986 (expand_builtin_strncat): Adjust.
16987 (expand_builtin_strncpy): Adjust and simplify.
16988 (expand_builtin_memset): Adjust.
16989 (expand_builtin_bzero): Adjust.
16990 (expand_builtin_memcmp): Adjust.
16991 (expand_builtin): Handle memcmp.
16992 (maybe_emit_chk_warning): Check strncat just once.
16994 2017-05-04 Martin Sebor <msebor@redhat.com>
16996 PR preprocessor/79214
16997 PR middle-end/79222
16998 PR middle-end/79223
16999 * builtins.c (check_sizes): Add inlining context and issue
17000 warnings even when -Wno-system-headers is set.
17001 (check_strncat_sizes): Same.
17002 (expand_builtin_strncat): Same.
17003 (expand_builtin_memmove): New function.
17004 (expand_builtin_stpncpy): Same.
17005 (expand_builtin): Handle memmove and stpncpy.
17007 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17009 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17010 which is not used any more.
17012 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17014 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17016 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17018 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17019 (cortexa53_tunings): Likewise.
17020 (cortexa57_tunings): Likewise.
17021 (cortexa72_tunings): Likewise.
17022 (cortexa73_tunings): Likewise.
17024 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17026 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17027 Set loop alignment to 8.
17029 2017-05-04 Martin Sebor <msebor@redhat.com>
17031 PR translation/80280
17032 * builtins.c (expand_builtin_object_size): Add missing quoting to
17033 %D and like directives.
17034 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17035 (hsa_type_for_tree_type): Same.
17036 (verify_function_arguments): Same.
17037 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17038 * varasm.c (get_section): Same.
17041 2017-05-04 Martin Sebor <msebor@redhat.com>
17043 PR translation/80280
17044 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17046 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17048 * config/aarch64/aarch64.c (generic_addrcost_table):
17049 Change HI/TI mode setting.
17051 2017-05-04 Martin Jambor <mjambor@suse.cz>
17053 PR tree-optimization/80622
17054 * tree-sra.c (comes_initialized_p): New function.
17055 (build_accesses_from_assign): Only set write lazily when
17056 comes_initialized_p is false.
17057 (analyze_access_subtree): Use comes_initialized_p.
17058 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17059 instead of testing for PARM_DECL.
17061 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17063 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17064 constraint on operand 0 to allow more general addressing modes.
17065 Adjust output template.
17066 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17068 * config/aarch64/aarch64-protos.h
17069 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17070 * config/aarch64/constraints.md (Dp): New address constraint.
17071 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17074 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17076 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17077 update use of estimate_ipcp_clone_size_and_time.
17078 (estimate_local_effects): Update use of
17079 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17080 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17081 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17082 Return nonspecialized time.
17084 2017-05-04 Richard Biener <rguenther@suse.de>
17086 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17087 for the last VUSE which def dominates the PHI. Directly call
17089 (get_continuation_for_phi_1): Remove.
17091 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17093 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17094 to explain the use of truncating division. Cap the number of
17095 iterations to the maximum given by nb_iterations_upper_bound,
17098 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17100 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17101 * configure: Regenerate.
17102 * config.in: Regenerate.
17103 * config/i386/driver-mingw32.c: new file.
17104 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17105 * config.host: Link driver-mingw32.o on MinGW host.
17106 * doc/install.texi: Document new --enable-mingw-wildcard configure
17109 2017-05-04 Marek Polacek <polacek@redhat.com>
17111 PR tree-optimization/80612
17112 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17114 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17115 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17117 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17118 (movt splitter): Likewise.
17119 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17120 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17121 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17122 block for Thumb-1 with MOVT.
17123 (thumb2_legitimate_address_p): Move code block ...
17124 (can_avoid_literal_pool_for_label_p): ... into this new function.
17125 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17127 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17128 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17129 "M-profile targets with the MOVT instruction".
17131 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17133 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17134 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17135 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17137 2017-05-04 Martin Liska <mliska@suse.cz>
17139 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17140 variable cond_code.
17142 2017-05-04 Richard Biener <rguenther@suse.de>
17144 * tree.c (array_at_struct_end_p): Handle arrays at struct
17145 end with flexarrays more conservatively. Refactor and treat
17146 arrays of arrays or aggregates more strict. Fix
17147 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17148 * tree.c (array_at_struct_end_p): Adjust prototype.
17149 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17150 * gimple-fold.c (get_range_strlen): Likewise.
17151 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17153 2017-05-04 Richard Biener <rguenther@suse.de>
17155 PR tree-optimization/31130
17156 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17158 (supports_overflow_infinity): Likewise.
17159 (is_negative_overflow_infinity): Likewise.
17160 (is_positive_overflow_infinity): Likewise.
17161 (is_overflow_infinity): Likewise.
17162 (stmt_overflow_infinity): Likewise.
17163 (overflow_infinity_range_p): Likewise.
17164 (usable_range_p): Remove as always returning true.
17165 (make_overflow_infinity): Remove.
17166 (negative_overflow_infinity): Likewise.
17167 (positive_overflow_infinity): Likewise.
17168 (avoid_overflow_infinity): Likewise.
17169 (set_value_range): Adjust accordingly.
17170 (set_value_range_to_nonnegative): Likewise, remove now unused
17171 overflow_infinity arg.
17172 (vrp_operand_equal_p): Adjust.
17173 (update_value_range): Likewise.
17174 (range_int_cst_singleton_p): Likewise.
17175 (operand_less_p): Likewise.
17176 (compare_values_warnv): Likewise.
17177 (extract_range_for_var_from_comparison_expr): Likewise.
17178 (vrp_int_const_binop): Likewise.
17179 (zero_nonzero_bits_from_vr): Likewise.
17180 (extract_range_from_multiplicative_op_1): Likewise.
17181 (extract_range_from_binary_expr_1): Likewise.
17182 (extract_range_from_unary_expr): Likewise.
17183 (extract_range_from_comparison): Likewise.
17184 (extract_range_basic): Likewise.
17185 (adjust_range_with_scev): Likewise.
17186 (compare_ranges): Likewise.
17187 (compare_range_with_value): Likewise.
17188 (dump_value_range): Likewise.
17189 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17191 (simplify_cond_using_ranges): Adjust.
17193 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17195 * brig-builtins.def: Added a builtin for class_f64.
17196 * builtin-types.def: Added a builtin type needed by class_f64.
17198 2017-05-03 Jason Merrill <jason@redhat.com>
17200 * timevar.def: Add TV_CONSTEXPR.
17202 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17204 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17206 2017-05-03 Martin Jambor <mjambor@suse.cz>
17208 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17209 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17210 * ipa-cp.c (ipcp_propagate_stage): Do not call
17211 ipa_update_after_lto_read.
17212 * ipa-inline.c (ipa_inline): Likewise.
17214 2017-05-03 Martin Jambor <mjambor@suse.cz>
17216 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17217 tag. Added a default constructor and a destructor.
17218 (ipa_edge_args_sum_t): New class;
17219 (ipa_edge_args_sum): Declare.
17220 (ipa_edge_args_vector): Remove declaration.
17221 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17222 (ipa_free_edge_args_substructures): Remove declaration.
17223 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17224 (ipa_edge_args_info_available_for_edge_p): Likewise.
17225 * ipa-prop.c (ipa_edge_args_vector): Removed.
17226 (edge_removal_hook_holder): Likewise.
17227 (edge_duplication_hook_holder): Likewise.
17228 (ipa_edge_args_sum): New variable.
17229 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17230 ipa_edge_args_vector.
17231 (ipa_free_edge_args_substructures): Likewise.
17232 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17233 ipa_edge_args_vector.
17234 (ipa_edge_removal_hook): Turned into method
17235 ipa_edge_args_sum_t::remove.
17236 (ipa_edge_duplication_hook): Turned into method
17237 ipa_edge_args_sum_t::duplicate.
17238 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17239 registering edge hooks.
17240 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17241 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17242 ipa_edge_args_sum instead of ipa_edge_args_vector.
17243 * ipa-profile.c (ipa_profile): Likewise.
17245 2017-05-03 Martin Jambor <mjambor@suse.cz>
17247 * symbol-summary.h (function_summary): New method exists.
17248 (function_summary::symtab_removal): Deallocate through release.
17249 (call_summary): New class.
17250 (gt_ggc_mx): New overload.
17251 (gt_pch_nx): Likewise.
17252 (gt_pch_nx): Likewise.
17254 2017-05-03 Jeff Law <law@redhat.com>
17256 PR tree-optimization/78496
17257 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17258 from simplify_cond_using_ranges. Split off code to walk
17259 backwards through casts into ...
17260 (simplify_cond_using_ranges_2): New function.
17261 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17262 (execute_vrp): After identifying jump threads, call
17263 simplify_cond_using_ranges_2.
17265 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17268 * ipa-inline.h (inline_summary): Add ctor.
17269 (create_ggc): Do not use ggc_cleared_alloc.
17271 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17272 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17274 * gcc.c (handle_braces): Support escaping in switch matching
17276 * doc/invoke.texi (Spec Files): Document it.
17277 Remove superfluous @code markup in items.
17279 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17281 * diagnostic-show-locus.c (struct column_range): New struct.
17282 (get_affected_columns): New function.
17283 (get_printed_columns): New function.
17284 (struct correction): New struct.
17285 (correction::ensure_capacity): New function.
17286 (correction::ensure_terminated): New function.
17287 (struct line_corrections): New struct.
17288 (line_corrections::~line_corrections): New dtor.
17289 (line_corrections::add_hint): New function.
17290 (layout::print_trailing_fixits): Reimplement in terms of the new
17292 (selftest::test_overlapped_fixit_printing): New function.
17293 (selftest::diagnostic_show_locus_c_tests): Call it.
17295 2017-05-03 Nathan Sidwell <nathan@acm.org>
17297 Canonicalize canonical type hashing
17298 * tree.h (type_hash_canon_hash): Declare.
17299 * tree.c (type_hash_list, attribute_hash_list): Move into
17300 type_hash_canon_hash.
17301 (build_type_attribute_qual_variant): Break out hash code calc into
17302 type_hash_canon_hash.
17303 (type_hash_canon_hash): New. Generic type hash computation.
17304 (build_range_type_1, build_array_type_1, build_function_type,
17305 build_method_type_directly, build_offset_type, build_complex_type,
17306 make_vector_type): Call it.
17308 2017-05-03 Richard Biener <rguenther@suse.de>
17310 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17311 When all DRs have unknown misaligned do not always peel
17312 when there is a store but apply the same costing model as if
17313 there were only loads.
17315 2017-05-03 Richard Biener <rguenther@suse.de>
17318 PR tree-optimization/80492
17319 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17320 compare_base_decls returning dont-know properly.
17322 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17324 * config/arm/iterators.md (CCSI): New mode iterator.
17325 (arch): New mode attribute.
17326 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17327 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17328 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17329 code iterator for success result mode.
17330 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17331 the corresponding new insn generators.
17333 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17336 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17337 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17339 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17341 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17342 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17343 (DDR_A): Wrap DDR argument in brackets.
17344 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17345 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17346 (DDR_REVERSED_P): Likewise.
17348 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17350 PR tree-optimization/79472
17351 * tree-switch-conversion.c (struct switch_conv_info): Add
17352 contiguous_range and default_case_nonstandard fields.
17353 (collect_switch_conv_info): Compute contiguous_range and
17354 default_case_nonstandard fields, don't clear final_bb if
17355 contiguous_range and only the default case doesn't have the required
17357 (check_all_empty_except_final): Set default_case_nonstandard instead
17358 of failing if contiguous_range and the default case doesn't have empty
17360 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17361 and only the default case doesn't have the required constants. Skip
17363 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17364 if default_case_nonstandard.
17365 (build_constructors): Build constant 1 just once. Assert that default
17366 values aren't inserted in between cases if contiguous_range. Skip
17368 (build_arrays): Skip virtual phis.
17369 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17370 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17371 Handle virtual phis.
17372 (gen_inbound_check): Handle default_case_nonstandard case.
17373 (process_switch): Adjust check_final_bb caller. Call
17374 gather_default_values with the first non-default case instead of
17375 default case if default_case_nonstandard.
17377 2017-05-02 Nathan Sidwell <nathan@acm.org>
17379 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17380 check. Fix formatting.
17382 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17384 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17385 errors when comparing specialized and unspecialized times.
17387 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17389 * diagnostic-show-locus.c
17390 (layout::should_print_annotation_line_p): Make private.
17391 (layout::print_annotation_line): Make private.
17392 (layout::annotation_line_showed_range_p): Make private.
17393 (layout::show_ruler): Make private.
17394 (layout::print_source_line): Make private. Pass in line and
17395 line_width, rather than calling location_get_source_line. Drop
17397 (layout::print_leading_fixits): New method.
17398 (layout::print_any_fixits): Rename to...
17399 (layout::print_trailing_fixits): ...this, and make private.
17400 Don't print newline fixits.
17401 (diagnostic_show_locus): Move logic for printing one row into...
17402 (layout::print_line): ...this new function. Move the
17403 location_get_source_line call and error-handling from
17404 print_source_line to here. Call print_leading_fixits, and rename
17405 print_any_fixits to print_trailing_fixits.
17406 (selftest::test_fixit_insert_containing_newline): Update now that
17407 newlines are partially supported.
17408 (selftest::test_fixit_insert_containing_newline_2): New test.
17409 (selftest::test_fixit_replace_containing_newline): Update comments.
17410 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17411 * edit-context.c (class added_line): New class.
17412 (class edited_line): Describe newline handling in comment.
17413 (edited_line::actually_edited_p): New method.
17414 (edited_line::print_content): Delete redundant decl.
17415 (edited_line::m_predecessors): New field.
17416 (edited_file::print_content): Call edited_line::print_content.
17417 (edited_file::print_diff): Update to support newlines.
17418 (edited_file::print_diff_hunk): Likewise.
17419 (edited_file::print_run_of_changed_lines): New function.
17420 (edited_file::print_diff_line): Convert to...
17421 (print_diff_line): ...this.
17422 (edited_file::get_effective_line_count): New function.
17423 (edited_line::edited_line): Initialize new field m_predecessors.
17424 (edited_line::~edited_line): Clean up m_predecessors.
17425 (edited_line::apply_fixit): Handle newlines.
17426 (edited_line::get_effective_line_count): New function.
17427 (edited_line::print_content): New function.
17428 (edited_line::print_diff_lines): New function.
17429 (selftest::test_applying_fixits_insert_containing_newline): New
17431 (selftest::test_applying_fixits_replace_containing_newline): New
17433 (selftest::insert_line): New function.
17434 (selftest::test_applying_fixits_multiple_lines): Add example of
17436 (selftest::edit_context_c_tests): Call the new tests.
17438 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17440 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17441 parameter cand. Update dump information.
17442 (get_computation_cost): Update uses.
17444 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17446 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17447 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17448 (get_computation_at, rewrite_use_address): Update use of
17449 get_computation_aff.
17451 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17453 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17454 (get_computation): Delete.
17455 (get_computation_cost): Implement like get_computation_cost_at.
17456 Use get_computation_at.
17457 (get_computation_cost_at): Delete.
17458 (rewrite_use_nonlinear_expr): Use get_computation_at.
17459 (rewrite_use_compare, remove_unused_ivs): Ditto.
17461 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17463 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17465 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17467 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17468 (ivopts_global_cost_for_size): Rename parameter and update uses.
17469 (iv_ca_recount_cost): Update uses.
17470 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17471 candidates seperately in n_invs and n_cands.
17472 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17474 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17476 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17477 (find_inv_vars_cb): New.
17478 (find_depends): Renamed to ...
17479 (find_inv_vars): ... this.
17480 (add_candidate_1, force_var_cost): Call find_inv_vars.
17481 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17483 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17485 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17486 inv_vars. Add inv_exprs.
17487 (struct iv_cand): Rename depends_on to inv_vars.
17488 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17489 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17490 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17491 (dump_cand): Dump inv_vars.
17492 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17493 (record_invariant, find_depends, add_candidate_1): Ditto.
17494 (set_group_iv_cost, force_var_cost): Ditto.
17495 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17496 (get_computation_cost_at, get_computation_cost): Ditto.
17497 (determine_group_iv_cost_generic): Ditto.
17498 (determine_group_iv_cost_address): Ditto.
17499 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17500 (determine_group_iv_costs): Ditto.
17501 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17502 (iv_ca_set_remove_invariants): Renamed to ...
17503 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17504 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17505 (iv_ca_set_add_invariants): Renamed to ...
17506 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17507 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17508 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17509 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17510 (create_new_ivs): Remove useless dump.
17512 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17514 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17516 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17517 (iv_ca_set_no_cp, create_new_iv): Ditto.
17519 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17521 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17523 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17525 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17526 function tree_check2.
17528 2017-05-02 Martin Liska <mliska@suse.cz>
17530 * doc/gcov.texi: Add missing preposition.
17531 * gcov.c (function_info::function_info): Properly fill up
17532 all member variables.
17534 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17536 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17538 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17540 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17542 2017-05-02 Martin Liska <mliska@suse.cz>
17545 * lto-streamer-in.c (lto_read_tree_1): Remove
17546 LTO_STREAMER_DEBUG.
17547 * lto-streamer.c (struct tree_hash_entry): Likewise.
17548 (struct tree_entry_hasher): Likewise.
17549 (tree_entry_hasher::hash): Likewise.
17550 (tree_entry_hasher::equal): Likewise.
17551 (lto_streamer_init): Likewise.
17552 (lto_orig_address_map): Likewise.
17553 (lto_orig_address_get): Likewise.
17554 (lto_orig_address_remove): Likewise.
17555 * lto-streamer.h: Likewise.
17556 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17557 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17559 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17561 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17562 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17563 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17564 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17565 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17566 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
17567 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
17568 (mm_maskz_sub_ss): New intrinsics.
17569 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17570 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17571 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
17572 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
17573 (__builtin_ia32_subss_mask_round): New builtins.
17574 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17575 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17576 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
17578 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
17579 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17581 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17584 2017-05-02 Martin Jambor <mjambor@suse.cz>
17586 PR tree-optimization/78687
17587 * tree-sra.c (access): New field parent.
17588 (process_subtree_disqualification): New function.
17589 (disqualify_candidate): Call it.
17590 (build_accesses_from_assign): Reset write flag if creating an
17592 (build_access_subtree): Fill in parent field and also prpagate
17593 down grp_write flag.
17594 (create_artificial_child_access): New parameter set_grp_write, set
17595 grp_write to its value.
17596 (propagate_subaccesses_across_link): Also propagate grp_write flag
17598 (propagate_all_subaccesses): Push the closest parent back to work
17599 queue if add_access_to_work_queue returned true.
17601 2017-05-02 Richard Biener <rguenther@suse.de>
17603 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17604 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17605 -fstrict-overflow documentation.
17606 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17607 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17608 flag_strict_overflow.
17609 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17610 * lto-opts.c (lto_write_options): Do not stream it.
17611 * lto-wrapper.c (merge_and_complain): Do not handle it.
17612 * opts.c (default_options_table): Do not set -fstrict-overflow.
17613 (finish_options): Likewise do not clear it when sanitizing.
17614 * simplify-rtx.c (simplify_const_relational_operation): Do not
17615 test flag_strict_overflow.
17617 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17619 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17620 using enabled attribute.
17621 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17622 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17623 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17624 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17625 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17626 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17627 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17628 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17629 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17630 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17632 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17634 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17636 2017-05-02 Richard Biener <rguenther@suse.de>
17638 PR tree-optimization/80591
17640 2017-04-10 Richard Biener <rguenther@suse.de>
17642 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17645 2017-05-02 Richard Biener <rguenther@suse.de>
17647 PR tree-optimization/80549
17648 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17649 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17650 headers if they do not have a preheader.
17652 2017-05-02 Martin Liska <mliska@suse.cz>
17655 * common.opt: Fix typo.
17656 * doc/invoke.texi: Likewise.
17658 2017-05-01 Jan Beulich <jbeulich@suse.com>
17660 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17661 swapping, add (x,x,m,x,n) alternative.
17663 2017-05-01 Nathan Sidwell <nathan@acm.org>
17665 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17666 unnecessary unadjusted_alignment check.
17668 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17671 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17673 * gimplify.c (gimplify_cilk_detach): New function.
17674 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17675 * tree-core.h: Document EXPR_CILK_SPAWN.
17676 * tree.h (EXPR_CILK_SPAWN): Define.
17678 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17680 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17681 to use new fixit_hint representation, using the "replace" logic.
17682 (get_line_span_for_fixit_hint): Likewise.
17683 (layout::print_any_fixits): Likewise.
17684 (selftest::test_one_liner_many_fixits): Rename to...
17685 (selftest::test_one_liner_many_fixits_1): ...this, and update
17686 comment and expected output to reflect that the multiple fix-it
17687 hints are now consolidated into one insertion.
17688 (selftest::test_one_liner_many_fixits_2): New test.
17689 (selftest::test_diagnostic_show_locus_one_liner): Update for
17691 (selftest::test_fixit_consolidation): Update for fix-it API
17693 * diagnostic.c (print_parseable_fixits): Likewise.
17694 * edit-context.c (edited_line::m_line_events): Convert from
17695 auto_vec <line_event *> to auto_vec <line_event>.
17696 (class line_event): Convert from abstract base class to a concrete
17697 class, taking over the role of replace_event.
17698 (class insert_event): Delete.
17699 (class replace_event): Rename to class line_event. Convert to
17701 (edit_context::add_fixits): Reimplement.
17702 (edit_context::apply_insert): Delete.
17703 (edit_context::apply_replace): Rename to...
17704 (edit_context::apply_fixit): ...this. Convert to half-open range.
17705 (edited_file::apply_insert): Delete.
17706 (edited_file::apply_replace): Rename to...
17707 (edited_file::apply_fixit): ...this.
17708 (edited_line::~edited_line): Drop deletion of events.
17709 (edited_line::apply_insert): Delete.
17710 (edited_line::apply_replace): Rename to...
17711 (edited_line::apply_fixit): ...this. Convert to half-open range.
17712 Update for change to type of m_line_events.
17713 * edit-context.h (edit_context::apply_insert): Delete.
17714 (edit_context::apply_replace): Rename to...
17715 (edit_context::apply_fixit): ...this.
17717 2017-05-01 Martin Sebor <msebor@redhat.com>
17719 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17722 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17725 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17726 __get_cpuid_max returns 0.
17727 (__get_cpuid_count): Ditto.
17729 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17731 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17732 replacement expression is another instance of one of its arguments.
17734 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17737 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17738 check for stack push/pop autoinc.
17739 * config/i386/i386.c (ix86_agi_dependent): Return false
17740 if the only reason why modified_in_p returned true is that
17741 addr is SP based and set_insn is a push or pop.
17743 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17745 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17748 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17751 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17752 (account_size_time): Use two predicates - exec_pred and
17754 (evaluate_conditions_for_known_args): Compute both clause and
17756 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17757 (inline_summary_t::duplicate): Update.
17758 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17760 (compute_inline_parameters): Likewise.
17761 (estimate_edge_size_and_time): Update caluclation of time.
17762 (estimate_node_size_and_time): Compute both time and nonspecialized
17764 (estimate_ipcp_clone_size_and_time): Update.
17765 (inline_merge_summary): Update.
17766 (do_estimate_edge_time): Update.
17767 (do_estimate_edge_size): Update.
17768 (do_estimate_edge_hints): Update.
17769 (inline_read_section, inline_write_summary): Stream both new predicates.
17770 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17772 (compute_inlined_call_time): Cleanup.
17773 (big_speedup_p): Update.
17774 (edge_badness): Update.
17775 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17776 (size_time_entry): Replace predicate by exec_predicate and
17777 nonconst_predicate.
17778 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17779 (estimate_edge_time): Return also nonspec_time.
17780 (reset_edge_growth_cache): Update.
17782 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17784 PR rtl-optimization/80491
17785 * ifcvt.c (noce_process_if_block): When looking for x setter
17786 with missing else_bb, don't check only the insn right before
17787 cond_earliest, but look for the last insn that x is modified in
17788 within the same bb.
17790 PR rtl-optimization/80491
17791 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17793 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17795 PR tree-optimization/80487
17796 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17798 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17800 PR tree-optimization/79697
17801 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17802 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17803 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17805 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17806 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17808 2017-04-28 Martin Sebor <msebor@redhat.com>
17810 PR tree-optimization/80523
17811 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17812 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17814 (maybe_warn, format_directive, parse_directive): Use new functions.
17815 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17817 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17819 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17821 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17823 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17824 target_header_dir): Set correctly.
17825 * configure: Regenerated.
17826 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17827 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17828 instead of SYSTEM_HEADER_DIR.
17830 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17832 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17833 (estimate_local_effects): Likewise.
17834 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17835 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17836 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17837 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17838 do_estimate_edge_time, estimate_edge_time): Likewise.
17839 * ipa-inline-analysis.c (estimate_node_size_and_time,
17840 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17841 (estimate_time_after_inlining): Remove.
17843 2017-04-28 Martin Liska <mliska@suse.cz>
17845 * doc/gcov.texi: Enhance documentation of gcov.
17847 2017-04-28 Martin Liska <mliska@suse.cz>
17849 * doc/gcov.texi: Sort options in alphabetic order.
17850 * doc/gcov-dump.texi: Likewise.
17851 * doc/gcov-tool.texi: Likewise.
17852 * gcov.c (print_usage): Likewise.
17853 * gcov-dump.c (print_usage): Likewise.
17854 * gcov-tool.c (print_merge_usage_message): Likewise.
17855 (print_rewrite_usage_message): Likewise.
17856 (print_overlap_usage_message): Likewise.
17858 2017-04-28 Martin Liska <mliska@suse.cz>
17860 PR gcov-profile/53915
17861 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17863 2017-04-28 Martin Liska <mliska@suse.cz>
17865 PR gcov-profile/79891
17866 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17867 is marked by compiler as living on a line.
17868 (get_cycles_count): Remove usage of the union.
17869 (output_intermediate_file): Likewise.
17870 (find_source): Fix GNU coding style.
17871 (accumulate_line_counts): Remove old non-all block mode.
17872 (output_lines): Remove usage of the union.
17873 * profile.c (output_location): Include all BBs, even if
17874 belonging to a same line (and file) as a previous BB.
17876 2017-04-28 Martin Liska <mliska@suse.cz>
17878 * gcov.c (process_args): Handle new argument 'w'.
17879 (read_graph_file): Assign ID to BBs.
17880 (output_branch_count): Display BB # if verbose flag is set.
17881 (output_lines): Likewise for arcs.
17882 (print_usage): Add '--verbose' option help.
17883 * doc/gcov.texi: Document --verbose (-w) option.
17885 2017-04-28 Martin Liska <mliska@suse.cz>
17887 * gcov.c (struct block_location_info): New struct.
17888 (process_file): Fill up the new structure.
17889 (read_graph_file): Replace usage of encoding by the newly added
17891 (add_line_counts): Likewise.
17892 (accumulate_line_counts): Remove usage of the union.
17893 (function_info::function_info): New function.
17894 (function_info::~function_info): Likewise.
17895 (process_file): Call delete instead of release_function.
17896 (release_function): Release the function.
17897 (release_structures): Call delete instead of release_function.
17898 (solve_flow_graph): Replace usage of num_blocks.
17899 (find_exception_blocks): Likewise.
17900 (output_lines): Fix GNU coding style.
17902 2017-04-28 Martin Liska <mliska@suse.cz>
17905 * coverage.c (coverage_remove_note_file): New function.
17906 * coverage.h: Declare the function.
17907 * toplev.c (finalize): Clean if an error has been seen.
17909 2017-04-28 Martin Liska <mliska@suse.cz>
17911 PR gcov-profile/80031
17912 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17913 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17914 * gcov.c (read_graph_file): Read just number of blocks.
17915 * profile.c (branch_prob): Do not stream 0 flags per a basic
17918 2017-04-28 Martin Liska <mliska@suse.cz>
17920 * gcov-dump.c (tag_*): Add new argument to declarations.
17921 (dump_gcov_file): Likewise.
17922 (tag_blocks): Add and use new argument depth.
17923 (tag_arcs): Likewise.
17924 (tag_lines): Likewise.
17925 (tag_counters): Likewise.
17926 (tag_summary): Likewise.
17927 (dump_working_sets): Use depth to do a proper indentation.
17929 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17932 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17933 * symtab.c (symtab_node::debug_symtab): Move definition here.
17935 2017-04-28 Richard Biener <rguenther@suse.de>
17937 * lto-streamer.h (LTO_major_version): Bump to 7.
17939 2017-04-28 Richard Biener <rguenther@suse.de>
17941 * tree-vrp.c (assert_info): New struct.
17942 (add_assert_info): New helper.
17943 (register_edge_assert_for_2): Refactor to add asserts to a vector
17945 (register_edge_assert_for_1): Likewise.
17946 (register_edge_assert_for): Likewise.
17947 (finish_register_edge_assert_for): New helper actually registering
17948 asserts where live on edge.
17949 (find_conditional_asserts): Adjust.
17950 (find_switch_asserts): Likewise.
17951 (evrp_dom_walker::try_find_new_range): Generalize.
17952 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17954 2017-04-27 Marek Polacek <polacek@redhat.com>
17957 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17958 arg10 and arg11 to itype.
17960 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17962 * doc/extend.texi (Object Size Checking): Improve grammar.
17964 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17967 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17968 that the logic for permitting reciprocal estimates matches that
17971 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17974 * tree.c (type_cache_hasher::equal): Only compare
17975 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17976 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17977 non-aggregate element types.
17978 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17979 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17981 2017-04-27 Richard Biener <rguenther@suse.de>
17983 PR middle-end/80533
17984 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17985 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17986 keeping a reference to a trailing array.
17988 2017-04-27 Richard Biener <rguenther@suse.de>
17990 PR middle-end/80539
17991 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17992 being in loop-closed SSA form conservatively.
17993 (chrec_fold_multiply_poly_poly): Likewise.
17995 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17997 PR middle-end/79665
17998 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17999 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18001 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18004 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18005 (aarch64_function_arg_alignment): Return unsigned int again, but still
18006 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18007 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18008 Don't emit -Wpsabi note.
18009 (aarch64_function_arg_boundary): Likewise.
18010 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18013 2017-04-26 Nathan Sidwell <nathan@acm.org>
18015 * tree.h (crc32_unsigned_n): Declare.
18016 (crc32_unsigned, crc32_unsigned): Make inline.
18017 * tree.c (crc32_unsigned_bits): Replace with ...
18018 (crc32_unsigned_n): ... this.
18019 (crc32_unsigned, crc32_byte): Remove.
18020 (crc32_string): Remove unnecessary braces.
18022 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18024 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18025 * ipa-inline-analysis.c (MAX_TIME): Remove.
18026 (account_size_time): Use sreal for time.
18027 (dump_inline_summary): Update.
18028 (estimate_function_body_sizes): Update.
18029 (estimate_edge_size_and_time): Update.
18030 (estimate_calls_size_and_time): Update.
18031 (estimate_node_size_and_time): Update.
18032 (inline_merge_summary): Update.
18033 (inline_update_overall_summary): Update.
18034 (estimate_time_after_inlining): Update.
18035 (inline_read_section): Update.
18036 (inline_write_summary): Update.
18037 * ipa-inline.c (compute_uninlined_call_time): Update.
18038 (compute_inlined_call_time): Update.
18039 (recursive_inlining): Update.
18040 (inline_small_functions): Update.
18041 (dump_overall_stats): Update.
18042 * ipa-inline.h: Include sreal.h.
18043 (size_time_entry): Turn time to sreal.
18044 (inline_summary): Turn self_time nad time to sreal.
18046 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18048 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18050 (sreal::stream_out, sreal::stream_in): New.
18051 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18053 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18055 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18058 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18061 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18063 (dimode_scalar_chain::compute_convert_gain): Ditto.
18064 (dimode_scalar_chain::make_vector_copies): Ditto.
18065 (dimode_scalar_chain::convert_reg): Ditto.
18066 (dimode_scalar_chain::convert_insn): Ditto.
18067 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18068 (VI248_AVX512BW_1): New mode iterator.
18069 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18070 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18073 2017-04-25 Martin Sebor <msebor@redhat.com>
18075 PR tree-optimization/80497
18076 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18077 constants are representable in HOST_WIDE_INT.
18078 (parse_directive): Ditto.
18080 2017-04-25 Martin Sebor <msebor@redhat.com>
18083 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18084 (new_zero_array): Adjust signature.
18085 (dom_info::dom_init): Used unsigned rather that size_t.
18086 (dom_info::dom_info): Same.
18088 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18089 Jakub Jelinek <jakub@redhat.com>
18092 * config/arm/arm.c: Include gimple.h.
18093 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18094 returns negative, increment ncrn only if it returned positive.
18095 (arm_needs_doubleword_align): Return int instead of bool,
18096 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18097 members, but if there is any such non-FIELD_DECL
18098 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18099 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18100 returns negative, increment nregs only if it returned positive.
18101 (arm_setup_incoming_varargs): Likewise.
18102 (arm_function_arg_boundary): Emit -Wpsabi note if
18103 arm_needs_doubleword_align returns negative, return
18104 DOUBLEWORD_ALIGNMENT only if it returned positive.
18106 2017-04-25 Marek Polacek <polacek@redhat.com>
18109 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18110 first argument to type.
18112 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18115 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18116 type checks to test for compatibility instead of equality.
18118 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18119 Jakub Jelinek <jakub@redhat.com>
18122 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18124 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18125 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18126 the alignment computation, but return their maximum in warn_alignment.
18127 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18128 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18130 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18131 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18134 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18136 * config/arc/simdext.md (dmpyh): Fix typo.
18138 2017-04-25 Richard Biener <rguenther@suse.de>
18140 PR tree-optimization/80492
18141 * alias.c (compare_base_decls): Handle registers with asm
18142 specification conservatively.
18143 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18144 compare_base_decls returning dont-know properly.
18146 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18148 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18149 (legitimate_offset_address_p): New function.
18150 (arc_legitimate_address_p): Use above function.
18152 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18154 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18156 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18158 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18159 ACCH registers whenever they are available.
18161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18163 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18164 double regs fix when not used.
18166 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18168 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18170 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18171 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18173 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18175 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18176 when emitting short ADD instructions.
18178 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18180 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18182 (cmpsi_cc_c_insn): Likewise.
18183 (cbranchsi4_scratch): Compute proper instruction length using
18184 compact_hreg_operand.
18185 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18187 2017-04-25 Richard Biener <rguenther@suse.de>
18189 PR middle-end/80509
18190 * passes.c (pass_manager::pass_manager): Initialize
18191 m_name_to_pass_map.
18193 2017-04-25 Richard Biener <rguenther@suse.de>
18195 PR tree-optimization/79201
18196 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18198 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18201 * config/s390/vector.md: Split MEM->GPR vector moves for
18202 non-s_operand addresses.
18204 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18207 * config/s390/predicates.md (reload_const_wide_int_operand): New
18209 * config/s390/s390.md ("movti"): Remove d/P alternative.
18210 ("movti_bigconst"): New pattern definition.
18212 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18215 * s390-protos.h (s390_expand_cs_hqi): Removed.
18216 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18217 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18218 modes as well as CCZ1mode and CCZmode.
18219 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18220 signature of s390_emit_compare_and_swap.
18221 (s390_expand_cs_hqi): Likewise, make static.
18222 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18223 compare-and-swap, in some cases.
18224 (s390_expand_cs): Wrapper function.
18225 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18227 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18228 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18229 patterns for small and large integers. Forbid symref memory operands.
18230 Move expander to s390.c. Require cc register.
18231 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18232 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18233 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18234 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18235 symref memory operands. Remove CC mode and call s390_match_ccmode
18237 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18239 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18241 * config/s390/s390.md (define_peephole2): New peephole to help
18242 combining the load-and-test pattern with volatile memory.
18244 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18246 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18247 with CCZmode for TARGET_Z196.
18249 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18251 PR rtl-optimization/80501
18252 * combine.c (make_compound_operation_int): Set subreg_code to SET
18253 even for AND with mask of the sign bit of mode.
18255 PR rtl-optimization/80500
18256 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18257 sum's initial value.
18259 2017-04-25 Julian Brown <julian@codesourcery.com>
18260 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18262 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18264 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18266 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18268 2017-04-25 Julian Brown <julian@codesourcery.com>
18269 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18271 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18272 (thunderx2t99_sha): New Reservation.
18274 2017-04-25 Julian Brown <julian@codesourcery.com>
18275 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18277 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18278 type for 1-element load.
18280 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18282 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18284 2017-04-24 Martin Jambor <mjambor@suse.cz>
18286 PR tree-optimization/80293
18287 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18288 char arrays not totally scalarizable if it is false.
18289 (analyze_all_variable_accesses): Pass correct value in the new
18290 parameter. Add a statistics counter.
18292 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18294 PR middle-end/79931
18295 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18297 2017-04-24 Richard Biener <rguenther@suse.de>
18299 PR tree-optimization/80494
18300 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18301 out for complex types.
18303 2017-04-24 Richard Biener <rguenther@suse.de>
18305 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18306 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18307 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18308 (DFS): Adjust and never fail.
18309 (sccvn_dom_walker::fail): Remove.
18310 (sccvn_dom_walker::before_dom_children): Adjust.
18311 (run_scc_vn): Likewise and never fail.
18312 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18313 (pass_fre::execute): Likewise.
18315 2017-04-24 Richard Biener <rguenther@suse.de>
18317 PR tree-optimization/79725
18318 * tree-ssa-sink.c (statement_sink_location): Return whether
18319 failure reason was zero uses. Move that check later.
18320 (sink_code_in_bb): Deal with zero uses by removing the stmt
18323 2017-04-24 Richard Biener <rguenther@suse.de>
18326 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18327 pointer-based references.
18329 2017-04-24 Richard Biener <rguenther@suse.de>
18332 * pass_manager.h (pass_manager::operator new): Remove.
18333 (pass_manager::operator delete): Likewise.
18334 * passes.c (pass_manager::operator new): Remove.
18335 (pass_manager::operator delete): Likewise.
18336 (pass_manager::pass_manager): Zero individual pass members.
18338 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18341 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18342 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18343 Check "XEXP (src, 1)" operand here.
18344 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18345 Check "XEXP (src, 1)" operand here.
18346 (dimode_scalar_chain::make_vector_copies): Detect count register
18347 of a shift instruction. Zero extend count register from QImode
18348 to DImode to satisfy vector shift pattern count operand predicate.
18349 Substitute vector shift count operand with a DImode copy.
18350 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18353 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18355 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18356 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18357 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18358 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18359 (UNSPEC_NOREX_MEM): Remove definition.
18361 2017-04-21 Richard Biener <rguenther@suse.de>
18363 PR tree-optimization/79547
18364 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18365 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18366 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18367 without any constraints.
18369 2017-04-21 Richard Biener <rguenther@suse.de>
18371 PR tree-optimization/78847
18372 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18374 2017-04-21 Richard Biener <rguenther@suse.de>
18376 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18377 (build_distinct_type_copy): Likewise.
18378 (build_variant_type_copy): Likewise.
18379 * tree.c (build_qualified_type): Pass down mem-stat info.
18380 (build_distinct_type_copy): Likewise.
18381 (build_variant_type_copy): Likewise.
18383 2017-04-21 Richard Biener <rguenther@suse.de>
18385 PR tree-optimization/80237
18386 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18388 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18389 for a simplified result.
18391 2016-04-21 Richard Biener <rguenther@suse.de>
18393 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18394 sth as strict as a simple_iv but a chrec without symbols and an
18395 operand defined in the loop we are peeling (and not some subloop).
18396 (propagate_constants_for_unrolling): Propagate all constants.
18398 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18401 * config/i386/i386.c (print_reg): Remove assert for disalowed
18402 regno values, call output_operand_lossage instead.
18404 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18407 * config/i386/constraints.md (Yc): New register constraint.
18408 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18409 Use Yc constraint for alternative 2 of operand 0. Remove
18410 preferred_for_speed attribute.
18412 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18414 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18415 lastprivate clauses in SIMT case.
18417 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18419 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18421 2017-04-20 Richard Biener <rguenther@suse.de>
18423 PR tree-optimization/57796
18424 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18425 as N scalar stores.
18426 (vect_model_load_cost): Cost gathers as N scalar loads.
18428 2017-04-20 Richard Biener <rguenther@suse.de>
18430 * ggc-page.c (ggc_allocated_p): Rename to ...
18431 (safe_lookup_page_table_entry): ... this and return the lookup
18433 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18435 2017-04-20 Richard Biener <rguenther@suse.de>
18437 PR tree-optimization/80453
18438 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18439 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18440 from the conditions.
18441 (vn_phi_eq): Pass them down.
18442 (vn_phi_lookup): Record them.
18443 (vn_phi_insert): Likewise.
18445 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18447 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18448 uninitialized variable warning to avoid buffer overrun.
18450 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18453 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18454 is suppressed for '{ 0 }' in C.
18456 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18458 * BASE-VER: Set to 8.0.0.
18460 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18462 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18463 priority .init_array and .fini_array section with SECTION_NOTYPE
18466 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18468 PR middle-end/80423
18469 * tree.h (build_array_type): Add typeless_storage default argument.
18470 * tree.c (type_cache_hasher::equal): Also compare
18471 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18472 (build_array_type): Add typeless_storage argument, set
18473 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18475 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18476 (build_array_type): Likewise. Add typeless_storage argument.
18478 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18479 Jakub Jelinek <jakub@redhat.com>
18481 PR tree-optimization/80426
18482 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18483 operation on symbolic operands, also compute the overflow for the
18484 invariant part when the operation degenerates into a negation.
18486 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18489 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18490 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18493 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18495 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18498 * config/avr/avr.c (tree.h): Include it.
18499 (cgraph.h): Include it.
18500 (avr_encode_section_info): Don't warn for uninitialized progmem
18501 variable if it's just an alias.
18503 2017-04-19 Richard Biener <rguenther@suse.de>
18506 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18507 when needed by AutoPGO.
18509 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18512 * doc/lto.texi: Remove an extra 'that'.
18514 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18516 PR rtl-optimization/80429
18517 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18518 are only used in debug insns.
18520 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18521 Vladimir Makarov <vmakarov@redhat.com>
18523 * config/sparc/predicates.md (input_operand): Add comment. Return
18524 true for any memory operand when LRA is in progress.
18525 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18527 2017-04-18 Jeff Law <law@redhat.com>
18530 * mips.md ({return,simple_return}_internal): Do not overwrite
18533 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18535 PR tree-optimization/80443
18536 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18537 instead of adding 1, subtract -1 and similarly instead of subtracting
18540 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18542 PR rtl-optimization/80357
18543 * haifa-sched.c (tmp_bitmap): New variable.
18544 (model_recompute): Handle duplicate use records.
18545 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18546 (free_global_sched_pressure_data): Free it.
18548 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18551 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18552 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18553 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18554 instead of SYSTEM_HEADER_DIR.
18556 2017-04-18 Jeff Law <law@redhat.com>
18558 PR middle-end/80422
18559 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18560 predecessors after walking up the insn chain.
18562 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18565 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18566 sizetype type into debug info.
18568 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18571 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
18572 unneeded test for TARGET_UPPER_REGS_SF.
18573 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
18575 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18578 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
18579 instead of gsi_after_labels.
18581 2017-04-18 Jeff Law <law@redhat.com>
18583 * regcprop.c (maybe_mode_change): Avoid creating copies of the
18587 2017-04-13 Jeff Law <law@redhat.com>
18588 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18589 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18591 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
18594 * config/avr/avr.c (intl.h): Include it.
18595 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
18597 2017-04-18 Martin Liska <mliska@suse.cz>
18599 PR gcov-profile/78783
18600 * gcov-tool.c (gcov_output_files): Validate that destination
18601 file is either removed by the tool or by a user.
18603 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18604 Guy Benyei <guybe@mellanox.com>
18606 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18607 block, and do not negate it, the stored id is already negative.
18609 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18611 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18613 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18616 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18617 masks of options that should be turned off if the VSX vector
18618 options are turned off.
18619 (OTHER_P8_VECTOR_MASKS): Likewise.
18620 (OTHER_VSX_VECTOR_MASKS): Likewise.
18621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18622 rs6000_disable_incompatible_switches to validate no type switches
18624 (rs6000_incompatible_switch): New function to disallow turning on
18625 other vector options if -mno-vsx, -mno-power8-vector, or
18626 -mno-power9-vector are specified.
18628 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18630 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18632 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18634 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18635 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18636 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18637 (ARG_POINTER_CFA_OFFSET): Likewise.
18639 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18641 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18642 conditions to take advantage of various optimizations.
18644 2017-04-13 Jeff Law <law@redhat.com>
18646 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18647 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18648 (zero_extendsidi2_dext): Likewise.
18650 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18653 * fold-const.c (fold_ternary_loc): Revert
18654 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18657 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18659 PR rtl-optimization/80343
18660 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18661 new scratch pseudo.
18663 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18666 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18667 to ubsan_encode_value.
18669 2017-04-13 Jeff Law <law@redhat.com>
18671 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18672 appearing in DEBUG_INSNs.
18674 2017-04-13 Martin Liska <mliska@suse.cz>
18676 PR gcov-profile/80413
18677 * gcov-io.c (gcov_write_string): Copy to buffer just when
18678 allocated size is greater than zero.
18680 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18683 * dwarf2out.c (decls_for_scope): Ignore declarations of
18684 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18686 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18689 * ipa-visibility.c (non_local_p): Fix typos.
18690 (localize_node): When localizing symbol in same comdat group,
18691 dissolve the group only when we know external symbols are going
18693 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18695 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18697 PR tree-optimization/79390
18698 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18699 order does not result in usable sequence, retry with reversed operand
18705 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18706 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18707 op0 instead of fold_convert_loc (loc, type, arg0).
18709 2017-04-12 Jeff Law <law@redhat.com>
18711 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18712 has a delay slot in the generated code.
18714 * config/cris/cris.md (cris_preferred_reload_class): Return
18715 GENNONACR_REGS rather than GENERAL_REGS.
18717 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18720 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18721 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18722 signedness of the result type.
18724 2017-04-12 Richard Biener <rguenther@suse.de>
18725 Jeff Law <law@redhat.com>
18727 PR tree-optimization/80359
18728 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18729 trim stores to TARGET_MEM_REFs.
18731 2017-04-12 Richard Biener <rguenther@suse.de>
18733 PR tree-optimization/79390
18734 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18735 threading case even more.
18737 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18740 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18741 for quad_address_p for TImode, instead of just not indexed_address.
18743 2017-04-12 Richard Biener <rguenther@suse.de>
18744 Bernd Edlinger <bernd.edlinger@hotmail.de>
18746 PR middle-end/79671
18747 * alias.c (component_uses_parent_alias_set_from): Handle
18748 TYPE_TYPELESS_STORAGE.
18749 (get_alias_set): Likewise.
18750 * tree-core.h (tree_type_common): Add typeless_storage flag.
18751 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18752 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18753 for types containing members with TYPE_TYPELESS_STORAGE.
18754 (place_field): Likewise.
18755 (layout_type): Likewise for ARRAY_TYPE.
18756 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18757 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18758 TYPE_TYPELESS_STORAGE.
18759 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18761 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18764 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18765 first argument to type.
18767 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18771 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18772 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18773 (rs6000_expand_binop_builtin): Likewise.
18774 (rs6000_expand_ternop_builtin): Likewise; also add missing
18775 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18776 vshasigma built-ins.
18777 * doc/extend.texi: Document that vec_xxpermdi's third argument
18778 must be a constant.
18780 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18782 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18783 Use shift_const cost parameter when calculating gain of STV shifts.
18785 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18787 PR rtl-optimization/70478
18788 * lra-constraints.c (process_alt_operands): Check memory for
18789 disfavoring memory insn operand.
18791 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18793 PR middle-end/80100
18794 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18795 left shift in unsigned HOST_WIDE_INT type.
18797 PR rtl-optimization/80385
18798 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18799 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18802 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18803 if they have any depend clauses.
18805 2017-04-11 Martin Liska <mliska@suse.cz>
18808 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18809 * ipa-split.c (split_function): Create a local comdat symbol
18810 if caller is in a comdat group.
18812 2017-04-11 Martin Liska <mliska@suse.cz>
18815 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18818 2017-04-11 Martin Sebor <msebor@redhat.com>
18820 PR middle-end/80364
18821 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18822 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18824 (directive::set_width, directive::set_precision, format_character):
18826 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18829 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18832 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18833 conflict, set target->arch_name instead of target->cpu_name.
18835 2017-04-11 Richard Biener <rguenther@suse.de>
18837 PR tree-optimization/80374
18838 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18839 build_zero_cst, remove fold_convertible_p check again.
18841 2017-04-11 Martin Liska <mliska@suse.cz>
18844 * ubsan.c (instrument_object_size): Do not instrument register
18847 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18850 * config/i386/i386-builtin-types.def
18851 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18852 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18853 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18854 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18855 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18856 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18857 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18858 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18859 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18860 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18861 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18862 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18863 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18864 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18865 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18866 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18867 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18868 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18869 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18870 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18871 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18872 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18873 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18874 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18875 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18876 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18877 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18878 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18879 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18880 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18881 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18882 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18883 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18884 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18885 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18886 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18887 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18888 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18889 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18890 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18891 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18892 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18893 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18894 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18895 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18896 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18898 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18899 flag to second_arg_count, handle 4 argument function type _COUNT
18900 aliases, handle second_arg_count on second argument rather than last.
18902 2017-04-10 Jeff Law <law@redhat.com>
18904 PR tree-optimization/80374
18905 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18906 record anything if we can not convert integer_zero_node to the
18909 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18912 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18913 Enhance special handling given to the TARGET_P9_MINMAX option in
18914 relation to certain other options.
18916 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18918 PR tree-optimization/80153
18919 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18920 remove POINTER_PLUS_EXPR's base part directly, rather than through
18923 2017-04-10 Richard Biener <rguenther@suse.de>
18924 Bin Cheng <bin.cheng@arm.com>
18926 PR tree-optimization/80153
18927 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18928 the first element of pointer type aff_tree. Build result expr in
18930 (add_elt_to_tree): Convert to type unconditionally. Remove other
18931 fold_convert calls.
18932 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18933 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18935 2017-04-10 Richard Biener <rguenther@suse.de>
18937 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18940 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18942 PR rtl-optimization/70478
18943 * lra-constraints.c (curr_small_class_check): New.
18944 (update_and_check_small_class_inputs): New.
18945 (process_alt_operands): Update curr_small_class_check. Disfavor
18946 alternative insn memory operands. Check available regs for small
18949 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18952 * config/mips/mips.opt (-mvirt): Update description.
18953 * doc/invoke.texi (-mvirt): Likewise.
18955 2017-04-10 Richard Biener <rguenther@suse.de>
18957 PR middle-end/80362
18958 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18959 looking for NEGATE_EXPR in -A / -B to A / B folding.
18961 2017-04-10 Martin Liska <mliska@suse.cz>
18963 PR gcov-profile/80224
18964 * gcov.c (print_usage): Fix usage string.
18965 (get_gcov_intermediate_filename): Remove.
18966 (output_gcov_file): Use both for normal and intermediate format.
18967 (generate_results): Do not initialize special file for
18968 intermediate format.
18970 2017-04-10 Richard Biener <rguenther@suse.de>
18972 PR tree-optimization/80304
18973 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18976 2017-04-10 Nathan Sidwell <nathan@acm.org>
18979 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18980 (rs6000_init_builtins): Use it.
18982 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18984 * config/arm/arm.md (<mrc>): Add mode to SET source.
18985 (<mrrc>): Likewise.
18987 2017-04-10 Richard Biener <rguenther@suse.de>
18989 PR middle-end/80344
18990 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18992 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18995 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18996 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18997 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18998 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18999 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19000 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19001 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19002 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19003 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19004 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19005 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19006 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19007 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19008 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19009 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19010 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19011 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19012 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19013 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19014 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19015 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19016 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19017 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19019 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19021 PR rtl-optimization/70478
19022 * lra-constraints.c: Reverse the last patch.
19024 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19026 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19027 Add comment for WCHAR_T.
19029 2017-04-08 Martin Liska <mliska@suse.cz>
19032 2017-04-07 Martin Liska <mliska@suse.cz>
19035 * ipa-split.c (split_function): Add function part to a same comdat
19038 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19041 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19043 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19045 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19046 (vec_pairing): ...to this.
19047 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19048 (rs6000_sched_init): Adjust for name change.
19049 (struct rs6000_sched_context): Likewise.
19050 (rs6000_init_sched_context): Likewise.
19051 (rs6000_set_sched_context): Likewise.
19053 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19059 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19061 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19062 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19063 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19065 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19067 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19069 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19071 PR rtl-optimization/70703
19072 * ira-color.c (update_conflict_hard_regno_costs): Use
19073 int64_t instead of HOST_WIDE_INT.
19075 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19077 PR rtl-optimization/70478
19078 * lra-constraints.c (process_alt_operands): Disfavor alternative
19079 insn memory operands.
19081 2017-04-07 Jeff Law <law@redhat.com>
19083 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19084 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19086 2017-04-07 Martin Liska <mliska@suse.cz>
19089 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19090 Show error message instead of an ICE.
19092 2017-04-07 Martin Liska <mliska@suse.cz>
19095 * ipa-split.c (split_function): Add function part to a same comdat
19098 2017-04-07 Richard Biener <rguenther@suse.de>
19100 PR middle-end/80341
19101 * tree.c (get_unwidened): Also handle ! for_type case for
19103 * convert.c (do_narrow): Split out from ...
19104 (convert_to_integer_1): ... here. Do not pass final truncation
19105 type to get_unwidened for TRUNC_DIV_EXPR.
19107 2017-04-07 Richard Biener <rguenther@suse.de>
19109 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19111 (aff_combination_const): Adjust.
19112 (aff_combination_scale): Likewise.
19113 (aff_combination_add_elt): Likewise.
19114 (aff_combination_add_cst): Likewise.
19115 (aff_combination_convert): Likewise.
19116 (add_elt_to_tree): Likewise. Remove unused argument.
19117 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19119 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19121 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19123 * config/arm/arm.c (arm_default_short_enums): Use
19124 ARM_DEFAULT_SHORT_ENUMS.
19125 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19127 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19130 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19131 members with redundant out-of-class redeclaration.
19133 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19136 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19137 * config/i386/i386.md (*zero_extendsidi2):
19138 Add (?*x,*x) and (?*v,*v) alternatives.
19140 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19143 * config/i386/i386.c (ix86_expand_builtin)
19144 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19145 mode from insn data. Convert operands to insn operand mode.
19146 Copy operands that don't satisfy insn predicate to a register.
19148 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19150 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19153 2017-04-06 Richard Biener <rguenther@suse.de>
19155 PR tree-optimization/80334
19156 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19157 preserve alignment of accesses.
19159 2017-04-06 Richard Biener <rguenther@suse.de>
19161 PR tree-optimization/80262
19162 * tree-sra.c (build_ref_for_offset): Preserve address-space
19164 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19165 Drop useless address-space information on MEM_REF offsets.
19167 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19169 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19171 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19173 PR rtl-optimization/70703
19174 * ira-color.c (update_conflict_hard_regno_costs): Use
19175 HOST_WIDE_INT instead of long.
19177 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19180 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19181 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19183 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19184 for x86_64 target. Handle -m3dnowa option.
19186 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19188 PR rtl-optimization/70703
19189 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19190 (update_conflict_hard_regno_costs): Use long instead of unsigned
19191 arithmetic for cost calculation.
19193 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19194 Bernd Edlinger <bernd.edlinger@hotmail.de>
19197 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19200 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19203 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19204 ptr_mode with Pmode throughout.
19205 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19206 into probe_stack_range and use DImode.
19208 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19211 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19212 call_eh_return is true.
19214 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19216 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19217 Initialize last_match_fntype_index.
19219 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19222 * tree-nvr.c: Include internal-fn.h.
19223 (pass_return_slot::execute): Ignore internal calls without
19226 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19227 Richard Biener <rguenther@suse.de>
19230 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19231 captures used multiple times, except for the last use.
19232 * generic-match-head.c: Include gimplify.h.
19234 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19236 PR tree-optimization/79390
19237 * target.h (struct noce_if_info): Declare.
19238 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19239 * target.def (noce_conversion_profitable_p): New target hook.
19240 * ifcvt.h (struct noce_if_info): New type, moved from ...
19241 * ifcvt.c (struct noce_if_info): ... here.
19242 (noce_conversion_profitable_p): Renamed to ...
19243 (default_noce_conversion_profitable_p): ... this. No longer
19245 (noce_try_store_flag_constants, noce_try_addcc,
19246 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19247 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19248 instead of noce_conversion_profitable_p.
19249 * config/i386/i386.c: Include ifcvt.h.
19250 (ix86_option_override_internal): Don't override
19251 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19252 (ix86_noce_conversion_profitable_p): New function.
19253 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19254 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19255 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19256 * doc/tm.texi: Regenerated.
19258 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19260 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19263 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19266 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19267 instructions for small multiply cores.
19269 2017-04-04 Jeff Law <law@redhat.com>
19271 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19273 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19274 that are not set by the loop over the elements.
19276 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19279 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19280 int mode, convert_modes it to mode as unsigned, otherwise use
19281 lowpart_subreg to mode rather than SImode.
19282 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19283 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19284 Use DImode instead of SImode for the shift count operand.
19285 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19288 2017-04-04 Richard Biener <rguenther@suse.de>
19290 PR middle-end/80281
19291 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19292 arithmetic done for the negate or the plus. Simplify.
19293 (A - (-B) -> A + B): Likewise.
19294 * fold-const.c (split_tree): Make sure to not negate pointers.
19296 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19298 PR rtl-optimization/60818
19299 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19300 a compare of comparisons with the thing compared if this results
19301 in a different machine mode.
19303 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19305 * alias.c (base_alias_check): Fix typo in comment.
19306 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19307 * cgraphunit.c (symbol_table::compile): Likewise.
19308 * collect2.c (maybe_run_lto_and_relink): Likewise.
19309 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19310 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19311 * config/avr/avr.c (avr_map_op_t): Likewise.
19312 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19313 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19314 * config/epiphany/epiphany.md (movcc): Likewise.
19315 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19316 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19318 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19319 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19320 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19321 * config/sh/sh.c (sh_rtx_costs): Likewise.
19322 * fold-const.c (fold_truth_andor): Likewise.
19323 * genautomata.c (collapse_flag): Likewise.
19324 * gengtype.h (struct type::u::s): Likewise.
19325 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19326 * input.c (FORMAT_AMOUNT): Likewise.
19327 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19328 (known_aggs_to_agg_replacement_list): Likewise.
19329 * ipa-inline-analysis.c: Likewise.
19330 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19331 * ipa-polymorphic-call.c
19332 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19333 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19334 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19336 * modulo-sched.c (apply_reg_moves): Likewise.
19337 * omp-expand.c (build_omp_regions_1): Likewise.
19338 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19339 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19340 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19341 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19342 * value-prof.c: Likewise.
19343 * var-tracking.c (val_reset): Likewise.
19345 2017-04-03 Richard Biener <rguenther@suse.de>
19347 PR tree-optimization/80275
19348 * fold-const.c (split_address_to_core_and_offset): Handle
19351 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19353 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19354 descriptors is at least equal to that of functions.
19356 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19358 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19360 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19363 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19364 (mov<IMOD4:mode>): New expander.
19365 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19367 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19369 PR rtl-optimization/79405
19370 * fwprop.c (propagations_left): New variable.
19371 (forward_propagate_into): Decrement it.
19372 (fwprop_init): Initialize it.
19373 (fw_prop): If the variable has reached zero, stop propagating.
19374 (fwprop_addr): Ditto.
19376 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19379 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19380 a FUNCTION_DECL, pass it as decl instead of origin to
19383 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19385 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19388 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19391 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19392 TARGET_VSX_SMALL_INTEGER.
19394 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19396 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19397 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19399 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19401 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19402 extraction from odd-numbered MSA register.
19404 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19406 PR middle-end/80173
19407 * expmed.c (store_bit_field_1): Don't attempt to create
19408 a word subreg out of hard registers wider than word if they
19409 have HARD_REGNO_NREGS of 1 for their mode.
19411 PR middle-end/80163
19412 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19413 conversions to integer types wider than word and pointer.
19416 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19417 (rtx_equal_for_cselib_p): Pass 0 to it.
19418 * cselib.c (cselib_hasher::equal): Likewise.
19419 (rtx_equal_for_cselib_1): Add depth argument. If depth
19420 is 128, don't look up VALUE locs and punt. Increment
19421 depth in recursive calls when walking VALUE locs.
19423 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19425 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19426 (make_gcov_file_name): Use the canonical path name for generating
19428 (read_line): Fix handling of files with ascii null bytes.
19430 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19432 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19433 to initialise a vector register instead
19434 of using a const_int.
19436 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19438 PR translation/80189
19439 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19440 diagnostic messages.
19442 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19445 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19446 (dfp_diex_<mode>): Update mode of operand 1.
19447 * doc/extend.texi (dxex, dxexq): Document change to return type.
19448 (diex, diexq): Document change to argument type.
19450 2017-03-30 Martin Jambor <mjambor@suse.cz>
19453 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19454 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19455 it reflects the signature changes performed at the callee side.
19456 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19457 to cgraph_build_function_type_skip_args.
19458 (build_function_decl_skip_args): Adjust call to the above function.
19460 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19463 * config/i386/sse.md
19464 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19465 register as dest whenever it is a MEM not rtx_equal_p to the
19466 corresponding dup operand, and when forcing into reg move the
19467 reg into the memory afterwards.
19468 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19469 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19470 for the force_reg mode.
19471 (avx512vl_vextractf128<mode>): Use register as dest either
19472 always when a MEM, or when it is a MEM not rtx_equal_p to the
19473 corresponding dup operand, or even not when it is a CONST_VECTOR
19474 depending on the mode and lo vs. hi.
19475 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19477 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19478 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19479 Likewise. Require that operands[2] is even.
19480 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19481 Remove extraneous parens. Require that operands[2] is a multiple
19483 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19484 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19485 disallow memory then.
19487 2017-03-30 Richard Biener <rguenther@suse.de>
19489 PR tree-optimization/77498
19490 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19491 to non-constants over backedges.
19493 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19495 PR rtl-optimization/80233
19496 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19497 as last_combined_insn. Do not test for BARRIER_P separately.
19499 2017-03-29 Andreas Schwab <schwab@suse.de>
19502 * calls.c (prepare_call_address): Convert funexp to Pmode before
19503 copying to temp reg.
19505 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19507 PR tree-optimization/80158
19508 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19509 Handle possible future case of more than one alternate
19511 (replace_rhs_if_not_dup): Likewise.
19512 (replace_one_candidate): Likewise.
19514 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19516 PR rtl-optimization/80193
19517 * ira.c (ira): Do not check allocation for LRA.
19519 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19521 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19522 (nvptx_output_simt_exit): Declare.
19523 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19524 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19525 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19526 (nvptx_declare_function_name): ...here. Emit declaration of local
19527 memory space buffer for omp_simt_enter insn.
19528 (nvptx_output_unisimt_switch): New.
19529 (nvptx_output_softstack_switch): New.
19530 (nvptx_output_simt_enter): New.
19531 (nvptx_output_simt_exit): New.
19532 * config/nvptx/nvptx.h (struct machine_function): New fields
19533 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19534 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19535 (UNSPECV_SIMT_EXIT): Ditto.
19536 (omp_simt_enter_insn): New insn.
19537 (omp_simt_enter): New expansion.
19538 (omp_simt_exit): New insn.
19539 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19541 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19542 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19543 (expand_GOMP_SIMT_EXIT): New.
19544 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19545 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19546 (GOMP_SIMT_EXIT): Ditto.
19547 * target-insns.def (omp_simt_enter): New insn.
19548 (omp_simt_exit): Ditto.
19549 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19551 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19552 (lower_rec_input_clauses): Likewise.
19553 (lower_lastprivate_clauses): Handle SIMT privatization.
19555 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19556 (ompdevlow_adjust_simt_enter): New.
19557 (find_simtpriv_var_op): New.
19558 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19559 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19561 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19562 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19563 (copy_decl_for_dup_finish): Ditto.
19565 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
19567 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
19570 * config/i386/i386.c (ix86_option_override_internal): Always
19571 allow -mpreferred-stack-boundary=3 for 64-bit targets.
19573 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19575 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
19577 2017-03-28 Bin Cheng <bin.cheng@arm.com>
19579 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
19580 mark new edge's irreducible flag accordign to it.
19581 (vect_do_peeling): Check loop preheader edge's irreducible flag
19582 and pass it to function slpeel_add_loop_guard.
19584 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
19586 PR tree-optimization/80218
19587 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19588 Update block frequencies and counts.
19590 2017-03-28 Richard Biener <rguenther@suse.de>
19592 PR tree-optimization/78644
19593 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
19594 of a simplification result we may not use it at all.
19596 2017-03-28 Richard Biener <rguenther@suse.de>
19599 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
19600 without arguments, generate default definition of a SSA name.
19602 2017-03-28 Richard Biener <rguenther@suse.de>
19604 PR middle-end/80222
19605 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19606 TYPE_REF_CAN_ALIAS_ALL references.
19607 * fold-const.c (fold_indirect_ref_1): Likewise.
19609 2017-03-28 Martin Liska <mliska@suse.cz>
19612 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19613 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19615 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19616 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19618 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19619 (EXTRA_SPECS): Define.
19620 (SUBTARGET_EXTRA_SPECS): Likewise.
19621 (SUBTARGET_CPP_SPEC): Likewise.
19622 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19623 SUBTARGET_EXTRA_SPECS.
19624 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19626 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19628 * config/arc/simdext.md (vst64_insn): Update pattern.
19629 (vld32wh_insn): Likewise.
19630 (vld32wl_insn): Likewise.
19631 (vld64_insn): Likewise.
19632 (vld32_insn): Likewise.
19634 2017-03-28 Marek Polacek <polacek@redhat.com>
19637 * fold-const.c (fold_comparison): Use protected_set_expr_location
19638 instead of SET_EXPR_LOCATION.
19640 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19642 * tree.c (add_expr): Avoid name lookup warning.
19644 2017-03-27 Jeff Law <law@redhat.com>
19646 PR tree-optimization/80216
19647 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19648 function name. Limit recursion depth.
19649 (record_temporary_equivalences): Corresponding changes.
19651 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19653 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19656 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19659 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19661 * cfgcleanup.c (reg_note_cfa_p): New array.
19662 (insns_have_identical_cfa_notes): New function.
19663 (old_insns_match_p): Don't cross-jump in between /f
19664 and non-/f instructions. If both i1 and i2 are frame related,
19665 verify all CFA notes, their order and content.
19667 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19670 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19671 HImode and SImode with zero extend to DImode to one insn.
19672 (bswap<mode>2_extenddi): Likewise.
19673 (bswapsi2_extenddi): Likewise.
19674 (bswaphi2_extendsi): Likewise.
19675 (bswaphi2): Combine bswap HImode and SImode into one insn.
19676 Separate memory insns from swapping register.
19677 (bswapsi2): Likewise.
19678 (bswap<mode>2): Likewise.
19679 (bswaphi2_internal): Delete, no longer used.
19680 (bswapsi2_internal): Likewise.
19681 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19682 store, and gpr<-gpr swap insns.
19683 (bswap<mode>2_store): Likewise.
19684 (bswaphi2_reg): Register only splitter, combine with the splitter.
19685 (bswaphi2 splitter): Likewise.
19686 (bswapsi2_reg): Likewise.
19687 (bswapsi2 splitter): Likewise.
19688 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19689 the insns into load, store, and register/register insns.
19690 (bswapdi2_ldbrx): Likewise.
19691 (bswapdi2_load): Likewise.
19692 (bswapdi2_store): Likewise.
19693 (bswapdi2_reg): Likewise.
19695 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19697 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19698 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19700 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19703 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19706 special handling for target option conflicts between dform
19707 options (-mpower9-dform, -mpower9-dform-vector,
19708 -mpower9-dform-scalar) and -mno-direct-move.
19710 2017-03-27 Richard Biener <rguenther@suse.de>
19712 PR tree-optimization/80181
19713 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19715 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19717 * config/arc/predicates.md (move_double_src_operand): Replace the
19718 call to move_double_src_operand with a call to address_operand.
19720 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19722 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19723 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19724 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19726 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19728 * config/arc/predicates.md (long_immediate_loadstore_operand):
19729 Consider scaled addresses cases.
19731 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19733 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19734 restored when in interrupt.
19735 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19736 doesn't have delay slot.
19738 2017-03-27 Richard Biener <rguenther@suse.de>
19741 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19742 inlined thunk clones.
19744 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19747 * asan.c (instrument_derefs): Copy over last operand from
19748 original COMPONENT_REF to the new COMPONENT_REF with
19749 DECL_BIT_FIELD_REPRESENTATIVE.
19750 * ubsan.c (instrument_object_size): Likewise.
19752 2017-03-27 Richard Biener <rguenther@suse.de>
19754 PR tree-optimization/80170
19755 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19756 sure DR/SCEV didnt fold in constants we do not see when looking
19757 at the reference base alignment.
19759 2017-03-27 Richard Biener <rguenther@suse.de>
19761 PR middle-end/80171
19762 * gimple-fold.c (fold_ctor_reference): Properly guard against
19763 NULL return value from canonicalize_constructor_val.
19765 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19768 * config/i386/i386.c (ix86_expand_builtin)
19769 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19770 flags reg setting and flags reg using instructions.
19771 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19772 clobbering instructions to zero extend op2.
19774 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19776 * doc/install.texi (Configuration) <--with-aix-soname>:
19777 Update link to AIX ld.
19779 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19781 PR rtl-optimization/80160
19782 PR rtl-optimization/80159
19783 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19784 reg_alternate_class into account.
19786 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19789 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19790 to consider in curr_insn_transform.
19792 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19794 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19795 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19796 and emit_mode_inner.
19798 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19800 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19801 argument to the overloaded builtin variants. Use the new flag to
19802 deprecate certain builtin variants.
19803 * config/s390/s390-builtin-types.def: Add new builtin types.
19804 * config/s390/s390-builtins.h: Support new flags field for
19805 overloaded builtins.
19806 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19807 (s390_macro_to_expand): Enable vector float data type.
19808 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19809 builtins by incrementing the __VEC__ version number.
19810 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19812 (s390_resolve_overloaded_builtin): Emit error messages depending
19813 on the builtin flags.
19814 * config/s390/s390.c (s390_expand_builtin): Support additional
19815 flags argument. Change error message to match the messages
19816 emitted in s390-c.c.
19817 * config/s390/s390.md: New UNSPEC_* constants.
19818 (op_type): Add new instruction types.
19819 * config/s390/vecintrin.h: Add new builtins and test data class
19821 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19822 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19823 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19824 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19825 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19826 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19828 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19829 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19830 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19831 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19833 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19834 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19835 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19836 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19838 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19839 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19840 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19841 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19842 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19843 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19844 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19846 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19847 ("vec_scatter_element<V_HW_4:mode>_DI")
19848 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19849 ("vec_fpint<mode>", "vflls")
19850 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19851 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19852 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19853 ("*vec_cmphe<mode>_cc"): ... these.
19855 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19856 mode constant instead of magic value.
19858 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19860 * config/s390/s390.c (s390_expand_vec_compare): Support other
19861 vector floating point modes than just V2DF.
19862 (s390_expand_vcond): Likewise.
19863 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19864 (s390_cannot_change_mode_class): Prevent mode changes between TF
19865 and V1TF in vector registers.
19866 * config/s390/s390.md (DF, SF): New mode attributes.
19867 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19868 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19869 SFmode support for VRs.
19870 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19872 (VFT, VF_HW): New mode iterators.
19873 (vw, sdx): New mode attributes.
19874 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19875 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19876 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19877 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19878 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19879 also the new vector floating point modes. Renaming to ...
19881 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19882 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19883 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19884 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19885 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19886 ("vec_unordered<mode>"): ... these.
19888 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19889 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19890 ("*vec_extendv2df"): New insn definitions.
19892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19894 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19895 ("mulditi3_2", "*muldi3_sign"): New patterns.
19896 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19897 rename the pattern definition.
19899 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19901 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19903 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19907 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19908 instruction if possible.
19909 * config/s390/vector.md (vec_halfnumelts): New mode
19911 ("*vec_vllezlf<mode>"): New pattern.
19913 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19915 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19916 ("popcountv4si2", "popcountv2di2"): Rename to ...
19917 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19918 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19920 ("popcount<mode>2_vxe"): New pattern.
19922 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19924 * common/config/s390/s390-common.c (processor_flags_table): Add
19926 * config.gcc: Add arch12.
19927 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19928 Default to arch12 for unknown CPU model numbers.
19929 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19930 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19931 PROCESSOR_max sanity check.
19932 * config/s390/s390-opts.h (enum processor_type): Add
19934 * config/s390/s390.c (processor_table): Add arch12.
19935 (s390_expand_builtin): Add check for B_VXE flag.
19936 (s390_issue_rate): Add PROCESSOR_ARCH12.
19937 (s390_get_sched_attrmask): Likewise.
19938 (s390_get_unit_mask): Likewise.
19939 (s390_sched_score): Enable z13 scheduling for arch12.
19940 (s390_sched_reorder): Likewise.
19941 (s390_sched_variable_issue): Likewise.
19942 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19944 (s390_tune_attr): Use z13 scheduling also for arch12.
19945 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19946 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19947 (TARGET_VXE_P): New macros.
19948 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19949 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19950 * config/s390/s390.opt: Add arch12 as processor_type.
19952 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19954 * config/s390/s390.md
19955 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19956 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19957 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19959 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19960 Rename expanders to ...
19962 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19963 ("fixuns_truncdddi2_emu"): ... these.
19965 ("fixuns_trunc<mode>si2_emu"): New expander.
19967 ("*fixuns_truncdfdi2_z13"): Rename to ...
19968 ("*fixuns_truncdfdi2_vx"): ... this.
19970 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19972 * config/s390/2964.md: Remove the single element vector compare
19973 instructions which are no longer used.
19974 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19976 (s390_canonicalize_comparison): Remove handling of DFmode
19978 (s390_expand_vec_compare_scalar): Remove function.
19979 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19980 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19982 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19986 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19988 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19989 will anyway by matched by mov<mode>_64dfp.
19991 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19993 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19994 vlef/vstef. Add missing operand to vleif.
19996 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19998 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19999 pair for all vector types with 64 bit elements.
20000 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20001 * config/s390/vector.md (V_HW_64): ... here.
20002 (V_128_NOSINGLE): New mode iterator.
20003 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20004 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20005 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20006 ("*vec_load_pairv2di"): Change to ...
20007 ("*vec_load_pair<mode>"): ... this one.
20009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20011 * config/s390/constraints.md: Add comments.
20012 (jKK): Reject element sizes > 8 bytes.
20013 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20015 * config/s390/s390.md: Add the s_operand checks formerly in
20016 s390_split_ok_p to various splitters where they are still
20018 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20019 for 128 bit vectors. Plus two splitters.
20021 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20023 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20026 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20029 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20030 error if the boundary argument is not constant.
20032 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20034 PR rtl-optimization/80112
20035 * loop-doloop.c (doloop_condition_get): Don't check condition
20036 if cmp isn't SET with IF_THEN_ELSE src.
20038 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20040 PR tree-optimization/80158
20041 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20042 replacing a candidate statement, also replace it for the
20043 candidate's alternate interpretation.
20044 (replace_rhs_if_not_dup): Likewise.
20045 (replace_one_candidate): Likewise.
20047 2017-03-24 Richard Biener <rguenther@suse.de>
20049 PR tree-optimization/80167
20050 * graphite-isl-ast-to-gimple.c
20051 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20053 (translate_isl_ast_to_gimple::get_rename): Likewise.
20055 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20058 handling of certain combinations of target options, including the
20059 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20060 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20062 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20065 * config/arm/arm.md (*load_multiple): Add reload_completed to
20066 matching condition.
20068 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20069 Richard Biener <rguenth@suse.de>
20071 PR tree-optimization/79908
20072 PR tree-optimization/80136
20073 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20074 been cast away, gimplify_and_add suffices.
20076 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20078 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20080 2017-03-23 Richard Biener <rguenther@suse.de>
20082 PR tree-optimization/80032
20083 * gimplify.c (gimple_push_cleanup): Forced unconditional
20084 cleanups still have to go to the conditional_cleanups
20087 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20089 PR tree-optimization/80072
20090 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20092 (next_operand_entry_id): Change type to unsigned int.
20093 (sort_by_operand_rank): Make sure to return the right return value
20094 even if unsigned fields are bigger than INT_MAX.
20095 (struct oecount): Change cnt and id type to unsigned int.
20096 (oecount_hasher::equal): Formatting fix.
20097 (oecount_cmp): Make sure to return the right return value
20098 even if unsigned fields are bigger than INT_MAX.
20099 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20102 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20103 TREE_READONLY on result if writing it more than once.
20106 * doc/invoke.texi (-fsanitize=thread): Document that with
20107 -fnon-call-exceptions atomics are not able to throw
20111 * tsan.c: Include tree-eh.h.
20112 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20113 maybe_clean_or_replace_eh_stmt where needed.
20114 (instrument_memory_accesses): Add cfg_changed argument.
20115 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20116 if it returned true.
20117 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20119 PR rtl-optimization/63191
20120 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20121 wrapper function, moved the whole old content into ...
20122 (ix86_delegitimize_address_1): ... this. New inline function.
20123 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20124 true as last argument instead of ix86_delegitimize_address.
20126 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20128 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20129 cortexa57_branch_cost.
20131 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20133 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20135 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20138 * doc/md.texi (Constraints): Document wA constraint.
20139 * config/rs6000/constraints.md (wA): New.
20140 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20141 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20142 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20143 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20145 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20148 * gimplify.c (is_oacc_declared): New function.
20149 (oacc_default_clause): Use it to set default flags for acc declared
20150 variables inside parallel regions.
20151 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20152 declared variables.
20153 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20154 declare attribute to any decl as necessary.
20156 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20159 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20160 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20161 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20162 (arm_arch_lpae): This.
20163 * config/arm/arm.c (arm_arch7ve): Rename into ...
20164 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20165 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20168 2017-03-22 Martin Liska <mliska@suse.cz>
20171 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20172 error message instead of an ICE.
20174 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20176 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20178 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20180 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20182 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20185 2017-03-21 Martin Sebor <msebor@redhat.com>
20187 * doc/extend.texi: Use "cannot" instead of "can't."
20188 * doc/hostconfig.texi: Same.
20189 * doc/install.texi: Same.
20190 * doc/invoke.texi: Same.
20191 * doc/loop.texi: Same.
20192 * doc/md.texi: Same.
20193 * doc/objc.texi: Same.
20194 * doc/rtl.texi: Same.
20195 * doc/tm.texi: Same.
20196 * doc/tm.texi.in: Same.
20197 * doc/trouble.texi: Same.
20199 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20202 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20203 (collect_checksum_attributes): Set it.
20204 (die_checksum_ordered): Use it.
20206 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20208 PR tree-optimization/79908
20209 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20210 change: For a VA_ARG whose LHS has been cast away, use
20211 force_gimple_operand to construct the side effects.
20213 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20215 PR translation/80001
20216 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20217 more amenable to translation.
20218 (oacc_loop_auto_partitions): Likewise.
20220 2017-03-21 Marek Polacek <polacek@redhat.com>
20221 Martin Sebor <msebor@redhat.com>
20223 PR tree-optimization/80109
20224 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20225 on INTEGRAL_TYPE_P.
20227 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20228 Segher Boessenkool <segher@kernel.crashing.org>
20231 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20232 check reg_used_between_p between insn and one of succ or succ2
20233 depending on if succ is artificial insn not inserted into insn
20236 2017-03-21 Martin Liska <mliska@suse.cz>
20238 PR gcov-profile/80081
20239 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20240 * doc/gcc.texi: Include gcov-dump stuff.
20241 * doc/gcov-dump.texi: New file.
20243 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20245 PR rtl-optimization/79150
20246 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20247 conditional jump, if the jump is the last insn of the loop.
20249 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20250 Richard Biener <rguenth@suse.de>
20252 PR tree-optimization/79908
20253 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20254 been cast away, use force_gimple_operand to construct the side
20257 2017-03-21 Martin Liska <mliska@suse.cz>
20259 PR libfortran/79956
20260 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20263 2017-03-21 Brad Spengler <spender@grsecurity.net>
20266 * plugin.c (htab_hash_plugin): New function.
20267 (add_new_plugin): Use it and adjust.
20268 (parse_plugin_arg_opt): Adjust.
20269 (init_one_plugin): Likewise.
20271 2017-03-21 Richard Biener <rguenther@suse.de>
20273 PR tree-optimization/80032
20274 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20275 if set force the cleanup to happen unconditionally.
20276 (gimplify_target_expr): Push inserted clobbers with force_uncond
20277 to avoid them being removed by control-dependent DCE.
20279 2017-03-21 Richard Biener <rguenther@suse.de>
20281 PR tree-optimization/80122
20282 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20283 va_arg_pack_len when the inlined call stmt requires pack
20285 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20287 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20290 * tsan.c (instrument_builtin_call): If the memory model argument
20291 is not a constant, assume it is valid.
20294 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20297 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20299 PR rtl-optimization/79910
20300 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20301 if its dest is used by an insn before I2 (other than the combined
20302 insns themselves, which are properly handled already).
20304 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20307 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20309 * combine.c (record_used_regs): New static function.
20310 (try_combine): Handle situations where there is an additional
20311 instruction between I2 and I3 which needs to have a LOG_LINK
20315 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20317 * combine.c (try_combine): Delete redundant i1 test. Call
20318 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20320 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20323 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20324 alternatives 13/14.
20326 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20328 PR tree-optimization/80054
20329 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20330 the optimization if a PHI or any of its arguments is not dominated
20331 by the candidate's basis. Use gphi* rather than gimple* as
20333 (replace_profitable_candidates): Clean up a gimple* variable that
20334 should be a gphi* variable.
20336 2017-03-20 Martin Sebor <msebor@redhat.com>
20339 * doc/extend.texi (attribute constructor): Document present limitation.
20341 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20344 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20345 __POWER9_VECTOR__ #ifdef control, change template definition to
20346 use Power9-specific built-in function.
20347 (vec_any_eq): Likewise.
20348 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20349 to control outcomes from this test.
20350 (vector_ae_<mode>p): For VEC_F modes, likewise.
20352 2017-03-20 Ian Lance Taylor <iant@google.com>
20354 * config/i386/i386.c (ix86_function_regparm): Save an extra
20355 register for -fsplit-stack with DECL_STATIC_CHAIN.
20357 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20360 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20361 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20363 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20365 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20367 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20370 2017-03-20 Marek Polacek <polacek@redhat.com>
20373 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20375 2017-03-20 Richard Biener <rguenther@suse.de>
20377 PR tree-optimization/80113
20378 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20379 allocate extra SSA name for PHI def.
20380 (add_close_phis_to_outer_loops): Likewise.
20381 (add_close_phis_to_merge_points): Likewise.
20382 (copy_loop_close_phi_args): Likewise.
20383 (copy_cond_phi_nodes): Likewise.
20385 2017-03-20 Martin Liska <mliska@suse.cz>
20387 PR middle-end/79753
20388 * tree-chkp.c (chkp_build_returned_bound): Do not build
20389 returned bounds for a LHS that's not a BOUNDED_P type.
20391 2017-03-20 Martin Liska <mliska@suse.cz>
20395 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20396 COMPLEX_CST and VECTOR_CST.
20398 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20401 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20402 target operand. A new splitter adds the clobber statement in case
20403 the target operand is dead anyway.
20405 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20407 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20408 to age-old versions of binutils and glibc.
20410 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20412 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20414 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20416 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20418 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20420 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20421 requirement for binutils 2.13.
20423 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20425 * combine.c (try_combine): Delete redundant i1 test. Call
20426 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20428 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20430 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20431 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20433 <riscv64-*-elf>: Re-arrange section
20434 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20435 <riscv32-*-linux>: Likewise.
20436 <riscv64-*-elf>: Likewise
20437 <riscv64-*-linux>: Likewise.
20439 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20442 * aarch64.opt(verbose-cost-dump): Fix typo.
20444 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20447 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20448 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20450 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20452 * reload.c (find_reloads): When reloading a nonoffsettable address,
20453 use RELOAD_OTHER for it and its address reloads.
20455 PR rtl-optimization/79910
20456 * combine.c (record_used_regs): New static function.
20457 (try_combine): Handle situations where there is an additional
20458 instruction between I2 and I3 which needs to have a LOG_LINK
20461 2017-03-17 Jeff Law <law@redhat.com>
20463 PR tree-optimization/71437
20464 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20465 conditional in the hash table first.
20466 (vrp_dom_walker::before_dom_children): Extract condition from
20467 ASSERT_EXPR. Record condition, its inverion and any implied
20468 conditions as well.
20470 2017-03-17 Marek Polacek <polacek@redhat.com>
20471 Markus Trippelsdorf <markus@trippelsdorf.de>
20473 PR tree-optimization/80079
20474 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20477 2017-03-17 Richard Biener <rguenther@suse.de>
20479 PR middle-end/80075
20480 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20481 Properly verify the LHS before the RHS possibly claims to be
20483 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20486 2017-03-17 Martin Jambor <mjambor@suse.cz>
20488 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20489 (List of -O2 options): Likewise.
20490 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20493 2017-03-17 Tom de Vries <tom@codesourcery.com>
20495 * gcov-dump.c (print_usage): Print bug_report_url.
20497 2017-03-17 Richard Biener <rguenther@suse.de>
20499 PR middle-end/80050
20500 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20501 (parser::peek): Likewise.
20503 2017-03-17 Richard Biener <rguenther@suse.de>
20505 PR tree-optimization/80048
20506 * sese.c (free_sese_info): Properly release rename_map and
20507 copied_bb_map elements.
20509 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20511 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20512 Add linked-list forward and backlinks. Insert on
20513 construction, remove on destruction.
20514 (class pass_store_merging): Add m_stores_head field.
20515 (pass_store_merging::terminate_and_process_all_chains):
20516 Iterate over m_stores_head list.
20517 (pass_store_merging::terminate_all_aliasing_chains):
20519 (pass_store_merging::execute): Check for debug stmts first.
20520 Push new chains onto the m_stores_head stack.
20522 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20525 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20526 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20527 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20529 2017-03-16 Jeff Law <law@redhat.com>
20531 PR tree-optimization/71437
20532 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20533 member function. Implementation moved into after_dom_children
20534 member function and into the threader's thread_outgoing_edges
20536 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20537 some code into new thread_outgoing_edges.
20538 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20539 definition. Simplify marker handling (do it here). Assume we always
20540 have the available expression and the const/copies tables.
20541 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20543 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20544 * tree-vrp.c (equiv_stack): No longer file scoped.
20545 (vrp_dom_walker): New class.
20546 (vrp_dom_walker::before_dom_children): New member function.
20547 (vrp_dom_walker::after_dom_children): Likewise.
20548 (identify_jump_threads): Setup domwalker. Use it rather than
20549 walking edges in a random order by hand. Simplify setup/finalization.
20550 (finalize_jump_threads): Remove.
20551 (vrp_finalize): Do not call identify_jump_threads here.
20552 (execute_vrp): Do it here instead and call thread_through_all_blocks
20555 PR tree-optimization/71437
20556 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20558 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20560 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20561 (dom_opt_dom_walker::thread_across_edge): Remove
20562 handle_dominating_asserts argument. All callers changed.
20563 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20564 changes. Remove calls to lhs_of_dominating_assert. Other
20565 uses of handle_dominating_asserts turn into unconditional code
20566 (simplify_control_stmt_condition_1): Likewise.
20567 (simplify_control_stmt_condition): Likewise.
20568 (thread_through_normal_block, thread_across_edge): Likewise.
20569 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
20570 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
20571 object if it is not an SSA_NAME.
20572 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
20573 before calling into the VRP specific simplifiers.
20574 (identify_jump_threads): Remove handle_dominating_asserts
20577 2017-03-16 Jakub Jelinek <jakub@redhat.com>
20580 * tree-diagnostic.c (default_tree_printer): No longer static.
20581 * tree-diagnostic.h (default_tree_printer): New prototype.
20583 2017-03-16 Tamar Christina <tamar.christina@arm.com>
20585 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
20586 Change ins into fmov.
20588 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20590 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
20591 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
20592 Use h_con constraint for operand 1.
20593 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
20594 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
20596 2017-03-15 Jeff Law <law@redhat.com>
20598 PR tree-optimization/71437
20599 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
20600 (record_temporary_equivalences): Use it.
20602 PR tree-optimization/71437
20603 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20604 tree-ssa-scopedtables.
20605 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20606 (record_conditions, record_cond, vuse_eq): Likewise.
20607 (record_edge_info): Adjust to API tweak of record_conditions.
20608 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20609 (record_temporary_equivalences, optimize_stmt): Likewise.
20610 (eliminate_redundant_computations): Likewise.
20611 (record_equivalences_from_stmt): Likewise.
20612 * tree-ssa-scopedtables.c: Include options.h and params.h.
20613 (vuse_eq): New function, moved from tree-ssa-dom.c
20614 (build_and_record_new_cond): Likewise.
20615 (record_conditions): Likewise. Accept vector of conditions rather
20616 than edge_equivalence structure for first argument.
20617 for the first argument.
20618 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20619 from tree-ssa-dom.c.
20620 (avail_exprs_stack::record_cond): Likewise.
20621 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20622 from tree-ssa-dom.c.
20623 (avail_exprs_stack): Add new member functions lookup_avail_expr
20625 (record_conditions): Declare.
20627 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20630 * lra-constraints.c (process_alt_operands): Increase reject for
20631 reloading an input/output operand.
20633 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20636 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20637 insns to convert from signed/unsigned char/short to IEEE 128-bit
20639 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20641 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20644 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20645 subreg of inner mode for values already in registers.
20647 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20649 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20650 iteration reg is used after the loop.
20652 2017-03-14 Martin Sebor <msebor@redhat.com>
20654 PR tree-optimization/79800
20655 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20656 precision in negative-positive range.
20657 (format_floating): Call non-const overload with adjusted precision.
20659 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20662 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20665 2017-03-14 Martin Sebor <msebor@redhat.com>
20667 PR middle-end/80020
20668 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20669 * builtins.def (aligned_alloc): Use it.
20672 * Makefile.in (GTFILES): Add calls.c.
20673 * calls.c: Include "gt-calls.h".
20675 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20677 PR rtl-optimization/79728
20678 * regs.h (struct target_regs): New field
20679 x_contains_allocatable_regs_of_mode.
20680 (contains_allocatable_regs_of_mode): New macro.
20681 * reginfo.c (init_reg_sets_1): Initialize it, and change
20682 contains_reg_of_mode so it includes global regs as well.
20683 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20684 rather than contains_regs_of_mode.
20686 2017-03-14 Martin Liska <mliska@suse.cz>
20688 * doc/invoke.texi: Document options that can't be combined with
20689 -fcheck-pointer-bounds.
20691 2017-03-14 Martin Liska <mliska@suse.cz>
20693 PR middle-end/79831
20694 * doc/invoke.texi (-Wchkp): Document the option.
20696 2017-03-14 Martin Liska <mliska@suse.cz>
20698 * Makefile.in: Install gcov-dump.
20700 2017-03-14 Martin Liska <mliska@suse.cz>
20702 * multiple_target.c (expand_target_clones): Bail out for
20703 an invalid attribute.
20705 2017-03-14 Richard Biener <rguenther@suse.de>
20707 * alias.c (struct alias_set_entry): Pack properly.
20708 * cfgloop.h (struct loop): Likewise.
20709 * cse.c (struct set): Likewise.
20710 * ipa-utils.c (struct searchc_env): Likewise.
20711 * loop-invariant.c (struct invariant): Likewise.
20712 * lra-remat.c (struct cand): Likewise.
20713 * recog.c (struct change_t): Likewise.
20714 * rtl.h (struct address_info): Likewise.
20715 * symbol-summary.h (function_summary): Likewise.
20716 * tree-loop-distribution.c (struct partition): Likewise.
20717 * tree-object-size.c (struct object_size_info): Likewise.
20718 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20719 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20720 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20721 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20722 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20723 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20724 (struct _stmt_vec_info): Likewise.
20726 2017-03-14 Martin Liska <mliska@suse.cz>
20729 * multiple_target.c (create_dispatcher_calls): Check that
20730 a target can create a function dispatcher.
20732 2017-03-14 Martin Liska <mliska@suse.cz>
20735 * multiple_target.c (expand_target_clones): Drop local.local
20736 flag for default implementation.
20738 2017-03-14 Richard Biener <rguenther@suse.de>
20740 PR tree-optimization/80030
20741 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20743 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20745 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20746 gcc_fallthrough() instead of __attribute__((fallthrough));
20748 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20750 * doc/gcc.texi: Remove "up" link to (DIR).
20751 * doc/gccint.texi: Ditto.
20753 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20755 * doc/install.texi (Specific) <avr>: Remove reference to
20758 2017-03-13 Jeff Law <law@redhat.com>
20760 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20761 attribute rather than comments.
20763 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20764 match_scratch operand is highest.
20766 2017-03-13 Martin Liska <mliska@suse.cz>
20768 PR middle-end/78339
20769 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20770 is a CHKP clone, use original declaration.
20772 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20774 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20775 (arc_conditional_register_usage): Use a different allocation order
20776 when optimizing for size.
20777 * common/config/arc/arc-common.c (arc_option_optimization_table):
20778 Section anchors default on when optimizing for size.
20780 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20782 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20784 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20786 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20787 * config/arc/arc.md (cpu_facility): Add cd variant.
20788 (*movqi_insn): Add code density variant.
20789 (*movhi_insn): Likewise.
20790 (*movqi_insn): Likewise.
20791 (*addsi3_mixed): Likewise.
20792 (subsi3_insn): Likewise.
20794 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20796 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20798 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20800 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20801 expressions with MINUS and UNARY ops.
20803 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20806 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20808 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20809 between vec_select and vector argument.
20810 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20811 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20812 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20813 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20814 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20815 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20817 2017-03-13 Richard Biener <rguenther@suse.de>
20820 * params.def (vect-max-peeling-for-alignment): Fix typo.
20822 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20824 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20825 issue that only occurred with binutils below 2.18.
20827 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20829 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20830 refer to binutils 2.11/2.12 minimum.
20832 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20834 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20835 ftp.kernel.org and simplify binutils requirement.
20837 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20839 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20841 (Optimize Options): Ditto. Also remove redundancy.
20843 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20845 PR translation/79848
20846 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20848 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20849 to G_ to avoid double translation.
20851 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20853 PR translation/79923
20854 * auto-profile.c (get_combined_location): Convert leading
20855 character of diagnostics to lower case and remove trailing period.
20856 (read_profile): Likewise for various diagnostics.
20857 * config/arm/arm.c (arm_option_override): Remove trailing period
20858 from various diagnostics.
20859 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20860 (msp430_expand_delay_cycles): Likewise.
20862 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20865 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20866 full command-line argument, rather than just "str".
20867 (aarch64_validate_march): Likewise.
20868 (aarch64_validate_mtune): Likewise.
20870 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20872 PR rtl-optimization/78911
20873 * lra-assigns.c (must_not_spill_p): New function.
20874 (spill_for): Use it.
20876 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20878 PR tree-optimization/79981
20879 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20880 ATOMIC_COMPARE_EXCHANGE ifn result.
20881 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20882 IFN_ATOMIC_COMPARE_EXCHANGE.
20884 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20887 * opts.c (parse_sanitizer_options): Add missing question mark to
20888 "did you mean" message.
20890 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20892 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20894 (VMULEUH_UNS): Likewise.
20895 (VMULOUB_UNS): Likewise.
20896 (VMULOUH_UNS): Likewise.
20897 * config/rs6000/rs6000.c (builtin_function_type): Remove
20898 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20900 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20903 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20904 x with result of extra_parsing_for_operand_code_0.
20905 (function_reader::extra_parsing_for_operand_code_0): Convert
20906 return type from void to rtx, returning x. When reading
20907 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20908 larger size containing struct block_symbol.
20910 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20912 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20913 -mfloat128-hardware without -m64.
20915 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20918 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20919 entries to the case statement that marks unsigned arguments to
20920 overloaded functions.
20922 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20925 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20927 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20930 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20931 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20933 2017-03-10 Martin Liska <mliska@suse.cz>
20937 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20938 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20941 2017-03-10 Olivier Hainque <hainque@adacore.com>
20943 * tree-switch-conversion (array_value_type): Start by resetting
20944 candidate type to it's main variant.
20946 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20948 PR rtl-optimization/79909
20949 * combine.c (try_combine): Use simplify_replace_rtx on individual
20950 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20951 of the whole CALL_INSN_FUNCTION_USAGE.
20953 PR tree-optimization/79972
20954 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20955 get_range_info on SSA_NAMEs. Formatting fixes.
20957 2017-03-10 Richard Biener <rguenther@suse.de>
20958 Jakub Jelinek <jakub@redhat.com>
20960 PR tree-optimization/77975
20961 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20962 edge to be constant.
20963 (get_val_for): For constant x return it. Formatting fix.
20964 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20965 would use the same bases as the current one.
20967 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20969 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20970 instead of vec_select for V1TImode.
20971 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20973 (VSX_LE_128): Add V1TI to this mode iterator.
20974 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20975 (*vsx_le_perm_store_<mode>): Likewise.
20976 (pre-reload splitter for VSX stores): Likewise.
20977 (post-reload splitter for VSX stores): Likewise.
20978 (*vsx_xxpermdi2_le_<mode>): Likewise.
20979 (*vsx_lxvd2x2_le_<mode>): Likewise.
20980 (*vsx_stxvd2x2_le_<mode>): Likewise.
20982 2017-03-09 Michael Eager <eager@eagercon.com>
20984 Correct failures with --enable-checking=yes,rtl.
20986 * config/microblaze/microblaze.c (microblaze_expand_shift):
20987 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20988 test for const0_rtx.
20989 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20990 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20992 2017-03-09 Richard Biener <rguenther@suse.de>
20994 PR tree-optimization/79977
20995 * graphite-scop-detection.c (scop_detection::merge_sese):
20996 Handle the case of extra exits to blocks dominating the entry.
20998 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21000 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21003 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21005 PR rtl-optimization/79949
21006 * lra-constraints.c (process_alt_operands): Check memory when
21007 trying to predict a cycle. Print about the overall increase.
21009 2017-03-09 Richard Biener <rguenther@suse.de>
21011 PR middle-end/79971
21012 * gimple-expr.c (useless_type_conversion_p): Preserve
21013 TYPE_SATURATING for fixed-point types.
21015 2017-03-09 Richard Biener <rguenther@suse.de>
21018 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21019 alignment of BLKmode params.
21021 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21024 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21025 (VALL_NO_V2Q): Likewise.
21027 * config/aarch64/aarch64-simd.md
21028 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21030 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21031 VALL_NO_V2Q mode iterator.
21032 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21034 2017-03-09 Martin Liska <mliska@suse.cz>
21036 PR tree-optimization/79631
21037 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21038 tree_int_cst_sign_bit just for INTEGER constants.
21040 2017-03-09 Martin Liska <mliska@suse.cz>
21044 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21047 2017-03-09 Marek Polacek <polacek@redhat.com>
21050 * tree.c (inchash::add_expr): Handle TREE_VEC.
21052 2017-03-09 Martin Liska <mliska@suse.cz>
21055 (chkp_narrow_size_and_offset): New function.
21056 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21057 (void chkp_parse_bit_field_ref): New function.
21058 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21059 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21061 2017-03-09 Martin Liska <mliska@suse.cz>
21064 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21065 (chkp_find_bounds_1): Remove gcc_unreachable.
21067 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21070 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21071 BUILT_IN_SYNC*, determine the access type from the size suffix and
21072 always build a MEM_REF with that type. Handle forgotten
21073 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21076 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21077 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21078 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21079 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21080 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21081 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21082 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21083 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21084 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21085 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21086 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21087 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21088 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21089 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21090 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21091 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21092 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21093 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21094 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21095 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21096 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21097 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21098 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21099 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21100 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21101 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21102 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21103 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21104 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21105 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21106 definitions outside of __OPTIMIZE__ guarded section.
21109 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21110 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21111 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21112 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21115 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21117 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21118 ("vfenez<mode>"): Add missing constraints.
21120 2017-03-08 Martin Sebor <msebor@redhat.com>
21123 * config/nds32/nds32.c (nds32_option_override):
21124 Fix misspelled diagnostic.
21126 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21129 * gimplify.c (gimplify_omp_for): Replace index var in outer
21130 taskloop statement with an artificial variable and add
21131 OMP_CLAUSE_PRIVATE clause for it.
21133 2017-03-08 Richard Biener <rguenther@suse.de>
21135 PR tree-optimization/79955
21136 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21137 for accesses that are completely outside of the variable.
21139 2017-03-08 Andrew Haley <aph@redhat.com>
21141 PR tree-optimization/79943
21142 * tree-ssa-loop-split.c (compute_new_first_bound): When
21143 calculating the new upper bound, (END-BEG) should be added, not
21146 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21148 * config/avr/avr.md (setmemhi): Make sure match_dup
21149 operand number comes before match_scratch.
21151 2017-03-08 Richard Biener <rguenther@suse.de>
21153 PR tree-optimization/79920
21154 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21155 with ncopies == 1 to ...
21156 (vect_transform_slp_perm_load): ... here. Properly compute
21157 all element loads by iterating VF times over the group. Do
21158 not handle ncopies (computed in a broken way) in
21159 vect_create_mask_and_perm.
21161 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21164 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21165 is a uniform vector, use uniform_vector_p return value instead of
21166 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21168 2017-03-07 Marek Polacek <polacek@redhat.com>
21170 PR middle-end/79809
21171 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21172 (alloca_call_type): Likewise.
21174 2017-03-07 Martin Liska <mliska@suse.cz>
21176 * gcov.c (process_args): Put comment to correct location.
21178 2017-03-07 Martin Liska <mliska@suse.cz>
21180 PR middle-end/68270
21181 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21182 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21183 (chkp_narrow_bounds_for_field): Likewise.
21184 (chkp_parse_array_and_component_ref): Pass one more argument to
21187 2017-03-07 Richard Biener <rguenther@suse.de>
21189 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21192 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21194 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21195 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21197 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21200 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21201 to end of description.
21202 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21204 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21206 PR rtl-optimization/79901
21207 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21209 (*avx512f_<code><mode>3<mask_name>): ... this.
21210 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21211 iterator instead of VI8_AVX2_AVX512BW.
21213 PR rtl-optimization/79901
21214 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21215 min/max expander, expand it using expand_vec_cond_expr.
21218 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21221 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21224 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21225 to void * for PCH reasons.
21226 * dwarf2out.c (output_loc_operands, output_die): Cast
21227 v.val_vec.array to unsigned char *.
21229 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21232 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21235 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21237 PR rtl-optimization/79571
21238 * lra-constraints.c (process_alt_operands): Calculate static
21239 reject and subtract it from overall when only addresses will be
21242 2017-03-06 Julia Koval <julia.koval@intel.com>
21245 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21246 incoming stack boundary to 128 for 64-bit targets.
21248 2017-03-06 Richard Biener <rguenther@suse.de>
21250 PR tree-optimization/79894
21251 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21252 to NULL after folding it.
21254 2017-03-06 Richard Biener <rguenther@suse.de>
21256 PR tree-optimization/79824
21257 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21258 check disabling peeling for gaps.
21260 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21262 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21263 attributes): Document gettimeofday.
21265 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21267 * config/s390/s390.c (s390_option_override_internal): Set
21268 PARAM_MIN_VECT_LOOP_BOUND
21270 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21272 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21273 * config/s390/s390.md: Likewise.
21275 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21278 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21279 (<avx2_avx512>_perm<mode>): Rename to ...
21280 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21282 (<avx512>_perm<mode>_mask): Rename to ...
21283 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21285 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21286 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21287 instead of VI8F_256_512.
21288 (avx512f_perm<mode>): New define_expand.
21289 (avx512f_perm<mode>_mask): Likewise.
21290 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21291 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21293 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21295 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21296 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21298 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21300 2017-03-06 Martin Liska <mliska@suse.cz>
21303 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21304 when having a SSA NAME w/o VAR_DECL assigned to it.
21306 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21308 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21309 msa_dpsub_<su>_d): Fix MODE for vec_select.
21311 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21313 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21315 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21317 2017-03-06 Richard Biener <rguenther@suse.de>
21319 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21320 * plugin.c (register_plugin_info): Likewise.
21321 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21323 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21325 * config/i386/sse.md (sse_storehps, sse_storelps,
21326 avx_<castmode><avxsizesuffix>_<castmode>,
21327 avx512f_<castmode><avxsizesuffix>_<castmode>,
21328 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21329 in condition that at least one operand is not a MEM.
21331 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21333 PR middle-end/79805
21334 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21335 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21337 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21338 gimple_call_nothrow_p flag based on whether original builtin can throw.
21339 If it can, emit following stmts on the fallthrough edge.
21340 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21341 don't create new bb if inserting just debug stmts on the edge, try to
21342 insert them on the fallthru bb or just reset debug stmts.
21344 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21347 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21348 restore recog_data (including the operand rtxes inside it) around
21349 the call to get_insn_template.
21351 2017-03-03 Martin Sebor <msebor@redhat.com>
21353 PR tree-optimization/79699
21354 * context.c (context::~context): Free MPFR caches to avoid
21355 a memory leak on program exit.
21357 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21359 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21360 Use wide_int::ulow () instead of .elt (0).
21362 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21364 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21365 (*pushxf): Limit oF constraint to 32bit targets and add oC
21366 constraint for 64bit targets.
21367 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21368 (*pushdf): Change rmF constraint to rmC.
21370 2017-03-03 Martin Liska <mliska@suse.cz>
21372 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21373 Remove unused variable.
21375 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21378 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21379 is a memory operand, increase num_memory.
21380 (ix86_expand_args_builtin): Likewise.
21382 2017-03-03 Jan Hubicka <jh@suse.cz>
21385 * ipa-devirt.c (maybe_record_node): Properly handle
21386 __cxa_pure_virtual visibility.
21388 2017-03-03 Martin Liska <mliska@suse.cz>
21390 PR tree-optimization/79803
21391 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21393 (pass_loop_prefetch::execute): Disabled optimization if an
21394 assumption about L1 cache size is not met.
21396 2017-03-03 Martin Liska <mliska@suse.cz>
21398 PR rtl-optimization/79574
21399 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21400 (hash_scan_set): Likewise.
21401 (dump_hash_table): Likewise.
21402 (hoist_code): Likewise.
21404 2017-03-03 Richard Biener <rguenther@suse.de>
21406 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21407 in place of uhigh ().
21408 (fixed_convert_from_real): Likewise.
21410 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21413 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21415 2017-03-03 Richard Biener <rguenther@suse.de>
21417 PR middle-end/79818
21418 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21419 TYPE_OVERFLOW_UNDEFINED check.
21421 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21423 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21425 (vector_ae_<mode>_p): Likewise.
21426 (vector_nez_<mode>_p): Likewise.
21427 (vector_ne_v2di_p): Likewise.
21428 (vector_ae_v2di_p): Likewise.
21429 (vector_ne_<mode>_p): Likewise.
21430 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21432 (vsx_tsqrt<mode>2_fe): Likewise.
21434 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21437 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21439 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21441 PR rtl-optimization/79780
21442 * cprop.c (one_cprop_pass): When second and further conditional trap
21443 in a single basic block is turned into an unconditional trap, turn it
21444 into a deleted note to avoid RTL verification failures.
21446 2017-03-02 Richard Biener <rguenther@suse.de>
21448 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21450 2017-03-02 Richard Biener <rguenther@suse.de>
21452 PR tree-optimization/79345
21454 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21455 param and abort the walk, returning -1 if it is hit.
21456 (walk_aliased_vdefs): Take a limit param and pass it on.
21457 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21458 defaulting to 0 and return a signed int.
21459 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21460 (check_defs): New helper.
21461 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21462 about uninitialized memory.
21463 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21464 bogus uninitialized warning.
21465 (fixed_convert_from_real): Likewise.
21467 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21469 PR tree-optimization/66768
21470 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21471 iv_use if base object can't be determined.
21473 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21475 PR tree-optimization/79345
21476 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21477 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21478 (get_pattern_stats): Initialize it.
21479 * genemit.c (gen_expand): Verify match_scratch numbers come after
21480 match_operand/match_dup numbers.
21481 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21482 match_scratch numbers.
21483 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21485 * config/s390/s390.md (trunctdsd2): Likewise.
21487 2017-03-02 Richard Biener <rguenther@suse.de>
21489 * wide-int.h (wide_int_storage::operator=): Implement in terms
21492 2017-03-02 Richard Biener <rguenther@suse.de>
21494 PR tree-optimization/79777
21495 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21496 the to insert expression to sth existing.
21498 2017-03-01 Martin Sebor <msebor@redhat.com>
21500 PR middle-end/79692
21501 * gimple-ssa-sprintf.c
21502 (directive::known_width_and_precision): New function.
21503 (format_integer): Use it.
21504 (get_mpfr_format_length): Consider the full range of precision
21505 when computing %g output with the # flag. Set the likely byte
21506 count to 3 rather than 1 when precision is indeterminate.
21507 (format_floating): Correct the lower bound of precision.
21509 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21511 * doc/invoke.texi: Document default code model for 64-bit Linux.
21513 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21516 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21517 udiv rather than div since input pattern is unsigned.
21519 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21521 * config/i386/i386.c (print_reg): Warn for values of
21522 unsupported size in integer register.
21524 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21527 * config/rs6000/predicates.md (current_file_function_operand): Do
21528 not allow self calls to be local if the function is replaceable.
21530 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21533 * config/rs6000/altivec.h (vec_ctz and others): Change the
21534 preprocessor macro that controls conditional compilation from
21535 _ARCH_PWR9 to __POWER9_VECTOR__.
21536 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21537 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21538 control (instead of _ARCH_PWR9 control) so that template
21539 definition uses power9-specific function.
21540 (vec_any_eq): Likewise.
21541 (vec_all_ne): Change macro definition to use a power9-specific
21542 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21543 _ARCH_PWR9 control).
21544 (vec_any_eq) Likewise.
21545 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21546 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21547 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21548 support for xvcmpnedp instruction.
21549 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21550 macro expansion so that Power9 implementation of vec_all_ne does
21551 not use the AltiVec predicate framework.
21552 (VCMPNEH_P): Likewise.
21553 (VCMPNEW_P): Likewise.
21554 (VCMPNED_P): Likewise.
21555 (VCMPNEFP_P): Likewise.
21556 (VCMPNEDP_P): Likewise.
21557 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21558 implementation of vec_any_eq to not use AltiVec predicate
21560 (VCMPAEH_P): Likewise.
21561 (VCMPAEW_P): Likewise.
21562 (VCMPAED_P): Likewise.
21563 (VCMPAEFP_P): Likewise.
21564 (VCMPAEDP_P): Likewise.
21565 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21566 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
21567 not use the AltiVec predicate framework.
21568 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
21569 of vec_any_eq to not use AltiVec predicate framework.
21570 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
21571 support for predefined __POWER9_VECTOR__ macro to indicate that
21572 Power9 instruction selection is enabled.
21573 (altivec_overloaded_builtins): Remove extraneous
21574 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
21575 function argument types RS6000_BTI_bool_V16QI and
21576 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
21577 entry for overloaded function argument types RS6000_BTI_bool_V4SI
21578 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
21579 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
21580 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
21581 Power9 for implementations of vec_cmpne. Change the signature for
21582 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
21583 (representing vec_all_ne) to remove the previously described first
21584 argument of type RS6000_BTI_INTSI, as this was an artifact of
21585 reliance on the AltiVec predicate framework, which is no longer
21586 used in the implementation of these functions. Add
21587 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
21588 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
21589 since, unlike the AltiVec predicate framework implementation, we
21590 do not share function descriptors between vec_alle and vec_anyeq.
21591 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
21592 set of modes that receive special treatment even when
21593 TARGET_P9_VECTOR is true. The special treatment emits code that
21594 does not depend on Power9 instructions.
21595 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
21596 define_expand to not rely on AltiVec predicate framework.
21597 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21599 (vector_ne_v2di_p): Change this define_expand to not rely on
21600 AltiVec predicate framework.
21601 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
21603 (vector_ne_<mode>_p): Change this define_expand to not rely on
21604 AltiVec predicate framework.
21605 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21607 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21608 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21609 define_insn pattern.
21610 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21611 define_insn pattern because the xvcmpne<VSs>. instruction is not
21613 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21614 instruction is not supported.
21616 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21618 * config/nvptx/nvptx.c: Include intl.h.
21620 2017-03-01 Martin Jambor <mjambor@suse.cz>
21623 * ipa-prop.h (ipa_bits): Removed field known.
21624 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21625 to pointers. Adjusted their comments to warn about their sharing.
21626 (ipcp_transformation_summary): Change bits to a vector of pointers.
21627 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21628 (ipa_get_ipa_bits_for_value): Declare.
21629 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21630 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21631 (ipa_bits_hash_table): Likewise.
21632 (ipa_vr_ggc_hash_traits): Likewise.
21633 (ipa_vr_hash_table): Likewise.
21634 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21635 being pointers and vr_known being removed.
21636 (ipa_set_jf_unknown): Likewise.
21637 (ipa_get_ipa_bits_for_value): New function.
21638 (ipa_set_jfunc_bits): Likewise.
21639 (ipa_get_value_range): New overloaded functions.
21640 (ipa_set_jfunc_vr): Likewise.
21641 (ipa_compute_jump_functions_for_edge): Use the above functions to
21642 construct bits and vr parts of jump functions.
21643 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21644 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21646 (ipcp_grow_transformations_if_necessary): Also allocate
21647 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21649 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21650 them. Fix too long lines.
21651 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21652 vr_known being removed.
21653 (ipa_read_jump_function): Use new setter functions to construct bits
21654 and vr parts of jump functions or set them to NULL.
21655 (write_ipcp_transformation_info): Adjust for bits being pointers.
21656 (read_ipcp_transformation_info): Likewise.
21657 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21659 Include gt-ipa-prop.h.
21660 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21662 (ipcp_store_bits_results): Likewise.
21663 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21664 Do not write to existing jump functions but use a temporary instead.
21666 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21669 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21670 attempt to use its first operand as BIT_FIELD_REF base.
21672 2017-03-01 Richard Biener <rguenther@suse.de>
21674 PR middle-end/79721
21675 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21676 interpolating formula in wrapping arithmetic.
21677 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21679 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21681 PR tree-optimization/79734
21682 * tree-vect-generic.c (expand_vector_condition): Optimize
21683 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21684 Handle VEC_COND_EXPR where comparison has different inner width from
21685 type's inner width.
21687 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21689 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21690 markup, and similar issues. Remove @opindex entries for things
21691 that aren't options. Add missing -mmpy-option entries.
21693 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21695 PR tree-optimization/79737
21696 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21697 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21698 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21699 instead of byte_size. Formatting fix.
21700 (shift_bytes_in_array_right): Formatting fix.
21702 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21705 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21706 condition on optimize for the leaf function test.
21708 2017-02-28 Martin Liska <mliska@suse.cz>
21711 * read-rtl-function.c (function_reader::handle_unknown_directive):
21712 Bail out when one uses -flto.
21714 2017-02-28 Martin Liska <mliska@suse.cz>
21716 * common.opt: Replace space with tabular for options of <number>
21718 * config/i386/i386.opt: Show <number> value for
21719 -mlarge-data-threshold.
21720 * opts.c (print_filtered_help): Do not display number in hexadecimal
21723 2017-02-28 Martin Liska <mliska@suse.cz>
21725 * common.opt: Fix --help=option -Q for options which are of
21728 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21730 * config/i386/i386.c (print_reg): Error out for values
21731 of 8-bit size in invalid integer register.
21733 2017-02-28 Martin Sebor <msebor@redhat.com>
21735 PR tree-optimization/79691
21736 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21738 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21741 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21742 gcc_unreachable with output_operand_lossage.
21744 2017-02-28 Richard Biener <rguenther@suse.de>
21746 PR tree-optimization/79740
21747 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21749 (visit_nary_op): Insert the nary into the hashtable if we
21750 pattern-matched sth.
21751 * tree-ssa-pre.c (eliminate_insert): Robustify.
21753 2017-02-28 Richard Biener <rguenther@suse.de>
21755 PR middle-end/79731
21756 * fold-const.c (decode_field_reference): Reject out-of-bound
21759 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21761 * config/i386/i386.c: Include intl.h.
21762 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21763 instead of just cond ? "..." : "...".
21764 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21765 * coverage.c (read_counts_file): Likewise.
21766 * omp-offload.c: Include intl.h.
21767 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21768 of just cond ? "..." : "...".
21769 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21770 of just cond ? "..." : "...".
21772 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21775 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21777 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21778 'tune for' CPU name.
21779 * config/arm/arm-cpu-data.h: Regenerated.
21781 2017-02-28 Richard Biener <rguenther@suse.de>
21783 PR tree-optimization/79732
21784 * tree-inline.c (expand_call_inline): Do not shadow var.
21786 2017-02-28 Richard Biener <rguenther@suse.de>
21788 PR tree-optimization/79723
21789 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21790 address-space properly.
21792 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21794 * doc/optinfo.texi (Optimization groups): Fix option used for
21796 * doc/invoke.texi (-fopt-info): Document "omp".
21797 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21798 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21799 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21800 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21801 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21803 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21805 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21806 OPTGROUP_OMP with "omp".
21808 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21811 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21812 for arithmetic shift of unsigned V2DI.
21814 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21816 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21817 arc/linux.h headers.
21818 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21819 (LINK_SPEC): Likewise.
21820 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21821 (EXTRA_SPECS): Likewise.
21822 (STARTFILE_SPEC): Likewise.
21823 (ENDFILE_SPEC): Likewise.
21824 (LIB_SPEC): Likewise.
21825 (TARGET_SDATA_DEFAULT): Likewise.
21826 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21827 (MULTILIB_DEFAULTS): Likewise.
21828 (DWARF2_UNWIND_INFO): Likewise.
21829 * config/arc/big.h: New file.
21830 * config/arc/elf.h: Likewise.
21831 * config/arc/linux.h: Likewise.
21832 * config/arc/t-uClibc: Remove.
21834 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21836 PR tree-optimization/77536
21837 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21838 (tree_transform_and_unroll_loop): Use above function to compute the
21839 estimated niter of unrolled loop and use it when scaling profile.
21840 Also use count info rather than frequency if it's non-zero.
21841 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21842 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21843 (vect_transform_loop): Call above function.
21845 2017-02-27 Richard Biener <rguenther@suse.de>
21847 PR tree-optimization/45397
21848 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21849 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21850 (visit_nary_op): Add pattern matching for CSEing sign-changed
21851 or truncated operations with wider ones.
21853 2017-02-27 Richard Biener <rguenther@suse.de>
21855 PR tree-optimization/79690
21856 * tree-vect-stmts.c (vectorizable_store): Use vector type
21857 built from the DR with address-space.
21859 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21861 * doc/invoke.texi (Optimize Options): Refine the description
21862 of asan-use-after-return.
21864 2017-02-25 Alan Modra <amodra@gmail.com>
21866 PR rtl-optimization/79584
21867 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21868 base, not ad->base_term, the reg within base. Remove assertion
21869 that ad->base == ad->base_term. Replace gen_int_mode using
21870 bogus mode with const0_rtx.
21872 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21874 PR middle-end/79396
21875 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21876 FMA_EXPR like tcc_binary or tcc_unary.
21878 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21881 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21883 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21884 (output_loc_operands): Handle DW_OP_call_ref and
21885 DW_OP_GNU_variable_value.
21886 (struct variable_value_struct): New type.
21887 (struct variable_value_hasher): Likewise.
21888 (variable_value_hash): New variable.
21889 (string_types): Remove.
21890 (copy_loc_descr): New function.
21891 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21892 (prepend_loc_descr_to_each): New function.
21893 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21894 instead of add_loc_descr_to_each if the first argument is single
21895 location list and the second has multiple.
21896 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21897 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21898 when looking for variable value which doesn't have other location info.
21899 (loc_list_from_tree): Formatting fix.
21900 (gen_array_type_die): Simplify DW_AT_string_length handling.
21901 (adjust_string_types): Remove.
21902 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21903 string_types. Call resolve_variable_values.
21904 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21905 (resolve_addr_in_expr): Likewise. Add A argument.
21906 (copy_deref_exprloc): Remove deref argument. Adjust for the
21907 original expression being DW_OP_GNU_variable_value with optionally
21908 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21909 optionally after it.
21910 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21911 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21912 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21913 (variable_value_hasher::hash, variable_value_hasher::equal): New
21915 (resolve_variable_value_in_expr, resolve_variable_value,
21916 resolve_variable_values, note_variable_value_in_expr,
21917 note_variable_value): New functions.
21918 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21921 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21924 * opts.h (handle_generated_option): Add GENERATED_P argument.
21925 * opts-common.c (handle_option): Adjust function comment.
21926 (handle_generated_option): Add GENERATED_P argument, pass it to
21928 (control_warning_option): Pass false to handle_generated_option
21930 * opts.c (maybe_default_option): Pass true to handle_generated_option
21932 * optc-gen.awk: Likewise.
21934 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21936 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21937 a REG, look at the REG it is a SUBREG of.
21938 (splitter for cmpeqsi_t): Ditto.
21940 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21942 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21943 the special USEs with the pattern of the insn, not the insn itself.
21945 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21948 * doc/invoke.texi: Document -mload-store-pairs.
21950 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21951 Sandra Loosemore <sandra@codesourcery.com>
21953 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21954 argument isn't a CONST_INT.
21955 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21956 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21957 (nios2_validate_compare): Bypass alternate compare logic if *op2
21958 is not a CONST_INT.
21959 (ldstwm_operation_p): Return false if first_base is not a REG or
21960 if first_offset is not a CONST_INT.
21962 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21964 * config/cris/cris.md: Use correct operand in a define_peephole2.
21966 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21968 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21970 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21972 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21973 this_insn if it is an INSN or JUMP_INSN.
21974 (force_offsettable): Look at base, not at addr.
21975 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21976 on things that aren't necessarily CONST_INTs.
21978 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21980 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21981 -mfpmath=sse is the default also for x86-32 targets with SSE2
21982 instruction set when @option{-ffast-math} is enabled
21984 2017-02-24 Jeff Law <law@redhat.com>
21986 PR rtl-optimizatoin/79286
21987 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21990 2017-02-24 Richard Biener <rguenther@suse.de>
21992 PR tree-optimization/79389
21993 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21996 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21998 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21999 function comment to reflect reality.
22000 (loop_exits_before_overflow): Fix typo in function description.
22002 2017-02-24 Richard Biener <rguenther@suse.de>
22004 PR tree-optimization/79389
22005 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22006 properly that a threading opportunity exists. Detect conditional
22007 copy/constant propagation opportunities.
22009 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22011 * config/visium/visium.md (type): Add trap.
22012 (b): New mode attribute.
22013 (*btst): Rename into...
22014 (*btst<mode>): ...this and adjust.
22015 (*cbranchsi4_btst_insn): Rename into...
22016 (*cbranch<mode>4_btst_insn): ...this and adjust.
22017 (trap): New define_insn.
22019 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22021 PR tree-optimization/79389
22022 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22023 (noce_reversed_cond_code): New function.
22024 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22025 reversed_comparison_code. Formatting fix.
22026 (noce_try_store_flag): Test rev_cond != NULL in addition to
22027 reversed_comparison_code.
22028 (noce_try_store_flag_constants): Likewise.
22029 (noce_try_store_flag_mask): Likewise.
22030 (noce_try_addcc): Use rev_cond if non-NULL instead of
22031 reversed_comparison_code.
22032 (noce_try_cmove_arith): Likewise. Formatting fixes.
22033 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22034 (noce_find_if_block): Initialize rev_cond.
22035 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22036 instead of false as last argument never attempt to reverse it
22039 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22041 PR tree-optimization/79663
22042 * tree-predcom.c (combine_chains): Process refs in reverse order
22043 only for ZERO length chains, and add explaining comment.
22045 2017-02-23 Jeff Law <law@redhat.com>
22047 PR tree-optimization/79578
22048 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22049 in call to operand_equal_p.
22051 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22054 * config/i386/cpuid.h: Fix another undefined behavior.
22056 2017-02-23 Richard Biener <rguenther@suse.de>
22058 PR tree-optimization/79683
22059 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22060 vector types for data-refs.
22062 2017-02-23 Martin Liska <mliska@suse.cz>
22064 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22066 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22068 PR middle-end/79665
22069 * internal-fn.c (get_range_pos_neg): Moved to ...
22070 * tree.c (get_range_pos_neg): ... here. No longer static.
22071 * tree.h (get_range_pos_neg): New prototype.
22072 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22073 are known to be in between 0 and signed maximum inclusive, try to
22074 expand both unsigned and signed divmod and use the cheaper one from
22077 2017-02-22 Jeff Law <law@redhat.com>
22079 PR tree-optimization/79578
22080 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22081 to compare base operands.
22083 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22086 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22087 gpc_reg_operand instead of fpr_reg_operand.
22089 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22091 * config/mips/mips.c (mips_return_in_memory): Force FP
22092 vector types to be returned in memory for o32 ABI.
22094 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22096 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22097 instead of DW_TAG_member for static data member declarations and don't
22098 set no_linkage_name for static inline data members.
22099 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22102 2017-02-22 Martin Liska <mliska@suse.cz>
22104 * doc/invoke.texi: Replace inequality signs with square brackets
22107 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22109 PR tree-optimization/68644
22110 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22112 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22115 * lra-constraints.c (simplify_operand_subreg): Handle
22116 WORD_REGISTER_OPERATIONS targets.
22118 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22121 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22122 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22123 elimination by swapping fld*.
22125 2017-02-22 Richard Biener <rguenther@suse.de>
22127 PR tree-optimization/79673
22128 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22129 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22130 irrelevant address-space qualifiers and avoiding a
22131 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22133 2017-02-22 Richard Biener <rguenther@suse.de>
22135 PR tree-optimization/79666
22136 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22137 to not symbolically negate if that may introduce undefined
22140 2017-02-22 Martin Liska <mliska@suse.cz>
22143 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22144 * data-streamer-out.c (streamer_write_gcov_count_stream):
22146 * value-prof.c (stream_out_histogram_value): Make assert more
22147 precise based on type of counter.
22149 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22152 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22153 Use nonimmediate_operand instead of memory_operand for operand 1.
22154 (float-extend standard_x87sse_constant_load splitter): Ditto.
22156 2017-02-21 Jeff Law <law@redhat.com>
22158 PR tree-optimization/79621
22159 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22160 blocks with edges to themselves.
22162 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22165 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22166 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22167 Use gimple_call_builtin_p.
22170 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22171 on temporarily removed DEBUG_INSNs.
22173 PR tree-optimization/79649
22174 * tree-loop-distribution.c (classify_partition): Give up on
22175 non-generic address space loads/stores.
22177 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22179 * doc/loop.texi (Loop manipulation): Remove nonexistent
22180 tree_ssa_loop_version from the documentation.
22181 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22183 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22186 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22187 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22188 * config/rs6000/rs6000.c: Include except.h.
22189 (rs6000_expand_split_stack_prologue): Call
22190 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22192 2017-02-21 Martin Jambor <mjambor@suse.cz>
22195 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22196 have been analyzed.
22198 2017-02-21 Martin Jambor <mjambor@suse.cz>
22200 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22201 for backward compatibility only.
22202 * doc/invoke.texi (Option Summary): Remove all references to
22203 -fipa-cp-alignment.
22205 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22209 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22211 * lra-constraints.c (curr_insn_transform): Handle
22212 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22214 2017-02-21 Martin Liska <mliska@suse.cz>
22216 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22218 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22220 PR translation/79638
22221 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22223 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22226 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22227 (arm_function_ok_for_sibcall): Return false for an indirect call by
22228 descriptor if all the argument registers are used.
22229 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22230 alignment of the function.
22232 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22234 PR tree-optimization/61441
22235 * simplify-rtx.c (simplify_const_unary_operation): For
22236 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22237 the sNaN unmodified.
22239 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22241 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22242 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22243 instead of SYSTEM_HEADER_DIR.
22245 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22246 Martin Liška <mliska@suse.cz>
22248 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22249 Fix typos and grammar, use active voice, and clarify.
22251 2017-02-20 Marek Polacek <polacek@redhat.com>
22253 PR middle-end/79537
22254 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22257 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22259 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22262 * config/i386/i386.c (ix86_expand_builtin): Handle
22263 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22264 ix86_builtins_isa[fcode].isa as a requirement of those
22265 flags and any other flag in the bitmask.
22266 (ix86_init_mmx_sse_builtins): Use 0 instead of
22267 ~OPTION_MASK_ISA_64BIT as mask.
22268 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22269 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22270 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22271 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22273 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22276 * lra-constraints.c (split_reg): Check requested split mode
22277 is supported by the register.
22279 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22281 * lra-constraints.c (simplify_operand_subreg): Remove early
22284 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22287 * lra-constraints.c (curr_insn_transform): Tighten condition
22288 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22290 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22293 * lra-constraints.c (curr_insn_transform): Handle
22294 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22296 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22299 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22301 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22303 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22306 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22309 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22311 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22314 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22316 * final.c (last_columnnum, override_columnnum): New variables.
22317 (final_start_function): Set last_columnnum, pass it to begin_prologue
22318 hook and pass 0 to dwarf2out_begin_prologue.
22319 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22320 to source_line debug hook.
22321 (notice_source_line): Compute last_columnnum and for debug_column_info
22322 return true on column changes.
22323 * debug.h (struct gcc_debug_hooks): Add column argument to
22324 source_line and begin_prologue hooks.
22325 (debug_nothing_int_charstar_int_bool): Remove prototype.
22326 (debug_nothing_int_int_charstar,
22327 debug_nothing_int_int_charstar_int_bool): New prototypes.
22328 (dwarf2out_begin_prologue): Add column argument.
22329 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22330 begin_prologue hooks.
22331 (debug_nothing_int_charstar_int_bool): Remove.
22332 (debug_nothing_int_int_charstar,
22333 debug_nothing_int_int_charstar_int_bool): New functions.
22334 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22335 through to dwarf2out_source_line.
22336 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22337 (dwarf2out_source_line): Add column argument, emit it if requested.
22338 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22340 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22341 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22342 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22343 through to dwarf2out_begin_prologue.
22344 (vmsdbgout_source_line): Add column argument, pass it through to
22345 dwarf2out_source_line.
22346 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22347 dbxout_source_line caller.
22348 (dbxout_source_line): Add column argument.
22350 * common.opt (gno-column-info, gcolumn-info): New options.
22351 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22352 (check_die): Also test for multiple DW_AT_decl_column attributes.
22353 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22354 DW_AT_decl_column if requested.
22355 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22357 (gen_variable_die): Likewise.
22358 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22359 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22362 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22363 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22364 (ix86_handle_option): Handle OPT_m3dnowa.
22365 * doc/invoke.texi (-m3dnowa): Document.
22366 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22367 -m3dnowa instead of -m3dnow -march=athlon.
22370 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22371 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22373 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22376 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22377 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22378 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22379 generator for vsx_xxpermdi_<mode>_be.
22380 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22381 force big-endian semantics.
22382 (vsx_xxpermdi_<mode>_be): New define_expand with same
22383 implementation as previous version of vsx_xxpermdi_<mode>.
22385 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22387 PR tree-optimization/79327
22388 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22389 variable, its initialization and use.
22391 2017-02-17 Julia Koval <julia.koval@intel.com>
22393 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22394 (OPTION_MASK_ISA_PKU_UNSET): New.
22395 (ix86_handle_option): Handle -mrdpid.
22396 * config/i386/cpuid.h (bit_RDPID): New.
22397 * config/i386/driver-i386.c (host_detect_local_cpu):
22398 Detect RDPID feature.
22399 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22400 * config/i386/i386-c.c (ix86_target_macros_internal):
22402 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22403 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22404 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22405 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22406 * config/i386/i386.md (define_insn "rdpid"): New.
22407 * config/i386/i386.opt Add -mrdpid.
22408 * config/i386/immintrin.h (_rdpid_u32): New.
22410 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22412 PR rtl-optimization/79541
22413 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22414 instead of transforming it into USE.
22416 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22418 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22419 If HONOR_SNANS (SFmode) force the input to a register.
22420 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22421 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22422 an frsp or similar insn.
22424 2017-02-17 Martin Liska <mliska@suse.cz>
22426 PR rtl-optimization/79577
22427 * params.def (selsched-max-sched-times): Increase minimum to 1.
22429 2017-02-17 Martin Liska <mliska@suse.cz>
22431 PR rtl-optimization/79574
22432 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22434 2017-02-17 Martin Liska <mliska@suse.cz>
22436 PR tree-optimization/79529
22437 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22438 ssa_defined_default_def_p to handle cases which are implicitly
22440 * tree-ssa.c (ssa_defined_default_def_p): New function.
22441 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22442 which are implicitly defined.
22443 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22445 2017-02-17 Richard Biener <rguenther@suse.de>
22447 PR middle-end/79576
22448 * params.def (max-ssa-name-query-depth): Limit to 10.
22450 2017-02-17 Richard Biener <rguenther@suse.de>
22452 PR tree-optimization/79552
22453 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22456 2017-02-17 Richard Biener <rguenther@suse.de>
22459 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22461 2017-02-17 Marek Polacek <polacek@redhat.com>
22463 PR middle-end/79536
22464 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22465 (fold_negate_expr): New wrapper.
22467 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22469 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22470 Correct terminology and de-emphasize pre-standard behavior.
22472 2017-02-16 Alan Modra <amodra@gmail.com>
22474 PR rtl-optimization/79286
22475 * ira.c (def_dominates_uses): New function.
22476 (update_equiv_regs): Don't create an equivalence for insns that
22477 may trap where the register def does not dominate the use.
22479 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22481 PR rtl-optimization/78127
22482 * lra.c (lra): Call lra_eliminate before finish the loop after
22485 2017-02-16 Richard Biener <rguenther@suse.de>
22487 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22489 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22490 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22491 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22492 (isl_val_int_from_wi): New function.
22493 (extract_affine_gmp): Rename to ...
22494 (extract_affine_wi): ... this, take a widest_int.
22495 (extract_affine_int): Just wrap extract_affine_wi.
22496 (add_param_constraints): Use isl_val_int_from_wi.
22497 (add_loop_constraints): Likewise, and extract_affine_wi.
22499 2017-02-15 Jeff Law <law@redhat.com>
22501 PR middle-end/79521
22502 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22503 ira_init_register_move_cost_if_necessary.
22505 2017-02-15 Martin Sebor <msebor@redhat.com>
22507 PR middle-end/32003
22508 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22509 removed in a prior commit.
22511 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22513 PR tree-optimization/79347
22514 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22515 counters during peeling.
22517 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22519 * Makefile.in (site.exp): Remove "set ISLVER".
22521 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22524 * real.c (real_from_integer): Call real_convert even for decimal.
22526 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22529 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22531 2017-02-14 Andrew Pinski <apinski@cavium.com>
22533 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22534 cores and change the partno/implementer to be correct.
22535 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22536 the 'B" as the implementer.
22537 * config/aarch64/aarch64-tune.md: Regenerate.
22539 2017-02-14 Carl Love <cel@us.ibm.com>
22541 * config/rs6000/rs6000.c: Add case statement entry to make the
22542 xvcvuxdsp built-in argument unsigned.
22543 * config/rs6000/vsx.md: Fix the source and return operand types so they
22544 match the instruction definitions from the ISA document. Fix typo
22545 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22548 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22551 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22552 member early_clobber_alts.
22553 * lra-lives.c (reg_early_clobber_p): New.
22554 (process_bb_lives): Use it.
22555 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22556 (debug_operand_data): Initialize early_clobber_alts.
22557 (setup_operand_alternative): Set up early_clobber_alts.
22558 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22559 alternatives to new_insn_reg.
22560 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22562 (lra_update_insn_regno_info): Pass the new arg.
22564 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22566 PR middle-end/79505
22567 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
22568 (new_oacc_loop_raw): Don't clear already cleared fields.
22571 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
22572 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
22573 _mm512_prefetch_i64gather_ps): New inline functions and macros.
22575 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
22578 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
22580 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
22583 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
22584 the extra instruction to the right place to store 128-bit constant
22587 2017-02-14 Martin Sebor <msebor@redhat.com>
22589 PR middle-end/79448
22590 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
22591 warning for strings of unknown length.
22593 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22595 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
22597 2017-02-14 Jeff Law <law@redhat.com>
22600 * ira-costs.c (scan_one_insn): Initialize register move costs
22601 for pseudos seen in USE/CLOBBER insns.
22603 PR tree-optimization/79095
22604 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22605 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22606 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22607 if the operands are known to be not equal, then the resulting range
22609 (intersect_ranges): If the new range is ~[0,0] and the old range is
22610 wide, then prefer ~[0,0].
22611 * tree-vrp.c (overflow_comparison_p_1): New function.
22612 (overflow_comparison_p): New function.
22613 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22614 if NAME is used in an overflow test.
22615 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22616 overflow check that can be expressed as an equality test, then adjust
22617 ops to be that equality test.
22619 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22621 * config/s390/s390-builtin-types.def: Remove flags argument.
22622 * config/s390/s390.c (s390_init_builtins): Likewise.
22624 2017-02-14 Martin Liska <mliska@suse.cz>
22626 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22627 vector. Fix trailing white spaces.
22629 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22631 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22634 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22636 PR rtl-optimization/68664
22637 * config/arm/arm.c (arm_sched_can_speculate_insn):
22638 New function. Declare prototype.
22639 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22641 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22643 PR rtl-optimization/68664
22644 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22646 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22648 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22650 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22651 max skip bytes for function, loop and jump.
22653 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22655 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22656 ABS_EXPR for gimple dump.
22658 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22661 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22663 PR tree-optimization/79408
22664 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22665 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22666 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22667 also if rhs1 is INTEGER_CST.
22669 2017-02-14 Richard Biener <rguenther@suse.de>
22671 PR middle-end/79432
22672 * tree-into-ssa.c (insert_phi_nodes): When the function can
22673 have abnormal edges rewrite SSA names with broken use-def
22674 dominance out of SSA and register them for PHI insertion.
22676 2017-02-13 Martin Sebor <msebor@redhat.com>
22678 PR middle-end/79496
22679 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22680 clearing info.nowrite flag when snprintf size argument is a range.
22682 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22684 * cprop.c (cprop_jump): Add missing space in string literal.
22685 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22686 (get_constraint_for_component_ref): Likewise.
22687 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22688 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22689 * lra-constraints.c (process_alt_operands): Likewise.
22690 * ipa-inline.c (inline_small_functions): Likewise.
22691 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22692 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22693 * trans-mem.c (diagnose_tm_1_op): Likewise.
22694 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22695 (grid_parallel_clauses_gridifiable): Likewise.
22697 * config/nvptx/mkoffload.c (process): Add space in between
22700 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22701 "MOD4_SSE_REGS" and "ALL_REGS".
22703 * spellcheck.c (test_data): Add , in between "foo" and "food".
22705 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22708 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22709 boundary crossing check and subsequent code generation agree.
22711 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22713 * config/aarch64/aarch64.c (has_memory_op): Delete.
22714 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22717 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22719 PR rtl-optimization/79388
22720 PR rtl-optimization/79450
22721 * combine.c (distribute_notes): When removing TEM_INSN for which
22722 corresponding dest has last value recorded, invalidate that last
22725 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22727 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22728 of explicit '@'. Add missing assembly comment marker on branch costs
22731 2017-02-13 Nathan Sidwell <nathan@acm.org>
22733 * gengtype-lex.l (<in_struct>): Add '/'.
22735 2017-02-13 Martin Liska <mliska@suse.cz>
22738 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22740 2017-02-13 Richard Biener <rguenther@suse.de>
22742 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22744 * configure: Re-generate.
22745 * config.in: Likewise.
22746 * graphite-dependences.c: Simplify as if
22747 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22748 * graphite-isl-ast-to-gimple.c: Likewise.
22749 * graphite-optimize-isl.c: Likewise.
22750 * graphite-poly.c: Likewise.
22751 * graphite-sese-to-poly.c: Likewise.
22752 * graphite.h: Likewise.
22753 * toplev.c: Include isl/version.h and use isl_version () for
22754 printing the ISL version.
22755 * doc/install.texi: Update ISL requirement.
22757 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22759 * doc/standards.texi (Standards): Update reference to
22762 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22764 * doc/extend.texi (Named Address Spaces): sourceware.org now
22766 * doc/install.texi (Binaries): Ditto.
22769 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22771 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22772 standard terminology "stringize"/"stringizing" throughout.
22773 * doc/cppinternals.texi: Likewise.
22775 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22777 * doc/extend.texi: Fix some spelling mistakes and typos.
22778 * doc/invoke.texi: Likewise.
22780 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22783 * params.def (inline-min-speedup) Change from 10 to 8.
22785 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22787 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22790 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22793 * ipa-inline-analysis.c (get_minimal_bb): New function.
22794 (record_modified): Use it.
22795 (remap_edge_change_prob): Handle also ancestor functions.
22797 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22799 * doc/contrib.texi (Contributors): Remove broken link into
22800 the Mauve CVS repository.
22802 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22804 PR middle-end/79454
22805 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22806 result computation whenever lhs doesn't have vector mode, not
22807 just when it has BLKmode.
22809 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22811 * doc/makefile.texi (profiledbootstrap): Refer to the
22812 installation instructions only in textual form.
22814 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22817 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22819 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22821 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22822 (Specific): Update mingw-w64 reference.
22824 (Specific): Remove broken link to Renesas RX processor.
22826 2017-02-10 Richard Biener <rguenther@suse.de>
22828 * toplev.c (process_options): Do not mention obsolete graphite
22829 options when printing sorry message about missing graphite support.
22830 Mention -floop-nest-optimize.
22832 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22834 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22835 (vtst_p16): Likewise.
22836 (vtstq_p8): Likewise.
22837 (vtstq_p16): Likewise.
22839 (vtstq_p64): Likewise.
22840 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22841 (vset_lane_p64): New.
22842 (vsetq_lane_p64): New.
22844 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22846 PR tree-optimization/79411
22847 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22848 stmt operands are SSA_NAMEs used in abnormal phis.
22849 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22852 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22855 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22858 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22860 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22862 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22864 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22868 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22869 not arbitrary TREE_CONSTANT.
22872 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22873 "omp declare target link" attribute unless is_global_var.
22874 * omp-offload.c (find_link_var_op): Likewise.
22876 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22877 Chung-Lin Tang <cltang@codesourcery.com>
22879 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22881 (gimplify_adjust_omp_clauses): Don't delete TILE.
22882 (gimplify_omp_for): Deal with TILE.
22883 * internal-fn.c (expand_GOACC_TILE): New function.
22884 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22886 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22887 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22889 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22890 avoid DIV for outermost collapse var.
22891 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22892 Remove out of date comments, fix whitespace.
22893 * omp-general.c (omp_extract_for_data): Deal with tiling.
22894 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22895 adjust OLF_DIM_BASE value.
22896 (struct omp_for_data): Add tiling field.
22897 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22898 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22899 for auto loops. Remove default auto determining, moved to
22900 oacc_loop_fixed_partitions.
22901 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22902 stmts, add e_mask field.
22903 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22904 (oacc_thread_numbers): Use oacc_dim_call.
22905 (oacc_xform_tile): New.
22906 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22907 (finish_oacc_loop): Adjust for ifns vector.
22908 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22909 add case for GOACC_TILE fns.
22910 (oacc_loop_xform_loop): Delete.
22911 (oacc_loop_process): Iterate over call list directly, and add
22912 handling for GOACC_TILE fns.
22913 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22915 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22916 vector partitioning to outer loops. Assign 2 partitions to loops
22917 when available. Add TILE handling.
22918 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22919 (execite_oacc_device_lower): Process GOACC_TILE fns,
22920 ignore unknown specs.
22921 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22922 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22923 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22925 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22927 * configure.ac (ACX_BUGURL): Update.
22928 * configure: Regenerate.
22930 2017-02-09 Richard Biener <rguenther@suse.de>
22932 PR tree-optimization/69823
22933 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22934 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22936 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22938 * config/arc/arc-c.def: Add __NPS400__ definition.
22939 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22940 (TARGET_NPS400): Define.
22942 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22944 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22946 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22947 pointer, arch_info.
22948 (arc_cpu_types): Fill the arch_info field with a pointer into the
22949 arc_arch_types table.
22950 (arc_selected_cpu): Declare.
22951 * config/arc/arc.c (arc_selected_cpu): Make global.
22952 (arc_selected_arch): Delete.
22953 (arc_base_cpu): Delete.
22954 (arc_override_options): Remove references to deleted variables,
22955 update access to arch information.
22956 (ARC_OPT): Update access to arch information.
22957 (ARC_OPTX): Likewise.
22958 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22959 (TARGET_ARC600): Update access to arch information.
22960 (TARGET_ARC601): Likewise.
22961 (TARGET_ARC700): Likewise.
22962 (TARGET_EM): Likewise.
22963 (TARGET_HS): Likewise.
22964 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22967 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22970 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22971 condition/operands for integer GE/LE/GEU/LEU operations.
22973 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22975 PR translation/79397
22976 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22979 2017-02-08 Martin Jambor <mjambor@suse.cz>
22982 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22983 whether allocation happened.
22984 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22985 nothing was allocated.
22987 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22989 PR tree-optimization/79408
22990 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22991 constant, but SSA_NAME with a known integer range, use the minimum
22992 of that range instead of op1 to determine if modulo can be replaced
22993 with its first operand.
22995 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22997 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22999 2017-02-08 Richard Biener <rguenther@suse.de>
23001 PR tree-optimization/71824
23002 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23003 Check all loops contained in the merged region.
23005 2017-02-07 Andrew Pinski <apinski@cavium.com>
23007 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23009 2017-02-07 Andrew Pinski <apinski@cavium.com>
23011 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23012 (thunderxt88): Likewise.
23013 (thunderxt81): Disable LSE and change v8.1 to v8.
23014 (thunderxt83): Likewise.
23016 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23017 Richard Biener <rguenther@suse.de>
23019 PR middle-end/79399
23020 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23021 type from int to size_t.
23022 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23024 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23026 PR rtl-optimization/79386
23027 * cprop.c (bypass_conditional_jumps): Initialize
23028 bypass_last_basic_block already before splitting bbs after
23029 unconditional traps...
23030 (bypass_conditional_jumps): ... rather than here.
23033 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23034 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23035 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23036 fix -masm=intel patterns.
23038 2017-02-07 Richard Biener <rguenther@suse.de>
23040 PR tree-optimization/79256
23041 PR middle-end/79278
23042 * builtins.c (get_object_alignment_2): Use min_align_of_type
23043 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23044 and ADJUST_FIELD_ALIGN.
23046 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23048 * doc/tm.texi: Regenerate.
23049 * stor-layout.c (layout_decl): Adjust.
23050 (update_alignment_for_field): Likewise.
23051 (place_field): Likewise.
23052 (min_align_of_type): Likewise.
23053 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23054 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23055 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23056 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23057 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23058 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23059 * config/i386/i386.c (x86_field_alignment): Likewise.
23060 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23061 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23062 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23063 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23064 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23065 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23069 2017-01-30 Richard Biener <rguenther@suse.de>
23071 PR tree-optimization/79256
23072 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23073 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23076 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23078 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23079 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23080 builtins to SImode and emit a zero-extend, if necessary.
23082 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23084 * docs/invoke.texi (RISC-V Options): Alphabetize.
23086 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23088 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23091 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23093 * config/riscv/riscv.c: New file.
23094 * common/config/riscv/riscv-common.c: Likewise.
23095 * config.gcc: Likewise.
23096 * config/riscv/constraints.md: Likewise.
23097 * config/riscv/elf.h: Likewise.
23098 * config/riscv/generic.md: Likewise.
23099 * config/riscv/linux.h: Likewise.
23100 * config/riscv/multilib-generator: Likewise.
23101 * config/riscv/peephole.md: Likewise.
23102 * config/riscv/pic.md: Likewise.
23103 * config/riscv/predicates.md: Likewise.
23104 * config/riscv/riscv-builtins.c: Likewise.
23105 * config/riscv/riscv-c.c: Likewise.
23106 * config/riscv/riscv-ftypes.def: Likewise.
23107 * config/riscv/riscv-modes.def: Likewise.
23108 * config/riscv/riscv-opts.h: Likewise.
23109 * config/riscv/riscv-protos.h: Likewise.
23110 * config/riscv/riscv.h: Likewise.
23111 * config/riscv/riscv.md: Likewise.
23112 * config/riscv/riscv.opt: Likewise.
23113 * config/riscv/sync.md: Likewise.
23114 * config/riscv/t-elf-multilib: Likewise.
23115 * config/riscv/t-linux: Likewise.
23116 * config/riscv/t-linux-multilib: Likewise.
23117 * config/riscv/t-riscv: Likewise.
23118 * configure.ac: Likewise.
23119 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23120 Waterman as RISC-V maintainers.
23121 * doc/install.texi: Add RISC-V entries.
23122 * doc/invoke.texi: Add RISC-V options section.
23123 * doc/md.texi: Add RISC-V constraints section.
23124 * configure: Regenerated.
23126 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23129 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23130 false values to be constant vectors with all 0 or all 1 bits set.
23131 (vcondu<mode><mode>): Likewise.
23132 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23134 (fpmask_comparison_operator): Update comment.
23135 (vecint_comparison_operator): New predicate.
23136 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23137 vector conditionals when the true and false values are constant
23138 vectors with all 0 bits or all 1 bits set.
23140 2017-02-06 Martin Sebor <msebor@redhat.com>
23142 PR tree-optimization/79376
23143 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23145 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23147 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23148 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23149 to simplify split condition.
23151 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23153 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23154 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23157 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23159 PR rtl-optimization/68664
23160 * target.def (can_speculate_insn): New hook.
23161 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23162 * doc/tm.texi: Regenerate.
23163 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23164 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23165 (rs6000_sched_can_speculate_insn): New function.
23167 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23169 PR tree-optimization/79284
23170 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23171 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23172 vectorizable_mask_load_store, vectorizable_operation,
23173 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23174 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23175 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23176 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23177 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23178 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23179 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23180 is_gimple_assign (stmt). Replace another such test with
23181 is_gimple_assign (stmt).
23183 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23186 * config/avr/avr.c (rtl-iter.h): Include it.
23187 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23188 (avr_legitimate_combined_insn): ...and implementation.
23190 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23192 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23193 * config/s390/s390.c (s390_const_operand_ok)
23194 (s390_canonicalize_comparison, s390_extract_part)
23195 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23196 (s390_contiguous_bitmask_p, s390_rtx_costs)
23197 (legitimize_pic_address): Likewise.
23198 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23199 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23200 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23201 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23202 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23204 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23206 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23207 REGNO($0) == REGNO($1).
23209 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23211 * config/s390/linux.h(SIZE_TYPE): Add comment.
23213 2017-02-06 Julian Brown <julian@codesourcery.com>
23214 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23215 Virendra Pathak <virendra.pathak@broadcom.com>
23217 * config/aarch64/aarch64-cores.def: Change the scheduler
23219 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23220 * config/aarch64/thunderx2t99.md: New file.
23222 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23224 * doc/standards.texi (Go Language): Update link to language
23227 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23229 * tree-eh.c (lower_resx): Sanitize profile.
23230 (cleanup_empty_eh_move_lp): Likewise.
23232 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23235 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23237 * cfgloopmanip.h (loop_version): Update prototype.
23238 * modulo-sched.c (sms_schedule): Update call of loop_version.
23239 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23240 * tree-parloops.c (gen_parallel_loop): Likewise.
23241 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23242 * tree-ssa-loop-split.c (split_loop): Likewise.
23243 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23244 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23246 2017-02-05 Martin Liska <mliska@suse.cz>
23249 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23251 (print_operand_address): Initialize a struct to zero.
23253 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23255 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23256 garbage collector only in textual form.
23258 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23260 * doc/extend.texi (x86 specific memory model extensions for
23261 transactional memory): Simplify a phrase.
23263 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23266 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23267 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23268 (atomic_storedi_1): Likewise.
23270 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23272 PR tree-optimization/79338
23273 * tree-parloops.c (gather_scalar_reductions): Don't call
23274 vect_analyze_loop_form for loop->inner before destroying loop's
23277 2017-02-03 Martin Sebor <msebor@redhat.com>
23279 PR tree-optimization/79327
23280 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23281 when precision has resulted in leading zeros.
23282 (format_integer): Adjust the likely counter to assume an unknown
23283 argument that may be zero is non-zero.
23285 2017-02-03 Jason Merrill <jason@redhat.com>
23288 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23289 avoid copying non-taken branch.
23291 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23293 PR tree-optimization/79340
23294 * tree-vect-loop.c (vectorizable_reduction): Release
23295 vec_defs elements after safe_splicing them into other vectors.
23298 PR tree-optimization/79327
23299 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23300 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23302 (format_integer): Use wide_int_to_tree instead of build_int_cst
23303 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23304 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23305 of shortest and longest sequence.
23307 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23309 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23310 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23312 2017-02-03 Walter Lee <walt@tilera.com>
23315 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23316 after initial stackframe link reg save.
23317 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23319 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23322 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23323 wu for stxssp alternative.
23325 2017-02-03 Martin Sebor <msebor@redhat.com>
23327 PR tree-optimization/79352
23328 * gimple-fold.c (get_range_strlen): Add argument.
23329 (get_range_strlen): Change return type to bool.
23330 (get_maxval_strlen): Pass in a dummy argument.
23331 * gimple-fold.h (get_range_strlen): Change return type to bool.
23332 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23333 * tree.h (array_at_struct_end_p): Add argument.
23334 * tree.c (array_at_struct_end_p): Handle it.
23336 2017-02-03 Martin Liska <mliska@suse.cz>
23339 * multiple_target.c (create_dispatcher_calls): Redirect edge
23340 from a caller of a dispatcher.
23341 (expand_target_clones): Make the clones local.
23342 (ipa_target_clone): Do both target clones and resolvers.
23343 (ipa_dispatcher_calls): Remove the pass.
23344 (pass_dispatcher_calls::gate): Likewise.
23345 (make_pass_dispatcher_calls): Likewise.
23346 * passes.def (pass_target_clone): Put as very first IPA early
23349 2017-02-03 Martin Liska <mliska@suse.cz>
23351 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23352 in case of a function with ifunc attribute.
23354 2017-02-03 Martin Liska <mliska@suse.cz>
23356 * cgraph.c (cgraph_node::dump): Dump function version info.
23357 * symtab.c (symtab_node::dump_base): Add missing new line.
23359 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23361 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23362 (ifcombine_ifandif): Use it.
23364 2017-02-03 Martin Liska <mliska@suse.cz>
23366 * doc/invoke.texi: Document default value for
23367 use-after-scope-direct-emission-threshold.
23369 2017-02-03 Martin Liska <mliska@suse.cz>
23371 PR tree-optimization/79339
23372 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23373 (format_floating): Likewise.
23375 2017-02-03 Martin Liska <mliska@suse.cz>
23378 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23380 (ipa_node_params_t::remove): Likewise.
23381 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23382 initialization from removed ipa_node_params_t::insert.
23383 (ipa_node_params::~ipa_node_params): Move from removed
23384 ipa_node_params_t::release.
23385 * symbol-summary.h (symbol_summary::m_released): New member.
23386 Do not release a summary twice. Do not allow to call finalizer
23387 for types of a summary that live in GGC memory.
23389 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23391 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23394 2017-02-02 Martin Sebor <msebor@redhat.com>
23396 PR middle-end/79275
23397 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23398 (format_string): Tighten up the range of output for non-constant
23399 strings and correct the expected range for wide non-constant strings.
23401 2017-02-02 Martin Sebor <msebor@redhat.com>
23403 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23405 PR middle-end/32003
23406 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23408 (-fdump-tree-@var): Add to index and document how to come up
23409 with pass-specific option and dump file names.
23410 (-fdump-passes): Clarify where to look for output.
23412 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23414 PR middle-end/77445
23415 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23416 statistics of the analyzed path; allow threading for speed when
23417 any of BBs along the path are optimized for speed.
23419 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23421 PR middle-end/78468
23422 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23423 settings of the virtual registers.
23426 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23428 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23429 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23432 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23434 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23435 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23437 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23439 * config/s390/s390.md: Add missing comments with the expanded
23441 * config/s390/vector.md: Likewise.
23442 * config/s390/vx-builtins.md: Likewise.
23444 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23447 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23448 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23449 conditions on a single line.
23451 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23453 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23454 __S390_VX__ to __VX__.
23456 2017-02-01 Andrew Pinski <apinski@cavium.com>
23458 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23459 stmt_info to record_stmt_cost.
23460 (vect_get_known_peeling_cost): Pass stmt_info if known to
23462 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23463 cpu_vector_cost field into
23464 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23465 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23466 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23467 splitting of scalar_stmt_cost and vec_stmt_cost.
23468 (thunderx_vector_cost): Likewise.
23469 (cortexa57_vector_cost): LIkewise.
23470 (exynosm1_vector_cost): Likewise.
23471 (xgene1_vector_cost): Likewise.
23472 (thunderx2t99_vector_cost): Improve after the splitting of the two
23474 (aarch64_builtin_vectorization_cost): Update for the splitting of
23475 scalar_stmt_cost and vec_stmt_cost.
23477 2017-02-01 Torvald Riegel <triegel@redhat.com>
23478 Richard Henderson <rth@redhat.com>
23480 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23481 conditional on existance of a fast atomic load.
23482 * optabs-query.c (can_atomic_load_p): New function.
23483 * optabs-query.h (can_atomic_load_p): Declare it.
23484 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23485 no fast atomic load is available for the particular size of access.
23486 (expand_atomic_compare_and_swap): Likewise.
23487 (expand_atomic_load): Likewise.
23488 (expand_atomic_store): Likewise.
23489 (expand_atomic_fetch_op): Likewise.
23490 * testsuite/lib/target-supports.exp
23491 (check_effective_target_sync_int_128): Remove x86 because it provides
23492 no fast atomic load.
23493 (check_effective_target_sync_int_128_runtime): Likewise.
23495 2017-02-01 Richard Biener <rguenther@suse.de>
23497 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23498 (graphite_transform_loops): Provide opt-info for optimized nests.
23499 * tree-parloop.c (parallelize_loops): Provide opt-info for
23500 parallelized loops.
23502 2017-02-01 Richard Biener <rguenther@suse.de>
23504 PR middle-end/79315
23505 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23506 was not set before.
23508 2017-02-01 Richard Biener <rguenther@suse.de>
23510 PR tree-optimization/71824
23511 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23512 Verify the loops are valid in the merged SESE region.
23513 (scop_detection::can_represent_loop_1): Check analyzing the
23514 evolution of the number of iterations in the region succeeds.
23516 2017-01-31 Ian Lance Taylor <iant@golang.org>
23518 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23519 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23521 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23523 PR preprocessor/79210
23524 * input.c (get_substring_ranges_for_loc): Replace line_width
23525 assertion with error-handling.
23527 2017-01-31 Richard Biener <rguenther@suse.de>
23529 PR tree-optimization/77318
23530 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23531 (create_pw_aff_from_tree): Take loop parameter.
23532 (add_condition_to_pbb): Pass loop of the condition to
23533 create_pw_aff_from_tree.
23535 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23537 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23538 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23540 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23544 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23546 (convert_int_to_float128): Likewise.
23547 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23548 (convert_int_to_float128): Likewise.
23549 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23550 (UNSPEC_IEEE128_CONVERT): Likewise.
23551 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23552 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23553 Use local variables for IBM extended format.
23554 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23555 (fix_trunc<mode>si2_fprs): Likewise.
23556 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23557 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23558 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23559 to know that we can now have integers of all sizes in vector
23561 (fix<uns>_<mode>di2_hw): Likewise.
23562 (float<uns>_<mode>si2_hw): Likewise.
23563 (fix_<mode>si2_hw): Likewise.
23564 (fixuns_<mode>si2_hw): Likewise.
23565 (float<uns>_<mode>di2_hw): Likewise.
23566 (float_<mode>di2_hw): Likewise.
23567 (float_<mode>si2_hw): Likewise.
23568 (floatuns_<mode>di2_hw): Likewise.
23569 (floatuns_<mode>si2_hw): Likewise.
23570 (xscvqp<su>wz_<mode>): Delete, no longer used.
23571 (xscvqp<su>dz_<mode>): Likewise.
23572 (xscv<su>dqp_<mode>): Likewise.
23573 (ieee128_mfvsrd_64bit): Likewise.
23574 (ieee128_mfvsrd_32bit): Likewise.
23575 (ieee128_mfvsrwz): Likewise.
23576 (ieee128_mtvsrw): Likewise.
23577 (ieee128_mtvsrd_64bit): Likewise.
23578 (ieee128_mtvsrd_32bit): Likewise.
23580 2017-01-31 Martin Liska <mliska@suse.cz>
23583 * ipa-prop.c (ipa_free_all_node_params): Call release method
23584 instead of ~sumbol_summary to not to trigger double times
23587 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
23589 PR tree-optimization/71691
23590 * bitmap.h (class auto_bitmap): New.
23591 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
23592 is_maybe_undefined instead of ssa_undefined_value_p.
23594 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23596 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23597 __S390_ARCH_LEVEL__ to __ARCH__.
23599 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23601 PR tree-optimization/79267
23602 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
23603 if should_remove_lhs_p is true.
23605 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23608 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23609 (add_alignment_attribute): New.
23610 (base_type_die): Add alignment attribute.
23611 (subrange_type_die): Likewise.
23612 (modified_type_die): Likewise.
23613 (gen_array_type_die): Likewise.
23614 (gen_descr_array_type_die: Likewise.
23615 (gen_enumeration_type_die): Likewise.
23616 (gen_subprogram_die): Likewise.
23617 (gen_variable_die): Likewise.
23618 (gen_field_die): Likewise.
23619 (gen_ptr_to_mbr_type_die): Likewise.
23620 (gen_struct_or_union_type_die): Likewise.
23621 (gen_subroutine_type_die): Likewise.
23622 (gen_typedef_die): Likewise.
23623 (base_type_cmp): Compare alignment attribute.
23625 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23628 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23629 (setb_unsigned) New pattern for setb with CCUNS.
23630 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23631 subfc./subfe sequence to avoid overflow problems. Generate a
23632 shorter sequence with cmpld/setb for power9.
23633 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23634 for generating subfc. instruction.
23635 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23636 now uses this instruction.
23638 2017-01-30 Ian Lance Taylor <iant@google.com>
23641 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23642 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23644 2017-01-30 Martin Sebor <msebor@redhat.com>
23646 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23647 Move constant to the right of a relational operator.
23648 (get_mpfr_format_length, format_character, format_string): Ditto.
23649 (should_warn_p, maybe_warn): Same.
23651 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23653 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23656 * asan.c (get_translation_unit_decl): Remove function.
23657 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23659 2017-01-30 Martin Liska <mliska@suse.cz>
23661 PR gcov-profile/79259
23662 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23663 -fprofile-generate.
23665 2017-01-30 Martin Liska <mliska@suse.cz>
23668 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23669 Initialize variables with NULL value.
23671 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23674 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23676 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23678 2017-01-30 Richard Biener <rguenther@suse.de>
23680 PR tree-optimization/79276
23681 * tree-vrp.c (process_assert_insertions): Properly adjust common
23682 when removing a duplicate.
23684 * gcc.dg/torture/pr79276.c: New testcase.
23686 2017-01-30 Richard Biener <rguenther@suse.de>
23688 PR tree-optimization/79256
23689 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23690 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23692 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23694 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23697 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23698 ("*r<noxa>sbg_<mode>_sll_bitmask")
23699 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23700 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23701 Use contiguous_bitmask_nowrap_operand.
23703 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23706 * config/rs6000/altivec.h (vec_xl): Revise #define.
23707 (vec_xst): Likewise.
23709 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23711 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23713 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23715 PR rtl-optimization/79194
23716 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23717 traps before call to bypass_conditional_jumps.
23719 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23721 PR tree-optimization/71374
23722 * lra-constraints.c (check_conflict_input_operands): New.
23723 (match_reload): Use it.
23725 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23728 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23729 account to calculate conflict_set.
23731 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23733 PR rtl-optimization/78559
23734 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23735 other_insn in combine.
23737 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23739 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23740 uint16_type_node for BT_UINT16.
23742 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23744 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23745 "RTL Tests" to menu.
23746 (GIMPLE Tests): New node.
23747 (RTL Tests): New node.
23749 2017-01-27 Richard Biener <rguenther@suse.de>
23751 PR tree-optimization/79245
23752 * tree-loop-distribution.c (distribute_loop): Apply cost
23753 modeling also to detected patterns.
23755 2017-01-27 Richard Biener <rguenther@suse.de>
23757 PR tree-optimization/71433
23758 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23759 (compare_assert_loc): New function.
23760 (process_assert_insertions): Sort and optimize assert locations
23761 to remove duplicates and push down identical assertions on
23762 edges to their destination block.
23764 2017-01-27 Richard Biener <rguenther@suse.de>
23766 PR tree-optimization/79244
23767 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23768 out SSA names even if abnormal.
23770 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23772 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23773 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23774 instead of MPFR_RNDN.
23776 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23779 * arm.c (arm_option_override): Don't call build_target_option_node
23780 until after doing all option overrides.
23781 (arm_valid_target_attribute_tree): Likewise.
23783 2017-01-27 Martin Liska <mliska@suse.cz>
23785 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23786 for {cd}tors and C++ {cd}tors.
23788 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23790 * config/s390/s390.md ("*setmem_long_and")
23791 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23793 2017-01-26 Martin Sebor <msebor@redhat.com>
23795 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23798 2017-01-26 Martin Sebor <msebor@redhat.com>
23800 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23801 HAVE_DFmode before using XFmode or DFmode.
23802 (parse_directive): Avoid using the z length modifier to avoid
23803 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23805 PR middle-end/78703
23806 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23807 to accept adjustment as an array.
23808 (get_int_range): New function.
23809 (struct directive): Make width and prec arrays.
23810 (directive::set_width, directive::set_precision): Call get_int_range.
23811 (format_integer, format_floating): Handle width and precision ranges.
23812 (format_string, parse_directive): Same.
23814 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23817 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23818 comdat_type_p set, just clone them, but keep the children in the
23822 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23823 which have direct callers with -fvar-tracking-assignments enabled
23825 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23826 inside of type units.
23828 2017-01-26 Martin Sebor <msebor@redhat.com>
23830 PR middle-end/78703
23831 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23833 (struct format_result): Replace number_chars, number_chars_min,
23834 and number_chars_max with a single member of struct result_range.
23836 (format_result::operator+=): Adjust.
23837 (struct fmtresult): Remove bounded. Handle likely and unlikely
23839 (fmtresult::adjust_for_width_or_precision): New function.
23840 (fmtresult:type_max_digits): New function.
23841 (bytes_remaining): Handle likely and unlikely counters.
23842 (min_bytes_remaining): Remove.
23843 (format_percent): Simplify.
23844 (format_integer, format_floating): Set likely and unlikely counters.
23845 (get_string_length, format_character, format_string): Same.
23846 (format_plain, should_warn_p): New function.
23847 (maybe_warn): Call should_warn_p. Update diagnostic messages
23848 and handle those for all directives, including plain strings.
23849 (format_directive): Handle likely and unlikely counters.
23850 Remove unnecessary quoting from diagnostics. Add an informational
23852 (add_bytes): Remove.
23853 (pass_sprintf_length::compute_format_length): Simplify.
23854 (try_substitute_return_value): Handle likely and unlikely counters.
23856 2017-01-26 Carl Love <cel@us.ibm.com>
23858 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23859 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23861 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23864 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23865 endianess for subregs into account.
23866 * lra-constraints.c (lra_constraints): Do risky transformations
23867 always on the first iteration.
23868 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23869 last_call_used_reg_set.
23870 (process_bb_lives): Define and use last_call_used_reg_set.
23871 * lra.c (lra): Always continue after lra_constraints on the first
23874 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23876 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23878 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23880 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23882 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23883 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23884 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23885 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23886 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23887 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23888 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23889 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23890 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23892 2017-01-26 Marek Polacek <polacek@redhat.com>
23895 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23896 for the third operand.
23898 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23900 PR middle-end/79236
23901 * omp-low.c (struct omp_context): Add simt_stmt field.
23902 (scan_omp_for): Return omp_context *.
23903 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23904 context to the _simt_ SIMD stmt.
23905 (lower_omp_for): For combined SIMD with sibling _simt_
23906 SIMD, make sure to use the same decls in _looptemp_
23907 clauses as in the sibling.
23909 2017-01-26 David Sherwood <david.sherwood@arm.com>
23911 PR middle-end/79212
23912 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23915 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23918 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23919 emit fld b; fld a; if possible.
23921 * brig-builtins.def: Update copyright years.
23922 * config/arm/arm_acle_builtins.def: Update copyright years.
23924 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23927 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23928 constraint instead of o for the stxsd instruction.
23930 2017-01-25 Carl Love <cel@us.ibm.com>
23932 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23933 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23935 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23937 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23939 2017-01-25 Richard Biener <rguenther@suse.de>
23941 PR tree-optimization/69264
23942 * target.def (vector_alignment_reachable): Improve documentation.
23943 * doc/tm.texi: Regenerate.
23944 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23946 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23947 earlier changes with respect to TYPE_USER_ALIGN.
23948 (vector_alignment_reachable_p): Likewise. Improve dumping.
23950 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23953 * config/arm/arm.md (xordi3): Force constant operand into a register
23956 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23958 * doc/invoke.texi (-fstore-merging): Correct default optimization
23959 levels at which it is enabled.
23960 (-O): Move -fstore-merging from list to...
23963 2017-01-25 Richard Biener <rguenther@suse.de>
23966 * omp-expand.c: Include debug.h.
23967 (expand_omp_taskreg): Make sure to generate early debug before
23968 outlining anything from a function.
23969 (expand_omp_target): Likewise.
23970 (grid_expand_target_grid_body): Likewise.
23972 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23975 * asan.c (get_translation_unit_decl): New function.
23976 (asan_add_global): Extract modules file name from globals
23977 TRANSLATION_UNIT_DECL name.
23979 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23982 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23983 for long calls with APCS frame and VFP.
23985 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23987 * cfg.c (original_copy_tables_initialized_p): New function.
23988 * cfg.h (original_copy_tables_initialized_p): New decl.
23989 * cfgrtl.c (relink_block_chain): Guard the call to
23990 free_original_copy_tables with a call to
23991 original_copy_tables_initialized_p.
23992 * cgraph.h (symtab_node::native_rtl_p): New decl.
23993 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23994 (symtab_node::needed_p): Don't assert for early assembly output
23995 for __RTL functions.
23996 (cgraph_node::finalize_function): Set "force_output" for __RTL
23998 (cgraph_node::analyze): Bail out early for __RTL functions.
23999 (analyze_functions): Update assertion to support __RTL functions.
24000 (cgraph_node::expand): Bail out early for __RTL functions.
24001 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24003 * function.h (struct function): Update comment for field
24005 * gimple-expr.c: Include "tree-pass.h".
24006 (gimple_has_body_p): Return false for __RTL functions.
24007 * Makefile.in (OBJS): Add run-rtl-passes.o.
24008 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24010 (gcc::pass_manager::get_clean_slate): New accessor.
24011 * passes.c: Include "insn-addr.h".
24012 (should_skip_pass_p): Add logging. Update logic for running
24013 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24014 property-provider override so it is only done for gimple passes.
24016 (skip_pass): New function.
24017 (execute_one_pass): Call skip_pass when skipping passes.
24018 * read-md.c (md_reader::read_char): Support filtering
24019 the input to a subset of line numbers.
24020 (md_reader::md_reader): Initialize fields
24021 m_first_line and m_last_line.
24022 (md_reader::read_file_fragment): New function.
24023 * read-md.h (md_reader::read_file_fragment): New decl.
24024 (md_reader::m_first_line): New field.
24025 (md_reader::m_last_line): New field.
24026 * read-rtl-function.c (function_reader::create_function): Only
24027 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24028 curr_properties. Set DECL_INITIAL to a dummy block.
24029 (read_rtl_function_body_from_file_range): New function.
24030 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24032 * run-rtl-passes.c: New file.
24033 * run-rtl-passes.h: New file.
24035 2017-01-24 Jeff Law <law@redhat.com>
24037 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24040 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24042 PR tree-optimization/79159
24043 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24044 (record_nonwrapping_iv): Improve boundary using above function if no
24045 value range information.
24047 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24048 Martin Jambor <mjambor@suse.cz>
24050 * brig-builtins.def: New file.
24051 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24052 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24053 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24054 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24055 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24056 * builtin-types.def (BT_INT8): New.
24057 (BT_INT16): Likewise.
24058 (BT_UINT8): Likewise.
24059 (BT_UINT16): Likewise.
24060 (BT_FN_ULONG): Likewise.
24061 (BT_FN_UINT_INT): Likewise.
24062 (BT_FN_UINT_ULONG): Likewise.
24063 (BT_FN_UINT_LONG): Likewise.
24064 (BT_FN_UINT_PTR): Likewise.
24065 (BT_FN_ULONG_PTR): Likewise.
24066 (BT_FN_INT8_FLOAT): Likewise.
24067 (BT_FN_INT16_FLOAT): Likewise.
24068 (BT_FN_UINT32_FLOAT): Likewise.
24069 (BT_FN_UINT16_FLOAT): Likewise.
24070 (BT_FN_UINT8_FLOAT): Likewise.
24071 (BT_FN_UINT64_FLOAT): Likewise.
24072 (BT_FN_UINT16_UINT32): Likewise.
24073 (BT_FN_UINT32_UINT16): Likewise.
24074 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24075 (BT_FN_INT_PTR_INT): Likewise.
24076 (BT_FN_UINT_PTR_UINT): Likewise.
24077 (BT_FN_LONG_PTR_LONG): Likewise.
24078 (BT_FN_ULONG_PTR_ULONG): Likewise.
24079 (BT_FN_VOID_UINT64_UINT64): Likewise.
24080 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24081 (BT_FN_INT8_INT8_INT8): Likewise.
24082 (BT_FN_INT16_INT16_INT16): Likewise.
24083 (BT_FN_INT_INT_INT): Likewise.
24084 (BT_FN_UINT_FLOAT_UINT): Likewise.
24085 (BT_FN_FLOAT_UINT_UINT): Likewise.
24086 (BT_FN_ULONG_UINT_UINT): Likewise.
24087 (BT_FN_ULONG_UINT_PTR): Likewise.
24088 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24089 (BT_FN_UINT_UINT_UINT): Likewise.
24090 (BT_FN_VOID_UINT_PTR): Likewise.
24091 (BT_FN_UINT_UINT_PTR: Likewise.
24092 (BT_FN_UINT32_UINT64_PTR): Likewise.
24093 (BT_FN_INT_INT_UINT_UINT): Likewise.
24094 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24095 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24096 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24097 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24098 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24099 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24100 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24101 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24102 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24103 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24104 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24105 * doc/frontends.texi: List BRIG FE.
24106 * doc/install.texi (Testing): Add BRIG tesring requirements.
24107 * doc/invoke.texi (Overall Options): Mention BRIG.
24108 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24110 2017-01-24 Richard Biener <rguenther@suse.de>
24112 PR translation/79208
24113 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24115 2017-01-24 Martin Jambor <mjambor@suse.cz>
24118 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24119 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24120 and known_contexts.
24122 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24124 PR middle-end/79123
24125 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24126 casts from signed to unsigned really don't have a range.
24128 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24130 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24131 GMP_RNDx for compatiblity.
24133 2017-01-24 Martin Liska <mliska@suse.cz>
24136 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24137 that would prevent us to call alloca with -1 as argument.
24139 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24141 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24142 Avoid -Wformat-security warning.
24144 2017-01-23 Andrew Pinski <apinski@cavium.com>
24146 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24149 2017-01-23 Martin Sebor <msebor@redhat.com>
24151 PR middle-end/78703
24152 * gimple-ssa-sprintf.c (warn_level): New global.
24153 (format_integer): Use it here and throughout the rest of the file.
24154 Use the same switch to compute sign as base.
24155 (maybe_warn): New function.
24156 (format_directive): Factor out warnings into maybe_warn.
24157 Add debugging output. Use warn_level.
24158 (add_bytes): Use warn_level.
24159 (pass_sprintf_length::compute_format_length): Add debugging output.
24160 (try_substitute_return_value): Same.
24161 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24163 PR middle-end/78703
24164 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24165 (struct fmtresult, format_integer, format_floating): Adjust.
24166 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24167 (get_string_length, format_string,format_directive): Same.
24168 (pass_sprintf_length::compute_format_length): Same.
24169 (try_substitute_return_value): Simplify slightly.
24171 PR middle-end/78703
24172 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24173 (fmtresult::operator+=): Outlined.
24174 (struct fmtresult): Add ctors.
24175 (struct conversion_spec): Rename...
24176 (struct directive): ...to this. Add and remove data members.
24177 (directive::set_width, directive::set_precision): New functions.
24178 (format_percent): Use fmtresult ctor.
24179 (get_width_and_precision): Remove.
24180 (format_integer): Make naming changes. Avoid computing width and
24182 (format_floating): Same. Adjust indentation.
24183 (format_character, format_none): New functions.
24184 (format_string): Moved character handling to format_character.
24185 (format_directive): Remove arguments, change return type.
24186 (parse_directive): New function.
24187 (pass_sprintf_length::compute_format_length): Move directive
24188 parsing to parse_directive.
24190 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24192 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24193 (assign_assembler_name_if_needed): ... this.
24194 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24195 (assign_assembler_name_if_needed): ... this.
24196 (free_lang_data_in_cgraph): Adjust callers.
24197 * cgraphunit.c (cgraph_node::analyze): Likewise.
24198 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24201 2017-01-23 Richard Biener <rguenther@suse.de>
24203 PR tree-optimization/79088
24204 PR tree-optimization/79188
24205 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24206 resetting loop bounds after last path deletion. Reset loop
24207 bounds of the target loop, make code match the comments.
24208 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24209 Make sure loops need no fixups.
24211 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24213 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24214 exponent support with double type for first argument.
24215 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24216 type returned by __builtin_vec_extract_sig,
24217 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24218 functions from "vector int" to "vector unsigned int" or from
24219 "vector long long int" to "vector unsigned long long int".
24220 Changed type returned by __builtin_vec_extract_exp,
24221 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24222 functions from "vector int" to "vector unsigned int" or from
24223 "vector long long int" to "vector unsigned long long int".
24224 Changed return type of __builtin_vec_test_data_class,
24225 __builtin_vec_test_data_class_sp, and
24226 __builtin_vec_test_data_class_dp from "vector int" to
24227 "vector bool int" or from "vector long long int" to "vector bool
24228 long long int" and changed second argument type from "unsigned
24229 int" to "int". Added new overloaded function forms "vector float
24230 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24231 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24232 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24233 double, vector unsigned long long int)" and "vector double
24234 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24235 long int)". Changed return type of
24236 __builtin_scalar_test_data_class and
24237 __builtin_scalar_test_data_class_sp and
24238 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24239 int" and changed second argument from "unsigned int" to "int".
24240 Changed type returned by __builtin_scalar_test_neg,
24241 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24242 from "int" to "bool int". Added new overloaded function form
24243 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24244 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24245 exponent double-precision with floating point first argument.
24246 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24247 documentation of scalar_test_data_class, scalar_test_neg,
24248 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24249 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24250 vec_test_data_class built-in functions to reflect refinements in
24251 their type signatures.
24253 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24255 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24257 (aarch64_elf_asm_destructor): Likewise.
24259 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24261 PR rtl-optimization/78634
24262 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24263 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24264 * ifcvt.c (noce_try_cmove): Add missing cost check.
24266 PR rtl-optimization/71724
24267 * combine.c (if_then_else_cond): Look for situations where it is
24268 beneficial to undo the work of one of the recursive calls.
24270 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24272 PR tree-optimization/70754
24273 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24274 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24275 combined stmt before it if not NULL.
24276 (combine_chains): Process refs reversely and compute dominance point
24279 2017-01-23 Martin Liska <mliska@suse.cz>
24281 PR tree-optimization/79196
24282 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24283 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24285 (strlen_optimize_stmt): Call the renamed function.
24287 2017-01-23 Michael Matz <matz@suse.de>
24289 PR tree-optimization/78384
24290 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24292 2017-01-23 Richard Biener <rguenther@suse.de>
24294 PR tree-optimization/79186
24295 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24296 both incoming edges before moving an assert.
24298 2017-01-23 Martin Jambor <mjambor@suse.cz>
24300 * ipa-prop.c (load_from_param_1): Removed.
24301 (load_from_unmodified_param): Bits from load_from_param_1 put back
24303 (load_from_param): Removed.
24304 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24305 with stmt. Reverted back to use of load_from_unmodified_param.
24307 2017-01-23 Martin Jambor <mjambor@suse.cz>
24310 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24311 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24312 field a pointer to garbage collected vector, mark lattices and
24313 ipcp_orig_node with GTY((skip)).
24314 (ipa_get_param_count): Adjust to descriptors being a pointer.
24315 (ipa_get_param): Likewise.
24316 (ipa_get_type): Likewise.
24317 (ipa_get_param_move_cost): Likewise.
24318 (ipa_set_param_used): Likewise.
24319 (ipa_get_controlled_uses): Likewise.
24320 (ipa_set_controlled_uses): Likewise.
24321 (ipa_is_param_used): Likewise.
24322 (ipa_node_params_t): Move into garbage collector. New methods insert
24324 (ipa_node_params_sum): Annotate wth GTY(()).
24325 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24327 (ipa_load_from_parm_agg): Adjust declaration.
24328 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24329 * ipa-profile.c (ipa_profile): Likewise.
24330 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24331 (ipa_populate_param_decls): Make descriptors parameter garbage
24333 (ipa_dump_param): Adjust to descriptors being a pointer.
24334 (ipa_alloc_node_params): Likewise.
24335 (ipa_initialize_node_params): Likewise.
24336 (load_from_param_1): Make descriptors parameter garbage collected.
24337 (load_from_unmodified_param): Likewise.
24338 (load_from_param): Likewise.
24339 (ipa_load_from_parm_agg): Likewise.
24340 (ipa_node_params::~ipa_node_params): Removed.
24341 (ipa_free_all_node_params): Remove call to delete operator.
24342 (ipa_node_params_t::insert): New.
24343 (ipa_node_params_t::remove): Likewise.
24344 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24345 copy known_csts and known_contexts vectors.
24346 (ipa_read_node_info): Adjust to descriptors being a pointer.
24347 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24349 (ipcp_transform_function): Make descriptors variable garbage
24352 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24354 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24355 * config/i386/avx512dqintrin.h: Ditto.
24356 * config/i386/avx512fintrin.h: Ditto.
24357 * config/i386/i386.c: Handle new builtins.
24358 * config/i386/i386-builtin.def: Add new builtins.
24359 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24360 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24362 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24363 Martin Liska <mliska@suse.cz>
24365 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24366 * asan.c (asan_expand_poison_ifn): Support stores and use
24367 appropriate ASAN report function.
24368 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24369 * internal-fn.def (ASAN_POISON_USE): Declare.
24370 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24371 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24372 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24373 ASAN_POISON calls w/o LHS.
24374 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24375 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24376 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24377 * gimplify.c (asan_poison_variables): Add attribute
24378 use_after_scope_memory to variables that really needs to live
24380 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24381 having the attribute.
24383 2017-01-23 Martin Liska <mliska@suse.cz>
24385 * asan.c (create_asan_shadow_var): New function.
24386 (asan_expand_poison_ifn): Likewise.
24387 * asan.h (asan_expand_poison_ifn): New declaration.
24388 * internal-fn.c (expand_ASAN_POISON): Likewise.
24389 * internal-fn.def (ASAN_POISON): New builtin.
24390 * sanopt.c (pass_sanopt::execute): Expand
24391 asan_expand_poison_ifn.
24392 * tree-inline.c (copy_decl_for_dup_finish): Make function
24394 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24395 * tree-ssa.c (is_asan_mark_p): New function.
24396 (execute_update_addresses_taken): Rewrite local variables
24397 (identified just by use-after-scope as addressable) into SSA.
24399 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24401 * doc/install.texi (Specific): opensource.apple.com uses https
24402 now. Remove trailing slash.
24404 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24406 * README.Portability: Remove note on an Irix compatibility issue.
24408 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24410 * gcov.c (INCLUDE_ALGORITHM): Define.
24411 (INCLUDE_VECTOR): Define.
24412 No longer include <vector> and <algorithm> directly.
24414 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24416 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24418 * doc/invoke.texi (Code Gen Options): Ditto.
24420 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24423 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24425 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24427 rtl-optimization/79125
24428 * cprop.c (local_cprop_pass): Handle cases where we make an
24429 unconditional trap.
24431 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24435 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24436 read from, for big endian.
24438 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24440 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24441 register pauth builtins for LP64 only.
24443 2017-01-20 Marek Polacek <polacek@redhat.com>
24446 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24449 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24451 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24453 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24454 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24455 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24457 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24460 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24461 in big-endian mode when they are not a single duplicated value.
24463 2017-01-20 Richard Biener <rguenther@suse.de>
24465 * BASE-VER: Bump to 7.0.1.
24467 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24469 * omp-low.c (omplow_simd_context): New struct. Use it...
24470 (lower_rec_simd_input_clauses): ...here and...
24471 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24472 references to idx, lane, max_vf, is_simt.
24474 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24476 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24479 2017-01-20 Martin Jambor <mjambor@suse.cz>
24481 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24482 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24484 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24485 (GTFILES): Rename hsa.c to hsa-common.c.
24486 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24487 * hsa-dump.c: Likewise.
24488 * hsa-gen.c: Likewise.
24489 * hsa-regalloc.c: Likewise.
24490 * ipa-hsa.c: Likewise.
24491 * omp-expand.c: Likewise.
24492 * omp-low.c: Likewise.
24493 * toplev.c: Likewise.
24495 2017-01-20 Marek Polacek <polacek@redhat.com>
24498 * doc/invoke.texi: Document -Wduplicated-branches.
24499 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24500 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24501 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24502 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24503 return 0 only when not OEP_LEXICOGRAPHIC.
24504 (fold_build_cleanup_point_expr): Use the expression
24505 location when building CLEANUP_POINT_EXPR.
24506 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24507 * tree.c (add_expr): Handle error_mark_node.
24509 2017-01-20 Martin Liska <mliska@suse.cz>
24512 * tree-profile.c (init_ic_make_global_vars): Do not call
24514 (gimple_init_gcov_profiler): Likewise.
24516 2017-01-20 Martin Liska <mliska@suse.cz>
24519 * cgraph.h (maybe_create_reference): Remove argument and
24521 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24523 * ipa-cp.c (create_specialized_node): Likewise.
24524 * symtab.c (symtab_node::maybe_create_reference): Handle
24525 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24527 2017-01-20 Martin Liska <mliska@suse.cz>
24529 * read-rtl-function.c (function_reader::create_function): Use
24530 build_decl instread of build_decl_stat.
24532 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24534 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24535 * config/i386/avx512dqintrin.h: Ditto.
24536 * config/i386/avx512fintrin.h: Ditto.
24537 * config/i386/i386-builtin-types.def: Add new types.
24538 * config/i386/i386.c: Handle new types.
24539 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24540 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24541 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24542 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24543 (__builtin_ia32_kshiftridi): New.
24544 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24546 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24550 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24551 define to rs6000_init_stack_protect_guard.
24552 (rs6000_init_stack_protect_guard): New function.
24554 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24555 Yunqiang Su <yunqiang.su@imgtec.com>
24557 * config.gcc (supported_defaults): Add madd4.
24558 (with_madd4): Add validation.
24559 (all_defaults): Add madd4.
24560 * config/mips/mips.opt (mmadd4): New option.
24561 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24563 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24565 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24566 (ISA_HAS_FUSED_MADD4): Likewise.
24567 * doc/invoke.texi (-mmadd4): Document the new option.
24568 * doc/install.texi (--with-madd4): Document the new option.
24570 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24572 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
24573 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
24574 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
24575 (aarch64_init_pauth_hint_builtins): New.
24576 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
24577 (aarch64_expand_builtin): Expand new builtins.
24579 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24581 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
24582 * combine-stack-adj.c (no_unhandled_cfa): Handle
24583 REG_CFA_TOGGLE_RA_MANGLE.
24584 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
24585 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
24586 info for return address signing.
24587 (aarch64_expand_epilogue): Likewise.
24589 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24591 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
24592 * config/aarch64/aarch64-protos.h
24593 (aarch64_return_address_signing_enabled): New declaration.
24594 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
24596 (aarch64_expand_prologue): Sign return address before it's pushed onto
24598 (aarch64_expand_epilogue): Authenticate return address fetched from
24600 (aarch64_override_options): Sanity check for ILP32 and ISA level.
24601 (aarch64_attributes): New function attributes for "sign-return-address".
24602 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
24603 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24604 ("*do_return"): Generate combined instructions according to key index.
24605 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24606 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24608 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24609 * config/aarch64/aarch64.opt (msign-return-address=): New.
24610 * doc/extend.texi (AArch64 Function Attributes): Documents
24611 "sign-return-address=".
24612 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24614 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24616 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24617 overall option summary.
24619 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24621 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24622 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24623 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24624 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24626 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24628 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24629 -mpower9-minmax by default for -mcpu=power9.
24630 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24631 128-bit floating point.
24633 2017-01-20 Alan Modra <amodra@gmail.com>
24635 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24636 optimizing for size.
24638 2017-01-20 Alan Modra <amodra@gmail.com>
24641 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24642 for strcmp and strncmp from corresponding builtin decl.
24644 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24646 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24647 instead of i386/rtems-64.h.
24648 * config/i386/rtems-64.h: Remove.
24650 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24654 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24656 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24658 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24660 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24661 Change int to HOST_WIDE_INT.
24662 * config/aarch64/aarch64-protos.h
24663 (aarch64_simd_gen_const_vector_dup): Likewise.
24664 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24666 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24668 * langhooks-def.h (lhd_type_for_size): New decl.
24669 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24670 * langhooks.c (lhd_type_for_size): New function, taken from
24673 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24675 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24676 define_bypass for CR latency.
24677 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24678 (power9-alu2): Add define_bypass for CR latency.
24680 (power9-mul): Update insn latency.
24681 (power9-mul-compare): Update insn latency, bypass latency and remove
24684 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24686 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24688 * config/aarch64/aarch64.md
24689 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24690 aarch64_nopcrelative_literal_loads.
24691 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24693 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24695 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24696 TARGET_LOONGSON_3A.
24697 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24699 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24702 * config.gcc (supported_defaults): Add lxc1-sxc1.
24703 (with_lxc1_sxc1): Add validation.
24704 (all_defaults): Add lxc1-sxc1.
24705 * config/mips/mips.opt (mlxc1-sxc1): New option.
24706 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24708 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24709 __mips_no_lxc1_sxc1.
24710 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24711 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24712 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24714 2017-01-19 Richard Biener <rguenther@suse.de>
24716 PR tree-optimization/72488
24717 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24718 sure to restore SSA info.
24719 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24721 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24723 PR rtl-optimization/79121
24724 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24725 of the inner type when shifting an extended value.
24727 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24730 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24731 interposable aliases.
24733 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24736 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24737 Use the evmergelohi instruction.
24738 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24739 (mov_sitf_e500_subreg8_2_be): Likewise.
24740 (mov_sitf_e500_subreg12_2_le): Likewise.
24741 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24742 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24743 (mov_sitf_e500_subreg8_2_le): Likewise.
24744 (mov_sitf_e500_subreg12_2_be): Likewise.
24746 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24748 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24749 attribute from vecsimple to vecperm.
24750 (altivec_vbpermq2): Likewise.
24752 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24755 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24757 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24758 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24759 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24760 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24761 case where N arg is SIZE_MAX.
24762 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24763 (cmpstrsi): Add pattern.
24765 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24767 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24768 __builtin_vec_revb builtins.
24769 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24770 built-in functions to support generation of the ISA 3.0 XXBR<x>
24771 vector byte reverse instructions.
24772 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24773 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24774 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24775 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24776 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24777 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24778 (P9V_BUILTIN_VEC_REVB): Likewise.
24779 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24780 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24781 (p9_xxbrq_v16qi): Likewise.
24782 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24783 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24784 (p9_xxbrh_v8hi): Likewise.
24785 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24786 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24787 vec_revb built-in functions.
24789 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24791 PR rtl-optimization/78952
24792 * config/i386/i386.md (any_extract): New code iterator.
24793 (*insvqi_2): Use any_extract for source operand.
24794 (*insvqi_3): Use any_shiftrt for source operand.
24796 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24798 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24800 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24802 2017-01-18 Matthias Klose <doko@ubuntu.com>
24804 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24806 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24808 * config/rs6000/altivec.h (vec_bperm): Change #define.
24809 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24810 (altivec_vbpermq2): New define_insn.
24811 (altivec_vbpermd): Likewise.
24812 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24813 function interface.
24814 (VBPERMD): Likewise.
24815 (VBPERM): New polymorphic function interface.
24816 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24817 Add entries for P9V_BUILTIN_VEC_VBPERM.
24818 * doc/extend.texi: Add interfaces for vec_bperm.
24820 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24822 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24823 first letter of error messages.
24824 (s390_resolve_overloaded_builtin): Likewise.
24825 * config/s390/s390.c (s390_expand_builtin): Likewise.
24826 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24827 (s390_valid_target_attribute_inner_p): Likewise.
24828 * config/s390/s390.md ("tabort"): Likewise.
24830 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24832 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24833 (ISA_AVOID_DIV_HILO): New macro.
24834 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24835 (ISA_HAS_DDIV): Likewise.
24837 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24839 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24841 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24843 * doc/invoke.texi (fabi-version): Spelling fix.
24845 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24848 * doc/invoke.texi (fabi-version): Mention mangling fix for
24851 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24854 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24856 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24859 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24862 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24864 (stack_protect_test): Ditto.
24865 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24866 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24868 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24869 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24870 -mstack-protector-guard-offset=.
24871 (RS/6000 and PowerPC Options): Ditto.
24873 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24875 * config/i386/i386.h (MASK_CLASS_P): New define.
24876 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24877 there are no registers from different register sets also when
24878 mask registers are used. Update function comment.
24879 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24880 to (*k/*r) and (*k/*km) alternatives.
24882 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24884 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24885 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24886 (EH_RETURN_HANDLER_RTX): New define.
24887 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24888 Force frame pointer in EH return functions.
24889 (aarch64_expand_epilogue): Add barrier for eh_return.
24890 (aarch64_final_eh_return_addr): Remove.
24891 (aarch64_eh_return_handler_rtx): New function.
24892 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24894 (aarch64_eh_return_handler_rtx): New prototype.
24896 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24898 * config/rs6000/altivec.h (vec_rlmi): New #define.
24899 (vec_vrlnm): Likewise.
24900 (vec_rlnm): Likewise.
24901 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24902 (UNSPEC_VRLNM): Likewise.
24903 (VIlong): New mode iterator.
24904 (altivec_vrl<VI_char>mi): New define_insn.
24905 (altivec_vrl<VI_char>nm): Likewise.
24906 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24908 (VRLDNM): Likewise.
24909 (RLNM): New polymorphic function entry.
24910 (VRLWMI): New monomorphic function entry.
24911 (VRLDMI): Likewise.
24912 (RLMI): New polymorphic function entry.
24913 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24914 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24915 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24918 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24921 * dwarf2out.c (field_byte_offset): Restore the
24922 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24923 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24924 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24927 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24930 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24932 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24935 * dwarf2out.c (add_data_member_location_attribute): For constant
24936 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24937 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24938 DW_AT_byte_size attributes.
24940 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24942 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24943 after forcing to constant memory when the code model is medium.
24945 2017-01-17 Julia Koval <julia.koval@intel.com>
24948 * config/i386/avx512fintrin.h
24949 (_mm512_i32gather_ps): Change __addr type to void const*.
24950 (_mm512_mask_i32gather_ps): Ditto.
24951 (_mm512_i32gather_pd): Ditto.
24952 (_mm512_mask_i32gather_pd): Ditto.
24953 (_mm512_i64gather_ps): Ditto.
24954 (_mm512_mask_i64gather_ps): Ditto.
24955 (_mm512_i64gather_pd): Ditto.
24956 (_mm512_mask_i64gather_pd): Ditto.
24957 (_mm512_i32gather_epi32): Ditto.
24958 (_mm512_mask_i32gather_epi32): Ditto.
24959 (_mm512_i32gather_epi64): Ditto.
24960 (_mm512_mask_i32gather_epi64): Ditto.
24961 (_mm512_i64gather_epi32): Ditto.
24962 (_mm512_mask_i64gather_epi32): Ditto.
24963 (_mm512_i64gather_epi64): Ditto.
24964 (_mm512_mask_i64gather_epi64): Ditto.
24965 (_mm512_i32scatter_ps): Change __addr type to void*.
24966 (_mm512_mask_i32scatter_ps): Ditto.
24967 (_mm512_i32scatter_pd): Ditto.
24968 (_mm512_mask_i32scatter_pd): Ditto.
24969 (_mm512_i64scatter_ps): Ditto.
24970 (_mm512_mask_i64scatter_ps): Ditto.
24971 (_mm512_i64scatter_pd): Ditto.
24972 (_mm512_mask_i64scatter_pd): Ditto.
24973 (_mm512_i32scatter_epi32): Ditto.
24974 (_mm512_mask_i32scatter_epi32): Ditto.
24975 (_mm512_i32scatter_epi64): Ditto.
24976 (_mm512_mask_i32scatter_epi64): Ditto.
24977 (_mm512_i64scatter_epi32): Ditto.
24978 (_mm512_mask_i64scatter_epi32): Ditto.
24979 (_mm512_i64scatter_epi64): Ditto.
24980 (_mm512_mask_i64scatter_epi64): Ditto.
24981 * config/i386/avx512pfintrin.h
24982 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24983 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24984 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24985 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24986 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24987 (_mm512_prefetch_i32scatter_ps): Ditto.
24988 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24989 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24990 (_mm512_prefetch_i64scatter_pd): Ditto.
24991 (_mm512_prefetch_i64scatter_ps): Ditto.
24992 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24993 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24994 * config/i386/avx512vlintrin.h
24995 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24996 (_mm_mmask_i32gather_ps): Ditto.
24997 (_mm256_mmask_i32gather_pd): Ditto.
24998 (_mm_mmask_i32gather_pd): Ditto.
24999 (_mm256_mmask_i64gather_ps): Ditto.
25000 (_mm_mmask_i64gather_ps): Ditto.
25001 (_mm256_mmask_i64gather_pd): Ditto.
25002 (_mm_mmask_i64gather_pd): Ditto.
25003 (_mm256_mmask_i32gather_epi32): Ditto.
25004 (_mm_mmask_i32gather_epi32): Ditto.
25005 (_mm256_mmask_i32gather_epi64): Ditto.
25006 (_mm_mmask_i32gather_epi64): Ditto.
25007 (_mm256_mmask_i64gather_epi32): Ditto.
25008 (_mm_mmask_i64gather_epi32): Ditto.
25009 (_mm256_mmask_i64gather_epi64): Ditto.
25010 (_mm_mmask_i64gather_epi64): Ditto.
25011 (_mm256_i32scatter_ps): Change __addr type to void*.
25012 (_mm256_mask_i32scatter_ps): Ditto.
25013 (_mm_i32scatter_ps): Ditto.
25014 (_mm_mask_i32scatter_ps): Ditto.
25015 (_mm256_i32scatter_pd): Ditto.
25016 (_mm256_mask_i32scatter_pd): Ditto.
25017 (_mm_i32scatter_pd): Ditto.
25018 (_mm_mask_i32scatter_pd): Ditto.
25019 (_mm256_i64scatter_ps): Ditto.
25020 (_mm256_mask_i64scatter_ps): Ditto.
25021 (_mm_i64scatter_ps): Ditto.
25022 (_mm_mask_i64scatter_ps): Ditto.
25023 (_mm256_i64scatter_pd): Ditto.
25024 (_mm256_mask_i64scatter_pd): Ditto.
25025 (_mm_i64scatter_pd): Ditto.
25026 (_mm_mask_i64scatter_pd): Ditto.
25027 (_mm256_i32scatter_epi32): Ditto.
25028 (_mm256_mask_i32scatter_epi32): Ditto.
25029 (_mm_i32scatter_epi32): Ditto.
25030 (_mm_mask_i32scatter_epi32): Ditto.
25031 (_mm256_i32scatter_epi64): Ditto.
25032 (_mm256_mask_i32scatter_epi64): Ditto.
25033 (_mm_i32scatter_epi64): Ditto.
25034 (_mm_mask_i32scatter_epi64): Ditto.
25035 (_mm256_i64scatter_epi32): Ditto.
25036 (_mm256_mask_i64scatter_epi32): Ditto.
25037 (_mm_i64scatter_epi32): Ditto.
25038 (_mm_mask_i64scatter_epi32): Ditto.
25039 (_mm256_i64scatter_epi64): Ditto.
25040 (_mm256_mask_i64scatter_epi64): Ditto.
25041 (_mm_i64scatter_epi64): Ditto.
25042 (_mm_mask_i64scatter_epi64): Ditto.
25043 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25044 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25045 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25046 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25047 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25048 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25049 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25050 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25051 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25052 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25053 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25054 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25055 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25056 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25057 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25058 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25059 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25060 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25061 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25062 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25063 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25064 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25065 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25066 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25067 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25068 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25069 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25070 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25071 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25072 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25073 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25074 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25075 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25076 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25077 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25078 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25079 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25080 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25081 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25082 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25083 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25084 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25085 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25086 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25087 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25088 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25089 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25090 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25091 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25092 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25093 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25094 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25095 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25096 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25097 definitions accordingly.
25099 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25100 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25103 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25106 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25109 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25110 hard regs for inner regno.
25112 2017-01-17 Martin Liska <mliska@suse.cz>
25115 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25116 assumption and add comment.
25118 2017-01-17 Nathan Sidwell <nathan@acm.org>
25120 * ipa-visibility.c (localize_node): New function, broken out of ...
25121 (function_and_variable_visibility): ... here. Call it.
25123 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25125 PR middle-end/77445
25126 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25127 correctly set frequency of oudgoing edge.
25128 (duplicate_thread_path): Fix profile updating.
25130 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25133 * configure.ac: Add GCC_BASE_VER.
25134 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25135 version from BASE-VER file.
25136 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25137 (gcc.o): Depend on $(BASEVER).
25138 * common.opt (dumpfullversion): New option.
25139 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25140 * doc/invoke.texi: Document -dumpfullversion.
25141 * doc/install.texi: Document --with-gcc-major-version-only.
25142 * configure: Regenerated.
25144 2017-01-17 Richard Biener <rguenther@suse.de>
25146 PR tree-optimization/71433
25147 * tree-vrp.c (register_new_assert_for): Merge same asserts
25148 on all incoming edges.
25149 (process_assert_insertions_for): Handle insertions at the
25152 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25154 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25155 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25157 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25160 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25163 2017-01-17 Alan Modra <amodra@gmail.com>
25166 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25167 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25168 symbolic stack limit when pic.
25170 2017-01-16 Martin Sebor <msebor@redhat.com>
25172 PR tree-optimization/78608
25173 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25175 2017-01-16 Jeff Law <law@redhat.com>
25178 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25179 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25180 for several include directories that may be relative to sysroot.
25181 * config/i386/x-mingw32 (gplus_includedir): Define.
25182 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25183 (native_system_includedir): Likewise.
25184 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25185 override if TARGET_SYSTEM_ROOT is defined.
25186 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25188 PR tree-optimization/79090
25189 PR tree-optimization/33562
25190 PR tree-optimization/61912
25191 PR tree-optimization/77485
25192 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25193 and computed trims into the dump file.
25195 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25197 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25199 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25202 * gimplify.c (gimplify_init_constructor): If want_value and
25203 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25207 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25208 sequence. Formatting fixes.
25209 (doloop_optimize): Formatting fixes.
25212 * gcc.c (debug_level_greater_than_spec_func): New function.
25213 (static_spec_functions): Add debug-level-gt spec function.
25214 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25216 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25217 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25218 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25219 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25220 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25221 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25223 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25225 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25226 QImode fixups to general and mask registers only.
25228 2017-01-16 Carl Love <cel@us.ibm.com>
25230 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25231 for built-in functions
25232 vector signed char vec_nabs (vector signed char)
25233 vector signed short vec_nabs (vector signed short)
25234 vector signed int vec_nabs (vector signed int)
25235 vector signed long long vec_nabs (vector signed long long)
25236 vector float vec_nabs (vector float)
25237 vector double vec_nabs (vector double)
25238 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25240 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25241 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25242 * doc/extend.texi: Update the documentation file for the new built-in
25245 2017-01-16 Martin Sebor <msebor@redhat.com>
25247 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25250 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25252 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25253 UNSPEC_VSX__XXSPLTD to require special splat handling.
25255 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25258 * system.h: Poison strndup.
25260 2017-01-16 Alan Modra <amodra@gmail.com>
25263 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25266 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25268 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25270 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25272 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25273 call recog here. Assert that INSN_CODE (insn) is non-negative.
25275 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25278 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25280 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25281 in the currently scheduled RTL fragment.
25283 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25285 PR rtl-optimization/78751
25286 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25289 2017-01-14 Jeff Law <law@redhat.com>
25291 PR tree-optimization/79090
25292 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25293 variable length stores.
25294 (compute_trims): Delete dead assignment to *trim_tail.
25295 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25298 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25300 PR rtl-optimization/78626
25301 PR rtl-optimization/78727
25302 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25303 of a block, and split such blocks after everything else is finished.
25305 2017-01-14 Alan Modra <amodra@gmail.com>
25308 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25309 target legitimate_combined_insn.
25310 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25311 (rs6000_legitimate_combined_insn): New function.
25312 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25314 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25315 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25316 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25318 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25320 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25322 2017-01-13 Jeff Law <law@redhat.com>
25324 PR tree-optimization/33562
25325 PR tree-optimization/61912
25326 PR tree-optimization/77485
25327 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25329 (delete_dead_assignment): Likewise.
25330 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25331 statement to delete_dead_call and delete_dead_assignment.
25333 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25336 * substring-locations.c (format_warning_va): Strengthen case 1 so
25337 that both endpoints of the substring must be within the format
25338 range for just the substring to be printed.
25340 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25342 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25343 * config/i386/i386.c (ix86_target_string): Add missing options
25344 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25345 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25346 flags_other and ix86_target_other to flags2_other. Display unknown
25348 (ix86_valid_target_attribute_inner_p): Add missing options and
25349 reorder options by implied ISAs, as in ix86_target_string.
25351 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25353 * hash-table.h (hash_table::too_empty_p): New function.
25354 (hash_table::expand): Use it.
25355 (hash_table::traverse): Likewise.
25356 (hash_table::empty_slot): Use sizeof (value_type) instead of
25357 sizeof (PTR) to convert bytes to elements. Shrink the table
25358 if the current size is excessive for the current number of
25361 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25363 * ira-costs.c (record_reg_classes): Break from the inner loop
25364 early once alt_fail is known to be true. Update outer loop
25365 handling accordingly.
25367 2017-01-13 Jeff Law <law@redhat.com>
25369 * tree-ssa-dse.c (decrement_count): New function.
25370 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25371 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25372 when we know the partially dead statement is a mem* function.
25374 PR tree-optimization/61912
25375 PR tree-optimization/77485
25376 * tree-ssa-dse.c: Include expr.h.
25377 (maybe_trim_constructor_store): New function.
25378 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25380 PR tree-optimization/33562
25381 PR tree-optimization/61912
25382 PR tree-optimization/77485
25383 * doc/invoke.texi: Document new dse-max-object-size param.
25384 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25385 * tree-ssa-dse.c: Include params.h.
25386 (dse_store_status): New enum.
25387 (initialize_ao_ref_for_dse): New, partially extracted from
25389 (valid_ao_ref_for_dse, normalize_ref): New.
25390 (setup_live_bytes_from_ref, compute_trims): Likewise.
25391 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25392 (maybe_trim_partially_dead_store): Likewise.
25393 (maybe_trim_complex_store): Likewise.
25394 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25395 Track what bytes live from the original store. Return tri-state
25396 for dead, partially dead or live.
25397 (dse_dom_walker): Add constructor, destructor and new private members.
25398 (delete_dead_call, delete_dead_assignment): New extracted from
25400 (dse_optimize_stmt): Make a member of dse_dom_walker.
25401 Use initialize_ao_ref_for_dse.
25403 PR tree-optimization/33562
25404 PR tree-optimization/61912
25405 PR tree-optimization/77485
25406 * sbitmap.h (bitmap_count_bits): Prototype.
25407 (bitmap_clear_range, bitmap_set_range): Likewise.
25408 * sbitmap.c (bitmap_clear_range): New function.
25409 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25411 2017-01-13 Martin Liska <mliska@suse.cz>
25414 * function.c (set_cfun): Add new argument force.
25415 * function.h (set_cfun): Likewise.
25416 * ipa-inline-transform.c (inline_call): Use the function when
25417 strict alising from is dropped for function we inline to.
25419 2017-01-13 Richard Biener <rguenther@suse.de>
25421 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25422 for dumping GIMPLE INTEGER_CSTs.
25424 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25426 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25427 to 201112L since C++17.
25429 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25432 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25433 if -fsanitize=kernel-address is present.
25435 2017-01-13 Richard Biener <rguenther@suse.de>
25437 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25438 as _Literal ( type ) number in case usual suffixes do not
25439 preserve all information.
25441 2017-01-13 Richard Biener <rguenther@suse.de>
25443 PR tree-optimization/77283
25444 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25445 and ssa-iterators.h.
25446 (is_feasible_trace): Implement a cost model based on joiner
25449 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25452 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25453 char or short to __float128/_Float128 directly.
25455 2017-01-12 Martin Sebor <msebor@redhat.com>
25457 to -Wformat-overflow.
25458 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25459 (min_bytes_remaining): Same.
25460 (get_string_length): Same.
25461 (format_string): Same.
25462 (format_directive): Same.
25464 (pass_sprintf_length::handle_gimple_call): Same.
25466 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25468 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25469 info.nowrite calls with no lhs that can't throw. Return bool
25470 whether gsi_remove has been called or not.
25471 (pass_sprintf_length::handle_gimple_call): Return bool whether
25472 try_substitute_return_value called gsi_remove. Formatting fix.
25473 (pass_sprintf_length::execute): Don't use gsi_remove if
25474 handle_gimple_call returned true.
25477 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25478 be removed due to side-effects, don't remove following barrier nor
25479 turn the successor edge into fallthru edge.
25481 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25484 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25485 element-reversing loads and stores as not swappable.
25487 2017-01-12 Nathan Sidwell <nathan@acm.org>
25488 Nicolai Stange <nicstange@gmail.com>
25490 * combine.c (try_combine): Don't ignore result of overlap checking
25491 loop. Combine overlap & asm check into single loop.
25493 2017-01-12 Richard Biener <rguenther@suse.de>
25495 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25496 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25498 2017-01-12 Richard Biener <rguenther@suse.de>
25500 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25501 and TS_TARGET_OPTION directly derive from TS_BASE.
25502 * tree-core.h (tree_optimization_option): Derive from tree_base.
25503 (tree_target_option): Likewise.
25505 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25507 * config/i386/i386.c (memory_address_length): Increase len
25508 only when rip_relative_addr_p returns false.
25510 2017-01-11 Julia Koval <julia.koval@intel.com>
25512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25513 (OPTION_MASK_ISA_SGX_SET): New.
25514 (ix86_handle_option): Handle OPT_msgx.
25515 * config.gcc: Added sgxintrin.h.
25516 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25517 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25518 * config/i386/i386.c (ix86_target_string): Add -msgx.
25520 (ix86_option_override_internal): Handle new options.
25521 (ix86_valid_target_attribute_inner_p): Add sgx.
25522 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25523 * config/i386/i386.opt: Add msgx.
25524 * config/i386/sgxintrin.h: New file.
25525 * config/i386/x86intrin.h: Add sgxintrin.h.
25527 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25530 * fold-const.c (maybe_nonzero_address): Return 1 for function
25532 (tree_single_nonzero_warnv_p): Don't handle function local objects
25536 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25539 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25542 * opts.c: Include "spellcheck.h"
25543 (struct string_fragment): New struct.
25544 (struct edit_distance_traits<const string_fragment &>): New
25546 (get_closest_sanitizer_option): New function.
25547 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25549 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25551 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25553 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25554 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25555 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25556 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25557 for initial die_offset if dwarf_split_debug_info.
25558 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25559 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25561 (output_skeleton_debug_sections): Formatting fix. Use
25562 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25563 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25565 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
25567 * config/arm/cortex-a53.md: Add bypasses for
25568 cortex_a53_r2f_cvt.
25569 (cortex_a53_r2f): Only use for transfers.
25570 (cortex_a53_f2r): Likewise.
25571 (cortex_a53_r2f_cvt): Add reservation for conversions.
25572 (cortex_a53_f2r_cvt): Likewise.
25574 2017-01-11 Tamar Christina <tamar.christina@arm.com>
25576 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
25577 to all inlined functions, change static to extern.
25579 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
25582 * config/arm/arm.c (legitimize_pic_address): Handle reference to
25584 (arm_assemble_integer): Likewise.
25586 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
25588 * config.gcc: Use new awk script to check CPU, FPU and architecture
25589 parameters for --with-... options.
25590 * config/arm/parsecpu.awk: New file
25591 * config/arm/arm-cpus.in: New file.
25592 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
25594 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
25596 * config/arm/t-arm: Update dependency rules.
25597 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
25598 of processing .def files.
25599 * config/arm/genopt.sh: Deleted.
25600 * config/arm/gentune.sh: Deleted.
25601 * config/arm/arm-cores.def: Deleted.
25602 * config/arm/arm-arches.def: Deleted.
25603 * config/arm/arm-fpus.def: Deleted.
25604 * config/arm/arm-tune.md: Regenerated.
25605 * config/arm/arm-tables.opt: Regenerated.
25606 * config/arm/arm-cpu.h: New generated file.
25607 * config/arm/arm-cpu-data.h: New generated file.
25608 * config/arm/arm-cpu-cdata.h: New generated file.
25610 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25613 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25615 (input_varpool_node): Unpack dynamically_initialized bit.
25617 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25619 PR rtl-optimization/79032
25620 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25621 the alignment of the adjusted memory reference against that of MODE,
25622 instead of the alignment of the original memory reference.
25624 2017-01-11 Martin Jambor <mjambor@suse.cz>
25626 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25628 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25629 decorated functions.
25631 2017-01-11 Richard Biener <rguenther@suse.de>
25633 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25634 set range/nonnull info for PHI results. Do not set it on
25635 stmts marked for removal.
25637 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25639 * expr.c (store_field): In the bitfield case, fetch the return value
25640 from the registers before applying a single big-endian adjustment.
25641 Always do a final load for a BLKmode value not larger than a word.
25643 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25646 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25647 that we correctly handle column numbers greater than
25648 LINE_MAP_MAX_COLUMN_NUMBER.
25650 2017-01-10 Martin Sebor <msebor@redhat.com>
25652 PR middle-end/78245
25653 * gimple-ssa-sprintf.c (get_destination_size): Call
25654 {init,fini}object_sizes.
25655 * tree-object-size.c (addr_object_size): Adjust.
25656 (pass_through_call): Adjust.
25657 (pass_object_sizes::execute): Adjust.
25658 * tree-object-size.h (fini_object_sizes): Declare.
25660 2017-01-10 Martin Sebor <msebor@redhat.com>
25662 PR tree-optimization/78775
25663 * builtins.c (get_size_range): Move...
25664 * calls.c: ...to here.
25665 (alloc_max_size): Accept zero argument.
25666 (operand_signed_p): Remove.
25667 (maybe_warn_alloc_args_overflow): Call get_size_range.
25668 * calls.h (get_size_range): Declare.
25670 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25672 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25673 from TI's devices.csv file as of September 2016.
25674 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25676 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25678 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25679 * doc/invoke.texi: Likewise.
25680 * doc/md.texi: Likewise.
25681 * doc/objc.texi: Likewise.
25683 2017-01-10 Joshua Conner <joshconner@google.com>
25685 * config/arm/fuchsia-elf.h: New file.
25686 * config/fuchsia.h: New file.
25687 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25688 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25690 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25692 2016-01-10 Richard Biener <rguenther@suse.de>
25694 PR tree-optimization/79034
25695 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25696 Propagate out degenerate PHIs in the joiner.
25698 2017-01-10 Martin Liska <mliska@suse.cz>
25700 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25701 (sort_congruence_classes_by_decl_uid): Likewise.
25702 (sort_congruence_class_groups_by_decl_uid): Likewise.
25703 (sem_item_optimizer::merge_classes): Sort class, groups in these
25704 classes and members in the groups by DECL_UID of declarations.
25705 This would make merge operations stable.
25707 2017-01-10 Martin Liska <mliska@suse.cz>
25709 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25710 usage of m_classes_vec.
25711 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25712 (sem_item_optimizer::get_group_by_hash): Likewise.
25713 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25714 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25715 (sem_item_optimizer::verify_classes): Likewise.
25716 (sem_item_optimizer::process_cong_reduction): Likewise.
25717 (sem_item_optimizer::dump_cong_classes): Likewise.
25718 (sem_item_optimizer::merge_classes): Likewise.
25719 * ipa-icf.h (congruence_class_hash): Rename from
25720 congruence_class_group_hash. Remove declaration of m_classes_vec.
25722 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25724 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25725 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25726 * config.gcc: Add avx512vpopcntdqintrin.h.
25727 * config/i386/avx512vpopcntdqintrin.h: New.
25728 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25729 * config/i386/i386-builtin-types.def: Add new types.
25730 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25731 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25732 __builtin_ia32_vpopcountq_v8di_mask): New.
25733 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25734 __AVX512VPOPCNTDQ__.
25735 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25736 (PTA_AVX512VPOPCNTDQ): Define.
25737 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25738 TARGET_AVX512VPOPCNTDQ_P): Define.
25739 * config/i386/i386.opt: Add mavx512vpopcntdq.
25740 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25741 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25743 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25745 PR middle-end/77484
25746 * predict.def (PRED_CALL): Set to 67.
25748 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25750 * expr.c (store_field): In the bitfield case, if the value comes from
25751 a function call and is of an aggregate type returned in registers, do
25752 not modify the field mode; extract the value in all cases if the mode
25753 is BLKmode and the size is not larger than a word.
25755 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25758 * config/i386/cpuid.h: Fix undefined behavior.
25760 2017-01-04 Jeff Law <law@redhat.com>
25762 PR tree-optimization/79007
25763 PR tree-optimization/67955
25764 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25765 conservative for pt.null when flag_non_call_exceptions is on.
25767 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25769 PR translation/79019
25770 PR translation/79020
25771 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25772 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25773 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25775 * config/avr/avr.opt (maccumulate-args): Likewise.
25776 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25777 * common.opt (freport-bug): Likewise.
25778 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25779 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25780 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25781 translatable string.
25782 * config/i386/i386.c (function_value_32): Likewise.
25783 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25784 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25786 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25787 * common/config/msp430/msp430-common.c (msp430_handle_option):
25789 * symtab.c (symtab_node::verify_base): Likewise.
25790 * opts.c (set_debug_level): Likewise.
25791 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25792 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25793 missing whitespace to translatable strings.
25794 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25795 * config/sh/superh.h: Likewise.
25796 * config/i386/xopintrin.h: Likewise.
25797 * config/i386/znver1.md: Likewise.
25798 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25799 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25800 * double-int.h (struct double_int): Likewise.
25801 * double-int.c (div_and_round_double): Likewise.
25802 * wide-int.cc: Likewise.
25803 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25804 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25805 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25806 (crossjumps_occurred): ... this.
25807 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25810 PR tree-optimization/78899
25811 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25812 returning bool return struct loop *, NULL for failure and the new
25814 (versionable_outer_loop_p): Don't version outer loop if it has
25815 dont_vectorized bit set.
25816 (tree_if_conversion): When versioning outer loop, ensure
25817 tree_if_conversion is performed also on the inner loop of the
25818 non-vectorizable outer loop copy.
25819 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25820 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25821 prevent vectorization of it.
25822 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25823 the outer loop vectorization of the non-scalar version is attempted
25824 before vectorization of the inner loop in scalar version. If
25825 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25826 vectorization of its inner loop.
25827 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25828 has 2 inner loops, rename also on edges from bb whose single pred
25829 is outer_loop->header. Fix typo in function comment.
25831 2017-01-09 Martin Sebor <msebor@redhat.com>
25834 * asan.c (asan_emit_stack_protection): Increase local buffer size
25835 to avoid snprintf truncation warning.
25837 2017-01-09 Andrew Pinski <apinski@cavium.com>
25839 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25840 to reference thunderx2t99 for the tuning structure
25841 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25843 (thunderx2t99_extra_costs): This.
25844 * config/aarch64/aarch64-tune.md: Regenerate.
25845 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25846 (vulcan_addrcost_table): This.
25847 (vulcan_regmove_cost): Rename to ...
25848 (thunderx2t99_regmove_cost): This.
25849 (vulcan_vector_cost): Rename to ...
25850 (thunderx2t99_vector_cost): this.
25851 (vulcan_branch_cost): Rename to ...
25852 (thunderx2t99_branch_cost): This.
25853 (vulcan_tunings): Rename to ...
25854 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25855 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25857 2017-01-09 Martin Jambor <mjambor@suse.cz>
25861 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25862 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25863 (propagate_vr_accross_jump_function): Use the above function for all
25864 value range computations for pass-through jump functions and type
25865 converasion from explicit value range values.
25866 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25867 parameters from TYPE_ARG_TYPES.
25868 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25869 (ipa_write_node_info): Stream type of the actual argument.
25870 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25872 2017-01-09 Martin Liska <mliska@suse.cz>
25875 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25876 (lookup_compiler): Do not show error message with have_E.
25878 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25880 PR tree-optimization/78938
25881 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25882 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25883 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25884 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25887 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25889 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25892 2017-01-09 Richard Biener <rguenther@suse.de>
25894 PR tree-optimization/78997
25895 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25896 name condition properly.
25898 2017-01-09 Richard Biener <rguenther@suse.de>
25901 * dwarf2out.c (is_cxx): New overload with context.
25902 (is_naming_typedef_decl): Use it.
25904 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25906 * invoke.texi (Option Summary): Correct spacing in option lists
25907 and add line breaks to fix over-long lines.
25909 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25911 PR middle-end/17660
25913 * extend.texi (Common Variable Attributes): Add xref to GCC
25914 Internals manual to explain mode attribute keywords.
25916 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25919 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25920 and Preprocessor Options.
25921 (Options for Linking): Document -pthread here....
25922 (RS/6000 and PowerPC Options): ...not here.
25923 (Solaris 2 Options): ...or here.
25924 * doc/cppopts.texi: Document -pthread.
25926 2017-01-08 Martin Sebor <msebor@redhat.com>
25928 PR middle-end/77708
25929 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25930 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25931 New member functions.
25932 (format_directive): Used them.
25934 (pass_sprintf_length::handle_gimple_call): Same.
25935 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25936 to avoid truncation for any argument.
25937 (extract_affine_mul): Same.
25938 * tree.c (get_file_function_name): Same.
25940 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25942 PR middle-end/77484
25943 * predict.def (PRED_INDIR_CALL): Set to 86.
25945 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25947 PR preprocessor/54124
25948 * doc/cppopts.texi: Reformat -d subtable to list the full name
25949 of the options. Add cross-reference to the docs for the general
25950 compiler -d options.
25951 * doc/invoke.texi (Developer Options): Add cross-reference to the
25952 preprocessor-specific -d option documentation.
25954 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25956 PR preprocessor/13498
25957 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25958 redudant material, and reflect new command-line options.
25959 (System Headers): Likewise.
25961 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25963 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25964 -isystem, and -idirafter. Copy-edit.
25965 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25966 default for -ftrack-macro-expansion. Delete obsolete and
25967 badly-formatted implementation details about -fdebug-cpp output.
25968 * doc/cppwarnopts.texi: Copy-edit.
25970 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25973 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25974 that the transition from a max line width >= 1<<10 to narrower
25975 lines works correctly.
25977 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25979 * doc/options.texi (PerFunction): New.
25980 * opt-functions.awk (switch_flags): Map both Optimization and
25981 PerFunction to CL_OPTIMIZATION.
25982 * opth-gen.awk: Test for PerFunction flag along with
25984 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25985 it only when the latter is present. Skip those that don't in
25986 the hash function generator.
25987 * common.opt (fvar-tracking): Mark as PerFunction instead of
25989 (fvar-tracking-assignments): Likewise.
25990 (fvar-tracking-assignments-toggle): Likewise.
25991 (fvar-tracking-uninit): Likewise.
25993 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25995 PR translation/79018
25996 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25999 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26002 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26003 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26004 TARGET_LONG_JUMP_TABLE_OFFSETS.
26005 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26006 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26007 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26008 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26009 * config/m68k/m68k.md (tablejump expander): Likewise.
26010 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26011 TARGET_LONG_JUMP_TABLE_OFFSETS.
26012 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26013 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26015 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26016 David Holsgrove <david.holsgrove@xilinx.com>
26018 * common/config/microblaze/microblaze-common.c
26019 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26020 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26022 * config/microblaze/microblaze.c (microblaze_must_save_register)
26023 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26025 (microblaze_eh_return): New function.
26026 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26027 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26028 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26029 * config/microblaze/microblaze.md (eh_return): New pattern.
26031 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26033 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26034 GCC_DIAGNOSTIC_STRINGIFY): Define.
26036 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26038 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26040 * config/arm/arm.md (<mcrr>): New.
26042 * config/arm/arm.c (arm_arch5te): New.
26043 (arm_option_override): Set arm_arch5te.
26044 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26046 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26047 (arm_mcrr_qualifiers): ... this. New.
26048 (MRRC_QUALIFIERS): Define to...
26049 (arm_mrrc_qualifiers): ... this. New.
26050 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26052 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26053 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26054 (MRRCI, mrrc, MRRC): New.
26055 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26056 VUNSPEC_MRRC2): New.
26058 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26060 * config/arm/arm.md (<mcr>): New.
26062 * config/arm/arm.c (arm_coproc_builtin_available): Add
26063 support for mcr, mrc, mcr2 and mrc2.
26064 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26065 (arm_mcr_qualifiers): ... this. New.
26066 (MRC_QUALIFIERS): Define to ...
26067 (arm_mrc_qualifiers): ... this. New.
26068 (MCR_QUALIFIERS): Define to ...
26069 (arm_mcr_qualifiers): ... this. New.
26070 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26072 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26073 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26074 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26075 VUNSPEC_MRC2): New.
26077 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26079 * config/arm/arm.md (*ldc): New.
26083 * config/arm/arm.c (arm_coproc_builtin_available): Add
26084 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26085 (arm_coproc_ldc_stc_legitimate_address): New.
26086 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26087 'qualifier_const_pointer'.
26088 (LDC_QUALIFIERS): Define to...
26089 (arm_ldc_qualifiers): ... this. New.
26090 (STC_QUALIFIERS): Define to...
26091 (arm_stc_qualifiers): ... this. New.
26092 * config/arm/arm-protos.h
26093 (arm_coproc_ldc_stc_legitimate_address): New.
26094 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26095 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26096 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26097 stc2, stcl, stc2l): New.
26098 * config/arm/constraints.md (Uz): New.
26099 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26100 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26101 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26102 VUNSPEC_STC2L): New.
26104 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26106 * config/arm/arm.md (<cdp>): New.
26107 * config/arm/arm.c (neon_const_bounds): Rename this ...
26108 (arm_const_bounds): ... this.
26109 (arm_coproc_builtin_available): New.
26110 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26111 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26112 (CDP_QUALIFIERS): Define to...
26113 (arm_cdp_qualifiers): ... this. New.
26115 (arm_expand_builtin_args): Add case for 6 arguments.
26116 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26117 (arm_const_bounds): ... this.
26118 (arm_coproc_builtin_available): New.
26119 * config/arm/arm_acle.h (__arm_cdp): New.
26121 * config/arm/arm_acle_builtins.def (cdp): New.
26123 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26124 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26125 'arm_const_bounds'.
26126 * config/arm/types.md (coproc): New.
26127 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26128 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26129 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26130 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26132 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26134 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26135 (UBINOP_QUALIFIERS): New.
26137 (acle_builtin_data): New. Change comment.
26138 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26139 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26140 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26141 arm_acle_builtins.def.
26142 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26143 (arm_init_acle_builtins): New.
26144 (CRC32_BUILTIN): Remove.
26145 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26146 crc32cb, crc32ch and crc32cw.
26147 (arm_init_crc32_builtins): Remove.
26148 (arm_init_builtins): Use arm_init_acle_builtins rather
26149 than arm_init_crc32_builtins.
26150 (arm_expand_acle_builtin): New.
26151 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26152 * config/arm/arm_acle_builtins.def: New.
26154 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26156 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26157 (arm_builtin_datum): ... this.
26158 (arm_init_neon_builtin): Rename to ...
26159 (arm_init_builtin): ... this. Add a new parameters PREFIX
26160 and USE_SIG_IN_NAME.
26161 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26162 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26163 'arm_builtin_datum'.
26164 (arm_init_vfp_builtins): Likewise.
26165 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26166 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26167 (arm_expand_neon_args): Rename to ...
26168 (arm_expand_builtin_args): ... this. Rename builtin_arg
26169 enum values and differentiate between ARG_BUILTIN_MEMORY
26170 and ARG_BUILTIN_NEON_MEMORY.
26171 (arm_expand_neon_builtin_1): Rename to ...
26172 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26173 values, arm_expand_builtin_args and add bool parameter NEON.
26174 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26175 (arm_expand_vfp_builtin): Likewise.
26176 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26178 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26180 PR middle-end/77484
26181 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26182 * predict.c (tree_estimate_probability_bb): Reverse direction of
26183 polymorphic call predictor.
26185 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26187 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26188 (determine_pass_name_match): ...this new function and...
26189 (should_skip_pass_p): ...this new function.
26191 2017-01-06 Nathan Sidwell <nathan@acm.org>
26193 * ipa-visibility.c (function_and_variable_visibility): Reformat
26194 comments and long lines. Remove extrneous if.
26195 * symtab.c (symtab_node::make_decl_local): Fix code format.
26196 (symtab_node::set_section_for_node): Fix comment typo.
26198 2017-01-06 Martin Liska <mliska@suse.cz>
26201 * lra-constraints.c: Rename invariant to lra_invariant.
26202 * predict.c (set_even_probabilities): Initialize e to NULL.
26204 2017-01-05 Martin Sebor <msebor@redhat.com>
26206 PR tree-optimization/78910
26207 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26208 (format_integer): Correct off-by-one error in the handling
26209 of precision with negative numbers in signed conversions..
26211 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26213 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26215 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26217 PR tree-optimization/71016
26218 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26219 factor_out_conditional_conversion. Formatting fix.
26220 (factor_out_conditional_conversion): Add cond_stmt argument.
26221 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26222 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26225 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26227 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26228 read-rtl-function.o, and selftest-rtl.o.
26229 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26230 (selftest::aarch64_test_loading_full_dump): New function.
26231 (selftest::aarch64_run_selftests): New function.
26232 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26233 selftest::aarch64_run_selftests.
26234 * config/i386/i386.c
26235 (selftest::ix86_test_loading_dump_fragment_1): New function.
26236 (selftest::ix86_test_loading_call_insn): New function.
26237 (selftest::ix86_test_loading_full_dump): New function.
26238 (selftest::ix86_test_loading_unspec): New function.
26239 (selftest::ix86_run_selftests): Call the new functions.
26240 * emit-rtl.c (maybe_set_max_label_num): New function.
26241 * emit-rtl.h (maybe_set_max_label_num): New decl.
26242 * function.c (instantiate_decls): Guard call to
26243 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26244 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26246 * gensupport.c (gen_reader::gen_reader): Pass "false"
26247 for new "compact" param of rtx_reader.
26248 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26249 rather than an empty string for NULL strings.
26250 * read-md.c: Potentially include config.h rather than bconfig.h.
26251 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26252 (have_error): New global, copied from errors.c.
26253 (md_reader::read_name): Rename to...
26254 (md_reader::read_name_1): ...this, adding "out_loc" param,
26255 and converting "missing name or number" to returning false, rather
26257 (md_reader::read_name): Reimplement in terms of read_name_1.
26258 (md_reader::read_name_or_nil): New function.
26259 (md_reader::read_string): Handle "(nil)" by returning NULL.
26260 (md_reader::md_reader): Add new param "compact".
26261 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26262 (md_reader::read_file): New method.
26263 * read-md.h (md_reader::md_reader): Add new param "compact".
26264 (md_reader::read_file): New method.
26265 (md_reader::is_compact): New accessor.
26266 (md_reader::read_name): Convert return type from void to file_location.
26267 (md_reader::read_name_or_nil): New decl.
26268 (md_reader::read_name_1): New decl.
26269 (md_reader::m_compact): New field.
26270 (noop_reader::noop_reader): Pass "false" for new "compact" param
26272 (rtx_reader::rtx_reader): Add new "compact" param.
26273 (rtx_reader::read_rtx_operand): Make virtual and convert return
26274 type from void to rtx.
26275 (rtx_reader::read_until): New decl.
26276 (rtx_reader::handle_any_trailing_information): New virtual function.
26277 (rtx_reader::postprocess): New virtual function.
26278 (rtx_reader::finalize_string): New virtual function.
26279 (rtx_reader::m_in_call_function_usage): New field.
26280 (rtx_reader::m_reuse_rtx_by_id): New field.
26281 * read-rtl-function.c: New file.
26282 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26283 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26284 (selftest::verify_three_block_rtl_cfg): New decl.
26285 * read-rtl-function.h: New file.
26286 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26287 For host, include function.h, memmodel.h, and emit-rtl.h.
26288 (one_time_initialization): New function.
26289 (struct compact_insn_name): New struct.
26290 (compact_insn_names): New array.
26291 (find_code): Handle insn codes in compact dumps.
26292 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26293 (bind_subst_iter_and_attr): Likewise.
26294 (add_condition_to_string): Likewise.
26295 (add_condition_to_rtx): Likewise.
26296 (apply_attribute_uses): Likewise.
26297 (add_current_iterators): Likewise.
26298 (apply_iterators): Likewise.
26299 (initialize_iterators): Guard usage of apply_subst_iterator with
26300 #ifdef GENERATOR_FILE.
26301 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26302 (md_reader::read_mapping): Likewise.
26303 (add_define_attr_for_define_subst): Likewise.
26304 (add_define_subst_attr): Likewise.
26305 (read_subst_mapping): Likewise.
26306 (check_code_iterator): Likewise.
26307 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26309 (one_time_initialization): New function.
26310 (rtx_reader::read_until): New method.
26311 (read_flags): New function.
26312 (parse_reg_note_name): New function.
26313 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26314 Handle reuse_rtx ids.
26315 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26316 Add parsing support for RTL dumps, mirroring the special-cases in
26317 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26318 values, and calling handle_any_trailing_information.
26319 (rtx_reader::read_rtx_operand): Convert return type from void
26320 to rtx, returning return_rtx. Handle case 'e'. Call
26321 finalize_string on XSTR and XTMPL fields.
26322 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26323 "(nil)" values were omitted. Call the postprocess vfunc on the
26325 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26326 class ctor. Initialize m_in_call_function_usage. Call
26327 one_time_initialization.
26328 * rtl-tests.c (selftest::test_uncond_jump): Call
26329 set_new_first_and_last_insn.
26330 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26331 * selftest-rtl.c: New file.
26332 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26333 (selftest::get_insn_by_uid): New decl.
26334 * selftest-run-tests.c (selftest::run_tests): Call
26335 read_rtl_function_c_tests.
26336 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26337 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26340 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26342 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26343 operands in a special way. Assert that pos+len <= mode precision.
26345 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26347 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26348 3 argument Alias with unlimited for the negative form.
26349 (fno-vect-cost-model): Removed.
26351 2017-01-05 Martin Liska <mliska@suse.cz>
26353 * hsa-gen.c (gen_hsa_divmod): New function.
26354 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26356 2017-01-05 Martin Liska <mliska@suse.cz>
26359 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26362 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26364 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26365 small constant length operands.
26367 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26369 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26370 between loop iterations.
26372 2017-01-05 Martin Liska <mliska@suse.cz>
26375 * gimplify.c (gimplify_decl_expr): Compare to
26376 asan_poisoned_variables instread of checking flags.
26377 (gimplify_target_expr): Likewise.
26378 (gimplify_expr): Likewise.
26379 (gimplify_function_tree): Conditionally initialize
26380 asan_poisoned_variables.
26382 2017-01-04 Jeff Law <law@redhat.com>
26384 PR tree-optimizatin/78812
26385 * rtl.h (contains_mem_rtx_p): Prototype.
26386 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26387 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26388 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26389 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26390 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26392 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26394 * input.c (assert_char_at_range): Default-initialize actual_range.
26396 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26398 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26399 to match the caller.
26401 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26403 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26404 insns after final jump in test to emit dummy move.
26406 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26408 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26409 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26411 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26413 * multiple_target.c (create_dispatcher_calls): Init e_next.
26414 * tree-ssa-loop-split.c (split_loop): Init border.
26415 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26418 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26423 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26424 (altivec_register_operand): Do not return true if the operand
26425 contains a SUBREG mixing SImode and SFmode.
26426 (vsx_register_operand): Likewise.
26427 (vsx_reg_sfsubreg_ok): New predicate.
26428 (vfloat_operand): Do not return true if the operand contains a
26429 SUBREG mixing SImode and SFmode.
26430 (vint_operand): Likewise.
26431 (vlogical_operand): Likewise.
26432 (gpc_reg_operand): Likewise.
26433 (int_reg_operand): Likewise.
26434 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26435 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26436 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26438 (rs6000_emit_move_si_sf_subreg): New helper function.
26439 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26440 fixup SUBREGs involving SImode and SFmode.
26441 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26442 numbers for the new peephole2 optimization.
26443 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26444 the GLIBC math library that do AND/IOR/XOR operations on single
26445 precision floating point.
26446 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26447 target macros to say whether we need to avoid SUBREGs mixing
26449 (TARGET_ALLOW_SF_SUBREG): Likewise.
26450 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26451 (UNSPEC_SI_FROM_SF): Likewise.
26452 (iorxor): Change spacing.
26453 (and_ior_xor): New iterator for AND, IOR, and XOR.
26454 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26455 (movdi_from_sf_zero_ext): Likewise.
26456 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26457 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26458 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26459 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26460 (fms<mode>4): Likewise.
26461 (fnma<mode>4): Likewise.
26462 (fnms<mode>4): Likewise.
26463 (nfma<mode>4): Likewise.
26464 (nfms<mode>4): Likewise.
26466 2017-01-04 Marek Polacek <polacek@redhat.com>
26469 * doc/invoke.texi: Document -Wpointer-compare.
26471 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26473 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26476 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26477 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26480 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26483 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26484 documentation of the powerpc_popcntb_ok attribute.
26485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26486 code to issue warning messages if a requested CPU configuration is
26487 not supported by the binary (assembler and loader) toolchain.
26488 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26489 made to define a built-in function that has been disabled.
26490 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26491 made to define a built-in function that has been disabled.
26492 (altivec_init_builtins): Add comment explaining why definition
26493 of the DST built-in functions is not preceded by an assertion
26494 check. Add assertions to prevent ICE if attempts are made to
26495 define an altivec predicate or an abs* built-in function that has
26497 (htm_init_builtins): Add comment explaining why definition of the
26498 htm built-in functions is not preceded by an assertion check.
26500 2017-01-04 Jeff Law <law@redhat.com>
26502 PR tree-optimizatin/67955
26503 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26504 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26505 the points-to solution does not include pt_null. Use DECL_PT_UID
26508 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26510 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26511 Use gen_int_mode instead of gen_lopwart for const_int operands.
26513 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26515 PR tree-optimization/71563
26516 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26517 out of range value - has low bits known to be zero.
26519 2017-01-04 Alan Modra <amodra@gmail.com>
26521 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26522 * configure: Regenerate.
26523 * config.in: Regenerate.
26525 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26528 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26529 a substring of the message, but strcmp with the whole message. Ifdef
26530 ENABLE_NLS, translate the message first using dgettext.
26532 2017-01-03 Jeff Law <law@redhat.com>
26534 PR tree-optimizatin/78856
26535 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26536 (mark_threaded_blocks): Remove code to truncate thread paths that
26537 cross multiple loop headers. Instead invalidate the cached loop
26538 iteration information and handle case of a thread path walking
26539 into an irreducible region.
26541 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26544 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26545 assertions. Add support for doing the signbit if the IEEE 128-bit
26546 floating point value is in a GPR.
26547 * config/rs6000/rs6000.md (Fsignbit): Delete.
26548 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26549 Update the length attribute if the value is in a GPR.
26550 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26551 the sign or zero extension instruction, since the value is always 0/1.
26552 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26555 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26556 extracting SImode to a GPR register so that we can generate a
26557 store, limit the vector to be in a traditional Altivec register
26558 for the vextuwrx instruction.
26560 2017-01-03 Ian Lance Taylor <iant@google.com>
26562 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26564 2017-01-03 Martin Sebor <msebor@redhat.com>
26566 PR tree-optimization/78696
26567 * gimple-ssa-sprintf.c (format_floating): Correct handling of
26568 precision. Use MPFR for %f for greater fidelity. Correct handling
26570 (pass_sprintf_length::compute_format_length): Set width and precision
26571 specified by asrerisk to void_node for vararg functions.
26572 (try_substitute_return_value): Adjust dump output.
26574 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
26576 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
26578 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
26580 * doc/invoke.texi (SPARC options): Document -mlra as the default.
26581 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
26582 -mlra/-mno-lra was passed to the compiler.
26584 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
26586 PR rtl-optimization/65618
26587 * emit-rtl.c (try_split): Move initialization of "before" and
26588 "after" to just before the call to emit_insn_after_setloc.
26590 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
26592 * doc/md.texi (Standard Names): Remove reference to Java frontend.
26594 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
26596 * dwarf2out.c (gen_enumeration_type_die): When
26597 -gno-strict-dwarf, add a DW_AT_encoding attribute.
26599 2017-01-03 Jakub Jelinek <jakub@redhat.com>
26601 PR tree-optimization/78965
26602 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
26603 Change first argument from const call_info & to call_info &. For %n
26604 set info.nowrite to false.
26606 PR middle-end/78901
26607 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26608 possibly throwing calls.
26610 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26611 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26612 and fns handling, rather than in a separate case SSA_NAME.
26614 2017-01-02 Jeff Law <law@redhat.com>
26616 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26617 fixes for first_period and second_period variables.
26619 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26622 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26623 (*insvqi_1): New insn pattern.
26624 (*insvqi_1_mem_rex64): Ditto.
26625 (*insvqi_2): Ditto.
26626 (*insvqi_3): Rename from *insvqi.
26628 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26630 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26632 * doc/cfg.texi (Edges): Remove reference to Java.
26633 (Maintaining the CFG): Ditto.
26635 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26637 PR middle-end/77674
26638 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26639 transparent aliases.
26641 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26643 PR middle-end/77484
26644 * predict.def (PRED_CALL): Update hitrate.
26645 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26646 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26647 into direct/indirect/polymorphic variants.
26649 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26651 Update copyright years.
26653 * gcc.c (process_command): Update copyright notice dates.
26654 * gcov-dump.c (print_version): Ditto.
26655 * gcov.c (print_version): Ditto.
26656 * gcov-tool.c (print_version): Ditto.
26657 * gengtype.c (create_file): Ditto.
26658 * doc/cpp.texi: Bump @copying's copyright year.
26659 * doc/cppinternals.texi: Ditto.
26660 * doc/gcc.texi: Ditto.
26661 * doc/gccint.texi: Ditto.
26662 * doc/gcov.texi: Ditto.
26663 * doc/install.texi: Ditto.
26664 * doc/invoke.texi: Ditto.
26666 Copyright (C) 2017 Free Software Foundation, Inc.
26668 Copying and distribution of this file, with or without modification,
26669 are permitted in any medium without royalty provided the copyright
26670 notice and this notice are preserved.